I have been gripped by the World Oddometer’s daily reports tracking the covid statistics. In a way it was a rejection of Stalin’s ‘one death is a tragedy; a million is a statistic’; I have watched with agony the paths of various countries.
But this statistician knew that the data reflected reported deaths from covid, and that there is considerable variation among countries’ diligence at collecting their statistics and definitions vary too..
A way around these problems is to use ‘excess death rates’ which is to look at the deaths above past trends. Until recently nobody did the exercise systematically and comprehensively enough to give a reasonable overview.
However, last week The Economist published figures for about sixty countries. There are some disappointing omissions – particularly China and India*.
It turns out that many countries are woefully under-reporting their covid deaths. But surprisingly, the excess deaths in some countries are substantially below the reported covid deaths and in some case (including New Zealand) they are negative. I shall not comment on the latter because it is outside my statistical skills; it is easy enough to think of reasons but one has no idea about each effect’s magnitude. My focus here is on the statistics..
Here is a tabulation of the top 25 countries ranked by their excess death rates but also showing their reported covid death rates and the ranking. (Note that the method/tabulation assumes that counties accurately count their dead; alas not always true. I have vague memories of grumpings about Brasil.)
One conclusion is that countries we tend to track – the UK and the US – report more rigorously than others. So they report 4th and 7th on the covid death measure, but on the excess death measure they are 17th and 23rd, a salutary reminder that for all the pain we have shared with them, there are countries who are suffering even more.
Excess and Reported Covid Death Rates to 18 February 18 (rates per 100,000)
Country* / Excess Death Rate/ Ranking/ Reported Death Rate/ Ranking
Bosnia & Ha 5100 1 1518 8
Bulgaria 2720 2 1415 13
Peru 2610 3 1338 15
Lithuania 2580 4 1166 19
Russia 2520 5 564 41
Serbia 2330 6 492 44
Mexico 2270 7 1372 14
Romania 2060 8 1026 26
Bolivia 2030 9 964 28
Ecuador 1990 10 867 32
South Africa 1890 11 815 33
Poland 1840 12 1099 24
Slovenia 1820 13 1806 2
Montenegro 1800 49 1474 9
Belgium 1770 14 1878 1
Portugal 1700 15 1548 6
Czechia 1670 16 1764 3
UK 1600 17 1753 4
Spain 1600 18 1426 12
Italy 1570 19 1571 5
N Macedonia1460 20 1449 11
Croatia 1460 21 1320 16
Hungary 1410 22 1455 10
USA 1380 23 1521 7
Moldova 1340 24 923 29
Kyrgyzstan 1270 25 220 48
If there is a conclusion it might be that there are ‘lies, damned lies and statistics’. More constructively, a statistician always needs to be cautious with sources – particularly cross-country comparisons.
Footnote. New Zealand. The Economist estimates New Zealand’s excess deaths at minus 200 in the covid era (in contrast to plus 26 reported covid deaths). The conclusion was earlier found by a team led by Nick Wilson of the Wellington Medical School. As I recall they conjectured that the greater flu vaccination roll out, greater personal hygiene and fewer car deaths because of less motoring during the lockdown may have had a role, but as far as I know detailed evaluation is not yet published
* The other countries/regions reported in the study were as follows. (The order reported here is the rankings by excess death rate from highest to lowest. The break between South Korea and Norway is to show those countries which have negative excess death rates.)
Switzerland; Kosova; Netherlands; Sweden; Austria; Brazil; France; Latvia; Slovakia; Chile; Ukraine; Germany; Egypt; Colombia; Belarus; Luxembourg; Uzbekistan; Istanbul; Estonia; Albania; Jakarta; Israel; Malta; Greece; Canada; Oman; Montenegro; Denmark; Finland; Qatar; South Korea;
Negative Excess Deaths
Norway; Singapore; Japan; Thailand; Iceland; Costa Rica; Cyprus; Philippines; Australia; Taiwan; Mongolia; New Zealand