Covid-19 Analysis

I’ve been trying to put together some statistics on the severity of the outbreak of Covid-19 in a selected group of countries. Published mortality rates are pretty useless because the denominator entirely depends on the extent of the testing programme. Likewise for the ‘reported’ number of cases.

The only reasonably reliable statistics are the actual number of deaths and the total population (and even with the former there may be problems in attributing the cause of death). So the most robust way of monitoring the severity and progression of the outbreak in a particular country is to take the number of deaths per million people. The results are in the table below.

From this analysis, which I’ve not seen published anywhere else, I am amazed at just how much more severe the outbreak is in Italy, and also in Spain, than anywhere else. After that Luxembourg is one of the worst affected countries, though a very long way behind. Germany is surprisingly low, and Poland perhaps with other East European countries hardly at all. Data is taken from the Johns Hopkins website here: https://coronavirus.jhu.edu/map.html . You could easily add other countries if you wish.

It could be interesting over time to rework this statistic on say a 10 day rolling average basis. This would indicate the progression of the outbreak in different countries. It would however be a statistic with a very long time lag, since ‘new’ deaths would only decline several weeks behind new ‘actual’ cases. What we can be reasonably confident about, and is hopeful, is that on the basis of China’s experience, as analysed here: https://medium.com/@tomaspueyo/coronavirus-act-today-or-people-will-die-f4d3d9cd99ca, the number of ‘actual’ cases will be begin to decline very quickly after a lockdown is imposed. (Of course the number of ‘actual’ cases as calculated, will still be an underestimate as the numbers of asymptomatic and very mild cases that are never even logged, maybe very high indeed, particularly among children.)

Covid-19 death rate
CountryPopulation (m)Deaths to 17/3ratio to 17/3Lockdown?
Italy59250342,4Y
Spain4662313,5Y
Luxembourg0,623,3Y
France 651482,3Y
Belgium11141,3Y
UK66711,1N
Ireland4,820,4Y
Germany82270,3Y
Poland3850,1?
USA3271000,3N
China138632002,3Y/N – easing
 Korea51911,8?

Clearly much of the variation is explained by different countries being in diffeent phases of the epidemic. China, admittedly made up of several  very different regions, is almost certainly coming out the other side; deaths are now expected to decline rapidly. The same applies to Korea. The US by contrast is only just waking up to the problem, and doesn’t have the public health infrastructure to deal with it, so we can expect the numbers there to rise quite steeply. Similarly I would expect the UK to get much worse yet. But Italy and Spain, now at the height of the epidemic, already have rates several times higher than China – remember these are cumulative figures. Mobility probably explains much of Luxembourg’s high-ish rate, and three generation households is probably a factor in Italy. But why the difference between say France and Germany?

You may also be interested in the following analysis of the first 1016  deaths in Italy by age group. Again the distribution is quite surprising.

Deaths in Italy to 12 March
Age Group Number
0 to 9 0
10 to 19 0
20 to 29 0
30 to 39 2
40 to 49 4
50 to 59 25
60 to 69 80
70 to 79 362
80 plus 543
Total 1016