Nice White Countries
The creation of statistical data is informed by basic questions of ethics and philosophy. How do we decide that something is a thing? How do we identify things as being in particular category? And which do we value more: beneficence or autonomy? What information should be gathered, and what should be allowed to be seen by the public? What are we trying to prove, or to say, about the world at large with our information?
And once we get those things settled as best we can, then there are practical considerations. Can the data we want actually be measured? Do we have the resources to do so? What parts of the data will we choose to best further whatever we have decided our agendas are?
At every step, underlying agendas and principles, philosophical or practical, are there.
This is all a lead-in to a discussion of data sets about COVID-19 from different political entities.
If you have an interest in playing with data, tweaking it and massaging it to try and find out how they are supposed to relate to “reality,” you must be finding the presentation of COVID-19 data sets very interesting.
Here we are going to look at one aspect of these basic sets from your browser and see if it can tell us anything about the most important part of this situation going forward: How do we get the right tools matched up with the right job?
You wouldn’t try to build a house using a shoe to drive the nails. Nor would you try to hang a picture using a sledgehammer. Using the wrong tool for a situation wastes energy and/or causes unnecessary destruction, while not even being very effective at the task it was supposed to do.
Governments around the world have been pressured to bring a lot of “tools” to bear of addressing COVID-19 infections. Yet a quick look at a search engine’s (Bing) information reveals a lot of asymmetry in terms of infection and fatality rates between different countries.
Can this data be used to parse out what interventions might actually offer some benefit? Or maybe to see if something else entirely is going on? This is important, because humans are very bad about confusing correlation with causation. In this case, just because an area did or did not do a certain intervention, does not mean that this action caused a change in viral infections rates or deaths. And we need to clarify this causation/correlation problem, because many of the quickly implemented, socially and politically influenced interventions that have been done have had very destructive “side-effects.”
In medicine, we often do use drugs or treatments with very destructive side effects, but we are careful to do our best to make sure that it’s “worth it.” The benefit from doing the treatment or using the drug needs to outweigh the damage from the intervention itself. Examples of this include cancer chemotherapy, or leg amputation for diabetes complication. We don’t want to cut off a leg that can be healed by less destructive measures, or give someone without cancer a course of cancer chemotherapy.
The analogy here is that governments, health scientists, and NGOs such as WHO need to look at data—not just with the idea of what should be done, but also look at background data to see what effect that is having. It might be all too easy to assume that because a particular government did or didn’t so X, X caused an observed result. And then we might think we have the right tool for the job, when actually we do not.
Let’s make a start.
Unfortunately, the world is full of bigotry, chauvinism and prejudice. People in, for example, the United States are therefore liable to be suspicious of data from non-Western countries; likely to attribute differences in reportage problems with data collection, resources, image issues or inefficiency. Therefore, we will concentrate here on “nice white countries,” that is, European ones. In order to smooth out the stats, we will limit this to countries of at least one million population.
Let’s compare the ten worst countries in Europe in terms of overall fatality rate versus reported cases of COVID-19 (all time, not going on right now), with the ten best. We will also look at some commonly invoked metrics to explain differences in health outcomes: age demographics, life expectancy, median household income, medical spending per capita and as a percent of GDP (gross domestic product,) and total “cases” reported as a percent of population.
Data for this was taken from public sources: Wikipedia, WHO (World Health Organization) and so on. As is usual it was not possible to get all the demographic data to be from the exact same year, but it is all in a 4 -5 spread at its most separated in time.
You’ll notice that the fatality rate and the percent of total population identified as “cases” are derived statistics, using raw numbers from the Bing search engine on 13 Sep 2020. Age demographics are additive from age pyramids.
These figures are not the CURRENT fatality rate or CURRENT percent positive cases. These numbers are cumulative, and therefore may not reflect what is going on in a particular country right now.
In terms of interpreting the % positive, remember that 0.1% means one in a thousand. So, for example, when Italy has a cumulative percent positive cases of 0.5%, that means that since this whole mess started, the total of 5 out of 1000 Italian residents have tested positive or been identified as having a COVID-19 infection. This does not mean that 5 out of 1000 is positive/sick now. That number could be closer to 1 out of 20,000-40,000.
Without further ado, here are Europe’s ten worst countries in fatality rates, for the time from identified onset to 13 Sep 2020, that is, the percentage derived by dividing reported deaths by reported cases. We also note the total population, percent of the total population with reported COVID-19, the percent of the total population over 60 years of age, the life expectancy of each country, the median household income, the percent of the gross domestic product the country spends on health care, and the amount spent per person in each country on health care.
Country fatality population percent percent life expect median income percent GDP US$(2015)
Rate% total pop positive 60y+ f/m per household spent health per cap on health
Italy 12.6 60Mil 0.5 % 30 85.5/81.7 20K (US$) 8.8 2700
UK 11.5 67Mil 0.5% 25 83.0/79.5 32K 9.8 4356
Belgium 10.9 27Mil 0.8% 25 83.8/79.8 27K 10.4 4228
France 8.3 65Mil 0.6% 25 85.4/79.5 31K 11.2 4026
N’lands 7.7 17.5Mil 0.5% 26 83.8/80.4 39K 9.9 4675
Sweden 6.7 10.4Mil 0.8% 26 84.4/80.9 51K 11.0 5600
Ireland 5.8 5 Mil 0.6% 24 83.7/80.4 25K 7.1 4757
Spain 5.3 47Mil 1.2% 29 85.4/80.9 22K 8.9 2354
Hungary 5.3 9.8Mil 0.1 % 26 80.1/73.1 12K 6.9 894
Romania 4.0 19Mil 0.5% 26 77.4/72.5 8K 5.3 442
We’ll come back to these, but first, the best:
Country fatality population percent percent life expect median income percent GDP US$(2015)
Rate% total pop positive 60y+ f/m per household spent health per cap on health
Slovakia 0.7 5.4Mil 0.1 23 80.8/73.8 17K 6.7 1108
Belarus 1.0 9.4Mil 0.7 22 79.4/69.4 15K 5.7 352
Czechia 1.3 10.7Mil 0.3 26 81.6/76.6 23K 7.5 1284
Croatia 1.6 4Mil 0.4 29 81.5/75.1 16K 6.8 852
Russia 1.7 147Mil 0.7 21 77.6/66.7 12K 5.3 524
Ukraine 2.1 42Mil 0.4 24 76.7/67.0 11K 7.1 584
Norway 2.2 5.4Mil 0.2 22 84.3/80.3 51K 10.2 7464
Greece 2.3 10.4Mil 0.1 28 84.5/79.6 18K 10.2 1505
Austria 2.3 9Mil 0.4 27 83.8/79.0 35K 10.3 4536
Serbia 2.3 6.9Mil 0.5 26 78.5/73.3 9K 8.0 491
Even with these simple charts there’s a lot to unpack.
First, to bring up an obvious problem, different countries have different ways of counting cases and fatalities. Some countries only count deaths within the context of obvious symptoms and a positive virus test. Others might count deaths that were suspected to be due to COVID-19 without confirmatory testing. Some might count all pneumonia deaths, or deaths with other potentially fatal conditions plus a positive virus test, and some may not. Some countries may even lump fatalities indirectly related to viral infection; such as suicide due to stress, or missed cancer treatments, in with COVID fatalities. Obviously, this must be borne in mind when comparing countries.
When you look at the “worst” you can also see that the last two countries, Hungary and Romania, do not match up with the first eight, economically.
Also, a note on Russia. At least in the United States, there is a tendency to look at Russia with suspicion; to think that they must be fudging their reportage, either due to incompetence or political pressure, because their numbers are better than the US figures. How can that sloppy, loopy bear be beating us?
But while it is true that the Russian government tends to lean into the beneficence side of the beneficence/autonomy debate, the Russian state is not incompetent. They may never be efficient, especially with human resources, but they can be terrifying effective. Just ask Hitler.
And Russia’s numbers actually make sense. Their reported infection rate is consistent with what other countries report. And their fatality rate makes sense given their demographics. They have a lower proportion of old people compared to most of Europe, and old people make up the vast majority of COVID-19 fatalities, and they also have a dramatic split between female and male life expectancy. This also makes a difference because mortality in males is somewhat higher than in females for COVID-19.
An aside about a point that might escape notice. These cumulative infection numbers for all these countries is not all that high, for a virus that, according to our media and governments, has been rampaging, surging (always surging), and running amok in the populace. Except for Spain, with 1.2%, in all these countries, 99% plus of this European population has not been identified as having a COVID-19 infection at any point in time. That’s pretty lame for a supposedly rampaging panepidemic. Seems COVID-19 is actually not that common a disease there.
It is of course, quite reasonable to assume that these figures only capture a portion of the cases, the ones that come to medical attention in some way. It might be quite reasonable to assume that anything up to 90% of cases are not detected. Nobody is really sure, and for some reason nobody seems to want to find out.
However, recently the state of Hawaii, for unclear reasons, but basically for the optics, did something called “surge testing.” They offered free (to the people being tested, the testing affair itself cost millions of dollars) no-questions-asked tests for current viral status, to the general public on the island of Oahu. As of September 13, 2020, the results were 267 positive tests out of 45,343 tests completed. Although this information led Bruce Anderson, the current Director of Hawaii’s health department to describe COVID-19 as “widespread on Oahu,” (Mr. Anderson is considered by many Hawaii locals to be incompetent even by Hawaii bureaucratic standards, which is saying a lot) these tests actually indicate the exact opposite. There is not a lot of COVID-19 infections running around undetected on Oahu right now. It’s hard to project how these acute test findings relate to total cumulative virus exposure over time. You would need antibody testing for that. But given that Oahu had virtually no reported cases until the middle of the summer of 2020, one might assume that the actual rate of people, past and present there who have had an immune response or infection is pretty close to reported case rate.
On the other hand, we have the example of antibody testing was done in the boroughs of New York City. On August 18, 2020 it was reported that 477,000 Queens residents had been tested for COVID antibodies, and that 28% of the tests showed the presence of these antibodies indicating previous infection or exposure. In addition, the Bronx had even higher prevalence, at about a third of those tested.
On yet a third “hand,” the esteemed scientific journal The Daily Mail, reported on 18 Sep 2020 that a London based population was tested and 6 % of those tested found to have antibodies to COVID viruses.
Remember, antibodies are a good thing, if you aren’t sick when the test was done. These results show that antibody prevalence, which presumably reflects either exposure or prior infection varies considerably by region.
And certainly, some of the variability we saw in the charts above might be due to testing capacity. In the Ukraine, for example, at least a few years ago, people outside the bigger cities had trouble obtaining even the simplest of medical laboratory tests. This testing and occult case problem could mean that actual exposure numbers could range from 1% to 12%. Still that means that overall 90% plus of these populations have failed to catch this very well advertised virus.
In terms of looking for patterns and correlations that might impact the differences between the best and worst countries, one thing you notice is that wealth and medical spending don’t seem to correlate all that well with outcomes. In fact, much of the wealth and high spending is found on the “bad” chart, not the “good” one. The highest fatality rate chart contains eight rich countries, and two poor ones. The lowest rate chart contains two wealthy countries and eight poor ones. Although we will discuss some less obvious reasons why that is so, at least at first glance, the answer to the problem isn’t money.
Demographics are more suggestive. It has repeatedly been demonstrated that, like most respiratory viruses, this virus is hard on the aged. About ninety percent of all deaths have been in people over the age of sixty, and the likelihood of dying rises linearly with age. That is a 65-year-old who catches COVID-19 might statistically have a five percent chance of dying, and an 85 year-old a 15% chance. You can see that overall, the “bad” countries are older than the “good” ones. (Just in case this is confusing, the 90% refers to all deaths, and the five and 15 % is the individual’s chance of dying of their particular infection. 90% of old people who catch COVID-19 don’t die, but they do make up the vast majority of the people who do die.)
The lower fatality rate countries have populations with the percent of people over 60 ranging from 21 to 29 percent with an average of 25. (with rounding) The higher rate countries range from 24-30 percent with a rounded average of 26 percent. A percentage point may not seem like much, but when you are talking millions of people, it adds up. And the impact on the over 60s is not uniform, but increases with increasing age. Another factor involving aging also comes into play.
Here is a further age breakdown: the percentage of population 85 and older each country. (all estimates from 2020, except Russia which is 2017)
Highest fatality Rate List
Country percent pop 85+ Male/female ratio
Italy 3.6 .50
UK 2.6 .67
Belgium 2.9 .45
France 3.3 .52
N’lands 2.2 .57
Sweden 2.6 .53
Ireland 1.3 .63
Spain 3.4 .49
Hungary 2.0 .33
Romania 2.0 .54
Lowest Fatality Rate List
Country %population 85+ Male/female ratio
Slovakia 1.4 .40
Belarus 1.7 .21
Czechia 1.9 .36
Croatia 2.5 .32
Russia 1.4 .27
Ukraine 1.4 .27
Norway 2.0 .54
Greece 3.6 .67
Austria 2.4 .50
Serbia 1.5 .50
In the naughty list, we see again two scenarios—countries that have high numbers of oldest old people, but relatively good resources, and Hungary and Romania, which probably represent a somewhat different reality. These are countries with much less health care spending, and demographics that may reflect that, and last these country’s relative poor outcomes may also reflect some lack of resources for treatment currently as well.
And this goes back to the seeming disconnect between health spending, wealth and outcomes. If you’ve lived to be 85 in a country with less than optimal health services, you’re a tough cookie. Conditions that can be nursed along in a more resource rich environment, such as heart failure, atherosclerosis, diabetes, renal failure and so on have in your area already killed off the people who had them. Therefore, in these countries, the 85 year and up population could be healthier overall than a similar population in a country that has more resources to commit to health care.
Here’s another factor to consider. Timing.
Timing, as they say, is everything. Here is a list of the best and worst performers in terms of fatality rate—giving the date (per Bing again) that they reached at least 100, 1000 and 10,000 reported cases, all dates for 2020. Please note, that these are the dates that these cases were available to media sources, not the dates that the infections actually started, were tested, or were diagnosed.
Country 100+ 1000+ 10,000+ Time 100+-10,000+ recognized cases (days)
Italy 2/23 2/29 3/10 16
UK 3/5 3/14 3/26 21
Belgium 3/6 3/16 3/29 23
France 2/29 3/8 3/19 19
N’lands 3/6 3/15 3/29 23
Sweden 3/6 3/15 4/10 35
Ireland 3/14 3/23 4/13 30
Spain 3/4 3/7 3/15 11
Hungary 3/21 4/10 9/10 173
Romania 3/13 3/26 4/23 41
Country 100+ 1000+ 10,000+ Time 100+-10,000+ recognized cases (days)
Slovakia 3/18 4/16 N/A N/A
Belarus 3/28 4/8 4/26 29
Czechia 3/12 3/22 6/16 95
Croatia 3/19 4/3 8/30 162
Russia 3/17 3/27 4/9 23
Ukraine 3/24 4/3 4/30 37
Norway 3/6 3/15 8/20 167
Greece 3/12 3/28 8/30 171
Austria 3/9 3/16 4/1 23
Serbia 3/19 4/1 5/9 51
In the worst list, all the “rich” countries had 100 cases on or before 3/6, except for Ireland at 3/14. They all had 1000 cases on or before 3/23. In the best list only Norway had 100 cases on or by 3/6, and only Norway, Austria and Czechia reached 1000 cases by that 3/23, and Czechia was at 3/22.
Why is this important?
One of the reasons that this time difference important is that effectively reducing COVID-19 fatalities in Europe has required a significant paradigm shift in terms of the typical European health care system’s approach to lower respiratory tract infections in the elderly. All these countries have some form of government sponsored and funded health care. Therefore, spending on health care is seen as an expenditure, not a money generating part of the economy. One of the basic ways that these countries try to control costs is by limiting acute medical care for the oldest residents. Of course, this decision reflects a great many societal attitudes about the old; that they are not as valuable as the young, that there are more productive ways to spend that money, ableism, etcetera that we don’t have room to discuss here.
Sometimes these policies shade into explicit geronticide, as with the use of lethal injections given in some countries with or without patient consent, or withholding fluids and food from a hospital patient until they die in others. Sometimes, it’s a bit more covert. For example, and we will come back to this, in Sweden centrally determined health policy is that supplemental oxygen not be administered in an assisted-living type facility. They then also promote policies to discourage transfer to hospitals from such facilities. Respiratory infections in the old in Europe were seen less as something to be treated aggressively and more of “Well, it’s time to say goodbye.”
As a side note, if done rationally, i.e. by comparing outcomes for interventions or lack thereof, policies about medical care for the elderly such as these can overall be fairly benign. Although they may result in unnecessary death or suffering for some individuals, they can also reduce exposure to futile painful, debilitating and unhelpful procedures for others. In the United States, for example, medical treatment of the elderly often involves the worst of both worlds, at great expense. Elderly patients in the United States are routinely over-medicated, tortured with ineffective and dangerous hospital and ICU stays, and offered unnecessary medical screenings. And at the same time, they are subject to excess morbidity and mortality from neglect in assisted-living environments, and withholding of useful treatments “because they are too old.”
So, where timing could have a vital impact here is not so much in being able to gather more physical resources, because it’s not really enough time to do that, but more in giving care providers and providing organizations a “head’s up” that the paradigm of treatment was changing. Instead of reacting to these respiratory infections with restricted access, palliative measures, or perhaps euthanasia, it was now society’s decision that aggressive treatment was called for.
Let’s put both factors together. Here is a diagram combining both timing and percent of population 85-years-old + The percent after each country’s name is the deaths reported divided by the number of cases reported in each country as of 13 Sep 2020.
Earlier onset (100+cases date) Later onset (100+ cases after 3/15)
Oldest old 3.0%+ Italy 12.6% (2/23)
Greece 2.3% (3/13)
Spain 5.3% (3/4)
France 8.3% (2/29)
Oldest old 2.3-2.9% UK 11.5% (3/5)
Belgium 10.9% (3/6)
Sweden 6.7% (3/14)
Austria 2.4% (3/9) Croatia 1.6% (3/19)
Oldest old 1.6.-2.2 % Netherlands 7.7% (3/6) Belarus 1.0% (3/28)
Czech Republic 1.9% (3/12) Hungary 5.3% (3/21)
Romania 4.0% (3/13)
Norway 2.2% (3/6)
Oldest Old1.0-1.5% Ireland 5.8% (3/14)
Slovakia 0.7% (3/18)
Russia 1.7% (3/17)
Ukraine 2.1% (3/24)
Serbia 2.3 % (3/19)
Putting it in a diagram really can make things pop, can’t it?
First, notice the case of Greece. They had a relatively early onset, and have a very old population, and not much wealth, but they report amazingly good numbers. Unfortunately, although you will undoubtedly see media articles praising their response and so on, what’s really going on when one set of data points doesn’t match up with what we would expect from experience with usual findings, is that there is something wrong with the data. For example, in the medical field, we know, from the results from hundreds and hundreds of trials, that when you compare about 6 weeks of antidepressant use to 6 weeks of placebo treatment, about 40% of people taking the antidepressant improve and about 20% of those in the placebo arm do. If a study comes out claiming that a new antidepressant led to an 80% improvement in the treatment arm, well, that antidepressant might be the greatest thing since sliced bread, but it is more likely there is something wrong with how the study was done, and thus the data is not valid. This has, in fact, turned out to be the case with every study reporting exceptional results for antidepressants.
So, Greece’s data don’t add up. Their reported results are not credible. It may be that they are using very different criteria from most other places to identify deaths from COVID-19 infections, or they are experiencing problems with data collection.
You can also see, that Hungary, relative to their demographics and onset date, has struggled.
Also, surprisingly, Ireland. Ireland has a very young population, by Western European standards and yet their fatality rate is much higher than any country on our lists with similar population pyramid findings for the oldest old. Their onset was not particularly early as well. Quite possibly, there have been remarks made or articles published boasting that Ireland has a much lower fatality rate than the UK. It should be even lower. Given the demographics and time of onset involved, in reality, Ireland has done a relatively poor job of addressing their COVID problem.
Interesting, too, to see how many less of the oldest old the Netherlands has compared to some neighbor countries. They are known for their enthusiasm for euthanasia. Perhaps they have found a final solution to the problem of old people?
It is beyond the capacities of our magazine to perform formal regressions to determine the precise amount of correlation between population age, timing and resulting fatality rate. But even just a look at the rough data shows that these two factors, without any consideration of any kind of intervention whatsoever, have a very pronounced effect on fatalities seen.
Okay, so we’ve spent over 3,000 words to demonstrate that having less old people in your country and more time to react to COVID-19 leads to better outcomes in terms of case fatality rate. Thank you, Captain Obvious? Well, unfortunately, this thinking is not as obvious as it needs to be. It is vitally important when looking at outcomes data to consider these underlying factors and not just the showy political interventions that make the news. Otherwise, we run the risk of, again, selecting the wrong tool for the job.
It is vital to remember that not all countries are starting on a level playing field. Wealth might be a factor, but more important here is underlying demographics, and the amount of time a region had to make the paradigm shift regarding treatment, as the hardest hit segment of the population was one that traditionally receives very rationed care (at best) in European countries.
This later point, as illustrated by the case of Sweden, also highlights the importance of interventions that are targeted, and targeted to the right population. When Sweden allowed its general population to move about relatively freely, they had more “cases,” but for most of the population this was no more than an inconvenience. They had no symptoms, or a “cold” and got on with their lives. But as in other countries, the increased exposure to COVID-19 allowed by this freedom was a disaster for the vulnerable aged population, especially when combined with longstanding government policies designed to reduce their access to care.
If Sweden had had the foresight to combine their policies for the general population with a strict quarantine and immediate access to medical care for the elderly, their figures might tell a very different story. As it was, even if they swung it in the opposite direction, they were still taking a one-size-fits-all sledge hammer to the problem. Our governments and bureaucracies need to realize that no, we are not all in this together. Different people in a country have different medical situations. The sledgehammer lockdowns, which cause irreparable harm to segments of the population at very low risk for any significant complications from a particular virus, should be reserved as a last ditch “nuclear” option, rather than the default reaction they appear to be today. But…if you are going to let the low risk population keep on with their lives, then, you have to be cognizant of what’s going on in other spheres, where there are more consequences for infection, and adjust plans and policies accordingly. Basically, if you are going to use a scalpel, you are going to have to think creatively about exactly where you need to use it.
If only we could learn this, the correct lesson, from the Swedish experience/experiment.
And in order for governments and bureaucracies to make these plans effectively, they need accurate information about effects on outcomes that may not be as publicized as showy lockdowns and travel bans. In order to evaluate effectiveness, we need to consider where every country started the race, not how soon or how fast they got to the finish line.