More covid cherry picking

There are so many statistics on the covid-19 pandemic that there is usually one that helps a given narrative. Back in June I wrote about one example from the right-wing ‘news’ outlet PJMedia (where one notable Sad Puppy often writes and very prone to covid-conspiracy mongering) https://camestrosfelapton.wordpress.com/2020/06/11/a-current-snapshot-of-covid-19/

The same writer I linked to in that earlier post has another attempt at trying to minimise the perilous state of things in the US. His focus this time is the case fatality rate i.e. the proportion of confirmed deaths to confirmed cases. The article is correct in pointing out that the CFR for the US is relatively low compared to many other countries. Except, while an interesting statistic the CFR combines both the issues with confirmed cases as a statistic (e.g. testing policy, accuracy and reporting) and the issues with deaths as a statistic (e.g. which deaths are counted). The main site I use for covid-19 graphs (OurWorldinData.org) has a long explainer on it here https://ourworldindata.org/covid-mortality-risk that is worth a read.

At the end of the article the author points to his earlier article from March and states:

“That data didn’t show the United States as number one for deaths or cases.”

Here he means number one on a per-capita basis and like myself he has been using the OurWorldinData.org figures. Personally I don’t think the deaths statistic is working well for cross national comparisons. Similar countries, with similar health systems and similar histories with the pandemic show quite different figures for population adjusted death rates. Maybe there’s some real difference in mortality there but if there is, nobody is clear on what it might be. With the US mortality statistics there is an odd fact that the deaths are skewing younger than other countries (https://www.economist.com/graphic-detail/2020/06/24/when-covid-19-deaths-are-analysed-by-age-america-is-an-outlier ) one possible reason for that might be that deaths of older people are either being under-counted in the US or over-counted in other countries (or both).

I’ve tended to stick with looking at cases per capita. Again, big flaws with using that for comparisons because testing regimes change the figures and also there are differences in who gets counted (e.g. if you get tested twice and a positive both times is that 1 person in the count or 2?). Still, the PJMedia article in both cases points to cases per capita as a statistic where the US is not the worst in the world.

So how is the US doing on that figure? You can look at the figures in a table here https://ourworldindata.org/grapher/total-confirmed-cases-of-covid-19-per-million-people?tab=table

Overall, the US is 11th in the list. 10 countries are worse on this stat! But…most of those ten are relatively small countries or actually VERY small countries (eg the Vatican) and a single outbreak sends a per capita figure soaring. Filter out those nations with populations less than 5 million and you more clearly see nations with proportionally large numbers. This is the top 15. The US is not number one…it’s number 2 because Chile is worse. The top 4 are all in the Americas. Sweden and Singapore are both special cases in their own way and may not be that comparable. After the table, I’ll look at a way both Sweden and Singapore are observably quite different.

CountryCases per millionPopulation
Chile17,081.7918,952,038
United States11,020.20329,064,917
Peru10,479.7632,510,453
Brazil9,627.09211,049,527
Singapore8,111.155,804,337
Sweden7,652.1410,036,379
Saudi Arabia7,061.8734,268,528
Belarus6,961.569,452,411
United Arab Emirates5,704.729,770,529
South Africa5,692.1558,558,270
Israel5,483.088,519,377
Belgium5,478.9611,539,328
Russia5,245.07145,872,256
Bolivia4,806.1211,513,100

As I’ve said, the cases stat is questionable because if you don’t do any testing then you don’t find cases, likewise the number of infections will also lead to an increase in testing and hence finding more cases. However, I’ve found watching the trajectory of cases per million to be an effective way of seeing what is going on. It’s an imperfect indicator but one of the better ones (as always, never depend on one statistic.

I said Sweden and Singapore are different and I think the graph of the top 6 nations in that table illustrates that and shows what I mean by trajectories.

As nations, Sweden and Singapore are chalk-and-cheese in terms of location, demographics, methods of funding healthcare and even pandemic response policies. However, their current paths are similar. Whether their high cases-per-million stat is an accounting issue or due to effective testing policies catching a greater proportion of asymptomatic people, I don’t know but what we can see is that cases aren’t accelerating.

The same cannot be said for the USA. It really is genuinely a very bad situation that it is in looking at the spread of cases. It really isn’t media exaggeration.

Advertisement

21 responses to “More covid cherry picking”

  1. When the best you can say for your policies or practice is “we’re not the worst in the world,” you’re conceding that you’re doing badly.

    Test numbers here in the US seem frighteningly flimsy, between the explicit intent to keep the number of tests down so fewer cases will be reported, insufficient supplies of reagents, and a certain amount of what seems like general incompetence. A friend of mine was supposed to get tested today, made an appointment, was told which lab to use, and discovered when she got there that it was closed.

    Liked by 2 people

  2. US is testing about twice as much as Sweden and two months ago most likely four times as many – at least – so I don’t think you can get anything of value whatsoever from just comparing number of cases.

    Even more so, I think it isn’t relevant in any way to compare US, more or less a continent, with small countries. The illness has shown that it doesn’t really catch on unless there is a critical mass of population. More or less all large Swedish cities, there aren’t many, has had their outbreaks, so country wise, it will take much shorter time before the pandemic reaches the whole country. We are now seeing a drop in absolutely every region and have no excess deaths according to Euromomo.

    https://www.euromomo.eu/graphs-and-maps

    US is a much larger country with a lot more possible epicenters. When one has passed the crisis, another will start, thus giving a longer, flatter curve for the country as a whole. So while Sweden still has more excess death than US (if US statistics can be trusted), I expect that will change.

    https://www.ft.com/content/a26fbf7e-48f8-11ea-aeb3-955839e06441

    Like

    • “Oh, they’re just rearranging the deck chairs on the Titanic.” First of all, that is a terrible metaphor. This administration is not sinking. This administration is soaring. If anything, they are rearranging the deck chairs on the Hindenburg!”

      Liked by 5 people

      • At least with the Hindenburg, they cared about the “oh the humanity.”

        Honestly, to be running around saying, “At least we have fewer deaths per capita than Chile!” is morbid absurdism. Our meager and scattershot lockdown and covid prevention and testing attempts, in the absence and outright sabotage of federal action under the Republicans, slowed rates slightly and allowed the initial hotspots to start flattening the curve and just barely preserve their health systems from collapsing. But what they didn’t get us was the main thing that a full and federally financed lockdown was supposed to get us — time to get better PPE, testing and tracing procedures in place to fully slow and contain the virus as other countries have done. Because we never had the full and federally financed lockdown, just a cash grab for corporate execs and crony con artists, like always under this regime. So instead of slowing down, the rates have rocketed up, the rich have skimmed the cream off the aid and we’re going to run out of refrigerator trucks as portable morgue storage.

        We also went fully into an economic recession which might have at least been delayed if again there had been a full funding by the federal government for people to stay home, get tested and be tracked while they isolate when sick, or even if they’d just mandated masks early on. And we’re going to have a massive eviction and food scarcity crisis develop over the next few months, again because of inaction from and outright sabotage by the federal government. Doing what has worked in other countries and what we know we desperately need, however, would show that government works and that disproves forty years of hard effort by the right to dismantle government, then claim the mangled remains don’t work and dismantle it some more. People dying from that is part of the fun for those folks.

        So I don’t really get why any of them are even bothering to minimize death rates. They were reveling in them not that long ago, with the some people are just going to have to die and we should all sacrifice our grandparents for the Dow Jones and we have to live with 40,000 people dying a month from it, that’s just the way it is. Why do they care if the U.S. is number one in death rates, under any rubric? It’s what they wanted. It’s the devastating collapse they cheered for and said would help them build a “new” autocratic country. It’s what they felt was a perfectly fine price for getting their nails done and had the added bonus of killing Black and brown people more often since they’d blocked them from healthcare access. They’re having Covid parties to prove how happy they are about it; next they’ll be dancing in the streets for death parades. Why bother lying about the death rates when they just contradict themselves the next second?

        So, yeah, rearranging deck chairs on the Death Star, whatever.

        Like

  3. I love the graphic circulating on Twitter, most notably by a user calling themselves “Georgia Person”, demonstrating how shifting categories make a Georgia case map look as though there’s no increase in COVID cases at all (i.e., the same color being used for 621 to 1070 cases one week, and then 955 to 1050 two weeks later). It’s been suggested that this is just incompetence in chart-making rather than outright lying, but the net effect is the same: the public gets misled.

    Liked by 4 people

    • I suspect incompetence here (if they charts were indeed published together). There are a lot of similar charts on Wikipedia, where the binning seems to be determined algorithmically from the data rather than by human decision (note unround numbers being used for bin boundaries), and which change in this fashion. These are suitable for showing the geographic distribution of cases. (What they fail at is distinguishing old hotspots from new hotspots.)

      Is there evidence that these images were used in a misleading way?

      Liked by 1 person

      • As I understand it, the charts are from the Georgia Department of Public Health, but issued at different days. You can find a thread about them here:

        I agree that the error here seems like something that is likely to happen if one relies too much on semi-automatic tools to make the charts. But the level of incompetence here in not just making this error a few times, but repeatedly over days, with no-one in the organization noticing it and reacting to it, is staggering. So while it’s certainly plausible that this is pure incompetence and not malice, I don’t think that really makes them look any better.

        Liked by 1 person

  4. Because different countries (and different States within the U.S.) have widely-varying and inconsistent policies about when a death is classified as COVID-19, I’m convinced that right now the only feasible way to compare per-capita deaths by country and state is by using the excess mortality data.

    Liked by 3 people

  5. …if you don’t do any testing then you don’t find cases…
    This is now the entirety of the Trump administration’s Covid-19 strategy.

    Like

  6. If you treat US states separately (and the rest of the world as countries), US states have 5 of the 6 highest per capita official COVID death rates in the world, with San Marino having the 4th. On the other hand, if you break up other countries by region, Madrid, Castile and Leon, and Castilla-La Mancha in Spain, and Lombardy in Italy, are in the same range.

    As the fatality rate sensitively depends on age the US, with an appreciably younger population, should have significantly lower numbers of deaths for a given intensity of pandemic than western European countries.

    Liked by 1 person

  7. I’m quite worried about my elderly relatives in the Chattanooga area, which is right on the border with Georgia. All three of my siblings have already had COVID-19 and recovered.

    On the bright side, President Trump seems to have decided, just today, that Masks Are Great Again, and told people that wearing one is patriotic. I find myself not wanting to hassle his supporters too much over this–it’s more important to me that mask wearing become the norm for everyone than it is for me to score a few points.

    Liked by 1 person

%d bloggers like this: