14 ms·

No country except Iceland or South Korea knows how many are infected, everyone else's numbers are horribly wrong. But death numbers and hospitalization numbers

No country except Iceland or South Korea knows how many are infected, everyone else's numbers are horribly wrong. But death numbers and hospitalization numbers don't lie. Sure, different countries are counting deaths differently, making it hard to compare them to each other, but the trajectories don't lie. And if you look at growth trajectories, no country is experiencing uncontrolled exponential growth at this point: https://ourworldindata.org/grapher/covid-daily-deaths-trajectory-per-million?country=BEL+DNK+FRA+ISL+ITA+NOR+SWE+GBR+USA https://ourworldindata.org/grapher/covid-daily-deaths-trajec... If deaths are on a downward trajectory, it means that infections have been on a downward trajectory for weeks. And that's how we can know that it's controlled, without knowing how many have it or have had it, while being asymptomatic.

wbl 7y ago
The timeline matches aggressive control measures that reduced infection.
tapland 7y ago
> but the trajectories don't lie Except for when they do. Easter just passed and the the trajectories are plotted up to and including today, when Sweden for example is adding deaths to dates two weeks back every day, and the last 5-7 days _ALWAYS_ looks like a downward trajectory. The last little spike there from before easter has grown by 35% since after easter and is still growing daily. You can't draw conclusions based on the trajectory of the last few days when an absolute majority of cases are added days, and up to weeks after.
- henrikschroder 7y ago
  Sure, but in that chart Italy for example has had a declining graph for over two weeks, which means you can be reasonably sure it's not growing exponentially there. And in Sweden's case, if you also look at the number of ICU admissions, they've been flat for weeks, and those numbers have very little lag. I'm not saying everything is fine, I'm not saying we're over the peak, but I am saying that it's not the case that any country is experiencing uncontrolled exponential growth and is going to turn into a zombie wasteland in a couple of days.
  - matthewdgreen 7y ago
    You have to ask what phenomenon causes an exponential increase in deaths/day and does not increase ICU utilization. The answer I’ve seen seems to be “yes, the disease is killing a steadily growing number of people but they’re older and aren’t going to the ICU before they die.” This is where Italy was at one point; they just got there much later.
    - riffraff 7y ago
      As someone pointed out to me, the euromomo website is useful for this, it tracks the total mortality rate across Europe and compares it with the historical data. By ignoring the hospitalized/homestuck difference one can have a clearer idea of what is happening. Sadly I don't know of a way to get the data for the whole world.
      - innomatics 7y ago
        Euromomo data is really interesting to me because of these two points: 1) the spike in death rate caused by Covid-19 this year is of the same magnitude as the normal flu season spike in 2016 2) this year's flue season spike was low compared to recent years. Maybe due to a warmer summer? One interpretation could be that a large number of vulnerable people escaped death from regular ILIs this European season, only to sadly succome to this pandemic.
        acqq 7y ago
        But all the countries did introduce various measures to slow down the spread. Otherwise nobody could claim that the numbers remained compatable.
        innomatics 7y ago
        Yes, which is wonderful. I am wondering what caused the large 2016 spike. Another interesting thing is the effect any behaviour changes are going to have on regular flu numbers. It's already looking to have a big effect in then southern hemisphere.
robbiep 7y ago
Austria did a relatively large scale antibody sample of its pop (0). Seems to discredit the notion that infections are vastly different than modelling. (0) https://amp.theguardian.com/world/2020/apr/10/less-than-1-of-austria-infected-with-coronavirus-new-study-shows https://amp.theguardian.com/world/2020/apr/10/less-than-1-of...
- SpicyLemonZest 7y ago
  This article says it was not an antibody sample.
  - robbiep 7y ago
    my bad
- djrogers 7y ago
  FTA: The study, conducted from 1-6 April, tested 1,554 people. It did not involve antibody tests, which can tell whether a person was previously infected and is therefore probably immune. Future studies should involve antibody tests, the government has said, to inform policy as the country starts to reopen shops from 14 April.
zizee 7y ago
> No country except Iceland or South Korea knows how many are infected, everyone else's numbers are horribly wrong. Is South Korea still consideered an outlier WRT testing regimes? Not to throw shade on their great efforts in leading the way, but many other countries now have extensive testing.
- ImaCake 7y ago
  Australia has highest per-capita testing in the world. New Zealand is doing well. A number of other developed nations have been testing people since the start, but Australia and NZ have some added benefits of distance and relatively small populations that are helping them counter the pandemic better than most.
  - hombre_fatal 7y ago
    I don't see those facts in: https://en.wikipedia.org/wiki/COVID-19_testing#Virus_testing_by_country https://en.wikipedia.org/wiki/COVID-19_testing#Virus_testing... https://ourworldindata.org/covid-testing#testing https://ourworldindata.org/covid-testing#testing I'd be wary of handing out superlatives just yet in any case.
    - robryan 7y ago
      In Australia we are testing people with symptoms (recently expanded in most places to hae no additional criteria). Some counties with a bigger outbreak/ lesser lockdowns are going to have more people with symptoms to test. 1.7% positive rate is one of the lowest in the world and combined with high testing suggests we are detecting most symptomatic cases.
      - mkl 7y ago
        NZ is mostly just testing people with symptoms too. That's pretty worrying to me given data like this article's. I think we should also be testing big random samples from the population. We've started testing hundreds of apparently-healthy volunteers, which is a good start: https://www.stuff.co.nz/national/health/coronavirus/121074955/coronavirus-wider-covid19-testing-starts-in-christchurch-community https://www.stuff.co.nz/national/health/coronavirus/12107495...
D-Coder 7y ago
> death numbers and hospitalization numbers don't lie. Death numbers are close but still have some uncertainty. New York for example is reporting much higher than usual numbers of people found dead at home. Not clear if they have been tested, if they are victims, or if maybe they had unrelated medical problems but couldn't/wouldn't go to a hospital under these circumstances.
- Spooky23 7y ago
  Key thing in New York is that the protocol is about fever. You won’t be allowed in a doctors office unless confirmed negative, and you usually need a fever to get tested. There are many scenarios where people in need of care won’t get it. It’s a vicious cycle and it’s fueling the ignorant rhetoric downplaying the broader issues.
- thomquaid 7y ago
  More people are found dead at home because during a lockdown-at-home you're simply more likely to be at home at any given point in time overall, obviously.
- iso947 7y ago
  UK deaths for week ending 3rd April was 60% above normal for that week. The rest of the year uk deaths was slightly below normal. Only 1/3 of those excess deaths were put down to covid.
Jordanpomeroy 7y ago
> But death numbers and hospitalization numbers don't lie. Well... there is some indication hospital staff are being directed to assume all cases are COVID19 until proven different. Patients that die with presumptive COVID19 are reported as COVID19 deaths. So agree, numbers don’t lie. But... handling emergencies effectively can lead to problematic numbers.
- jeremyjh 7y ago
  This is so tiresome. Someone has cancer for six months, gets COVID and dies 17 days later and now Tucker Carlson is here telling me not to count that death. How often do you really think it happens that someone is hospitalized with COVID (otherwise, they'd not even be getting tested most places in the world) and dies of something else a week later? I'm sure it has happened a few times, but what has also been happening hundreds of times every day is people die untested and are not included in the official numbers. Its ridiculous to even have this discussion, and the motivations for it are absolutely disgusting.
  - ngcc_hk 7y ago
    It could be an information problem. Is the medical system even allow 2 causes of death. Underlying condition and ...
    - nradov 7y ago
      Most death report forms have fields for a primary cause and contributing causes.
      - kevin_thibedeau 7y ago
        For influenza, the CDC includes deaths by pneumonia during flu season as part of the official death count because the certificates don't reliably list the true cause of death. We see this happening now in the US with unofficial deaths getting lumped into the C19 deaths, some of which may very well have been flu.
EGreg 7y ago
Sweden knows after sampling fecal matter
AmericanChopper 7y ago
> But death numbers and hospitalization numbers don't lie. They most certainly do lie. I know people who work in a particular police department, and they have been exercising a policy of not testing people they find dead. Situations like somebody found dead at home in bed, perhaps some blood that looks like it’s been coughed onto a pillow. Whoever’s in charge of policy wants the numbers lower, so they don’t do a test, and the death is recorded as something non-Covid related. Conversely you have NY who recently decided to declare all cases of death where Covid could have been a plausible cause as Covid-19 deaths. Not only are death statistics reliant on methodology and testing policies, but they are equally as open to manipulation by political actors as the infection statistics are. This isn’t unique to this situation either. These kinds of statistics are manipulated all the time in lots of different situations. For example you could never smoke a cigarette in your life, but if you manage to die of lung cancer, there’s a reasonably good chance you’ll be recorded in smoking statistics. Then you have things that are even more nebulous, like trying to figure out how many people died as a result of the Chernobyl accident. Lots of different people will try to answer that one for you, and the gap between the lower estimates and the higher estimates is enormous.
- KaiserPro 7y ago
  you'll see a spike in the overall mortality for a region. If you look at the mortality figures from the UK (published by the ONS) you can clearly see a peak, compared to the average range for that week over the last (10?) years. That peak matches the shape of the numbers of victims of covid (the number is higher because nursing homes are horribly affected)
  - AmericanChopper 7y ago
    A recession can increase mortality rates more than the numbers published by the ONS.
iso947 7y ago
I don’t believe either of those countries has been doing statistically random testing, not mass antibody testing. Testing large numbers of self selecting population means you can’t draw any conclusions.
bart_spoon 7y ago
South Korea hasn't done much antibody testing, and they are largely testing people with symptoms. They are doing a lot of it, but they still have a potentially large proportion of infected going under the radar.

wbl 7y ago

The timeline matches aggressive control measures that reduced infection.

tapland 7y ago

> but the trajectories don't lie Except for when they do. Easter just passed and the the trajectories are plotted up to and including today, when Sweden for example is adding deaths to dates two weeks back every day, and the last 5-7 days _ALWAYS_ looks like a downward trajectory. The last little spike there from before easter has grown by 35% since after easter and is still growing daily. You can't draw conclusions based on the trajectory of the last few days when an absolute majority of cases are added days, and up to weeks after.

henrikschroder 7y ago
Sure, but in that chart Italy for example has had a declining graph for over two weeks, which means you can be reasonably sure it's not growing exponentially there. And in Sweden's case, if you also look at the number of ICU admissions, they've been flat for weeks, and those numbers have very little lag. I'm not saying everything is fine, I'm not saying we're over the peak, but I am saying that it's not the case that any country is experiencing uncontrolled exponential growth and is going to turn into a zombie wasteland in a couple of days.
- matthewdgreen 7y ago
  You have to ask what phenomenon causes an exponential increase in deaths/day and does not increase ICU utilization. The answer I’ve seen seems to be “yes, the disease is killing a steadily growing number of people but they’re older and aren’t going to the ICU before they die.” This is where Italy was at one point; they just got there much later.
  - riffraff 7y ago
    As someone pointed out to me, the euromomo website is useful for this, it tracks the total mortality rate across Europe and compares it with the historical data. By ignoring the hospitalized/homestuck difference one can have a clearer idea of what is happening. Sadly I don't know of a way to get the data for the whole world.
    - innomatics 7y ago
      Euromomo data is really interesting to me because of these two points: 1) the spike in death rate caused by Covid-19 this year is of the same magnitude as the normal flu season spike in 2016 2) this year's flue season spike was low compared to recent years. Maybe due to a warmer summer? One interpretation could be that a large number of vulnerable people escaped death from regular ILIs this European season, only to sadly succome to this pandemic.
      - acqq 7y ago
        But all the countries did introduce various measures to slow down the spread. Otherwise nobody could claim that the numbers remained compatable.
        innomatics 7y ago
        Yes, which is wonderful. I am wondering what caused the large 2016 spike. Another interesting thing is the effect any behaviour changes are going to have on regular flu numbers. It's already looking to have a big effect in then southern hemisphere.

robbiep 7y ago

Austria did a relatively large scale antibody sample of its pop (0). Seems to discredit the notion that infections are vastly different than modelling. (0) https://amp.theguardian.com/world/2020/apr/10/less-than-1-of-austria-infected-with-coronavirus-new-study-shows https://amp.theguardian.com/world/2020/apr/10/less-than-1-of...

SpicyLemonZest 7y ago
This article says it was not an antibody sample.
- robbiep 7y ago
  my bad
djrogers 7y ago
FTA: The study, conducted from 1-6 April, tested 1,554 people. It did not involve antibody tests, which can tell whether a person was previously infected and is therefore probably immune. Future studies should involve antibody tests, the government has said, to inform policy as the country starts to reopen shops from 14 April.

zizee 7y ago

> No country except Iceland or South Korea knows how many are infected, everyone else's numbers are horribly wrong. Is South Korea still consideered an outlier WRT testing regimes? Not to throw shade on their great efforts in leading the way, but many other countries now have extensive testing.

ImaCake 7y ago
Australia has highest per-capita testing in the world. New Zealand is doing well. A number of other developed nations have been testing people since the start, but Australia and NZ have some added benefits of distance and relatively small populations that are helping them counter the pandemic better than most.
- hombre_fatal 7y ago
  I don't see those facts in: https://en.wikipedia.org/wiki/COVID-19_testing#Virus_testing_by_country https://en.wikipedia.org/wiki/COVID-19_testing#Virus_testing... https://ourworldindata.org/covid-testing#testing https://ourworldindata.org/covid-testing#testing I'd be wary of handing out superlatives just yet in any case.
  - robryan 7y ago
    In Australia we are testing people with symptoms (recently expanded in most places to hae no additional criteria). Some counties with a bigger outbreak/ lesser lockdowns are going to have more people with symptoms to test. 1.7% positive rate is one of the lowest in the world and combined with high testing suggests we are detecting most symptomatic cases.
    - mkl 7y ago
      NZ is mostly just testing people with symptoms too. That's pretty worrying to me given data like this article's. I think we should also be testing big random samples from the population. We've started testing hundreds of apparently-healthy volunteers, which is a good start: https://www.stuff.co.nz/national/health/coronavirus/121074955/coronavirus-wider-covid19-testing-starts-in-christchurch-community https://www.stuff.co.nz/national/health/coronavirus/12107495...

D-Coder 7y ago

> death numbers and hospitalization numbers don't lie. Death numbers are close but still have some uncertainty. New York for example is reporting much higher than usual numbers of people found dead at home. Not clear if they have been tested, if they are victims, or if maybe they had unrelated medical problems but couldn't/wouldn't go to a hospital under these circumstances.

Spooky23 7y ago
Key thing in New York is that the protocol is about fever. You won’t be allowed in a doctors office unless confirmed negative, and you usually need a fever to get tested. There are many scenarios where people in need of care won’t get it. It’s a vicious cycle and it’s fueling the ignorant rhetoric downplaying the broader issues.
thomquaid 7y ago
More people are found dead at home because during a lockdown-at-home you're simply more likely to be at home at any given point in time overall, obviously.
iso947 7y ago
UK deaths for week ending 3rd April was 60% above normal for that week. The rest of the year uk deaths was slightly below normal. Only 1/3 of those excess deaths were put down to covid.

Jordanpomeroy 7y ago

> But death numbers and hospitalization numbers don't lie. Well... there is some indication hospital staff are being directed to assume all cases are COVID19 until proven different. Patients that die with presumptive COVID19 are reported as COVID19 deaths. So agree, numbers don’t lie. But... handling emergencies effectively can lead to problematic numbers.

jeremyjh 7y ago
This is so tiresome. Someone has cancer for six months, gets COVID and dies 17 days later and now Tucker Carlson is here telling me not to count that death. How often do you really think it happens that someone is hospitalized with COVID (otherwise, they'd not even be getting tested most places in the world) and dies of something else a week later? I'm sure it has happened a few times, but what has also been happening hundreds of times every day is people die untested and are not included in the official numbers. Its ridiculous to even have this discussion, and the motivations for it are absolutely disgusting.
- ngcc_hk 7y ago
  It could be an information problem. Is the medical system even allow 2 causes of death. Underlying condition and ...
  - nradov 7y ago
    Most death report forms have fields for a primary cause and contributing causes.
    - kevin_thibedeau 7y ago
      For influenza, the CDC includes deaths by pneumonia during flu season as part of the official death count because the certificates don't reliably list the true cause of death. We see this happening now in the US with unofficial deaths getting lumped into the C19 deaths, some of which may very well have been flu.

EGreg 7y ago

Sweden knows after sampling fecal matter

AmericanChopper 7y ago

> But death numbers and hospitalization numbers don't lie. They most certainly do lie. I know people who work in a particular police department, and they have been exercising a policy of not testing people they find dead. Situations like somebody found dead at home in bed, perhaps some blood that looks like it’s been coughed onto a pillow. Whoever’s in charge of policy wants the numbers lower, so they don’t do a test, and the death is recorded as something non-Covid related. Conversely you have NY who recently decided to declare all cases of death where Covid could have been a plausible cause as Covid-19 deaths. Not only are death statistics reliant on methodology and testing policies, but they are equally as open to manipulation by political actors as the infection statistics are. This isn’t unique to this situation either. These kinds of statistics are manipulated all the time in lots of different situations. For example you could never smoke a cigarette in your life, but if you manage to die of lung cancer, there’s a reasonably good chance you’ll be recorded in smoking statistics. Then you have things that are even more nebulous, like trying to figure out how many people died as a result of the Chernobyl accident. Lots of different people will try to answer that one for you, and the gap between the lower estimates and the higher estimates is enormous.

KaiserPro 7y ago
you'll see a spike in the overall mortality for a region. If you look at the mortality figures from the UK (published by the ONS) you can clearly see a peak, compared to the average range for that week over the last (10?) years. That peak matches the shape of the numbers of victims of covid (the number is higher because nursing homes are horribly affected)
- AmericanChopper 7y ago
  A recession can increase mortality rates more than the numbers published by the ONS.

iso947 7y ago

I don’t believe either of those countries has been doing statistically random testing, not mass antibody testing. Testing large numbers of self selecting population means you can’t draw any conclusions.

bart_spoon 7y ago

South Korea hasn't done much antibody testing, and they are largely testing people with symptoms. They are doing a lot of it, but they still have a potentially large proportion of infected going under the radar.