Artillery Row

Coronavirus: are international comparisons helpful?

And do Europe’s small countries give us the most accurate statistics?

There will be no Olympic Games medal table this summer. Our quadrennial snapshot of international sporting virility must wait for another year.

The innocuous pleasure of such comparison has, meanwhile, been replaced with a grizzly chart of international vulnerability. Now, we have the unwanted competition of countries, differentiated by coloured lines snaking daily up a graph of coronavirus victims. China, the early pacesetter, has surrendered its lead to Italy and Spain. The UK finds itself jostling in the middle of a chasing, predominantly European, pack.

When the 2016 Olympic Games came to their close, a statistician promoted an alternative medal table. It showed that the most successful nations were not the United States, the UK and China as displayed by total medals won, but the Bahamas and Grenada (on the basis of medals per head of population). If we do likewise for coronavirus incidence it would show China hardly registering as having a problem compared to the most blighted nations of Europe. It might also highlight something interesting that is missed when looking at the total number of deaths in absolute terms.

There are limitations whichever way the statistics are tabulated. International death comparisons are not exactly comparable because there is no uniform definition of what constitutes a death from coronavirus. Some countries attribute the cause of death to the underlying health condition to which covid-19 provided the fatal boost, rather than to the virus itself. But, while noting this variation, there is still sufficient similarity in attribution for international comparisons to have value.

Attempting international comparisons of the incidence of confirmed cases risks considerably greater hazard. The more a country tests, the higher the likelihood of carriers being identified. What may therefore be measured is not the incidence of the virus, but the ability of a country’s healthcare system to keep tabs on it.

This could explain the low incidence of positive cases in several large population countries in Southeast Asia. For example, in Vietnam only 200 cases and no deaths from coronavirus have been recorded. Is this down, as my Vietnamese friends assure me, to the timely and precautionary planning of their Socialist Republic? It would indeed be a remarkable achievement if a country of 95 million people whose capital city is scarcely 100 miles from the Chinese border had fewer cases than San Marino (population 33,000) which is a distant and highly indirect 5,300 miles away from Wuhan. It is so remarkable that it is incredible.

To state this is not to fire a cheap shot at Vietnam, or any of the other Asian countries that inexplicably have far lower confirmed cases than Europe. The number of positive tests being recorded in most major European countries (Germany’s high testing rate being an exception) may also reveal a fraction of the real number of carriers.

On 31 March the number of tests in the UK passed the 143,000 mark (with just over 25,000 testing positive). That means only 0.2 percent of the British population has so far been tested. What is more, the testing that has taken place is skewered towards those most at risk or who are already showing symptoms. It is thus far from being a representative sample.

only 0.2 percent of the British population has so far been tested.

Keeping this in mind, it is therefore tempting to disregard an international league table of confirmed cases. Nevertheless, take a look at which states make up the top ten countries by coronavirus incidence, as measured as confirmed cases per one million of population:

  1. Vatican City 7,491
  2. San Marino 6,774
  3. Andorra 4,789
  4. Luxembourg 3,479
  5. Faeroe Islands 3,459
  6. Iceland 3,326
  7. Gibraltar 2,048
  8. Spain 2,019
  9. Switzerland 1,870
  10. Liechtenstein 1,783

Of course this is all a bit like claiming the Rio Olympics were won by the Bahamas. But two common factors leap out from this table. Firstly, all top ten countries are in Europe. Secondly, that – with the exception of Spain (46.6 million inhabitants) and Switzerland (8.5 million) – they are all either micro-states or very small countries. Are you really at far greater risk of contracting the coronavirus if you’re trying to sit it out in the Faeroe Islands than almost anywhere in China outside Wuhan?

this is all a bit like claiming the Rio Olympics were won by the Bahamas.

Certainly you are at greater risk if you are in the Holy See, but the Vatican City is so miniscule that just six cases can be made to look significant. For sensible statistical analysis, the state that comes top should therefore be excluded. Most of the others though are, whilst very small, big enough to produce an interesting sample.

If all these tiny countries were densely populated city-states and international travel hubs, then we might conclude that the combination of cheek-by-jowl living and air miles accumulation made them especially vulnerable. Yet except (arguably) for Gibraltar, none of these small states are high-density metropolises with over-worked air traffic controllers. Indeed, city-states that do meet that criteria, like Hong Kong (ranked 81, with an incidence of 95 cases per million inhabitants) and Singapore (ranked 61; incidence of 158), are doing relatively well at keeping the virus at bay despite their greater geographical and other connections to mainland China.

So, it is hard to identify any plausible reason for why European mini-countries should proportionately have been more exposed to coronavirus than larger countries.

A far more likely explanation is that comprehensive monitoring of a population for viruses is easier to accomplish in very small countries. There are tiny distances to cover, a high ratio of healthcare facilities to population, and equally good per head ratios of resources, wealth and comparably little ignorance or compliance avoidance.

Therefore, whilst the recorded positive test rate in large countries may not accurately capture the true picture nationwide, in very small countries there is a reasonable likelihood of a significantly greater correlation between the official incidence rate and the actual rate.

Germany’s testing is vastly more comprehensive than France’s. Yet, their recorded incidence rate is virtually the same.

If the UK’s true rate was the same as Luxembourg (which went into lockdown a week earlier than the UK) then instead of 0.04 percent of the British population being currently diagnosed with having (or recovered from) covid-19, the actual figure would exceed 0.3 percent. In absolute numbers this is the difference between about 27,000 and 230,000 cases. It should be noted that even if the latter was the actual rate, it is far below the levels of infection upon which high ‘herd immunity’ hypotheses like that from Oxford University are modelled.

Indeed, extrapolating from small European countries to large European countries is highly problematic if done without reference to other factors. Sometimes the incidence rate between big and small is in harmony rather than conflict. Tiny Gibraltar has almost the same incidence rate per head as Spain. The more easily monitored territory of The Rock transpires to be no different than its vast neighbour. To take another example, San Marino’s rate greatly exceeds the Italian national rate, but is not so out-of-kilter with the high incidence in the Italian provinces that neighbour it.

Indeed, the more the variation is studied between European countries of all sizes and proximity to one another, the less clear-cut any assumption becomes. Germany’s testing is vastly more comprehensive than France’s. Yet, its recorded incidence rate (Germany 814 per million; France 799 per million) is virtually the same. So, perhaps the rate of testing does not shine such a penetrating light upon darkness as we are now relentlessly told that it must?

Nor is the varying proportion of elderly people necessarily the determinant. Italy’s high death rate at over 11 percent of covid-19 cases is often explained as a consequence of it having Europe’s highest proportion of elderly people. This may be part of the explanation, but it can be over-stated. The Italian median age is 45.4. In Germany the median age is 45.9. Yet, Germany’s death rate is 1 percent. The difference in median age is insufficient by itself to explain this expanse of gulf in death rates. To complicate matters further, at 40.5, the British median age is far below that of Germany, but the death rate in the UK is currently running at 7 percent.

In this battle of the rival statistics it is still early days. And nothing, convincingly, adds up. Beware, therefore, those who claim your future is already in the stats.

Enjoying The Critic online? It's even better in print

Try three issues of Britain’s newest magazine for £5

Critic magazine cover