Sunday, 12 September 2021

COVID World Review, Part 6: Deaths per Case


This is the sixth of my review reports on COVID-19 worldwide as at August 20th 2021. This report will cover deaths per case; calculable for all 191 reporting countries. And it is a good method, by which to judge a country’s performance against the virus. For a high deaths per case ratio may indicate a failure to find and record a significant fraction of the cases which have occurred; or a poor quality of treatment within the healthcare system; or both.

Europe

I’ll start, as before, in Europe. I’ll show two graphs for each group of countries. First, cumulative deaths per case as measured over the course of the epidemic. And second, the actual deaths per case ratios, with the numbers of cases always offset by 21 days, which is the mean length of time between a case being reported, and death supervening (if it does). At the end, I’ll show ordered lists of the percentage of cumulative deaths per case, and of the current deaths per case ratios among all the countries in the region.


The first graph shows convergence towards a roughly constant total of deaths per case for most countries (though the UK, and one or two others, have decreased). The second shows the peaks and troughs since October 2020; each peak maybe representing a new variant.


Here, in several of the countries, after the first trough the deaths per case have increased; particularly in Hungary, Slovakia and Romania. The cumulative deaths per case to date are both generally higher, and further apart from each other, than in countries to the west.


Here, the cumulative deaths per case are even wider spread, with Bosnia and Herzegovina at the top, and Cyprus as a significant outlier at the bottom.


In this group of mostly small countries, cumulative deaths per case to date are even lower than in most of the countries than in the other groups. They are all below 2%!

So, here is the ordered list of European countries by cumulative deaths per case:

All the countries near the top, except Italy, are in Eastern Europe. I’ll add Bosnia and Herzegovina, and Italy, to my list of countries deserving follow-up for their poor performance. For the other end of the scale, I’ll add Cyprus, Denmark and Finland. And the UK and Ireland, because of their significant recent improvements in cumulative deaths per case.

Here is the ordered list of current deaths per case in Europe:

There’s a strong east-to-west gradient in this list. And most of the major West and North European countries are near the bottom of the list. Iceland, Denmark, Norway, Sweden, and Finland are all Nordic countries; and Estonia is all but. Of the major non-Nordic countries, Ireland, the Netherlands, Belgium, Austria, Switzerland, the UK, Cyprus, Spain, Luxembourg, Portugal and France are all under 0.5%.

I’ll add North Macedonia to my follow-up list; that 35% ratio of current deaths per case is amazingly bad!

Americas

As I don’t find the deaths per case spaghetti graphs particularly informative, I’ll only show the cumulative deaths per case here (except for North/Central America). I’ll show the deaths per case (with a 21-day offset) in the report on the individual countries I pick for further study.

This looks much like Western Europe, except for the enormously high outlier of Mexico. Because of this outlier, I’ll show the deaths per case over time graph for this area:

Mexico (dark blue) has been way above the rest almost all the time since the very first wave!

A big spread; and big trouble in Peru and Ecuador. Venezuela is at the bottom; given the régime, I don’t believe that. I think they deserve to go on the “probable naughty boys” list!

In general, much like Eastern Europe, or South America except for Peru and Ecuador.

The top three in this list are all on my list for further follow-up. As is Grenada near the bottom, and (now) Venezuela, whose figures trigger my bullshit meter, just as Nicaragua’s and Egypt’s did in earlier reports.

Jamaica is worth adding to the follow-up list. Cuba, too: are their figures believable?

Middle East and North Africa

A wide spread, but comparable with other places; except for Syria, which is a war zone.

Except for Yemen, which is another war zone and excluded from the above graph altogether, there are far lower deaths per case here; these countries must be doing something right! Saudi Arabia is an outlier.

Comparable with other countries, except for Sudan and Egypt; both of which have had their problems in recent years.


Sudan deserves a closer look. So would Egypt, if I could believe its figures; but as I told in the previous report, I can’t do that. Qatar and Bahrain at the bottom are also worth another look.

Sub-Saharan Africa

These figures seem low in comparison to other countries, even to Western Europe. Not only does the virus seem to spread less easily in Central Africa than elsewhere; but it also seems to be less lethal per case.

Somalia is the high outlier; another war zone. Burundi, the Seychelles and Eritrea are low outliers. Otherwise, these figures are comparable with norms in other countries.

Here, Mauritius is the low outlier, and Zimbabwe the high one.

Not much to see here, except the recent upward trend in Guinea, which doesn’t look good.


Tanzania deserves another look, because its deaths per million are very low, yet its cumulative deaths per case are the second highest in Africa.

Rest of Asia


These countries divide into two subgroups: China and Taiwan, and the rest. China and Taiwan are quite close together in cumulative deaths per case. The Chinese figures are not credible, but I have no reason to believe that Taiwan’s are not. The other countries are low to middle in the “normal” range.


Two clusters here: Kazakhstan, Kyrgyzstan, Azerbaijan and Georgia the upper, Tajikistan, Uzbekistan and Mongolia the lower. Both are low by the standards of many other countries.


All are low by world standards, but Sri Lanka doesn’t look good at all.


These range from very low to middling in cumulative deaths per case to date. Again, there are recent upticks in Myanmar, Indonesia and Vietnam; probably the effect of the delta variant.



Taiwan certainly deserves another look, as does Sri Lanka. At the bottom, Laos deserves another look, as the second lowest after Singapore.

That weird figure for current deaths per case in Kazakhstan comes from an enormous negative adjustment (more than 59,000) in the number of new cases on July 24th, which caused the weekly averaged number of cases for the week centred on July 27th to go negative! At the other end of the table, Vietnam is worth adding to the list.

Australasia and Oceania


Vanuatu has had four cases and one death; which looks like sheer bad luck. All the others are within normal parameters, although I’m a bit surprised to find Australia, rather than Papua New Guinea or Fiji, in second place.



Nothing particularly remarkable here.

Top and bottom 20

Here are the lists of the top and bottom 20 countries in cumulative deaths per case:



After Vanuatu, a recipient of bad luck, I have most of the high scorers, war zones excepted, on my list already. Also the three lowest scorers, and Qatar.


There is no point showing the bottom 20, as there are many countries with no recent deaths at all.

Recent improvements in deaths per case

For deaths per case, I’ve done some calculations, like those I did for hospital and ICU occupancy per case in the fourth report of this set. For 32 countries which supplied data on hospital or ICU occupancy, I compared deaths per case (21-day offset) as at August 20th with the same figure six months previously, on February 20th. I calculated the ratio of deaths per new case now to the deaths per new case six months ago for each of these countries, then grouped the countries by this ratio, with the following results:

0%: Iceland (which had no deaths around either of the dates).

1-9%: Denmark, Ireland.

10-19%: Austria, Estonia, Germany, Netherlands, Norway, UK.

20-29%: Belgium, Hungary, Luxembourg, Slovakia, Sweden.

30-39%: France, Israel, Italy, Switzerland.

40-49%: Croatia, Czechia, Spain.

50-74%: Finland, Portugal, Poland.

75-99%: Cyprus, Lithuania, USA.

100-199%: Bulgaria, Canada, Latvia, Romania, Slovenia.

It seems that Nordic countries, and countries close to the core of Western Europe, have done best. The USA has not improved much in deaths per case over the course of six months, and Canada has got worse. In Eastern Europe, the northern countries have tended to do better than the southern ones.

The follow-up list

So, here’s the list of countries I plan to follow up on in a later report. It is culled from this report and the previous one.

Country

Reason

Peru

Highest deaths per million: Highest average excess mortality: Highest cumulative deaths per case

Hungary

High deaths per million

Brazil

High deaths per million

Bhutan

Low deaths per million

Singapore

Low deaths per million: Lowest cumulative deaths per case

Grenada

Low deaths per million: Low cumulative deaths per case

Georgia

Highest current daily deaths per million

Fiji

High current daily deaths per million

Sri Lanka

High current daily deaths per million: High cumulative deaths per case

Cuba

High current daily deaths per million: Figures may be suspect

Ecuador

High average excess mortality: High cumulative deaths per case

Mexico

High average excess mortality: High cumulative deaths per case

Norway

Lowest average excess mortality: Nordic

Iceland

Low average excess mortality: Low current excess mortality: Nordic

Finland

Low average excess mortality: Low cumulative deaths per case: Nordic

Denmark

Low average excess mortality: Low cumulative deaths per case: Nordic

Sweden

Low average excess mortality: Nordic

Japan

Low average excess mortality

Hong Kong

Low average excess mortality

Oman

Highest current excess mortality

Colombia

High current excess mortality

Liechtenstein

Lowest current excess mortality

Cyprus

Low current excess mortality: Low cumulative deaths per case: Low current deaths per case

Switzerland

Low current excess mortality: Low current deaths per case

Bosnia and Herzegovina

Highest cumulative deaths per case in region

Italy

Highest cumulative deaths per case in core of Europe

UK

Significant recent improvement in cumulative deaths per case: Low current deaths per case

Ireland

Significant recent improvement in cumulative deaths per case

North Macedonia

High current deaths per case

Venezuela

Low cumulative deaths per case: Figures may be suspect

Jamaica

Highest current deaths per case in region

Cuba

Low current deaths per case: Figures may be suspect?

Sudan

High current deaths per case

Qatar

Low current deaths per case

Bahrain

Low current deaths per case

Tanzania

High deaths per case for the region: Figures may be suspect

Taiwan

Highest cumulative deaths per case in region

Sri Lanka

High cumulative deaths per case for region

Laos

Low cumulative deaths per case

Vietnam

High current deaths per case

That’s 40 countries to be followed up in detail; more than a fifth of the 191. To this list, I’ll add any more countries which I find are worth another look because of their record on testing or lockdowns. My plan now is to report on the testing and lockdown data first, then to follow up on all the individual countries in a final report. That will also give me the opportunity to update the countries in my list with another month or so of data.


Saturday, 4 September 2021

COVID-19 World Review, Part 5: Deaths and Excess Mortality

This is the fifth of my review reports on COVID-19 worldwide as at August 20th 2021. This report will cover deaths attributed to COVID per million population, and excess mortality (from all causes), in the 191 countries which have reported COVID cases.

Excess Mortality

Deaths per million from COVID are reported daily by all countries which have reported at least one COVID case. Excess mortality figures, on the other hand, are only reported by a subset of 89 countries – about 47% of the total. Excess mortality in a period is defined as the number of deaths from all causes in that period, divided by the average number of deaths from all causes in the same period over a number of years which together define a reference period, then converted to a percentage, and 100% subtracted from the result. The reference period used for the Our World in Data feed is the years 2015 to 2019 inclusive. The reporting period for excess mortality figures can vary from country to country, but it is usually either a week or a month; though it can occasionally change, or there may be gaps.

Thus an excess mortality of, say, 100% in a period means that the deaths in the period were twice the average level for the same time of year over the course of the reference period. An excess mortality of minus 5% means that deaths in the period were 95% of the average level for the reference period.

One advantage of excess mortality as a metric is that it is calculated completely independently from COVID statistics such as cases, tests and deaths. It is also, in most countries, calculated by a separate governmental function from the COVID statistics. Thus, where provided, it should enable inter-country comparisons without regard to variations in parameters such as COVID test kit availability, health system quality, or propensity to fudge COVID figures in particular countries.

My “magic spreadsheets” can give me two different measures of excess mortality in a country. One, the current excess mortality, is simply the latest excess mortality figure which has been reported for that country. The other, the average excess mortality, is the simple average of all the excess mortality figures over the course of the epidemic. Unlike lockdown stringencies, I make no attempt to weight excess mortality figures by how many days each figure covers.

Europe

I’ll start, as usual, in Europe. Here, I’ll show daily deaths per million (weekly averaged) and average excess mortality graphs next to each other for each of my four groups of European countries. Many of the deaths per million graphs are similarly shaped, at least over relatively short periods, to the cases per million graphs in my second report, offset by an average of about 21 days (which is the mean time between a case being reported and death supervening). However, deaths per million are also affected by the ratio of deaths per new case, which can vary at different times of the epidemic, even within one country. I plan to discuss deaths per case in the next (sixth) report.


You can pick out some of the same features on both graphs, such as early peaks in Spain and Belgium, and somewhat later peaks in Portugal and the UK. The excess mortality also shows clearly the decline in the general level of mortality through 2021 in many of the countries.


These graphs both show quite clearly that in most of these countries, the epidemic didn’t really “get going” until October 2020. And that since then, mortality peaks have been declining for most of the time.


Here again, it looks as if the epidemic started spreading in Eastern Europe only when the “first wave” in Western Europe was over. The recent uptick in deaths doesn’t show in the excess mortality, because many countries’ figures haven’t been calculated yet.


Because these are mostly small countries, the deaths per million graph doesn’t show anything very much. The excess mortality graph is far more instructive, particularly the downward slope of the peaks.

The four excess mortality graphs, taken together, show that the countries hit hardest at the peaks – Spain, Belgium, the UK, Poland, Czechia, North Macedonia, Bulgaria, Albania, San Marino, Andorra, Liechtenstein – have all suffered excess mortalities above 100% at some point in the epidemic. And Hungary had two large waves, both of which were not far under 100%. So much for those who try to make out that COVID is no more lethal than a ’flu!

Americas


Unlike in Europe, different countries don’t move together here. But Mexico, El Salvador and Panama have both been above 100% excess mortality; Panama almost to 200. And Mexico twice over.


Excess mortality in South America seems to be far worse than further north. Perhaps they got a more lethal variant than anyone else? Or, perhaps, their health care systems are worse than those of other countries? Ecuador, Bolivia and Peru have all been over 250%, and Peru has had four peaks. Paraguay and Colombia, too, have been over 100%.


Deaths per million in the West Indies are comparable with North America; lower than South. And the excess mortality data for both Cuba and Jamaica trails off before the end of 2020.

Middle East and North Africa


Peaks of deaths per million here are close to those in North America, except for Lebanon. And Armenia suffered a big peak in excess mortality near the end of 2020.


These graphs tend to suggest that COVID is still building up in this region! Israel, Jordan, Palestine, Israel again, Jordan again, Bahrain and Oman have had significant peaks in deaths. And even in Israel, which has done a lot of vaccinating, the epidemic isn’t over yet.


Those rising deaths in Tunisia don’t look very good. But Egypt looks worse. Their daily deaths per million have never been above 1, and their reported cases have only reached 0.28% of the population. Yet they have twice had excess mortality above 70%? My bullshit meter fires immediately.

The Egyptian deaths monitoring system is reputed to be as good as any in Africa; so, I suspect the excess mortality figures may be correct, and they have been missing a lot of COVID cases. And deaths, too. Even the media are starting to notice this: https://www.dw.com/en/authoritarian-states-obscuring-covid-death-tolls-study-shows/a-58771888.

Rest of Asia


Both the deaths and excess mortality figures (apart from one excursion in Hong Kong) here are way lower than in Europe or the Americas. But the excess mortality data doesn’t yet cover the most interesting phases of the epidemic in this region. And which is the only country in this group which isn’t providing excess mortality figures? You’ve guessed it. China. Ah, China.


That big peak in excess mortality in Azerbaijan doesn’t really tally with the small number of deaths per million compared with the countries around it. Another authoritarian state fudging COVID data, perhaps? And Georgia’s data peters out at the end of 2020.

There are no excess mortality figures for any of these countries. So, my best guess is that the epidemic still has a long way to run in this part of the world, except in the Maldives.


On the daily deaths per million curves, I get the same feeling here as I did for South Asia. Except, perhaps, in the case of Singapore, which seems to have done a very good job so far in controlling the virus; only 1.13% of the population have become cases, and there have only been 8 deaths per million. The excess mortality data here shows up the relative lack of COVID deaths in this region.

Sub-Saharan Africa


It’s a pity the Seychelles excess mortality figures peter out at the end of 2020, since the epidemic there didn’t really start until 2021!


Mauritius’ excess mortality data has the same problem as the Seychelles did; it ends before the period of greatest interest (from June 2021) begins.

In short, there really isn’t much to see in this part of the world.

Australasia and Oceania


In common with other relatively remote places, the epidemic in this part of the world doesn’t yet seem to have got its boots on. The Australian and New Zealand excess mortality data is interesting, though; it gives an idea of what the excess mortality elsewhere might have been without COVID! And of a “natural variability,” of the order of plus or minus around 10% to 12%. Note that mortality has tended to be lower in the local winter and higher in the local summer than it was during the reference period 2015 to 2019.

Top and bottom 20

Here are the lists of the top and bottom 20 countries in cumulative deaths per million:


This suggests some countries which may well be worth a closer look. Peru and Hungary, obviously. Brazil will be interesting, too. Bhutan, Singapore and Grenada are three (so far) well-performing countries, which I haven’t looked at before; and since none of them are in Africa, I can probably believe their data.

Here are the lists of top daily deaths per million currently – the “hot spots” of COVID mortality as of August 20th 2021:

Geographically, that’s a bit of a mixed bag; but Georgia, Fiji, Sri Lanka and Cuba seem worth a closer look.

There’s no point giving the bottom 20, because each week there are many countries recording no deaths at all in the week. So, I’ll go on to the top and bottom 20 in average excess mortality:

You can see, right there, which parts of the world have been hit hardest by the virus over the course of the epidemic. Eight of the top 10 are in Central or South America! And they include Nicaragua, which has by a long way the lowest reported case counts in the region. The Egyptians are not the only régime to fudge the COVID figures, only to be caught out by their very own deaths recording system!

I'll add Ecuador and Mexico to my list, too.

The Seychelles figures are so out-of-date that they are misleading. But Norway? Negative excess mortality, when considered over the epidemic as a whole? They will have to go on my list for a closer look. And note that Iceland, Denmark, Finland and Sweden are in the same bottom 20! Japan and Hong Kong are also worth putting on the list, as being two of the places nearest to China whose data is likely to be reliable.

And here are the lists of the top and bottom 20 in current excess mortality (from among those reporting):


Oman, Bolivia and Colombia are worth adding to the list, too.

Those numbers for Liechtenstein, Iceland and Switzerland are amazing! They will all have to go on the list for further detail. It’s interesting to see Norway, Sweden and Finland all joining Iceland in the bottom 20. It’s also heartening to see that Hungary, Israel, Belgium, and Czechia, four countries which have been hit particularly hard by the virus, now have significantly negative excess mortality figures. That gives a warm feeling that, for these countries at least, herd immunity may now not be so far away.

I see that I have added so many countries to my list, that I do not have space to include more detail in this particular report. I also expect to find (at least) a few more countries needing more investigation as a result of looking at the deaths per case statistics. I now therefore plan to do these follow-ups as a separate report, following the deaths per case report.

To sum up

South and Central America have been hit hardest by the COVID-19 virus, so far. Both in terms of deaths per million, and of excess mortality. Europe is in second place, and the Middle East and North America vie for third.

But in those places like Africa, Asia and Australasia, which initially dealt well with the virus, it looks as if it’s now starting to get going.

Hearteningly, some of the countries worst hit by the virus now have negative excess mortality. This is something you would expect to be a harbinger of approaching herd immunity.