Talk:COVID-19 pandemic in Sweden/Archive 2

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

"Corona in sweden" listed at Redirects for discussion

An editor has asked for a discussion to address the redirect Corona in sweden. Please participate in the redirect discussion if you wish to do so. _signed, Rosguill ^talk 19:40, 30 March 2020 (UTC)

Infobox

I removed two parameters from the infobox because both had numbers that failed verification. I tried to find fresh numbers from reliable sources but I couldn't find any, I hope someone else knows where to look. Reasons for removal: Recovered cases uses an unreliable source, has been tagged as such for long enough. Severe cases: Reliable source, but it says "306 intensivvårdsfall". This does not translate to "306 severe cases. Best thing would be to change to a label that explicitly says "Intensive care cases" but I don't know how to do that. Kittens n thugs 💬 13:36, 30 March 2020 (UTC)

a official source: https://portal.icuregswe.org/siri/report/inrapp-corona ; currently hospitalized persons (including ICU) — Preceding unsigned comment added by 155.4.92.139 (talk) 19:00, 30 March 2020 (UTC)

Is it possible to filter on total + severe? Kittens n thugs 💬 19:29, 30 March 2020 (UTC)

Sweden reports ICU-cases, not severe cases. They are practically interchangeable, yet it might need clarification in the infobox. Carl Fredrik ^talk 21:01, 30 March 2020 (UTC)

Inconsistencies in table "New COVID-19 cases in Sweden by county" and graphs "Number of Deaths" and "Deaths and Recoveries" for March 30 and 31, and other graphs for April 6 and 7

The table "New COVID-19 cases in Sweden by county" shows for today, March 31, that 34 deaths occurred; however, the graph "Number of Deaths" for March 31 shows 23. Also, the graph just beneath titled "Deaths and Recoveries" seems to indicate that the greater number of deaths occurred on the 31st and fewer deaths on the 30th, while "Number of Deaths" shows the greater number occurring on the 30th with 42 and 23 deaths on the 30th. Would try to make changes myself but (1.) I don't know which is accurate or if they are inconsistencies in reported data. (2.) I don't know whether they are yet to be finalized due to the recent nature of the data. — Preceding unsigned comment added by Caracoid (talk • contribs) 18:05, 31 March 2020 (UTC)

Your help is very much appreciated! As of now the statistics section is mess. I think there's pretty much concensus in using Folkhälsomyndigheten as a source for number of cases etc.. See discussions above. Kittens n thugs 💬 23:50, 31 March 2020 (UTC)

I don't like the lack of consistency. But I think this will be the reality for a time forward at least. The Folkhälsomyndigheten has even changes values on previous days complicating the situation even further. We should stick to Folkhälsomyndigheten and try prioritize the correctness of the values instead of focusing on having quick updates. We should try to use Folkhälsomyndigheten. We can at least find correct values for each date for number of cases here. I would really like to find the same kind of data for deaths but they state here that even they don't have complete numbers regarding the number of deaths or recoveries for each day. I think we will have to make due with the numbers we have for now and try to at least have correct number from this date. They publish number of deaths each day so this should be possible. Mrconter1 (talk) 09:34, 1 April 2020 (UTC)

For April 7: - Table "New COVID-19 cases in Sweden by county": Total 190 new cases. - Graph "COVID-19 cases in Sweden": 8419 - 7693 = Total 726 new cases. This discrepancy is big, and represents important misinformation. Kind advice for the person(s) who are editing this Wikipedia article: It is better to put no data at all, than to put inconsistent data; every day when you put new data you should check consistency at least for following 4 objects: table and graph mentioned above, plus graphs "New cases per day" and "Number of deaths" - if you cannot make these objects consistent, do not refresh them (wait a day longer and then try again).
— Preceding unsigned comment added by 195.29.137.106 (talk) 04:44, 9 April 2020 (UTC)

Death statistics

Writers you must be aware that swedish death statistic figueres dosn´t give figures always for the right day. I Region Stockholm they have begun to controll the deathcases back in time to include all that have died -even those who were not tested before death. As many of you probably know italian death statistic are very unreliabel. There are investigations of lokal death figures in Lombardy much higher than official death cases. When you write "Death toll hit more than 50 new deaths in one day" the 1:st of April, it is not a fact because all these deaths was not on one day even if they now are registerad in one day. You should go over to give figures for seven days and see the trend in them instead.--Klostergårdaren (talk) 13:22, 1 April 2020 (UTC)

If you want to se i diagram with medium of seven days you can look at this place. It dosn´t give day figures but the trend is obvious. See --https://platz.se/coronavirus/?--Klostergårdaren (talk) 21:07, 1 April 2020 (UTC)

Regarding the figure "Stefan Löfvén speaking at Almedalsveckan in 2016"

The figureis in my opinion not related enough to the wiki page. If a user wants to see figures regarding the Almedalen week they can go to Almedalsveckan. — Preceding unsigned comment added by Laboz125 (talk • contribs) 16:37, 2 April 2020 (UTC)

Chart gallery

What purpose does a logarithmic chart fill for the average reader? Or any reader at all, because the section offers zero context to why they're even there in the first place. If there's a good reason to include them in this chart gallery, it needs to be explained in the non-existent text body. Which will make the section too detailed anyway. The section is confusing as it is, with a zillion charts showing the same information found in the table (the table is really good). — Preceding unsigned comment added by Kittens n thugs (talk • contribs) 22:42, 2 April 2020 (UTC)

Contacting ECDC

I checked other statistics sites, and it seems everyone is grinding to a halt in covering daily deaths in Sweden. But did find this updated information from the European Centre for Disease Prevention and Control, which has current total confirmed deaths in Sweden as of 4/3 at a significantly different 282. I've emailed them to see if they can send us their source material. — Preceding unsigned comment added by Caracoid (talk • contribs) 12:51, 3 April 2020 (UTC)

@Caracoid Their source is likely the Swedish Public Health Agency. According to ECDC, the figures in the latest update was collected between 6:00 and 10:00 CET. Swedish data is published daily at noon, therefore the ECDC numbers were correct (although only for a few hours). — Preceding unsigned comment added by Kittens n thugs (talk • contribs) 16:15, 3 April 2020 (UTC)

Of course, I see it now. Sorry for that. --Caracoid (talk) 17:06, 3 April 2020 (UTC)

Map of deaths cases in Sweden

Hello, I would suggest to create a map that shows the death cases by subnational divisions.

Thanks. — Preceding unsigned comment added by 83.39.208.224 (talk) 13:21, 28 March 2020 (UTC)

@Anonymous suggester - I thought there already is a map by counties. Are you saying you would like to see a map with individual locations for all cases? If that's the case, I think it is a lot of work and the data might not exist. Do we even know in which city Sweden's Patient Zero lives? If they are from a small area where few people live, users might be able to determine their exact identity.

Nevertheless, I don't have any experience with a Wikipedia update of that size. My friend is supposed to be sending me a breakdown of the most affected areas of Stockholm. Do you think a Stockholm map would be useful? MXMLLN (talk) 13:28, 28 March 2020 (UTC)

83.39.208.224 has correctly pointed out that the map we use shows cases, not deaths. Hookandloop — there is no data on individual locations spare media reports for among the first 200 cases.
A map over cases as they distribute in Stockholm could be useful, yes, but a map over the distribution of deaths in the country could easily be made, as that data exists. The Public Health Agency is releases an infobank this weekend where we could probably reliably pull that data. Carl Fredrik ^talk 13:42, 28 March 2020 (UTC)

@Carl Fredrik - Are we sure the map being suggested doesn't already exist somewhere? Such a map would need constant updating. How have we been doing with updating all the other data on this page? There's a limit to how much work can feasibly be done by the community. I have not been keeping up with all the other projects on this page, but last I checked there were so many planned tasks that it became a bit too overwhelming. What do you think, CF? — Preceding unsigned comment added by Hookandloop (talk • contribs) 13:50, 28 March 2020 (UTC)

Whether the map exists elsewhere is irrelevant unless it is under a compatible license. And there is frankly no limit to what the community here is able to do — and if something isn't updated, it can be removed later on. Carl Fredrik ^talk 13:55, 28 March 2020 (UTC)

Another problem would be the cluttering of the page. Some of us already feel like we have enough diagrams. Mrconter1 (talk) 14:10, 28 March 2020 (UTC)

We don't have enough, we have far too many. The problem is that we have several diagrams that show the same thing — some of which we could remove, without losing anything. (I'm going to see if I can combine or clean up any of them tomorrow.)
The reason why it would be useful to add this type of map, is that it brings new information. Carl Fredrik ^talk 21:22, 28 March 2020 (UTC)

Support. But please show numbers per capita, not absolute numbers. The counties are really different in size. Tomastvivlaren (talk) 21:19, 29 March 2020 (UTC)

Agreed. Use per capita, per the precedent consensus at the main pandemic article. {{u|Sdkb}} ^talk 00:01, 4 April 2020 (UTC)

Should we use numbers from Folkhälsomyndigheten press conference?

I have not been able to find numbers on a day to day basis for number of deaths on the official Folkhälsomyndigheten website. However they did show a diagram that displays number of deaths each day on today's press conference. I was wondering if we could use this diagram to update the numbers number of deaths? This could be used in both the diagrams that presents deaths.

• Here is the diagram from the press conference.
• Here is a spreadsheet containing the extracted values and the image showing the method of extraction of numbers.

What do you think? Mrconter1 (talk) 18:07, 1 April 2020 (UTC)

those numbers make more sense (also with the backporting of nummbers in mind, see section 'Death statistics'), and its a official source, source referencing might be a bit difficult, lets use the numbers untill they provide a better source like they did with the cases — Preceding unsigned comment added by 155.4.92.139 (talk) 18:43, 1 April 2020 (UTC)

Do you mean that we should be using the numbers from the presentation or the daily update numbers? Mrconter1 (talk) 09:53, 4 April 2020 (UTC)

FMH includes a Tidsserie: nya fall av avlidna on the daily reports, we should use that as a consistent source for backporting, looks like it uses the day of dead, not the day of reported dead — Preceding unsigned comment added by 155.4.92.139 (talk) 08:44, 5 April 2020 (UTC)

"Debate" section

On 27 March, Sweden didn't have more deaths than all neighboring countries combined. - that is not worded well and it is unclear as to what the author's purpose was. Should we expect Sweden to have more deaths than all neighboring countries combined, so that it is a surprise if it does not? And what is significant about 27 March?2602:306:CFEA:170:5DDD:CA79:BC87:2B (talk) 22:22, 3 April 2020 (UTC)

Original wording was "Sweden has more deaths than all neighboring countries combined", added earlier today with three sources from late March. However the sources didn't support this sentence. Adding "didn't" didn't made it less unclear, but it did make it in-line with the sources. Until whoever put this in the "International public response" subsection of "Debate" shows up and explain the purpose. Btw can anyone explain what an international public response is? — Preceding unsigned comment added by Kittens n thugs (talk • contribs) 23:12, 3 April 2020 (UTC)

Hungary, also part of the EU has almost the same population as Sweden, yet 1/10th of the COVID-19 deaths and cases, even though Hungary isn't nearly as wealthy. One difference is the partial curfew in Hungary and the other being all vaccinations, including BCG have been mandatory and actively enforced in Hungary since 1954, instead of merely recommended. All in all, something is off in Sweden. 94.21.160.90 (talk) 22:28, 4 April 2020 (UTC)

First, on Wikipedia we use sources to suggest edits — mere assertions are totally irrelevant.

Second, your point that Hungary isn't nearly as wealthy is apt — as it is also the reason why Sweden has many cases — because of high travel to Italy in February. There is no point comparing countries in that way before we've seen the full progression in Hungary. So far, all there is evidence of is that Hungary is later on the same curve. Carl Fredrik ^talk 09:02, 5 April 2020 (UTC)

Using new sources from Folkhälsomyndigheten?

The agency has now published day to day numbers for both number of cases and number of deaths. These official numbers can be found here. I have also made a copy of them here. The problem is that the last days don't have all the reports. This means that the curve appear to stagnate. But up to the last 5 days the values should be really accurate. Now to the question:

• Should we combine the accurate day to day data from Folkhälsomyndigheten up to the last 5 days and then continue with the day to day reports?
• Should we simply use their data?
• Should we continue to use (the now widely inaccurate data) that is reported for each day?
• Should we use the official day to day data but simply cut the last 5 days?

What do you think? Mrconter1 (talk) 07:52, 5 April 2020 (UTC)

Wait till 14:00 for updated statistics. Carl Fredrik ^talk 08:59, 5 April 2020 (UTC)

What do you mean? The statistics will not be corrected at 14.00. I am discussing the fact that we now are using the values that FM has given each day. These are wrong according the statistics they released today.Mrconter1 (talk) 10:29, 5 April 2020 (UTC)

@User:Mrconter1 The article is now updated with the correct values. Folkhälsomyndigheten does daily updates during the afternoon, meaning the article needs to be updated daily to be up to date — Preceding unsigned comment added by Kittens n thugs (talk • contribs) 10:54, 5 April 2020 (UTC)

Deaths per 100K

In the table "New COVID-19 cases in Sweden by county," "Deaths per 100K" currently (Apr. 5, 2020) shows 33.4 deaths per 100,000. Using 10.12M as the current population of Sweden, gives us 101.2 segments of 100,000 population (10.12M / 100,000) making up the whole population of Sweden. 401 deaths divided by 101.2 gives us 3.96 deaths per 100,000. Not knowing the population of the various counties of Sweden but observing the deaths per 100,000 seems to indicate a similar problem. However, I could be misinterpreting something (again), but thought I would check.--Caracoid (talk) 19:05, 5 April 2020 (UTC)

Disruptive behavior

Mercer17, please be kind and stop making repeated disruptive unsourced edits in the short paragraph about international media coverage, as seen here, here and here. — Preceding unsigned comment added by Kittens n thugs (talk • contribs) 10:35, 7 April 2020 (UTC)

Number of available ventilators and intensive care units

The main reason to abandon the goal of herd immunity in many countries was the lack of enough ventilators available for intensive care. For example in Austria there were only 2700 ventilators available. Also it turned out that many of these were specialized for a certain use case, that in fact less than 1000 of them were free for COVID-19 use. It should be quite clear that the limiting factor for Corona herd immunity is not the amount of hospital beds, but the amount of the ventilators. Even if only 10% of the population would become infected the need for them would be by more than a magnitude higher than available. These are the numbers for Austria. How about Sweden? How many ventilators and intensive care units having them are available there? — Preceding unsigned comment added by 188.22.225.147 (talk) 12:45, 7 April 2020 (UTC)

"Hospital beds" in this context doesn't refer to literal beds, it refers to the capacity of the health system by the number of patients that can receive in-hospital care at any given point. — Preceding unsigned comment added by Kittens n thugs (talk • contribs) 13:42, 7 April 2020 (UTC)

inconsistent data in graph "no. of new cases"

the data shown and written is different from that in other parts, at least for the last few days — Preceding unsigned comment added by 5.54.186.113 (talk) 12:09, 8 April 2020 (UTC)

Recoveries in the graphs

Should we remove recoveries from the graphs? The numbers look obviously incorrect so I think it is better to not have them there at all. Jeltz talk 13:25, 12 April 2020 (UTC)

This has been discussed before. I will remove the recoveries later unless there are objections. I will also remove the active cases which appear not be useful as well. They can always be restored if there are fresh up-to-date data on them. Hzh (talk)

Someone appears to have adjusted the graphs and added some more, so unless the questions of reliability and sourcing of the data arise again, I'll leave them as they are for the time being. Hzh (talk) 12:53, 14 April 2020 (UTC)

Debate section - Deleting ref to the Imperial College study

I removed in its entirety this little section:

Ignoring the fact that a study by the Imperial College London has found that as of March 28, 3.1% of Sweden's population had the virus vs just 0.41% for it's neighbour Norway, which had taken much stronger actions to stop the spread.

The information is taken accurately from the study, however its interpretation is not at all clear, and it seems to have been added to make a point ("Ignoring the fact that ...").

With all respect,

Springnuts (talk) 13:02, 14 April 2020 (UTC)

It makes perfect sense that he ignored Norway when responding to someone who comparing Sweden to the US. Jokes aside, I also interpret it as they were trying to make a point, and it wasn't the first time Norway made a sudden appearance in the section. Norway might actually be relevant for the section though, as it gets mentioned a lot in the debate, and there is a lot of debate to mention. I'm working on expanding the section and started with the international coverage, main subjects and will do my best to give equal weight to "both sides", and I'll try to weave Norway in to it. hopefully they will be satisfied

Government policy

The section on government policy should also mention the results of "trusting the experts" leading to 40-50 % of school children being held at home by their parents, even though elementary schools are not in lockdown. Sweden has mandatory education for children (home schooling is prohibited), and when school children do not attend class their parents risk being reported to the social authorities. The section also mentions, that the Swedish society has "a very high level of trust in its authorities from both the people and the politicians", which in general terms perhaps might be true, but there are many opinions amongst medical and nursing home workers on the authorities in relation to the approach on Covid-19 that does not have a high level of trust. 94.191.143.31 (talk) 22:48, 16 April 2020 (UTC)

Where did you find "40-50 % of school children being held at home by their parents"? Couldn't find it in your source. Tried to find some data myself, but I could only found reports for individual municipalities, which varied wildly. For instance, the total absence in comprehensive/compulsory schools in Stockholm was 25% during week 14. This number includes all sickleave, absenteeism etc. We can of course speculate how many of those had good reason to stay home, but that would go under WP:NOR. Maybe we need to wait with this one? But the massive criticism from nursery home workers should absolutely be covered in the article (same for home care workers, hospital personnel etc). Also, there is recent data regarding trust in authorities, could be relevant.

Charts and figures

Personally, I like the charts, especially the logarithmic chart of total cases (and deaths). They may be a bit esoteric, but they mean a lot to me. They tell us what the growth rate is, when countermeasures make it no longer exponential, and give us some idea of what the saturation case numbers and time will be (extrapolation, but still important). I make my own for several countries and US cities, but I didn't make one for Sweden because there was a nice one here. I hope someone puts it back. - Zeke, April 14

I have no idea what is going on with the figures, but they are not making sense, and some haven't been update. One says the number of new cases for 5 April is 84, while another says it is 341, yet another suggests it is just under 400, and another over 500. This is not "slightly different".  Hzh (talk) 17:09, 8 April 2020 (UTC)

@Hzh: I fully agree. I tried to start a discussion about this a week ago, sadly nobody joined in. But as there are now two of us, I guess two is a party and we can reach a consensus. What do you think?

I say remove every single one. While I do think charts add a lot to this article, any inaccurate information must be removed Immediately (and fact-checking is easy in this case). As said, I think charts make the article better, but until the numbers are accurate they simply has no place in the article. Kittens n thugs 💬 22:16, 9 April 2020 (UTC)

Hi, I don't think that charts are inaccurate. I think they just use different definitions. The main issue is the difference between the date when the death was declared and the date when death occurred. Swedish authorities correct the dates of death after a few days, whereas other countries only give the number of new deaths reported everyday (see: https://twitter.com/MarkkuPeltonen/status/1247019498642825216 ). So I don't see any reason to remote Template:2019–20 coronavirus pandemic data/Sweden medical cases chart from the page. This chart is present on the Wikipedia page of all countries in the world I checked, so it is really useful to be able to compare countries. What do you think? Best, Antoine (in Sthlm :) ) A455bcd9 (talk) 10:19, 10 April 2020 (UTC)

@A455bcd9:You have a point, I guess the graph would we accurate if we rename it to something like "increase in announced cases and deaths per day", with a note like "the figures in the chart doesn't reflect the actual number of cases and deaths on a specific day". And perhaps "contrary to what the graphics suggest, Covid-19 isn't less contagious during weekends".

But as it's easy work to make a chart showing number of actual cases per death, I think than one would be more relevant. Only problem for me is I don't know how to do it. When I press E for "edit" all I see is a lot of confusing code. Is there a GUI somewhere? Had this been a spreadsheet I would have fixed it within minutes.

Also, some of the data used for the graph is from an unreliable source (this one has been discussed on the talk page, and I don't think anybody opposed it), and the figure shouldn't be near the article as long as the data is still there. Kittens n thugs 💬 11:06, 10 April 2020 (UTC)

Since the outbreak escaped China, I have been using the "Template:2019-20..." charts for each country as a means of keeping tabs on each country and as a "generally reliable" source of data for tracking the virus spread so that I may analyze and test my various hypothesis. In short, the omission of this particular table from an article I find to be a serious disservice, (unless it is so egregiously inaccurate as to make even a rough gage of trends completely misleading. Even then, an empty window with link & disclaimer ought to be inserted in its place).MarkSonntag (talk) 10:52, 10 April 2020 (UTC)

Why not put in the official data from fohn (with the backports), its available for download in a xls from https://www.folkhalsomyndigheten.se/smittskydd-beredskap/utbrott/aktuella-utbrott/covid-19/bekraftade-fall-i-sverige some xls magic should make it not that hard? — Preceding unsigned comment added by 155.4.92.139 (talk) 12:49, 10 April 2020 (UTC)

Agreed. I just came here to get a feel for what's going on in Sweden, and the main graph has gone. Surely a decent source is available? Arcturus (talk) 14:06, 10 April 2020 (UTC)

@Arcturus: Yes.Here is the official data. Updated daily.

I've restored the main graph for the time being, maybe get a consensus first before removing it. Hzh (talk) 20:54, 10 April 2020 (UTC)

A few questions -

1. Is there a preferred data set (or sets?) that should be used? As mentioned, the numbers are different for all the graphs, maybe the number are being constantly updated in at least one of the sources, so what number you use depends on when you access the numbers, or the different data sets are updated at different times, which resulted in an apparent discrepancy in numbers which may not be actually real. Sticking to just one set of data may be better? Hzh (talk) 21:54, 10 April 2020 (UTC)
2. Which graph or graphs should be used? The three graphs in the chart section appears not to be regularly updated, and I'm not sure which data set they used (because the numbers don't agree).
3. Is there any point in having "confirmed cases" and "active cases" in the cases graph? Hzh (talk) 21:54, 10 April 2020 (UTC

@Hzh: Or perhaps a consensus should be reached before you restored something removed per WP:INACCURATE. However, this has been discussed at length several times before, and it's possible a consensus had already been reached before you reached a new one with yourself. Please see the archived discussions. But as I belive more people will come here and ask why there are half a dozen graphs and a table all showing different numbers, perhaps someone needs to step in. I'll leave it to others after giving my opinion.

1. In my opinion it would indeed be a good idea to stick to one set of data. I would have gone for the official data (as linked by several users on this talk page), as opposed to "alternative facts" or, to quote MarkSonntag above, "generally reliable". In my opinion a source can only be either reliable or unreliable. Also regarding sources, I'll again ask you to read the archived discussions. This is what I had to say about Worldmeters, which seems to be a household name among some users, and keeps making its way back to the article: Apart from having errors in their stats, they are at best tertiary sources, and sometimes even quaternary sources [...] one of the sources links a tweet from a TV station in Gibraltar. It might be a good source, I have no idea, however the tweet is about a new Corona case in Gibraltar. Even worse, an update about "167 new cases in Sweden" links to yet another obscure site called Platz, which in turns links a myriad of different sources for those 167 cases, which obviously makes it impossible to verify. So: here we have an unreliable source, using another unreliable site as source, which in turn links to updates from established news media, reporting new information from official press releases.

2. Again, please look at the previous discussions. My opinion: Graph #1 (black and red): Yes, and this isn't really up for discussion. Graph #2 (yellow and red): I'm not sure what I'm looking at, there's almost no difference between confirmed and active? Graph #3 (don't remember the colors): This is a total mess. That's why I removed it. Surprised to see nobody put it back in already. Graph #4 (red): Is this really needed when we have the first graph? Graph #5 (black and baby blue): See graph #2. Btw what's the source for recovered cases? Graph #6 and #7: I have no idea. I've actually asked this question previously: What purpose does a logarithmic chart fill for the average reader? Or any reader at all, because the section offers zero context to why they're even there in the first place. If there's a good reason to include them in this chart gallery, it needs to be explained in the non-existent text body. Which will make the section too detailed anyway. The section is confusing as it is, with a zillion charts showing the same information found in the table (the table is really good).

3. See above. Kittens n thugs 💬 01:04, 11 April 2020 (UTC)

Hi,

All the pages I checked for other countries have the same graphs. I don't see any reason to remove them from the Swedish page. Then obviously we need to make sure the source is official and serious and the data is accurate. But this point (data accuracy) can be fixed in the template of the different graphs or in the content of the page. The most reliable global source is probably the European CDC (used by Our World in Data): https://www.ecdc.europa.eu/en/geographical-distribution-2019-ncov-cases It is up-to-date, has the same definition for all countries and is available in several languages (including English). An alternative could be to rely on the Swedish authorities' websites directly.

Regards, A455bcd9 (talk) 08:26, 11 April 2020 (UTC)

@Kittens n thugs: It would help if you can link to the discussion where there is a consensus to remove the graphs. The issue with the data set used is different from the graphs and charts. The data in the graphs can always be adjusted if necessary, so if you are unhappy with the data given, then change the data rather than delete the graphs. I see that there is some agreement that the data from folkhalsomyndigheten should be used, then stick to that, and adjust the graphs and charts appropriately. It seems that the numbers from folkhalsomyndigheten may change as they are updated, then perhaps they can be adjusted later and avoid using incomplete figures, add a hidden note if necessary. However, I don't see a cumulative daily figure given there, is there one anywhere else? If other charts use other data set, then discuss if they should be deleted. There is also the question which data set it is that they are using.

Graphs are useful for visualising data, it is hard to tell what the trends are from the mass of data in the table, but it is clear from the graphs. The semi-log graphs may be useful in telling us how the outbreak is progressing, although I think it may not be necessary (a similar one was deleted from the main article). I guess the daily new cases may be also useful in giving an idea of the trend (the red and black one doesn't list daily new cases separately, which I assume is the reason there is a separate graph). Hzh (talk) 10:26, 11 April 2020 (UTC)

On a separate note, the referecing is a mess, there are many missing references. Someone needs to tidy them up. Hzh (talk) 10:26, 11 April 2020 (UTC)

wp:notnews we do not need a constant tally. I would suggest one edit every 24 houses at around 1200 gmt using the previous days tally.Slatersteven (talk) 10:46, 11 April 2020 (UTC)

I haven't yet studied all the sources mentioned above, but I would make the following general observations. Of all the subsidiary articles, this one is currently one of the most important, due to the unique way Sweden is handling the situation. As such we do need the main graph, and we need a completely reliable source. I know this latter point is stating the obvious, but I mention it here just as a reminder. Arcturus (talk) 11:44, 11 April 2020 (UTC)

@Arcturus: Had to check. This article is currently the 134th most viewed page on Wikipedia! If someone could fix this, they deserve a virtual medal on their talk page. Do you know if such awards exist? For someone with the skills, I believe it could even be automated. The updates it is, not the medals Kittens n thugs 💬 21:26, 11 April 2020 (UTC)

Not arguing with that, we just need to make sure we wait until a given days figures can be verified (and at a minimum that means 24 hours, and it might be argued longer).Slatersteven (talk) 12:21, 11 April 2020 (UTC)

@Slatersteven: Updating the Swedish stats with reliable sources is actually very easy. The chief epidemiologist even brags about Sweden having the world's most precise Covid-19 statistics. The stats are updated daily at 11:30 CEST, and made available in .xls online while simultaneously being announced to the public on a daily press briefing at 14:00 CEST. As punctual as when the Nobel committee announces the winner at 12:00, but probably with much better ratings.

The number of new confirmed cases and deaths during the past 24 hour period is reported, but this number is obviously distributed over two days (as it's always updated just before noon). Because of this, there has never any "new daily cases" to begin with. Additionally, as the cases are reported immediately, the cases are often backdated as a they are yet to be properly registered by the authorities.

Registering a death on the wrong date would be unthinkable in Sweden. The data is automatically shared between the Public Health Agency and and the Health and Welfare Agency, the Swedish Statistics Agency, and some 500 other government agencies, regional and municipal registers and countless of other databases. Future researchers will have no problem to study Covid-19 mortality among female former pilots who didn't pay their TV-license in 1973.

TL;DR. If our chief state epidemiologist is to be believed, this article should have Wikipedia's most precise Covid-19 statistics. Kittens n thugs 💬 15:00, 11 April 2020 (UTC)

All that needs to be done is to update the figures when the official source updates the numbers, it does not mean what's given in the article is necessarily inaccurate, since the numbers in the official source also change. Hzh (talk) 15:15, 11 April 2020 (UTC)

No agree (for figures), the official data comes out once a day every day at 14:00 (cet) with data upto 11:30 (as you can read on the page of fohm and, is mentioned in all the notes everywhere), the totals don't change, only the dates when cases or deaths occured are backported to earlier dates, the figures should represent that data, not the data from the day before — Preceding unsigned comment added by 155.4.92.139 (talk) 12:30, 11 April 2020 (UTC)

I'm not sure what you mean. I assume you are the person who updated the table, so for 8 April, the total was first given as 8,419, but the next day, the number for 8 April had changed to 8,924, then 8,957 the following day, and now, 8,958. The total for 8 April obviously has constantly changed. The new cases seem to be always much lower on the day they are first reported. I have no problem with the figures changing according to the official source as they get updated, we can always add a note to the graphs that the numbers may change when the figures get updated so that people won't start wondering why the numbers are different and keep changing. It also means that the numbers for individual dates in the graphs may need updating for a few days afterwards. Hzh (talk) 13:18, 11 April 2020 (UTC)

sorry, with "figures" I only ment the right top "2020 coronavirus pandemic in Sweden" figure with the map, it should represent the latest official data available (which are the totals upto 11:30 that day), the charts (which is what this discussion is mainly about) are a different story, they can/should lag 24 hours behind and should be updated with the official (backported) data — Preceding unsigned comment added by 155.4.92.139 (talk) 13:29, 11 April 2020 (UTC)

Ah, you meant the infobox figures. We are actually discussing the figures in the graphs and charts, which are not consistent. I think we might be getting to the root of the problem, which is that the numbers get updated at different times, and they changed. So the numbers are not inaccurate as such, just taken at different time of the day or different days. We can just explain why the numbers might change over time in the Charts section. Do you have the link that explains why the numbers change? Hzh (talk) 13:49, 11 April 2020 (UTC)

@Hzh: Actually, official stats is only updated once a day, at 14:00 UTC+02:00. The root of the problem as I see it can be seen in my reply to Slatersteven above (15:00 today).

Inconsistency between different graphs is something different to inconsistency within one graph. Had everything been sourced correctly there would be a separate source for each update (as date+time would be needed). The problem is caused by people making quick edits to the graphs just by adding data, without bothering about consistency. By doing it this way, each update comes from a new data set (and consequently, a new source). Had all the charts and tables actually been sourced properly we would need a separate source for each edit (date+time) and we would have several hundreds of additional entries in the reference section. Not to mention the number of non-unique references. I have to admit it would be fun to see how long it will take til we reach the 5-digit amount of non-unique references in an otherwise well sourced article with daily updates of stats in all those those charts and tables. Kittens n thugs 💬 16:27, 11 April 2020 (UTC)

Is there no one place where you have all the updated data? For example, here - https://experience.arcgis.com/experience/09f821667ce64bf7be6f9f87457ed9aa in the chart for new cases per day you can see the numbers for all the days apart for the earliest, and the numbers seem to fit with the ones given in the table here in the article. You can see that the number for the latest date is likely not the final number, but people (including IP user 155.4.92.139) have been doing a good job updating the numbers in the table (presumably they have access to the correct numbers). We can just use the numbers from the table for the graphs, we just need to explain that the numbers can change when they get updated. Hzh (talk) 18:29, 11 April 2020 (UTC)

the (same) numbers can be found here in a xls: https://www.folkhalsomyndigheten.se/smittskydd-beredskap/utbrott/aktuella-utbrott/covid-19/bekraftade-fall-i-sverige (the link: "Data som statistiken ovan bygger på kan laddas ner här (Excel)" then use a spreadsheet to calc the totals en percentages needed (this is how the table is compiled) The problem is not the data, but that someone has to fix it and keep it updated, which with the backports means (a lot) more work, and keep an eye on ppl only updating the last day which would invalidate the data — Preceding unsigned comment added by 155.4.92.139 (talk) 18:41, 11 April 2020 (UTC)

If all the data can be found in that source, then we only need to use just one single source, and derive the charts from those numbers in the table. Thanks for your work updating the table. There will be a lot of work updating the data in the charts, especially the main red and black one which appears to use data that have not been updated. There is just the question of recovered cases and active cases (which I assume is confirmed cases minus recovered cases), where do those numbers come from? Perhaps we should remove them? Hzh (talk) 19:08, 11 April 2020 (UTC)

What about the comment in the graph template; The numbers in this chart doesn't reflect the actual number of confirmed cases and deaths for a given day. For accurate numbers and % increase in daily confirmed cases and deaths, see 2020 coronavirus epidemic in Sweden. The wiki-link just takes you to a different part of the article, where the summary data is the same. I checked the current figures with the source from the Swedish PHA and everything is fine. Should we remove the comment? Presumably we should also agree just to update the graph and other data once a day, when the 14:00hrs data is available? Arcturus (talk) 19:29, 11 April 2020 (UTC)

I'm a bit confused, which data is the same as which? The data set in the main chart (red and black one) is not the same as the one in table except the latest numbers which have not been updated (you can check the numbers before today). the data in the table is the same as the PHA one. The comment in the main chart can be removed once the numbers in the chart have been updated. Hzh (talk) 22:15, 11 April 2020 (UTC)

"Data is compiled by Folkhälsomyndigheten at 11:30 (UTC+02:00) each day. Reports of new cases and deaths to Folkhälsomyndigheten might be delayed by up to several days, especially around weekends, possibly introducing delays in reported number of cases for the last few days." so why not just wait a few days until we know for sure?Slatersteven (talk) 13:44, 15 April 2020 (UTC)

I see no problem with it as long as it is explained, so that people reading it know that the figures are not the final numbers. The explanation would also help them when they see the latest figures in official website to understand that even official numbers may not be the final figures. Hzh (talk) 11:01, 16 April 2020 (UTC)

It seems that the death stats are not properly updated when FOHM revises them up to days later. The actual current numbers are much higher than what is shown in the graph, e.g. 114, 170, and 130 for April 14th, 15th, and 16th, according to Johns Hopkins Npc239 (talk) 11:18, 17 April 2020 (UTC)

Then Johns Hopkins clearly isn't a reliable source, because the actual current numbers are 56, 45, 38. The charts and tables are from yesterday, that's why they are lower. Every number in the article is sourced

Invitation to edit

You are cordially invited to edit Draft:Mismanagement of the 2019-20 COVID-19 pandemic. Calmecac5 (talk) 20:24, 27 April 2020 (UTC)

Recoveries

Just out of curiosity, why is the number of recoveries not included in the information box and the charts for the pandemic, unlike in the case of other countries? Is it because accurate data is lacking (due to what constitutes a "recovery" being difficult to define)? — Preceding unsigned comment added by 95.43.30.163 (talk) 20:45, 2 May 2020 (UTC)

The official Swedish data source FOHM COVID-19, arcgis.com, does not seem to provide recoveries: the main data tabs are Sjudkdomsfall/dag, Sjudkdomsfall/dag kumulativt, Intensivårdade and Avlidna/dag, none of which is recoveries. By contrast, Worldometers does provide recoveries for Sweden, but it is unclear what their data source is.

As to why the Swedish data source does not provide recoveries, I don't know. I for one think that putting the number of recoveries next to number of deaths as Worldometers does at the top is grossly misleading, and without proper interpretation will cause the wrong picture in the mind of the unsuspecting non-expert reader. In general, recoveries seem to be harder to track and detect clearly, and compared to deaths they would suffer from a considerable time lag. I think the lay reader is better off ignoring recoveries altogether and only focusing on the shape of the curve of the new cases per day and the shape of the curve of new deaths per day. An article supporting the recovery-tracking difficulty: Most people recover from Covid-19. Here's why it's hard to pinpoint exactly how many, Apr 5, cnn.com. --Dan Polansky (talk) 06:40, 3 May 2020 (UTC)

I am the same user who posed the question! Many thanks for responding, great explanation and I agree with your reasoning! Stay safe & healthy and I hope that things everywhere will get better soon!

You're more than welcome. Note that Worldometers:Sweden now has a spike of recoveries on May 4, over 3000 recoveries on May 4. That data is more noise than signal to me. --Dan Polansky (talk) 07:53, 5 May 2020 (UTC)

Excellent point, I wish a miraculous spike in recoveries like that could really happen, but such numbers should always be viewed through the proper prism! — Preceding unsigned comment added by 95.43.30.163 (talk) 20:35, 5 May 2020 (UTC)

Copyright of Swedish all-cause death graph by FHOM

There is a vitally important graph in Week 19's report by FHOM, folkhalsomyndigheten.se, Figur 10. The graph shows weekly all-cause deaths in Sweden. Does anyone know whether there is something in the Swedish copyright law that would put the graph into public domain, or whether there is another copyright-law-based justification for including the graph in Wikipedia, after uploading it to Commons? Like, could it be that all publications by a state agency such as FHOM are in public domain? --Dan Polansky (talk) 06:58, 22 May 2020 (UTC)

Does Commons:Template:PD-Sweden-URL9 apply? The template says "This document is in the public domain in Sweden because it is a reproduction of a law, decision, or report issued by a Swedish public authority (svensk myndighet) or an official translation of such a text. Lag (1960:729) om upphovsrätt till litterära och konstnärliga verk 9 §. This does not include maps, visual art, musical works, or poetic works."

FHOM seems to be a public authority by its name (it is a "svensk myndighet"); the graph is an element taken from a report and the graph is neither a map, nor visual art, nor musical work nor poetic work. --Dan Polansky (talk) 10:13, 22 May 2020 (UTC)

As far as I am aware, it is probably legally fine to use the image. In the remote chance that it is not, the agency can easily just ask us to remove it again. David A (talk) 11:09, 22 May 2020 (UTC)

@Dan Polansky: There's no need really as the data is easily accessible: Number of deaths per day 2015-2020, Number of Covid deaths per day, Excess mortality in Sweden.bladjur (talk) 11:24, 22 May 2020 (UTC)

Thank you both. Euromomo shows z-score for Sweden while the FHOM report shows absolute death numbers, much better for the lay reader, I think. And Euromomo graphs are not in public domain (or are they?), while it seems the FHOM report is in public domain, and could therefore be used in Wikipedia. --Dan Polansky (talk) 11:57, 22 May 2020 (UTC)

@Dan Polansky:

Line chart vs non-stacked area chart showing weekly deaths in 2020 and 2015-2019 average. Weekly numbers can be replaced with daily numbers, and each year can have a line of its own

Perhaps a line chart showing 1 Jan to today for 2015-2020 and the average, with more weight (thickness, color) to the average and 2020 line?

Line chart vs non-stacked area chart showing the number of reported deaths vs excess mortality. This might be interesting as well. Haven't checked yet, but it might be possible to split reported deaths into confirmed and non-confirmed.

Note that this was quickly done using the simplest template, so I haven't double-checked the numbers or given much thought into the design or the esthetics (I simply wrote the names of some colors). — Preceding unsigned comment added by Kittens n thugs (talk • contribs) 13:38, 22 May 2020 (UTC)

Here we go:

It would be good to be able to plot this ourselves, where 1) the y-axis should start at zero and 2) the data should go many years back if possible, ideally back to the 1968 Asian flu, but any more years back would be good. Furthermore, 3) plotting the actual deaths without any expected deaths and standard deviation lines would suffice and could be even preferable: the raw data speaks clearly enough on its own. --Dan Polansky (talk) 15:00, 22 May 2020 (UTC)

Picture

I fail to see how painting it red is designed to scare people (after all it kills peoples, which is rather more scary than the colour red).Slatersteven (talk) 14:39, 23 May 2020 (UTC)

We are talking about File:2019-nCoV-CDC-23312 without background.png. Red is scary, it is obvious. Anecdotally, this particular image looks super scary to me. The red in the image does not serve to represent anything; it does not talk to the high-level congitive capacity of the viewer but rather to the low-level vulnerabilities. Furthermore, this article is not about the virus but rather about the epidemic, and therefore, the image does not even properly belong there. --Dan Polansky (talk) 14:47, 23 May 2020 (UTC)

It serves to highlight what is different areas look like, it is done all the time where the objective is to give an easy to interpret visual representation.Slatersteven (talk) 14:51, 23 May 2020 (UTC)

Color can be used to highlight different parts of an object, but it does not need to be the scary color. It is obvious that mass media chose the red color in their visuals to create maximum scare effect; Wikipedia should do all that it can to abstain from that kind of tabloid behavior, and maintain its status as a NPOV encyclopedia. I recently saw some media make rectifications in that area, and that is good news.

And the shape of the virus is not even relevant to the spread of the epidemic; it is completely ancillary information. --Dan Polansky (talk) 14:54, 23 May 2020 (UTC)

Who said they did, this is just your opinion. Lets ask others, is "RED!" scary, so scary using it violates NPOV?Slatersteven (talk) 14:57, 23 May 2020 (UTC)

Me and you obviously disagree on the scary effect; please respond to the point that the virus image, even if toned down, does not properly belong to an article on the epidemic, that is, the spread and demographic effects of the virus. --Dan Polansky (talk) 14:59, 23 May 2020 (UTC)

I have asked others to chime in. I will wait until others respond as the key issue is NPOV, not undue.Slatersteven (talk) 15:04, 23 May 2020 (UTC)

I have raised a separate point (img off topic in an epidemic article), and I got an evasive non-answer. I raise both issues: 1) too scary and tabloid and therefore not NPOV, and 2) not fit per article topic. --Dan Polansky (talk) 15:06, 23 May 2020 (UTC)

I have no view on this one way or the other. I do not see why we need it, I do not see why we cannot have it.Slatersteven (talk) 15:10, 23 May 2020 (UTC)

Incidentally, in Bill Maher was 'absolutely bang on' when he derided COVID-19 media hysteria, youtube.com, Bill Maher says "stop showing us this; you know everything looks scary when you magnify it thousand times" in relation to a picture that looks very similar to what we are discussing here. Now I am not passing Bill Maher as a reliable resource for Wikipedia purposes, but at least one another person agrees with me that the picture looks scary. --Dan Polansky (talk) 18:51, 23 May 2020 (UTC)

Maybe we can just solve this by painting it in a different color. bladjur (talk) 07:23, 24 May 2020 (UTC)

New cases

No new cases since May, 1st? Is Covid-19 finished in Sweden? — Preceding unsigned comment added by 92.209.18.63 (talk) 16:23, 3 May 2020 (UTC)

25-30,000 cases so far as of May 10th in Sweden. In a country of 10.3 million people, Swedish health officials assumed 1.3 million were exposed to Covid-19 so far, plus a quarter of Stockholm County (over 600k out of 2.4 million). And 3,000 plus deaths, although unfortunate, Swedish society believes drug treatments, antiviral research and in the future, a vaccine, showed they have a huge trust in science to manage a pandemic without lockdowns they feel are disruptive to society, the economy and human evolution to be immune to a new disease if it doesn't disappear from the earth (but SARS-1 did vanish in 2004 and MERS had 3 small outbreaks in the 2010s). 2605:E000:100D:C571:7D82:A683:E434:DB3D (talk) 06:22, 10 May 2020 (UTC)

The Swedish Ministry of Health announced even without a national or regional lockdown, the number of daily cases are in decline and the rate of hospitalizations plateaued. International criticism said Sweden is trying the "herd immunity" approach when it involves young and healthy people getting exposed to a coronavirus similar to the flu and common cold, but the cooperative public knows to control themselves with social distancing (although seen as an option best suited in certain situations) and other measures to not left Covid-19 spread out of control to cause more deaths and overwhelm hospitals with too many patients for their limited numbers of ICUs. An odd but apparently working situation, the WHO (World Health Organization) recently praised Sweden's no-lockdown approach to end the wave. 2605:E000:100D:C571:7D82:A683:E434:DB3D (talk) 06:19, 10 May 2020 (UTC)

Is there a reason that the new cases by county chart stops on May 7th? It doesn't look as if somebody has been updating this by hand, but Folkhälsomyndigheten has continued to produce statistics by region, here for instance https://experience.arcgis.com/experience/09f821667ce64bf7be6f9f87457ed9aa Just checking to see if some screen-scraping script is not working as intended, since I wasn't able to figure out how the spreadsheet was being populated. Seems to be fixed now. Lacreighton (talk) 00:12, 15 May 2020 (UTC)

Let us plot fresh daily cases from Folkhalsomyndigheten_Covid19-2.xlsx, arcgis.com, tab "Antal per dag region":

Let us also plot data for Stockholm area, from the same source:

--Dan Polansky (talk) 07:13, 22 May 2020 (UTC)

It all over the place.Slatersteven (talk) 16:10, 22 May 2020 (UTC)

A note on interpretation of the above, although it is stating the obvious: the epidemic is receding in Stockholm area, but since it is raising in some other parts of Sweden, it is on a plateau in Sweden overall. Graphs with similar information for other parts of Sweden, albeit using bar charts, can be found in the FHOM weekly report[1] by searching for "Antal laboratoriebekräftade fall av covid-19 per region". --Dan Polansky (talk) 08:50, 24 May 2020 (UTC)

statistics

in the right bottom corner of the big table, we have composition percentage for death counts, and ICU counts. I would suggest change to death rate(death/infected) and ICU rate(ICU/infected). any support or objection? Jackzhp (talk) 07:02, 20 April 2020 (UTC)

Like this? and we need a abbr for percent and rate with a bit more explanation about what it represents

In at least two places (the summary at top right, and the charts at the bottom - search for "hospitali"), this page conflates "hospitalized" with "ICU". This is wrong: many (most?) patients who are hospitalized never enter ICU. Eg, right now the summary at the top right says "Hospitalized cases: 1,158", when the linked source (in Swedish) lists 1,158 as the ICU number. I suggest changing the wording from "Hospitalized cases" to "ICU cases", and "Hospitalisations" in the charts to "ICU hospitalisations". Unless someone knows a good source for the number hospitalized (not just in ICU) - I don't.

Who is counting the victims of lockdown in order to present the fair comparison of approaches? — Preceding unsigned comment added by 188.195.216.206 (talk) 11:18, 24 May 2020 (UTC)

Numbers manipulation

What's the point of everyday changing the old numbers? To hide the fact that the situation is not improving? Wikipedia should not serve as a propaganda machine to political parties. — Preceding unsigned comment added by 193.52.24.15 (talk) 09:45, 2 May 2020 (UTC)

German Wikipedia shows two daily death charts (de:COVID-19-Pandemie in Schweden#Todesfälle), which makes it a little more clear what is going on. The thing is, one has to distinguish the death occurrence date from death report date (the terms chosen by me; the terminology may vary), the latter being generally later, accounting for a delay. As a result, when new deaths are reported on a day, that impacts the graph columns of past days of the graph that shows death occurrence dates. Either way, the two graphs in the linked German Wikipedia section both shows that, on a per day basis, the situation of the deaths is improving with daily deaths going down, and that is apparent on both graphs. Daily new cases are not improving but they are approximately constant, and therefore, the cumulative case growth is linear, and that is manageable, and is rather fine.

The English Wikipedia only shows the death occurrence date graphs, which looks smoother (no regular spikes and gaps), but has the last several columns updated as new data arrives; no fraud going on here. --Dan Polansky (talk) 11:59, 2 May 2020 (UTC)

That said, arguably, the smooth graph has a slightly deceptive effect: because of the registration delay, the rightmost columns are always incomplete while the older columns are complete, and therefore, the right end of the smooth graph is artificially going down a little compared to reality. Ideally, the last columns should have a different color suggesting lack of completeness, or there should be other means in the graph to warn of the problem. Another option is to do the same thing as the German Wikipedia: show two graphs of daily deaths. --Dan Polansky (talk) 06:04, 3 May 2020 (UTC)

Why are the numbers here so completely different from those at https://ourworldindata.org/grapher/covid-daily-deaths-trajectory-per-million?yScale=linear&country=SWE Skeptical Realist (talk) 05:56, 6 May 2020 (UTC)

The ourworldindata.org link above shows daily deaths per million (not absolute count), and furthermore, it shows 7-day rolling average (moving average), so their curve is smooth but not because it is death occurrence date but rather because it is death report date curve that was smoothed via moving average (Worldometers started to show moving-average curves as well). The last 5 data points from the ourworldindata.org page: May 1, 2020: 7.992; May 2, 2020: 7.08671; May 3, 2020: 6.74714; May 4, 2020: 6.86029; May 5, 2020: 7.00186. If we reckon 10.23 million pop for Sweden, 7 daily deaths per million corresponds to 71.61 absolute daily deaths. That is quite a bit more than what the current Wikipedia graph shows, which is in part caused by the fact that the rightmost columns in death occurrence date graph are by necessity incomplete: it is the slightly deceptive effect of this graph I mentioned above, and in part by the time lag introduced by the moving average, where the higher old values drag the last value up via the averaging. The nice thing about the ourworldindata.org moving average graph is that it does not seem to have this deceptive effect. On the other hand, moving averages introduce a time lag into the smoothed value so any real improvement is slowed down by them. Maybe the English Wikipedia should have two graphs like the German Wikipedia (de:COVID-19-Pandemie in Schweden#Todesfälle) does, the 2nd graph using the death report date. Either way, Swedish deaths per ourworldindata.org show an optimistic trend as well. The ourworldindata.org number scale is corroborated if we use the jaggy graph from German Wikipedia and for May 4 we calculate the moving average: (85+90+10+16+67+124+107)/7.0 = 71.3, which approximately fits the previously calculated 71.61 absolute daily deaths. --Dan Polansky (talk) 06:59, 6 May 2020 (UTC)

One thing has to be granted: the current WP daily death graph is highly misleading for the readers since it shows an unrealistic downward sloping trend at the end. Replacing it with the other type of graph or at least adding the other type of graph would seem imperative. --Dan Polansky (talk) 07:12, 6 May 2020 (UTC)

I took the data from de WP and calculated a 7-day moving average:

--Dan Polansky (talk) 08:04, 6 May 2020 (UTC)

On the other hand, the 7-day moving average graph above suffers from double delay: 1) the day of report is delayed compared to day of occurence; 2) the moving average causes further delay.

Let us make a graph of daily deaths by date of occurrence, ignoring last 7 days to prevent the deceptive effect of registration delay:

The above has no deceptive end (we dropped last 7 days) and it peaks earlier than the 7-day moving average of deaths by day of report.

The above suggests to me that providing both graphs would be preferable, and it would be nice to indicate in one of the graphs that, say, last 7 values are very unreliable.

--Dan Polansky (talk) 12:38, 6 May 2020 (UTC)

Update: Let us make the above graph with fresh data from Folkhalsomyndigheten_Covid19-2.xlsx, arcgis.com, tab "Antal avlidna per dag", this time dropping last 9 days rather than last 7 days (when I dropped 7 days, I saw at the end something that looked like an artifact of late registration, and such is seen in the previous graph above as well):

--Dan Polansky (talk) 06:39, 22 May 2020 (UTC)

Update from the same source[2], dropping last 9 days:

--Dan Polansky (talk) 11:43, 31 May 2020 (UTC)

Dropping last 9 days in the daily death graph

I somewhat boldly dropped the last 9 days in the daily death graph, to avoid the deceptive downward slope at the end. I propose to keep this practice. Data source: Folkhalsomyndigheten_Covid19-2.xlsx, arcgis.com, tab "Antal avlidna per dag". --Dan Polansky (talk) 19:47, 22 May 2020 (UTC)

Dropping 7 days seemed not enough, by my impression; maybe we should round it up, and drop last 10 days. I quote Anders Tegnell from Coronavirus: Anders Tegnell, State Epidemiologist of Sweden, on herd immunity - BBC HARDtalk, youtube.com, at around 1:46: "[...] And we do it in that way that we don't look too much at the last 10 days because we know they are very unsure [...]" --Dan Polansky (talk) 20:43, 22 May 2020 (UTC)

I don't think it should be changed without consensus as there's a consensus on doing it the way it was before, with numerous users doing the updates. And it's also consistent with how it's done on other articles about the pandemic in different countries. bladjur (talk) 07:23, 24 May 2020 (UTC)

People opposing this need to come up with substantive objections, not just "consensus". Other countries use different death day reporting methods; their daily death graphs did not suffer from the deceptive downward slope at the end. The problem with things being misleading urgently needs to be addressed, and dropping the last 9 days addresses the problem. --Dan Polansky (talk) 08:18, 24 May 2020 (UTC)

Let me add that dropping last 9 items from a series is very easy for any editor to do; there is no fancy statistical correction involved. Furthermore, File:All-cause weekly deaths - Sweden.png shows how FHOM week 19 report has graph ending in week 17, meaning they drop 2 weeks of data in the report, further confirming that FHOM themselves are aware of the problem. --Dan Polansky (talk) 08:28, 24 May 2020 (UTC)

The reason for the delay is more likely that excess mortality numbers aren't immediately available for the agency. They actually have a policy to share all data immediately, even when it's preliminary. bladjur (talk) 11:20, 25 May 2020 (UTC)

Sure. When a death occurs, on, say May 21, the record may reach the agency on, say, May 25. That record is kept in the graph on the May 21 date. As a result, say, May 24 column is underrepresented on May 25 since the records of May 24 deaths are going to arrive on May 25, or May 26, or May 27, etc. Therefore, there is a systematic effect that makes the rightmost part of the chart misleading: the appearance is that the deaths are going down much faster than they really are. When I first saw the chart, I was really confused by this, and I know other people were; it is being discussed in the media. Fortunately, this misleading effect can be removed by dropping last 9 items. --Dan Polansky (talk) 16:22, 25 May 2020 (UTC)

You should not drop the chart entirely if it continues a data set, as the new data gets added that graph will modify (or not, we don't know, until the data is in). As it stands now, this is a chart conveying officially reported data from the source website. --Balance66 (talk) 16:44, 25 May 2020 (UTC)

Why should I not? We report the data accurately per the source, but we post-process the source by dropping last 9 items. That is both accurate and non-misleading, which is good. Another solution would be to modify the chart to containing a warning or disclaiming element, but that is what no-one did ever since discussion of this problem started at #Numbers manipulation. --Dan Polansky (talk) 16:49, 25 May 2020 (UTC)

May I ask a newbie question then, is this standard practice for all time series charts? -- Balance66 (talk) 16:52, 25 May 2020 (UTC)

It is a practice that can be seen in the FOHM report in that their report for week 19 has a chart ending in week 17; they dropped two weeks of data to plot their chart. --Dan Polansky (talk) 16:54, 25 May 2020 (UTC)

I mean chart File:All-cause_weekly_deaths_-_Sweden.png, which is from this report[3]. --Dan Polansky (talk) 16:57, 25 May 2020 (UTC)

I think you could make the case that removing the last 9 days data of the daily deaths, while leaving the chart right above it, with the accumulated deaths runs the risk of misinforming the readers. You should back off the last 9 days of both charts so that both seen together convey an accurate picture of the same measurements. --Balance66 (talk) 16:57, 25 May 2020 (UTC)

Or we could drop the accumulated death graph entirely as not so interesting as the daily death graph. But having the daily death graph non-misleading is a very good thing on its own. --Dan Polansky (talk) 16:59, 25 May 2020 (UTC)

As for the raised point: these are two separate graphs, ending on different dates. I guess the last 9 items should better be dropped from the accumulated death graph as well; I do not object to that. --Dan Polansky (talk) 17:09, 25 May 2020 (UTC)

Either way would be better. The total accumulated deaths doesn't really convey any useful information as much as the daily rate, the rate of change or the rate per 100k --Balance66 (talk) 17:17, 25 May 2020 (UTC)

I also think that it is a more honest statistical presentation to include the recently diminishing death numbers. David A (talk) 17:20, 25 May 2020 (UTC)

I went ahead and dropped the last 9 items from the accumulated deaths as well.

Why is it more honest given these final columns do not reflect reality? What is dishonest about dropping last items to prevent misleading effect? Is FOHM dishonest by ending their chart with week 17 in a report from week 19? -Dan Polansky (talk) 17:24, 25 May 2020 (UTC)

Did you notice that EuroMOMO uses correction techniques at the rightmost parts of their graphs to address exactly this problem, and that they use yellow marking for regions affected by the problems even after they used the correction techniques? --Dan Polansky (talk) 17:26, 25 May 2020 (UTC)

It seems intellectually/scientifically dishonest to me to only include the statistics for the period when the death rate was increasing, but stop when it began to decrease. Omission of the full context easily gives a misleading impression. David A (talk) 18:29, 25 May 2020 (UTC)

But the criterion is not "drop when it starts to decrease" but rather "drop when the registration-delay-caused distortion becomes too big". And in fact, even when the 9 points are dropped, there is a mild downward sloping trend, contrary to "[...] only include the statistics for the period when the death rate was increasing [...]". What is your response to my FOHM and EuroMOMO points, both professional data processors? --Dan Polansky (talk) 18:54, 25 May 2020 (UTC)

There is a much more clearly displayed downward slope displayed if we include the last 9 days, and, as far as I have understood, including all available statistical data is the practice that Wikipedia uses for other articles about this pandemic regarding other countries. It is important to not hide this information from the readers for Sweden alone, especially given the widespread mainstream media criticism towards how the Swedish government has handled the crisis. Wikipedia should be neutral when it comes to presenting matter-of-fact information, period. David A (talk) 12:01, 26 May 2020 (UTC)

This isn't misleading as it's clearly explained in the note right below the chart. And the note serves as the "warning or disclaimer" you suggest. bladjur (talk) 19:01, 25 May 2020 (UTC)

Removing 9 days won't make it accurate either, as reports might be delayed for up to 30 days (further, if you want to be really strict about having the true number of people who died of Covid in Sweden, this chart won't be accurate even 30 years from now, see note). But for "real-time" data, this is the best option, and this is most likely how it's done in most (or all) Wikipedia articles about the pandemic in different locations. bladjur (talk) 19:06, 25 May 2020 (UTC)

Come on. 1) It is misleading or else it would not need a note. 2) It is misleading since people have been mislead and report confusion. It is on record. 3) Many readers do not read notes and only glance at a chart. You may think they should read the notes, but that's how things are; many don't. 4) Removing 9 days removes the worst effect. We must not remove the worst effect since some small effect still remains? Makes no sense. 5) Removing last items is consistent with what professional processor FOHM does. 6) Other charts for other countries do not suffer from this since they use different dating technique. And 7) don't show me distortions of reality accompanied by disclaimers; show me a reasonably accurate picture of reality instead, and then I don't need your disclaimers (I don't really mean your in particular; I mean general "your"). 8) If you absolutely must show the most recent data, act like professional data processor EuroMOMO and correct the data for registration delay using some statistical technique. --Dan Polansky (talk) 19:25, 25 May 2020 (UTC)

We could always wait until there is a final total, but what we should have is consistency. So either remove all but the 100% conformed figures or lets have a daily update.Slatersteven (talk) 12:07, 26 May 2020 (UTC)

I strongly support the daily update option. David A (talk) 12:12, 26 May 2020 (UTC)

That is the kind of foolish consistency that makes no sense. Should FOHM reports remove all deaths from their excess death charts that are not 100% free from registration delay? They only drop 2 weeks, and no more, and yet there are some small registration delay effects even for a month. What is the basis for the above proposal? Like, what is the publication that makes the proposal, or what is the publisher that uses that proposal? --Dan Polansky (talk) 16:47, 30 May 2020 (UTC)

Even I would say over a week is a bit long, why has there been no update?Slatersteven (talk) 18:03, 28 May 2020 (UTC)

My points above have been left without response, especially about FOHM and EuroMOMO. If you can address my points, please go ahead and do so. What qualifications do the above editors have to override publication practices of FOHM and EuroMOMO, and dump misleading raw data on the unsuspecting reader instead? What kind of encyclopedic content is that?

Let us make it a bit more vivid how the registration effect looks like. It is like taking the underlying death counts and multiplying the series by (1, 1, 1, ..., 1, 1, 1, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1). The numbers I gave are very approximate, but give the first idea. In fact, the effect goes much further back in time, but the fundamental principle is correct: the several last days are massively affected, and the more you go back in time, the less significant the registration delay effect is. What I just said can be verified by looking at the histories of the data, e.g. in Wikipedia page history, and see how the upward corrections looked like. It is right there to see for anyone who is more insterested in facts than opinions.

People really misunderstand the graph, as witnessed by another editor's statement "You don't actually know that the slope will change"; if he understood the mechanism, he would not say that. Most lay readers do not have the skill and experience to read the notes, understand what they mean, and apply them to what is visually presented to them, a graph. Graphs must not be misleading. --Dan Polansky (talk) 15:32, 30 May 2020 (UTC)

(outdent) Let me show you something else: I will take the full 2020 data from SCB[4], no dropping of last 9 days, and plot it together with 2019 data:

You can see how the final part of the all-cause daily deaths heavily drops underneath the 2019 numbers; that is the effect of registration delay. (Something seems wrong with the x-axis rendering, though.) --Dan Polansky (talk) 11:31, 31 May 2020 (UTC)

Furthermore, how many days are worst impacted by registration delay is seen in a chart from SCB[5], Tabell 8, where you can see different states of statistics in different colors. It suggests that while dropping last 5 days would remove the very worst effect in some weeks, the yellow/orange line suggests better remove 9 days and so does the blue line. For this purpose, this is an excellent kind of chart they made. --Dan Polansky (talk) 15:55, 31 May 2020 (UTC)

All-cause daily deaths from SCB

Thanks to a link posted above, we can plot daily all-cause deaths, from which daily excess death is apparent.

The source of data and the idea for the graph: Number of deaths per day 2015-2020, scb.se, Tabell 1.

I removed Feb 29 from the source data to remove a glitch in the average series: there is no value on Feb 29 for multiple years. And I cut the data after May 7 to remove registration delay effect; 8 values were removed from the 2020 series.

--Dan Polansky (talk) 10:53, 23 May 2020 (UTC)

Using the same data, we can also plot individual years rather than the average:

And here the total deaths from Jan 1 to May 7 (including) excluding Feb 29, calculated from the same source:

--Dan Polansky (talk) 07:22, 23 May 2020 (UTC)

@Dan Polansky: This is really nice work! bladjur (talk) 07:23, 24 May 2020 (UTC)

Thank you for discovering and providing the link to the data; without the link we would not have this. --Dan Polansky (talk) 07:57, 24 May 2020 (UTC)

Let us redo one of the graphs by smoothing it via exponential moving average with factor 0.8:

The above seems to have better cognitive value because of the removed high-frequency noise; on the other hand, some people may prefer to see raw data, especially when they fear that the data processing (smoothing, etc.) technique is dubious. Moving averages are a standard smoothing tool, used e.g. by Google Ngram Viewer. I wonder whether readers would be more comfortable with the plain multi-day arithmetic average (say 7-day average) rather than the exponential moving average. --Dan Polansky (talk) 08:07, 24 May 2020 (UTC)

Hi. What do you think about showing the number of deaths in dependence to the age, like I found it in de.wiki here to Switzerland. Like this:

|caption=Mortality displacement age 65 years and above (orange, up) and up to 64 years (orange, down) compared to expected (blue) |width=500}}

With best greetings from Germany. Heavytrader-Gunnar (talk) 12:29, 29 May 2020 (UTC)

---

@Heavytrader-Gunnar: Using Number of deaths per day 2015-2020, scb.se, Tabell 2, we can plot the graph you requested.

Daily death count in Sweden in 2020, split into age groups, smoothed via exponencial moving average (0.8):

We can also plot the 0-64y age group separately to see the minor changes more pronounced:

--Dan Polansky (talk) 11:16, 31 May 2020 (UTC)

Thanks Dan Polansky!

It turns out as expected: the mortality is hardly worth mentioning for people under 65. Heavytrader-Gunnar (talk) 18:12, 31 May 2020 (UTC)

China-Virus vanished?

No new cases since 2020-05-20? — Preceding unsigned comment added by 88.75.198.237 (talk) 12:08, 29 May 2020 (UTC)

What has this to do with Sweden?Slatersteven (talk) 12:10, 29 May 2020 (UTC)

The question was. Has the China-Virus vanished in Sweden, because there are no new figures for end of May. — Preceding unsigned comment added by 188.109.68.88 (talk) 09:53, 1 June 2020 (UTC)

This article is not about anything called the China-Virus.Slatersteven (talk) 18:56, 1 June 2020 (UTC)

If you have nothing productive to say, just ... — Preceding unsigned comment added by 188.109.66.208 (talk) 08:23, 2 June 2020 (UTC)

All-cause daily deaths in Stockholm county per SCB

From SCB[6], Tabell 3, we can plot the following.

All-cause daily deaths in Stockholm county per SCB:

No smoothing this time; I am too tired. --Dan Polansky (talk) 18:17, 31 May 2020 (UTC)

Updated. --Dan Polansky (talk) 10:30, 19 June 2020 (UTC)

And here are the total deaths for Stockholm county from Jan 1 to Jun 3 excluding Feb 29:

--Dan Polansky (talk) 18:39, 31 May 2020 (UTC)

That should be updated, if new data is available. — Preceding unsigned comment added by 92.209.196.140 (talk) 07:33, 8 June 2020 (UTC)

Updated. --Dan Polansky (talk) 10:30, 19 June 2020 (UTC)

Latest numbers after June 18?

Are there no more updates of daily numbers after 18. June 2020? Folkhälsomyndigheten arcgis does not show anything new. --Traut (talk) 10:04, 22 June 2020 (UTC)

Traut, the Swedish authorities have decided to stop reporting data on the weekends. The numbers have just been updated here. Deaths: 5,122; Infected: 58,932; ICU: 2,354. No data on recoveries. Do you understand Swedish? I'm hoping to find a breakdown of cases and deaths per day of the weekend before we update the tables. --Spaastm (talk) 15:08, 22 June 2020 (UTC)

Notice of Current COVID template addition

I was able to add {{Current COVID}} in top of the page to reflect this situation of ongoing coronavirus (COVID-19) pandemic per country or region, such as COVID-19 pandemic in Sweden. 2A02:2F01:6504:6200:1F6:9D55:4D4:C5D5 (talk) 19:39, 22 June 2020 (UTC)

All-cause death totals chart

The following chart I added was removed at some point:

The total deaths from Jan 1 to May 7 (including) excluding Feb 29, calculated from SCB:

I believe it has its value and is not visually redundant. It shows the areas under the curves of the other graph, and these are valuable provided the readers cannot do reliable visual integration (in the mathematical sense) themselves: I for one can't. --Dan Polansky (talk) 04:17, 25 May 2020 (UTC)

I was the one who removed that graph (and the out-of-date excess deaths table which also showed the same thing).

I believe the average reader will get a picture after reading that "from late-March and onwards, there was en excess mortality in Sweden" in the prose, accompanied by the nice graph you made. While the graph might be complicated to some, it's easy to see that there's a blue line suddenly jumping and leaving all the green lines below. And isn't this all we want to explain?

If you think more information is needed, adding a third chart isn't the way to go. Rather improve the first chart. But I think it's perfect, a line going up is the simplest chart there is. Also, I don't really see the purpose of showing the total number of all-cause deaths since 1 January for 2015-2020 in the context of the pandemic as it didn't start to affect Swedish death statistics before late March. And as the prose also says something like "there has been 3500 excess deaths in Sweden during the pandemic", I think that is pretty self-explanatory. bladjur (talk) 11:06, 25 May 2020 (UTC)

I moved the all-case deaths chart to provide context with the prose. I also put it a frame together with the source and the headline (above the chart) for context together with the prose, consistency within the article, and to avoid confusion for the readers. However I think the sentence "smoothed by applying exponential moving average with factor 0.8" needs clarification, or at least links to other articles. bladjur (talk) 12:15, 25 May 2020 (UTC)

I am not entirely happy with the edits, but they are essentially acceptable as long as the charts are still somewhere on the page. The frames around charts are not a standard publication practice in the publications I am acquainted with, and do not present an improvement, in my view, but they are tolerable.

As for "there has been 3500 excess deaths in Sweden during the pandemic", you will note that the discussed chart shows that, in Jan 1 to May 7, there were about 800 deaths more in 2020 than in 2018; that is information that the 3500 excess death figure does not and cannot capture. And the visuals are super important to show relations of quantities; text does not serve that purpose so well.

As for "exponential moving average", it can be linked to W:exponential moving average if required; on the other hand, the kind reader can use search box as well. --Dan Polansky (talk) 16:32, 25 May 2020 (UTC)

Good point. I edited the text to clarify that the published 3500 (now 4000) number concern the pandemic period March–May. — Preceding unsigned comment added by Blådjur (talk • contribs) 21:29, 25 May 2020 (UTC)

(Outdent) Multiple graphs of a similar type but with many more years can be obtained from tweet by a HaraldofW, Jun 8, twitter.com. Harald uses SCB as the source. In one of the graphs, Harald plots seasons rather than years, and therefore, the counting does on start on Jan 1 of, say, 2020, but rather on Nov 1 of 2019 for the 2019/2020 season. The deaths plotted are per 100,000 pop. Can anyone find the SCB page with data from which we could make plots for so many years? --Dan Polansky (talk) 12:32, 30 June 2020 (UTC)

I can get a long series of annual (per year) deaths from 1968 to 2019 from http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101/, by going to http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101__BE0101I/DodaFodelsearK/. But I have not found daily or weekly deaths; I speak no Swedish. For the type of chart dealt with in this section, at least weekly granularity would be required. --Dan Polansky (talk) 13:16, 30 June 2020 (UTC)

Deaths per month from 1851 to 2019 can be downloaded from http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101__BE0101G/ManadFoddDod/, as referenced in the image in tweet by a HaraldofW; an Excel spreadsheet can be downloaded. Now we would need population data per year, and the image indicates these could be obtained from SCB. --Dan Polansky (talk) 12:46, 1 July 2020 (UTC)

Meanwhile, new tweet by a HaraldofW, June 28, contains updated graphs. --Dan Polansky (talk) 14:28, 1 July 2020 (UTC)

Daily cases up and daily deaths down - how can it be

Looking at daily case and daily death charts, daily cases go up while daily deaths go down overall. That seems like a discrepancy; how can it be?

The likely answers can be seen in weekly FOHM report:

• 1) The recent increase of daily cases was caused by increased testing, as follows from Figur 1A. Andel positiva fall bland provtagna individer per vecka, which shows the ratio of daily cases to tests was around 12% in weeks 20-25 rather than increasing.
• 2) Figur 1C. Antal bekräftade fall av covid-19 per vecka i Sverige fördelat på lindriga och allvarliga fall shows weekly confirmed cases separated into mild cases (lindriga) and severe cases (allvarliga). The figure shows how the daily new covid-positive severe cases are going down while the mild cases are going up. The figure shows that the recent expansion of testing is disproportionally more catching the mild cases.

The above explains at least in part why daily deaths are not going up: mild cases are much less likely to lead to death than the severe cases, and it is the daily mild cases that account for the growth in daily cases.

Disclaimer: The above is very tentative.

--Dan Polansky (talk) 18:33, 2 July 2020 (UTC)

Also, when they are catching the cases earlier the potential severe cases are treated earlier so they don't become so severe. It is also easier to protect vulnerable groups better as mild cases and exposed are quarantined / self-quarantined by increased testing. --Zache (talk) 16:01, 3 July 2020 (UTC)