10 ms·

Coronavirus Country Comparator

109 points by j4yav 7y ago

arcticbull 7y ago
This is incredibly pretty and deserves a serious round of applause. The UI is gorgeous and very usable, the only thing that threw me is when you hover over a line, I wish it would show which country it corresponds to. With that said I still find a log-log plot much easier to interpret, and so far, the gold standard on trend clarity visualization remains Aatish Bhatia's (https://aatishb.com/covidtrends/ https://aatishb.com/covidtrends/). Maybe a way to switch to a similar log/log scale? Otherwise it's very difficult to figure out when a given country approaches the knee.
- amelius 7y ago
  Btw, the prettiest presentation I've seen so far is this one: https://www.covid.is/data https://www.covid.is/data (though only for Iceland)
  - tabs_masterrace 7y ago
    Looks nice and clean, not sure about the bouncy animations though. I mean these are charts of people dying, it doesn't fell all that appropriate. Simple ease out would be more fitting.
- blitmap 7y ago
  I agree. This is the best one I've seen and I love that I can add/remove countries and switch between linear and log scale. Definitely add log-log. (Thank you, this is great!)
- silon42 7y ago
  Seems to have some issues in Firefox for me. Anyone else?
- MiguelVieira 7y ago
  One thing this chart highlights is the unsung success stories among the US states. Some isolated and sparsely populated states are doing quite well: Alaska, Hawaii, Wyoming, Montana, Idaho, Vermont.
  - paul_f 7y ago
    I would say some large states are also doing well per-capita - Texas, California, Florida - all three less than one-third the per-capita death rate for the country as a whole. California is 3X Montana, New York is 86X Montana.
    - arcticbull 7y ago
      That could just be density or demographics, or fewer people with co-morbidities. I doubt it's anything to do with different approaches taken by states -- after all Wyoming didn't even have lockdowns until a week or two ago. Nebraska, Iowa, the Dakotas and Arkansas aren't under any form of lockdown at all.
amelius 7y ago
Some missing features (imho): - overlaying the graphs from the moment of the first case (shifting the data in time) - ICU beds available/used - overlay of measures taken (e.g. when did social distancing start) - easily showing differences in time (e.g. dragging between two dates shows the number of days between them)
- the8472 7y ago
  > - overlaying the graphs from the moment of the first case (shifting the data in time) That's available in the bottom left. You can align to either deaths or confirmed cases.
pjc50 7y ago
What went wrong with Belgium??
- DanBC 7y ago
  Belgium counts confirmed deaths in hospitals, but also counts suspected deaths outside hospitals -- so it's including care homes and nursing homes. These deaths can be about 50% of the total deaths. https://www.theguardian.com/world/2020/apr/24/is-comparing-covid-19-death-rates-across-europe-helpful-?CMP=Share_iOSApp_Other https://www.theguardian.com/world/2020/apr/24/is-comparing-c... > But such comparisons can be misleading. Unlike the UK, Italy or Spain, Belgium counts all coronavirus deaths outside hospitals in its daily statistics: deaths in care homes account for 53% of the total. Belgium’s official toll also includes people suspected of having died of coronavirus, without a confirmed diagnosis. Nearly all deaths in care homes (94%) are suspected Covid-19 cases, rather than confirmed – an approach that has led some to complain Belgium is overestimating the number of fatalities. (They probably aren't over counting. Covid-19 is devastating for nursing homes and care homes).
  - sgt101 7y ago
    Yup, serious work to put the numbers on a solid footing required in the next months and years
  - Luc 7y ago
    In some measure they have _less_ excess deaths than an average of the last years (after subtracting the counted Covid-19 cases), which would be an indication that they are over-counting.
    - Lewton 7y ago
      Are you looking at the correct demographics? We’re seeing a pretty big drop in all cause mortality outside risk groups due to lockdowns
  - raphaelj 7y ago
    The government actually issues the two numbers [1]. Current (Apr 26) toll is 7,094 if you include suspected deaths, but is 3,596 if you only count positive cases. -- [1] https://covid-19.sciensano.be/sites/default/files/Covid19/Derni%C3%A8re%20mise%20%C3%A0%20jour%20de%20la%20situation%20%C3%A9pid%C3%A9miologique.pdf https://covid-19.sciensano.be/sites/default/files/Covid19/De...
ekianjo 7y ago
Nice, but such comparisons are kind of useless since the testing and reporting varies widely by country, so there's nothing you can compare when you look down into details.
- tomgaga 7y ago
  Yeah, I really don't understand why people are still comparing these numbers like this. It's much more objective to just compare "excess number of deaths compared to last year per million". Like the New York times did: https://www.nytimes.com/interactive/2020/04/21/world/coronavirus-missing-deaths.html https://www.nytimes.com/interactive/2020/04/21/world/coronav...
  - easytiger 7y ago
    That's also not useful. Compare with a season of flu outbreaks, which last year wasn't
    - henrikschroder 7y ago
      Euromomo shows data back to 2016, so you can see how 2020 compares to a bad flu year (2017) and a "good" flu year (2019): https://euromomo.eu/graphs-and-maps/ https://euromomo.eu/graphs-and-maps/
    - kolinko 7y ago
      In some countries the excess deaths are higher that during a worst flew year in the last 5 years. But also, since the virus is growing exponentially, differences in countries is by 10-100x, while the differences in the number of tests performed seems to be closer to 2-10x.
  - ferzul 7y ago
    the answer is simple: the data is there. you can't compare vietnam and germany on that page - if you can access it. (if i have to signup to read or buy, i'm not really interested. anonymous access used to be possible with paper, and it's still possible today)
cvladan 7y ago
covidly.com ?
ghego1 7y ago
Very nice UI indeed, although the data does not seem to be up to date
catchmilk 7y ago
UI is gorgeous. Anyone happen to know if they're using a charting library?
- sk0g 7y ago
  Vue.JS/ Ruby/ D3 is what Wappalyzer picks up for this. D3 is very powerful, but there's a lot to learn too!
  - catchmilk 7y ago
    Much appreciated. Any D3 learning resources you would personally recommend?
    - sk0g 7y ago
      I never got fully through it, but Amelia Wattenberger's book [0] on it is very good! There's a fairly soft ramp up in complexity, and the later chapters are complex enough that even the previews make you excited to reach and implement them. If you just want to dip your toes in though, the official website's linked gallery [1] is a pretty cool resource. No set up or data loading required! [0] https://www.newline.co/fullstack-d3 https://www.newline.co/fullstack-d3 [1] https://observablehq.com/@d3/gallery https://observablehq.com/@d3/gallery
lettergram 7y ago
I’d recommend not using “cases”, but rather deaths. Most countries don’t have the testing capacity. In fact, if you look at places like the USA testing capacity is effectively flat. Yet 20% of the tests are coming back positive... Italy on the other hand has seen a decrease in positive rate as they are increasing testing. That’s what happens as you’re nearing Or over a peak. All countries seem to have some method of testing / adding suspect cases of people who died. As such, that’s probably the most accurate (albeit lagging) method of comparison. https://github.com/lettergram/covid19-analysis https://github.com/lettergram/covid19-analysis
- Tepix 7y ago
  Good idea. May not work well with countries reporting wrong numbers- even of deaths- on purpose, like Brazil.
- dede4metal 7y ago
  Moreover in each country the definition of recovered is very different. The most strict is in Italy were you need two swabs both positives within 24h
  - Tepix 7y ago
    You mean negative? That does't sound like the strictest to me. I have heard Prof. Drosten talk about tests further apart
    - dede4metal 7y ago
      Yeah negative sorry
- chvid 7y ago
  Deaths are also counted differently.
  - dilap 7y ago
    Death attribution can vary; I think deaths themselves -- as in, how many people have died, from any cause -- should be fairly reliable and comparable. So for the clearest, most unbiased picture of COVID impact, calculate % increase over normal death rate by country, and then compare that. (I guess. Just spitballing.)
zelphirkalt 7y ago
The one thing I find to criticize is, that if you don't want to be tracked on that website, and disable Google fonts, the whole UI for the left side panel is messed up. This should have been tested and alternatives should have been provided. It is a classical and very common UI and usability mistake to simply assume that everyone will load Google fonts. Aside from that, the visualization looks very nice.
mrtksn 7y ago
Isn't it interesting that there isn't much of interest for the most successful major European-Asian country out there? I am talking about Turkey, where death rates are even lower than Germany and the curfews are partial. There's an abundance of IC units despite the 100K cases and the spread curve is already going down and the country is in a position to offer substantial help to other major countries. There have been some CNN[0] and BBC reporting on this but with the numbers THAT good, I would have expected much higher interest. I know about the democracy and free speech situation but the country is not North Korea, Turkey is actually doing that good in managing the situation. Just switch Turkey on in this website and you'll see how much better Turkey is doing. I am beginning to suspect that the Eurocentric mindset is literally killing Europeans at this time. There's a hugely successful country that is integrated with Europe, has some of its territory in Europe, it has a population of a large European country and yet nobody seems to think about looking at Turkey for an example. As a quick note, Turkey manages to do this through aggressive contact tracing through thousands of tracking teams around the country. It was revealed that in Istanbul alone there are 1200 contact tracing teams consisting of doctors, nurses and etc. that track down every infection. It's not as cool as having Tech giants make an app but it works. According to the health ministry, these teams use purpose built apps to coordinate and the entire health system is well integrated and they monitor every single patients status - be it at home or in the hospital and have very good picture of the situation. Turkey has many problems but the health system is not one of them and maybe it's worth taking a look at. [0]: https://edition.cnn.com/videos/world/2020/04/21/inside-turkey-icu-coronavirus-arwa-damon-pkg-cnni-vpx.cnn/video/playlists/coronavirus-intl/ https://edition.cnn.com/videos/world/2020/04/21/inside-turke...
- renox 7y ago
  How about Czech Republic? I think that's an even better example.
  - mrtksn 7y ago
    Sure, why not, they are doing great too. However, Turkey has a population of 82 million(and more than 5 million refugees) which is in par with the UK, France, Germany. You don't have the excuse of "It's a small country" when it comes to Turkey. Thousands of citizens were repatriated, large PPE deliveries to the UK, Spain, Italy and so on, the death rate is at 2.5, the hospital bed occupancy rate was at about %50, ICU occupancy rate at about 60% last time I checked. The country is doing something right and is doing it right at scale.
MiguelVieira 7y ago
I prefer this one. Use "New Deaths, 1 Wk. Avg." Plots US states as well as countries. http://91-divoc.com/pages/covid-visualization/ http://91-divoc.com/pages/covid-visualization/
Fiahil 7y ago
Your data source is compromised. France's confirmed case count is 124k not 161k. You're off by 40k. Source: https://dashboard.covid19.data.gouv.fr/ https://dashboard.covid19.data.gouv.fr/
doodlebugging 7y ago
Sweden has an interesting pattern with a period of about one week. It appears that they have some activity that occurs weekly that results in new cases. There is a persistent 2-3 day plateau before the next caseload increase.
- detaro 7y ago
  You can see that for some other places too, e.g. Germany. I suspect it's just the weekend and reporting chains being slower then (it certainly plays a role in German data). The data then appears over the next week, but if you just daily calculate the difference between the numbers announced on that day for your graphs, you don't see that it is being backfilled in the new week, even if the country releases that detail.
stillbourne 7y ago
Can this also be weighted by number of tests per million?

109 points by j4yav 7y ago

This is incredibly pretty and deserves a serious round of applause. The UI is gorgeous and very usable, the only thing that threw me is when you hover over a line, I wish it would show which country it corresponds to. With that said I still find a log-log plot much easier to interpret, and so far, the gold standard on trend clarity visualization remains Aatish Bhatia's (https://aatishb.com/covidtrends/ https://aatishb.com/covidtrends/). Maybe a way to switch to a similar log/log scale? Otherwise it's very difficult to figure out when a given country approaches the knee.

amelius 7y ago
Btw, the prettiest presentation I've seen so far is this one: https://www.covid.is/data https://www.covid.is/data (though only for Iceland)
- tabs_masterrace 7y ago
  Looks nice and clean, not sure about the bouncy animations though. I mean these are charts of people dying, it doesn't fell all that appropriate. Simple ease out would be more fitting.
blitmap 7y ago
I agree. This is the best one I've seen and I love that I can add/remove countries and switch between linear and log scale. Definitely add log-log. (Thank you, this is great!)
silon42 7y ago
Seems to have some issues in Firefox for me. Anyone else?
MiguelVieira 7y ago
One thing this chart highlights is the unsung success stories among the US states. Some isolated and sparsely populated states are doing quite well: Alaska, Hawaii, Wyoming, Montana, Idaho, Vermont.
- paul_f 7y ago
  I would say some large states are also doing well per-capita - Texas, California, Florida - all three less than one-third the per-capita death rate for the country as a whole. California is 3X Montana, New York is 86X Montana.
  - arcticbull 7y ago
    That could just be density or demographics, or fewer people with co-morbidities. I doubt it's anything to do with different approaches taken by states -- after all Wyoming didn't even have lockdowns until a week or two ago. Nebraska, Iowa, the Dakotas and Arkansas aren't under any form of lockdown at all.

amelius 7y ago

Some missing features (imho): - overlaying the graphs from the moment of the first case (shifting the data in time) - ICU beds available/used - overlay of measures taken (e.g. when did social distancing start) - easily showing differences in time (e.g. dragging between two dates shows the number of days between them)

the8472 7y ago
> - overlaying the graphs from the moment of the first case (shifting the data in time) That's available in the bottom left. You can align to either deaths or confirmed cases.

pjc50 7y ago

What went wrong with Belgium??

DanBC 7y ago
Belgium counts confirmed deaths in hospitals, but also counts suspected deaths outside hospitals -- so it's including care homes and nursing homes. These deaths can be about 50% of the total deaths. https://www.theguardian.com/world/2020/apr/24/is-comparing-covid-19-death-rates-across-europe-helpful-?CMP=Share_iOSApp_Other https://www.theguardian.com/world/2020/apr/24/is-comparing-c... > But such comparisons can be misleading. Unlike the UK, Italy or Spain, Belgium counts all coronavirus deaths outside hospitals in its daily statistics: deaths in care homes account for 53% of the total. Belgium’s official toll also includes people suspected of having died of coronavirus, without a confirmed diagnosis. Nearly all deaths in care homes (94%) are suspected Covid-19 cases, rather than confirmed – an approach that has led some to complain Belgium is overestimating the number of fatalities. (They probably aren't over counting. Covid-19 is devastating for nursing homes and care homes).
- sgt101 7y ago
  Yup, serious work to put the numbers on a solid footing required in the next months and years
- Luc 7y ago
  In some measure they have _less_ excess deaths than an average of the last years (after subtracting the counted Covid-19 cases), which would be an indication that they are over-counting.
  - Lewton 7y ago
    Are you looking at the correct demographics? We’re seeing a pretty big drop in all cause mortality outside risk groups due to lockdowns
- raphaelj 7y ago
  The government actually issues the two numbers [1]. Current (Apr 26) toll is 7,094 if you include suspected deaths, but is 3,596 if you only count positive cases. -- [1] https://covid-19.sciensano.be/sites/default/files/Covid19/Derni%C3%A8re%20mise%20%C3%A0%20jour%20de%20la%20situation%20%C3%A9pid%C3%A9miologique.pdf https://covid-19.sciensano.be/sites/default/files/Covid19/De...

ekianjo 7y ago

Nice, but such comparisons are kind of useless since the testing and reporting varies widely by country, so there's nothing you can compare when you look down into details.

tomgaga 7y ago
Yeah, I really don't understand why people are still comparing these numbers like this. It's much more objective to just compare "excess number of deaths compared to last year per million". Like the New York times did: https://www.nytimes.com/interactive/2020/04/21/world/coronavirus-missing-deaths.html https://www.nytimes.com/interactive/2020/04/21/world/coronav...
- easytiger 7y ago
  That's also not useful. Compare with a season of flu outbreaks, which last year wasn't
  - henrikschroder 7y ago
    Euromomo shows data back to 2016, so you can see how 2020 compares to a bad flu year (2017) and a "good" flu year (2019): https://euromomo.eu/graphs-and-maps/ https://euromomo.eu/graphs-and-maps/
  - kolinko 7y ago
    In some countries the excess deaths are higher that during a worst flew year in the last 5 years. But also, since the virus is growing exponentially, differences in countries is by 10-100x, while the differences in the number of tests performed seems to be closer to 2-10x.
- ferzul 7y ago
  the answer is simple: the data is there. you can't compare vietnam and germany on that page - if you can access it. (if i have to signup to read or buy, i'm not really interested. anonymous access used to be possible with paper, and it's still possible today)

cvladan 7y ago

covidly.com ?

ghego1 7y ago

Very nice UI indeed, although the data does not seem to be up to date

catchmilk 7y ago

UI is gorgeous. Anyone happen to know if they're using a charting library?

sk0g 7y ago
Vue.JS/ Ruby/ D3 is what Wappalyzer picks up for this. D3 is very powerful, but there's a lot to learn too!
- catchmilk 7y ago
  Much appreciated. Any D3 learning resources you would personally recommend?
  - sk0g 7y ago
    I never got fully through it, but Amelia Wattenberger's book [0] on it is very good! There's a fairly soft ramp up in complexity, and the later chapters are complex enough that even the previews make you excited to reach and implement them. If you just want to dip your toes in though, the official website's linked gallery [1] is a pretty cool resource. No set up or data loading required! [0] https://www.newline.co/fullstack-d3 https://www.newline.co/fullstack-d3 [1] https://observablehq.com/@d3/gallery https://observablehq.com/@d3/gallery

lettergram 7y ago

I’d recommend not using “cases”, but rather deaths. Most countries don’t have the testing capacity. In fact, if you look at places like the USA testing capacity is effectively flat. Yet 20% of the tests are coming back positive... Italy on the other hand has seen a decrease in positive rate as they are increasing testing. That’s what happens as you’re nearing Or over a peak. All countries seem to have some method of testing / adding suspect cases of people who died. As such, that’s probably the most accurate (albeit lagging) method of comparison. https://github.com/lettergram/covid19-analysis https://github.com/lettergram/covid19-analysis

Tepix 7y ago
Good idea. May not work well with countries reporting wrong numbers- even of deaths- on purpose, like Brazil.
dede4metal 7y ago
Moreover in each country the definition of recovered is very different. The most strict is in Italy were you need two swabs both positives within 24h
- Tepix 7y ago
  You mean negative? That does't sound like the strictest to me. I have heard Prof. Drosten talk about tests further apart
  - dede4metal 7y ago
    Yeah negative sorry
chvid 7y ago
Deaths are also counted differently.
- dilap 7y ago
  Death attribution can vary; I think deaths themselves -- as in, how many people have died, from any cause -- should be fairly reliable and comparable. So for the clearest, most unbiased picture of COVID impact, calculate % increase over normal death rate by country, and then compare that. (I guess. Just spitballing.)

zelphirkalt 7y ago

The one thing I find to criticize is, that if you don't want to be tracked on that website, and disable Google fonts, the whole UI for the left side panel is messed up. This should have been tested and alternatives should have been provided. It is a classical and very common UI and usability mistake to simply assume that everyone will load Google fonts. Aside from that, the visualization looks very nice.

mrtksn 7y ago

Isn't it interesting that there isn't much of interest for the most successful major European-Asian country out there? I am talking about Turkey, where death rates are even lower than Germany and the curfews are partial. There's an abundance of IC units despite the 100K cases and the spread curve is already going down and the country is in a position to offer substantial help to other major countries. There have been some CNN[0] and BBC reporting on this but with the numbers THAT good, I would have expected much higher interest. I know about the democracy and free speech situation but the country is not North Korea, Turkey is actually doing that good in managing the situation. Just switch Turkey on in this website and you'll see how much better Turkey is doing. I am beginning to suspect that the Eurocentric mindset is literally killing Europeans at this time. There's a hugely successful country that is integrated with Europe, has some of its territory in Europe, it has a population of a large European country and yet nobody seems to think about looking at Turkey for an example. As a quick note, Turkey manages to do this through aggressive contact tracing through thousands of tracking teams around the country. It was revealed that in Istanbul alone there are 1200 contact tracing teams consisting of doctors, nurses and etc. that track down every infection. It's not as cool as having Tech giants make an app but it works. According to the health ministry, these teams use purpose built apps to coordinate and the entire health system is well integrated and they monitor every single patients status - be it at home or in the hospital and have very good picture of the situation. Turkey has many problems but the health system is not one of them and maybe it's worth taking a look at. [0]: https://edition.cnn.com/videos/world/2020/04/21/inside-turkey-icu-coronavirus-arwa-damon-pkg-cnni-vpx.cnn/video/playlists/coronavirus-intl/ https://edition.cnn.com/videos/world/2020/04/21/inside-turke...

renox 7y ago
How about Czech Republic? I think that's an even better example.
- mrtksn 7y ago
  Sure, why not, they are doing great too. However, Turkey has a population of 82 million(and more than 5 million refugees) which is in par with the UK, France, Germany. You don't have the excuse of "It's a small country" when it comes to Turkey. Thousands of citizens were repatriated, large PPE deliveries to the UK, Spain, Italy and so on, the death rate is at 2.5, the hospital bed occupancy rate was at about %50, ICU occupancy rate at about 60% last time I checked. The country is doing something right and is doing it right at scale.

MiguelVieira 7y ago

I prefer this one. Use "New Deaths, 1 Wk. Avg." Plots US states as well as countries. http://91-divoc.com/pages/covid-visualization/ http://91-divoc.com/pages/covid-visualization/

Fiahil 7y ago

Your data source is compromised. France's confirmed case count is 124k not 161k. You're off by 40k. Source: https://dashboard.covid19.data.gouv.fr/ https://dashboard.covid19.data.gouv.fr/

doodlebugging 7y ago

Sweden has an interesting pattern with a period of about one week. It appears that they have some activity that occurs weekly that results in new cases. There is a persistent 2-3 day plateau before the next caseload increase.

detaro 7y ago
You can see that for some other places too, e.g. Germany. I suspect it's just the weekend and reporting chains being slower then (it certainly plays a role in German data). The data then appears over the next week, but if you just daily calculate the difference between the numbers announced on that day for your graphs, you don't see that it is being backfilled in the new week, even if the country releases that detail.

stillbourne 7y ago

Can this also be weighted by number of tests per million?