Some preliminary COVID statistics and thoughts

So I was scrolling around the blogs, as I do, when this piece by Bryan Caplan caught my eye. In particular, something that he links to:

Americans entered the Covid pandemic in much poorer health than citizens of other developed countries. For instance, over 27,000 U.S Covid deaths list diabetes as a comorbidity, accounting for 16% of total Covid-related fatalities. But what if instead of having the highest diabetes rate among rich countries the U.S. had the same rate as Australia, with less than half the U.S. level? 

Then this comment:

There’s an issue here, which is that the US doesn’t actually have a high COVID mortality rate. The US has a lot of COVID deaths because it has a lot of COVID cases, but the actual fatality rate is actually very low in the US. Here are some comparisons:
Italy – 11.51%
UK – 9.57%
Canada – 5.87%
France – 5.71%
Spain – 4.20%
Germany – 3.28%
Australia – 3.27%
USA – 2.87%

That is interesting. What do the numbers look like? If we just follow the zeitgeist it seems that COVID is an American problem (and Victorian problem too).

So I went to the COVID Worldometer and downloaded the data and had a play.

Okay – so some heroic assumptions:

  1. The data are clean.
  2. The data are consistent.

Both these assumptions are unlikely to hold. Data collection varies in quality across countries and definitions are likely to be very fluid (i.e. dying of COVID and dying with COVID etc.). But working with what we have got …

First thing, I calculated an infection rate (number of cases divided by population) and then a death rate (number of deaths divided by cases). Bear in mind the least dodgy number there is the population data.

I then plotted the data. (Australia, US and World average are pointed out). I dropped Yemen from the graph as it was a huge outlier on the y-axis (28.8%).

I then created an excess infection variable and an excess death variable. The excess infection variable is simply whether the infection rate in the particular country was above the world average infection rate (including Yemen) and the excess death variable is simply whether the death rate is above the world death rate (including Yemen).

I then classify countries into four categories:

  1. Excess Infection and Excess Death
  2. Excess Infection but No Excess Death
  3. No Excess Infection but Excess Death
  4. No Excess Infection and No Excess Death

Clearly the best place to be is the in 4. followed by 2.  I suppose 3. is better than 1. But that boils down to an empirical question.

So now, who falls where?

In category 1, I find a lot of European countries. Including Sweden. Now I have not yet folded in economic performance, I am just looking at health outcomes.  Right now my view is that countries that fall into this category have failed from a public health perspective and a public policy response perspective.

In category 2, I find countries like Chile, South Africa, Austria, Singapore, and Israel with excess infection rates but no excess death rates. This I classify as poor public health but good policy response. This group of countries includes the United States.

In category 3, I find countries like China, Afghanistan, Syria, Tanzania, Angola, Indonesia, and Germany  with no excess infection rates but excess death rates. This I classify as good public health but poor policy response. In this group disproportionately more people died than were infected. This group of countries includes Australia. Bear in mind about 90% of Australian deaths come from Victoria.

In category 4 I find countries like Norway, Estonia, Taiwan, New Zealand, and South Korea.

Quick conclusion: Relative to the number of infections fewer people have died in the US than in Australia. So for all the talk about not letting us have a US style health system, it has performed relatively better than ours did. True, Australian base levels of health are higher than US levels. Relatively fewer people got infected here than in the US. (We can argue about why that happened in the comments – probably policy errors in themselves at the State level in the US). Once infected, however, it looks like the chances of survival was higher in the US.

This entry was posted in COVID-19. Bookmark the permalink.

44 Responses to Some preliminary COVID statistics and thoughts

  1. Entropy

    I think the uS general population got infected more than Australia, where it got into old people’s homes in particular but did not get out of them.

  2. Roberto

    You should split Australia into two ‘countries’: VIC and ‘all the other states/territories’. VIC’s COVID mortality rate is 25 times the rate in the rest of Australia.

  3. Epidermis

    All very good but when the one major variable (deaths with or from) is really indeterminate the outcome is GIGO. So are we talking about rubbish? WADR.

  4. pbw

    CDC current best estimate for Infection Fatality Rate as at 10th September.

    My attention was drawn to this by Ron Paul. Docter Ron Paul, that is.

  5. Roger

    Quick conclusion: Relative to the number of infections fewer people have died in the US than in Australia. So for all the talk about not letting us have a US style health system, it has performed relatively better than ours did.

    Not all mortalities are equal.

    Reportedly, many of our infected aged folk in Victoria were offered nothing more than palliative care.

  6. Ben

    You are assuming ‘cases’ whatever that means is a good proxy for the infection rate in the population. In the UK we know this isn’t true because ‘cases’ are spiking due to increased testing. Seeing as in the UK it’s not possible to compare ‘cases’ across time because of changes in the testing regime I don’t see how you can meaningfully compare cases across countries.

  7. Leo G

    Quick conclusion: Relative to the number of infections fewer people have died in the US than in Australia. So for all the talk about not letting us have a US style health system, it has performed relatively better than ours did.

    Keep in mind that the US has experienced a real epidemic with complex patterns of propagation whilst the Australian situation is one of isolated outbreaks with limited propagation. Both countries have had “protected” COVID-susceptible citizens infected as a result of government misfeasance, but in Australia cases associated with those events were a much higher proportion of the total cases than in the US.
    A different profile of infection rates by age group leads to a different overall death rate.

  8. HGS

    Why not just deaths per population, relative to past 25 years or so. Any rate above norm might show the China Flu damage and or lockdown damage.

    And , good public health in Angola, Afghanistan etc. Just what do we mean by good public health?

  9. Tel

    World average should be world cases divided by world population, right?

    So for all the talk about not letting us have a US style health system, it has performed relatively better than ours did. True, Australian base levels of health are higher than US levels.

    Most of the supposed “bad outcomes” of the US health system are directly coming from car accidents and gunshot wounds. Car accidents because a lot more people drive, and overall they drive more often and longer distances. Gunshot wounds almost entirely coming from cities like Chicago and Detroit and we can argue all day about how to fix that problem. In either case, it has nothing to do with the quality of the doctors and hospitals, although they are doing their best to re-define various things as health problems and then after that redefinition they go and blame the hospital system. Clever huh?

  10. Tel

    Tanzania’s approach was to tell the World Health Organization to piss off, then say a prayer and get on with life. The median age in Tanzania is approx 18 years, most of the population have nothing to worry about.

    There’s a heartbreaking Tom Woods episode where they discuss the “lockdown” in Africa.

    https://tomwoods.com/ep-1743-lockdowns-are-crushing-the-developing-world/

    Children starving for no reason because their government decided to save all the old people … conclusion is that unless you adjust for age the whole statistic is meaningless.

  11. C.L.

    Fascinating numbers crunching.
    Imagine how Australia would look if even two other state governments – or, dear God, three or four – were as criminally negligent as Victoria. Just imagine.

    As far as the US goes, I see its higher infection rate as both a corollary of its freedom culture (comparatively speaking) but also – more importantly – its governmental complexity. Fifty states + the Feds. The other countries in category 2 have fairly unitary political systems (the provinces of South Africa and Chile notwithstanding).

  12. H B Bear

    There is unlikely to be a less reliable data set than Covid stats. I believe that at one stage the UK was not counting deaths in old people’s homes, only deaths in hospitals.

  13. A Lurker

    I think the real question should be: How many died of COVID-1984, and how many died with COVID-1984. The answer to that question would be very illuminating.

  14. Sean

    Lol, yeah Sweden failed from a policy response.

    You can’t just look at a pandemic using numbers like this. How can you model how an outbreak will occur and spread given a novel virus is going to get a jump on you? Hence, how can you make a blanket judement on policy responsed between all nations. A connected European country vs an isolated island…

  15. Fair Shake

    The underlying assumptions are data collation processes are comparable. We know the Victorian stats are flawed full of duplications which then get reversed. The co-morbidity factor throws the clean data into the mud. How the heck do we know? The data is corrupted and being used to further agendas.
    I’m sure if I was tested my blood cells would show a significant trust deficiency. No amount of VicGov ad stimulus will cure.

  16. Chris M

    Thanks for this Sinc. You talk about Australia when what you mean is Victoria. To many of the rest of us there is a difference, it’s really different.

    US style health system, it has performed relatively better than ours

    Ours = Victoria where hundreds died needlessly. Either the Vic health system is incompetent or it’s a direct result of the medical drugs that Dan deliberately withheld, or some combination of the two. I fully believe that Mao Tse Dan wants some daily deaths to continue the fear and obedience. Most are definitely political deaths, no doubt about that.

    But Vic being different they’ll probably vote him in again, assuming elections will still be a thing there.

  17. John Brumble

    Your Singapore assessment is waaay out. Singapore had a high number of infections due to the high number of young Malaysian workers. The low death rate way is far more likely to be due to the age of said workers than much else.

  18. candy

    I tend to think that cultural factors specifically might be included. People who don’t speak English very well, where measures to communicate infection control were not addressed properly by the Health Department and contact tracing becomes difficult. It may be that Melbourne has more of a problem in that respect, compounded by incompetent government officials.
    A broad look at Australia or Victoria compared to America does not really hone in.

    Once it is a nursing home it is quite deadly to those terribly frail people.
    The Ruby Princess had a outbreak but the control of that and contact tracing perhaps was much easier due to the demographics involved.

  19. RobK

    Thanks Sinc,
    You can only try to make sense out of non-sense. Good try.

  20. Knuckle Dragger

    ‘I dropped Yemen from the graph as it was a huge outlier’

    No no no.

    Outrider. It’s outrider.

  21. Aethelred

    Hmm, you probably need to consider the age distribution of the population, rather than just the total number.

  22. Matt

    How have you factored in testing? Case fatality rate is very different to the infection fatality rate.

  23. Pyrmonter

    Are averages without some adjustment for the age structure of the population useful? Australians and Germans tend to be older and their populations have many more people over 70 than in historically lower income/lower longevity countries. The comparison between Germany and Austria – similar age, wealth profiles – is interesting.

    There’s wide variation within the US as well – by far the worst states are in the northeast and Illinois. Worldometer has a state-by-state breakdown, it would be interesting to do the same analysis against both world and US average levels.

  24. Statistical analysis, especially multiple regression modelling is right up there with wind-watching for being as much fun as you can have with your clothes on.
    Nice work Sinc!

  25. Tel

    Lol, yeah Sweden failed from a policy response.

    China’s approach (lie about the numbers) put them in a much better position on the chart.

  26. chrism

    the ‘dry-tinder’ of one or more prior years of lower than average ‘excess mortality’ in the Winter months may explain some of countries death-rates : looking at Australian case fatality rates (deaths/total cases) the trend is very strongly increasing by latitude ?? vitamin D levels :

  27. Steve

    The USA CDC ( Center for Disease Control ) released a report that basically said this :

    Of all “covid deaths”( deaths that allegedly caused by covid ):

    * 94% was due to pre-existing meidcal conditions, i.e. covid like any bad flu, pushed the body too hard and they died, but not of covid.

    * 6% was due to actual covid.

    So……if you take any “covid deaths”figuge and multiply it by 0.05, you get the true covid death rate.

    SO there is no medical reason for any lock down. Herr Andrws are you listeneing?

  28. Jannie

    1 million deaths globally. 12 Trillion $ lost production and costs.

    Thats $12,000,000 per death. A meaningless statistic. But a lot.

  29. Angus Black

    In the absence of an effective vaccine (and, let’s be honest, we’ve never seen an effective vaccine for this class of viruses), I’d suggest you’d prefer to be in category 2 provided that your health system wasn’t under the pump (and, since the very early days in Italy, no health service has been seriously stressed by COVID).

  30. Sparkle Motion

    What does the graph look like if you assess death rate using population rather than infection rate?

    Like you say “Bear in mind the least dodgy number there is the population data.”, it may be a better way to scale the overall death rate.

  31. Sinclair Davidson

    All good points.

    The things that interest me would be to follow up on why some places had massive infection rates and other didn’t. That sort of question goes to the underlaying health of the population (the original story that Caplan thought about), demographic factors, and private and public responses to the initial outbreak. Then why did some places have higher death rates. That also points to underlaying health of the population and demographic factors, and again policy responses. Overlaying that would be a regional response. So population homogeneity and heterogeneity can into play.

    Those are all interesting questions before we get into cost-benefit analysis of political-economy responses to the pandemic.

    … and of course, all of that depends on whether we have good data to play with.

  32. Sinclair Davidson

    What does the graph look like if you assess death rate using population rather than infection rate?

    Very small numbers on the y-axis.

  33. incoherent rambler

    The unreliable nature of the raw data for infections and deaths can be used to dismiss any conclusions one might be tempted to draw.

    “Very small numbers on the y-axis.”
    As good as it gets.

  34. MACK

    As several people have said, there are structural factors and there are policy factors. Structurally, high risk comes from high population density (by far the most important) as well as average population age, co-morbidities and the number of international travellers (northern Italy had 60,000 Chinese working there, and New York, London and Paris have millions of business and tourist travellers). Policy responses are important, but the most important and neglected effect in places with high literacy comes from voluntary actions by the people, especially physical distancing and had washing. The growth curve in Victoria shows this working effectively very early on. From the government, with a virus that can spread from non-symptomatic people, contact tracing is critical and has been a central cause of the Victorian disaster. Banning crowds especially indoors helps, and the concentration in aged care needs high priority. Correct use of PPE in health care is a major issue, even now in Victoria. All the rest of the shutdown stuff and using masks outside is just political posturing. So any quantitative analysis needs to contain lots of factors, with appropriate weighting.

  35. lotocoti

    why some places had massive infection rates and other didn’t.

    London’s poisoned dwarf seems to be hinting at some sort of systemic ism.

  36. Pyrmonter

    @ Mack

    If density is so important, where is the outbreak in HK and Tokyo? London, Paris, Berlin and NY are suburbs by comparison.

  37. Lizzi55

    The Covid test data is meaningless, it’s only purpose is to produce fear and justify the ongoing fake pandemic. Cross country comparison based on the Covid test data is meaningless. You are better off to narrow the scope of your data analysis, to a single country or a single state for example, and then compare total mortality relative to previous years, noting that lockdown will have an impact due to restricted medical services.

    On the other hand if you did want to analyse the Covid test data, you should analyse results from a random sampling of the population, in a particular state or country, tested using the same Covid test kit. You should also document the number of PCR cycles taken during the test (some say this relates to the viral load), and the condition of the subject (symptoms, comorbities, age, etc). And this process should be repeated over time, in different countries, to analyse in-country trends, and to peform cross-country comparisons. The data should be normalised to population, and ideally the Covid test sampling rate should be the same. You can’t do this analysis from worldometer data, the data doesn’t exist.

  38. Sinclair Davidson

    Lizzi55 – it’s worse than that. The data to derive clean results do not exist and the organisation that should have been coordinating the collection of that data is itself horribly compromised and incompetent.

  39. Aethelred

    All good points.

    The things that interest me would be to follow up on why some places had massive infection rates and other didn’t. That sort of question goes to the underlaying health of the population (the original story that Caplan thought about), demographic factors, and private and public responses to the initial outbreak. Then why did some places have higher death rates. That also points to underlaying health of the population and demographic factors, and again policy responses. Overlaying that would be a regional response. So population homogeneity and heterogeneity can into play.

    Those are all interesting questions before we get into cost-benefit analysis of political-economy responses to the pandemic.

    … and of course, all of that depends on whether we have good data to play with.

    Should be a paper or 2 in it

  40. calli

    I know I muck around on the OTs too much.

    This is the sort of stuff I come to the Cat to read and think about. Thanks Sinc and commenters.

  41. John A

    C.L. #3603718, posted on October 1, 2020, at 7:38 pm

    Fascinating numbers crunching.
    Imagine how Australia would look if even two other state governments – or, dear God, three or four – were as criminally negligent as Victoria. Just imagine.

    As far as the US goes, I see its higher infection rate as both a corollary of its freedom culture (comparatively speaking) but also – more importantly – its governmental complexity. Fifty states + the Feds. The other countries in category 2 have fairly unitary political systems (the provinces of South Africa and Chile notwithstanding).

    Overgoverned? Not sure. Excluding local government (that may be unfair, but…) the comparison is:
    Australia: 25.5 million under seven governments = 3.64 mill per
    USA: 331 million under 51 governments = 6.49 mill per
    Canada 37.7 million under 11 governments = 3.43 mill per
    Land area comparison: USA 9.8 m sq km vs Aus 7.7 m sq km vs Canada 10 m sq km

    I agree that our data is skewed by Diktator Dan and the PDR of Viktoriastan.

    As someone above said, we need to separate Viktoriastan from the rest of the continent.

  42. BoyfromTottenham

    Sinc, IMO the easiest way to reduce the effect of the various statistical errors that you mentioned is to simply quote the mortality rate per million of population (available from the same Worldometer source btw). This way, the only variable that is likely to have a significant level of error is the way that deaths are attributed to COVID-19.
    Using deaths/million population, Australia is currently looking good at around 35 deaths per million, the US is 642, Canada is 246, UK is 621, Sweden is 583.
    I am sure you will agree that the sloppy definitions of “COVID Deaths” being reported – e.g. ‘died of COVID’ vs ‘died with COVID’ make even this statistic unreliable, and IMO will be causing historians and bio-statisticians heartburn for many years to come.

  43. Roger W

    Maybe even better, what are the extra deaths (if any) compared with total deaths each year for maybe the last 10 years? Then will be interesting into the future to gain an idea of how many extra will die because of the lockdowns etc?

  44. Beachcomber

    All of this statistical jitter-bugging is pointless. It just gives credibility to the whole vicious charade 0f Covid-1984. It is based on meaningless numbers of results from tRNA-PCR testing that are falsely represented as “infections” or “cases”‘.

    There is no epidemic, no pandemic, no plague. Look at the overall death statistics in Victoria.

    It is the same across the world.

    There Is no Coronavirus Pandemic, So How About a Dose of Reality?

    ……… according to the World Life Expectancy tables, total deaths have reached over 44 million out of a total population of 7.734 billion people. If this number is extrapolated out for this year, total deaths worldwide will be approximately 58.7 million. Average deaths per year worldwide currently average close to 60 million deaths per year, so this year is certainly not abnormal, and in fact seems to be slightly less than normal. How can this be, given that we are told we are in the middle of one of the most deadly pandemic periods in history, so deadly that the world has been shut down?

Comments are closed.