Coronavirus Country Comparator | Hacker News Reader

Coronavirus Country Comparator

boogheta.github.io

109 points_teyd102 comments

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

Good idea. May not work well with countries reporting wrong numbers- even of deaths- on purpose, like Brazil.

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

You mean negative? That does't sound like the strictest to me. I have heard Prof. Drosten talk about tests further apart

Yeah negative sorry

Deaths are also counted differently.

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.)

Deaths even within the US have wildly different criteria for being counted:

https://www.washingtonpost.com/investigations/which-deaths-c...

The suggestion you are responding to is not suggesting counting covid deaths.

It's suggesting we track all deaths/ Then we look at the increase in average deaths. Based on the trends we should be able to estimate how many additional (likely covid) deaths we have incurred during the crisis

The obvious problem there is the safer at home laws and lack of socializing would have an impact on your average deaths in the opposite direction. So comparing average deaths in the past to now are apples and oranges.

Yes, the unadjusted number you get will be a lower-bound. You could then try to get closer to the real number by noticing, e.g., how much car crashes were reduced and adjusting for that.

Exactly the increase in "all cause mortality" is the most honest we have at this point. But they say that number often gets adjusted in the weeks after (not sure why that is happening or acceptable).

From all my calculations i find one thing: we've been lied to a lot in the past few weeks.

People normally don't care about realtime death stats, so in many places there is a delay on reporting because nobody ever cared about doing it any other way.

Excess deaths may work, considering we are in a relatively calm time when it comes to armed conflicts and such.

Yes they are. But the "deathcount-by-COVID" will be more accurate that "positive cases". The metric I am using and will be releasing once this thing blows over is "Deaths by COVID per million of population" (as measured in the latest census)(for each country)

I'll have it like (imagine spreadsheet columns): -Country name -County population (latest census) -Deaths by COVID -Deaths per 1 million (e.g. 100 deaths / 10m pop = 10) -Date of first death -Source of population data -Source of COVID death count

I am curious to see how countries that are "too big to fail" will perform (UK, USA) in comparison to other (Harari mentioned Greece on an interview).

I was checking and so far Greece has a ~13 Vs the UK which has ~320 ("ratios").

So congrats Boris Johnson!! You just killed x24 people!! (Great choice UK voters - keep up the wise choices!!!)

This isn't about BJ specifically, this is more to show how each country treated this situation and respecting its people (these are not just numbers, these are dead people, parents, grandparents, and some very few children). It would also be useful to then process these numbers and categorise the results (I am curious to see the ratios in the Triad USA, UK, Brazil Vs other similar sized countries with different-style leaders).

I also do not expect the truth from some countries (e.g. Russia, China, any Arab peninsula countries, NK etc.) or some African countries that are torn by civil wars or other conflicts. But I believe there will be a solid baseline and the outliers (below) will be easily pointed out as ineffective measured.

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.

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/coronav...

That's also not useful. Compare with a season of flu outbreaks, which last year wasn't

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/

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.

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)

I think it's only marginally useful though; for example you can't rationally make personal or policy decisions using this because the distortion Inthe data that the gp flagged is so large, and also we are so early in this event. In 18 months we will see!

Open it in a private browsing window and anonymity still is possible

Private browsing doesn't provide any form of anonymity: https://panopticlick.eff.org/

Heck, it is tricky to remain anonymous even over the Tor Browser.

https://en.m.wikipedia.org/wiki/Tor_(anonymity_network)#Weak...

It does in the sense that the comment which I replied to meant it: access without being logged in to an account.

Beware of comparing deaths to last year though, I'd rather take a 5 year average

At least this one doesn't include the case fatality rate which is the biggest bullshit number of them all. If nothing else, this pandemic has shown that bad data is so much worse than no data.

Edited to add: The daily statistic by population ratio is very spiky, and gets spikier the smaller the country is, which makes it hilariously useless. A rolling average would improve that a bit, but remember that some countries might be off with a factor of 2x or 3x when it comes to reported deaths, which is an insane level of uncertainty.

It does allow comparing trends though. Yes, the actual numbers are not reliable, but whether numbers in a country are growing exponentially, growing linearly or stabilizing is comparable.

You can also make some indications if a country is doing unusually well or bad: If the ratio of cases/population or deaths/population were significantly different from most other countries, this would warrant a closer look.

Nope. If the data is bad, looking more closely at the same data won’t help anything. In Covid 19 there are so many different variables in the way the data is reported that it’s completely pointless to draw any conclusions across counties, including where to focus further attention on.

Some trends are clearly visible nonetheless, like when numbers are going down (unless someone started to deliberately exclude the symptomatic from testing. And observed "high water marks" serve as a lower bound for worst case outcomes, this is where county level data would be far more interesting than country level data (it's no coincidence that tiny states like San Marino, Andorra or the Vatican dwarf all other normalized country level stats).

The trends you see may not means anything. You’re looking at garbage data.

Obligatory xkcd

https://xkcd.com/2295/

Are you telling me that e.g. right now, new cases in China could actually be doubling each day while the US might not have any - and we wouldn't be seeing it because of differences in reporting?

Also, how did we get from reporting differences to "garbage data"? What kind of data would you classify as non-garbage?

> If the data is bad, looking more closely at the same data won’t help anything.

I'm not arguing for looking more closely - on the contrary, I'm saying that the large-scale trends are meaningful - say, how growth rate or mortality/population develops over time.

This. Comparisons hospital admissions, icu admissions, and all-cause excess deaths make sense. Comparing the number of tested or even the number of “known fatalities” works within a country as a trend but not across countries.

I think only excess mortality makes sense. ICU admissions depends on ICU availabilty and has a ceiling when the system nears saturation so it's not very comparable, a country with very few ICU will have low numbers.

I would love to see testing capacity stats per country.

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/). 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.

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!)

Btw, the prettiest presentation I've seen so far is this one: https://www.covid.is/data

(though only for Iceland)

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.

Seems to have some issues in Firefox for me. Anyone else?

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.

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.

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.

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-turke...

How about Czech Republic? I think that's an even better example.

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.

That all depends on if you can trust the numbers coming out of Turkey.

I know but Turkey is not North Korea or Iran, there are many western journalists based in Turkey and they are indeed reporting that the situation is fine and the hospitals have plenty of capacity to go. There are some concerns on reporting methodology but these concerns are not different than those in the UK. Nothing that will change the picture substentially.

Turkey is essentially a dictatorship, if you want to believe in their numbers feel free to do so but I'm very skeptical about the accuracy of their reports.

You are right to be sceptical, it is indeed a dictatorship but it's not the kind of dictatorship that makes doctors disappear. There is no indication about any major cover-ups, the country is not insulated from the west, has visa-free travel with many Western countries, is part of many western institutions and has many wester journalists based in Turkey.

"I don't like Erdogan and his policies" is not good enough to dismiss Turkey as a closed society, the country is NOT like DPRK or Iran. Turkey is financially and institutionally integrated with Europe and the USA.

This(dismissing the country solely on its current political situation or lack of awareness beyond the stereotype and prejudice) is exactly the mentality that I am talking about and I suspect that it is killing many people in the west.

The whole thing is about to be so political that everything is buried in noise.

China's numbers cannot be trusted. Turkey's numbers cannot be trusted. WHO cannot be trusted.

Even Germany cannot be trusted; I saw a statement from an Italien doctor saying that reason for the difference between Germany and Italy was that Germany was systematically undercounting deaths (for political reasons).

I agree it's unrealistic to propose they're hiding bodies. But the death count could be suppressed by merely implementing unreasonable standards for what's reported as a coronavirus death. Articles I've read suggest that Turkey isn't doing particularly well when measured by excess deaths over the historical baseline.

(I agree with the obvious followup, that we should check to see if other countries we believe are doing well look good in the same metric.)

I suspect that too, the death rates in Istanbul are higher than usual and people are asking questions.

That said, it's no different than what's happening some other western countries. There are reports about the UK numbers being higher too, essentially the idea is that some deaths are not reported as COVID-19 deaths if it happens out of hospital or pre-testing.

People are not dying en masse. The mortality is increased and maybe the real numbers are %50 more than the official numbers but even then it's much better than most places.

No, Turkey's official mortality numbers are worse than most places. It's only in comparison to Western numbers that they look good. I checked Mexico (as a good country the US could more easily learn from) and they have over 3x fewer deaths per 100,000.

I don't say that to knock on Turkey, just to explain the perspective. People aren't ignoring Turkey because they don't trust Turkey specifically; it's because they don't trust any of these numbers. The idea that there's some core competency of pandemic response possessed by almost every country in the world but absent in most of the West seems... implausible.

I am a European citizen currently in Turkey, I have perspective on the UK, on Germany and some other EU countries as I lived in those.

I don't know about Mexico, I cannot comment about them but if they are doing even better than Turkey, good for them but I cannot comment on that, I don't know the reality on the ground.

You don't have to hide the bodies, simply don't count them. That's what France was doing with people dying in retirement houses up until a few weeks ago.

The recent freedom of press report [1] certainly does not help in building trust, no matter how many western institutions have a presence there.

[1] https://rsf.org/en/ranking

Doctors are all over Instagram, Twitter and so on. There would have been doctors telling things. The medical professionals in Turkey are well paid, they have very strong jobs security, they have influential societies that are quite hostile to the government.

They are overworked and are under the risk of antisocial patients but Silencing a doctor in Turkey is not as easy as silencing a journo. They are well educated, well paid and well organized. It is very unlikely that a doctor would participate in a cover-up in fear of politicians.

They put journalists and opponents in jail, what do you think would happen to them ? Who in his right mind would defend such a country ? I can't even begin to understand the logic behind that, sure it's not the lowest of the low like DPRK but comparing it to EU or the US ... I'm speechless.

https://en.wikipedia.org/wiki/List_of_arrested_journalists_i...

https://www.hrw.org/news/2018/03/27/turkey-crackdown-social-...

https://ahvalnews.com/germany-turkey/turkey-detains-german-n...

https://cpj.org/reports/2019/12/journalists-jailed-china-tur...

Doctors are not a journalist. Journalists in Turkey are low paid with limited employment options and with %90 of the media being held by Erdogan affiliated businessmen it is easy to purge them.

That's not the situation with the doctors. The doctors in Turkey are picked from Crème de la crème among the students, educated in some of the most prestigious institutions and paid very well. They have money, they have options, they have connections.

Doctors are not journalists, these lists are about journalists and it is correct that Erdogan destroyed the press freedom in Turkey but Doctors don't work for the TV channels and the newspaper and they have access to social media.

I don't think the salary effects who goes in jail or not. The only thing it could change is that people making a good living would be less likely to take risks and lose everything than people already at the bottom.

There isn't much to debate, Turkey is universally seen as a dictatorship and has a track record of doing very shady and questionable things (which are not even close to being acceptable in the EU or US). Feel free to believe whatever you want but you haven't provided any resources besides your own idea of what Turkey is and this view isn't backed by anyone besides Erdoğan supporters.

You don't go to jail just like that in Turkey. This is not DPRK.

They need to accuse you of something and have a trail. All can be unfair and it is often unfair but you don't disappear, you can fight in the process and it catches a lot of public attention. Being a doctor means that the public loves you, trusts you, you have the money and the connections to fight.

If there was a cover-up, I am sure that doctors would have been talking because there are many doctors openly critical to the government and they even have a society for that. This is not Iran, China, DPRK or whatever country where you can just disappear.

https://www.foxnews.com/world/turkey-sentences-doctors-to-pr...

You should probably just stop digging, information is available on internet for free and every single one of your claims can be refuted in a few seconds. They do send people to jail over the most ridiculous things, even doctors.

An expectation of the rule doesn't make it a rule. If you dig deeper I am sure you will stumble upon to doctors who went to jail for other political activities too. You don't understand the nuances here, you don't get public support if you say something against the nationalistic stance, if you dig you will find that doctors involved in the kurdish politics can also be jailed.

Whatever. Believe whatever you want, go ahead believe that I died from COVID-19 but the government is covering up by writing replies to you from my account.

Things are fine in Turkey(regarding the COVID-19 situation), Erdogan being a dictator doesn't change a thing about that, journalist in jail don't automatically count as COVID-19 deaths.

> This(dismissing the country solely on its current political situation or lack of awareness beyond the stereotype and prejudice) is exactly the mentality that I am talking about and I suspect that it is killing many people in the west.

Why would they tell the truth and make their country look bad if they can avoid it ? You won't make me believe Turkey has a better health care system, testing facilities and post mortem checks than EU countries. Do they send people in remote rural areas and check the causes of death from people who died in the last 3 months ? Do they count people with comorbidities ? Do they count people dying outside of hospitals ?

> stereotype and prejudice

It literally is a military dictatorship, like, what ?! do we have to accept everything no matter how fucked up they are these days ? The world isn't rainbow and sunshine, you won't make me trust Turkey on these kind of topics. I simply don't buy it, there is such a long trail of lies, deceptions, propaganda, &c. Look at what Trump gets away with while being under spotlight worldwide, now imagine what a guy like Erdoğan gets away with ...

It's alright, I made peace with people being wrong on the Internet.

I would suggest you do some research, like researching if it's a military dictatorship or not. Check out what happened to the military in 2016 when they attempted a coup. The results may surprise you.

Anyway, the Turkish health system is quite similar to NHS and the doctors are severely overworked so I wouldn't claim that Turkey has a better health system but I would claim that it is handling the COVID-19 response better than the most.

Turkey is not a closed society, there are many ways to verify these claims.

Actually, this deeply flawed understanding of the situation in Turkey is part of the problem anyway. If you can overcome this, you can learn a thing or two from the way Turkey is handling the situation. Take it or leave it.

> The results may surprise you.

The first link that pops when typing "turkey coup aftermath" in google starts with "The aftermath of the failed Turkey coup: Torture, beatings and rape". I'm not sure where I'm supposed to be surprised.

https://www.amnesty.org.uk/aftermath-failed-turkey-coup-tort...

Yes, these things happened to the military personnel but that's not a military dictatorship. In the military dictatorship, the coup succeeds and the military personnel rape and torture people.

In Turkey's case, the military attempted a coup, failed and the civilians did these things to the military.

Pro tip: The guys with the uniform are the military ones, they got beaten on the streets when attempted to establish a military dictatorship.

Why do you thing the coup failed ? The military isn't a single entity that decided to make a coup, otherwise who would have stopped them ? The great Erdoğan himself ?

Wikipedia even state their numbers: "97% of Turkish Armed Forces ranking officers and 66% of the military's top leadership" sided with Erdoğan. The coup was attempted by a little over 10k people.

I frankly don't think you researched any of what you're talking about.

And this makes Turkey a military dictatorship? In the UK 100% of the forces are loyal to the government, that must be one awful military dictatorship then. sight.

I don't claim that Turkey is a functioning democracy but this doesn't make it a military dictatorship. turkey has many problems, that doesn't make it wrong or bad in every other aspect too.

As someone with German perspective, I'm sure you have a point about Europe having a blind spot there.

Still, it looks like Turkey also are earlier in the curve, which would explain at least some of the better fatality numbers.

Why do you think Turkey is early in the curve, the new infections are falling and Turkey is doing more tests per capita than France or the UK.

That's probably an example of the blind eye.

I'm looking at data in the JHU dashboard. I see the first cases in the Daily Cases chart in late February for Germany, and around March 17 for Turkey. It looks like an about three-week time lag.

Have spoken to a doctor you know in Turkey about the situation (I'm assuming you are Turkish)?

The ones on the ground don't have such a rosy view.

I have relatives in the business(doctors and nurses) who work in some of the largest hospitals that deal with the pandemic.

Obviously it's not a rosy situation when a hunders people are dying every day but I am yet to see a credible report of a cover-up.

My doctor relative told me that she had to buy some of her PPE and that they are getting paranoid about catching the bug but she doesn't think that the situation is worse than the reported. She complained about some management issues like asking people to get a test but not providing them with a safe place to queue and so on but no, people are not dying en mass here in Istanbul.

sadly, many people refuse to set their prejudice aside and admit when someone(or country) is doing the right thing

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.

I prefer this one. Use "New Deaths, 1 Wk. Avg." Plots US states as well as countries.

http://91-divoc.com/pages/covid-visualization/

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/

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)

> - 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.

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.

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.

What went wrong with Belgium??

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-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).

Yup, serious work to put the numbers on a solid footing required in the next months and years

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.

Are you looking at the correct demographics? We’re seeing a pretty big drop in all cause mortality outside risk groups due to lockdowns

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/De...

Actually in Italy (as well as France) deaths outside hospitals are counted.

Yes, everywhere is also counting the deaths outside hospital. But Belgium is reporting those numbers every day, while Italy and France and the UK and some other countries are not.

Understand but the quote from the article is wrong.

Edit: actually it seems in Italy is also reported every day.

This is correct. Most other countries are undercounting Covid-19 deaths significantly because they are only counting confirmed cases.

The problem becomes obvious when the total death numbers for the last few months are compared to the historical average.

See for example this NYT analysis, which shows that Belgium's numbers line up with the actual increase in death rate, not the more-or-less made up numbers from most other countries: https://www.nytimes.com./interactive/2020/04/21/world/corona...

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

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

Vue.JS/ Ruby/ D3 is what Wappalyzer picks up for this.

D3 is very powerful, but there's a lot to learn too!

Much appreciated. Any D3 learning resources you would personally recommend?

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 [1] https://observablehq.com/@d3/gallery

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

covidly.com ?

Reply on news.ycombinator.com