One arguably unhelpful part of the Coronavirus crisis has been the media’s insistence on comparing international death rates.
Yesterday, some deemed the UK the “worst-hit European country” after the death toll in England and Wales reached 29,648.
It is a staggering, horrible figure – reminding us of the cruelty of this virus.
It’s not, however, an excuse to abandon statistical nuance – given we are dealing with enormously complicated data.
As Dominic Raab said in yesterday’s daily press briefing, it is too early to come to conclusions.
Professor Chris Whitty has explained that “all-cause mortality, age adjusted, and possibly seasonally adjusted” would be the most important measure – sentiment echoed by other Governmental advisers.
This is not them trying to avoid criticism, nor is it an ideological position.
There are many reasons direct comparisons between countries should be resisted, which ConservativeHome has previously touched upon.
To elaborate, here are some of them:
The first obvious factor that needs to be accounted for when it comes to analysing death rates is population size; the more people, the more cases of Covid-19.
Population sizes vary hugely across Europe – and, indeed, the world.
Russia and Germany have the highest (145 million and 83 million, respectively).
The UK follows with 67 million people.
This is why it might be fairer to measure deaths per hundred thousand or million, which help to normalise differences in population size.
If analysts use the latter metric, as of today Belgium has the worst death rate, followed by Spain, Italy and then the UK.
Even more important than population size is population density.
The more people living closely together, the more higher transmissions of the virus and deaths.
The UK, for example, has one of the highest population densities in Europe, with 727 people per square mile, and 83 per cent living in urban areas.
Germany has a population density of 623 people per square mile and the Republic of Ireland has 186 people per square mile.
So it’s unreasonable to compare them in the same way.
Furthermore, some countries have more capacity to socially distance than others.
That makes it far easier for citizens to self-isolate, even if the Government does not insist on it.
These nuances have to be controlled for in analyses.
It is incredibly important to understand demographic variables when understanding differences between countries.
We know for instance that the Coronavirus disproportionately affects black, Asian and minority ethnic people, men and older people.
Though we do not know the reasons why there are correlations between these groups and Covid-19, they need to be deciphered – and considered in analyses of death rates – to understand why some countries are more severely affected than others.
Countries that have older populations, for example, will suffer much worse, as the average age of deaths for Covid-19 is above 80.
The median age in Italy is 47, compared to Ireland (37), so this explains part of the reason we see big disparities in their Covid-19 death rates.
We need to explore these interactions to make valid assessments.
Covid-19 is more severe for people with certain underlying health conditions, such as diabetes, heart disease and high blood pressure.
Obesity, too, has been recognised as a big risk factor.
A study by the Coronavirus Clinical Characterisation Consortium has shown that almost 75 per cent of Covid-19 patients in intensive care are overweight – with death rates 37 per cent higher for obese patients.
That means that countries with higher rates of these will suffer worse when hit by Covid-19.
This, too, needs to be understood when assessing death rates across countries.
Countries have different testing regimes
Countries have different tests for Covid-19, and they also have different quantities of tests, and this affects how their death rate looks – particularly if it’s measured as “case-fatality ratio”.
Using this metric, Germany’s death rate appears much less severe than others.
This is because of its mass testing regime, which meant it detected much milder cases of Covid-19 – bringing the fatality part of the case-fatality ratio down.
The UK, on the other hand, had – up until a few days ago – a much smaller testing regime, mainly measuring hospital incidences.
This brings the fatality figure up, as it means only the most severely affected people (hence why they’re in hospital) get tested.
The more testing a country does, generally, the more it can get a sense of cases of Covid-19 compared to deaths from it.
This, in turn, can allow governments to understand if their strategy in regards to Covid-19 was worse than others’.
Differences between the ways deaths are recorded
Countries also have differences in the ways that they record deaths.
Some of these are set out in Politico, which shows a whole variety of approaches.
There are countries with narrow definitions, others with a mixed approach and the rest have a broad definition.
Austria has a “narrow definition”; it only counts “deaths of patients who have died ‘from” COVID-19 and have been reported in the country’s epidemiological reporting system”.
The same goes for the Netherlands, which only counts the deaths of patients “who tested positive for the disease and died in hospitals”.
As of April 29, the UK has included “deaths in all settings where there’s a positive COVID-19 test, for example, in care homes.”
Generally, this area has been quite opaque – in terms of understanding how different governments record figures, but it needs to be deciphered to make assessments.
Differences between healthcare systems
In years to come – God willing that the Coronavirus crisis will end – one suspects that people will become far more interested in international healthcare systems, and what we can learn from them.
Indeed, there is already much interest in South Korea and Germany, as they have been praised for controlling the virus well.
One advantage Germany has is 621 acute care hospital beds available per hundred thousand people (the second highest number in the whole of Europe).
Differences between healthcare systems will no doubt have an impact on death rates, and in due course, analysts will want to study them – in order to learn most of all.
The speed at which data is collected
In a Twitter post Nick Stripe, Head of Health Analytics at the UK’s Foreign Office for National Statistics, has warned against making comparisons between countries’ death rates, because some are quicker – or slower – at reporting data.
He said the UK death registrations data “is the fastest, most frequent and most in depth than any other stats agency”.
This should be considered in light of a recent study by EuroMOMO, a network of epidemiologists, who collect weekly reports on deaths from all causes in 24 European countries – and concluded the UK has the worst score.
While excess mortality is a good metric – “the gap between the total number of people who died from any cause, and the historical average for the same place and time of year” – it could also be affected by lags in reporting Covid-19 deaths.
Until we know that every nation is up to speed on reporting figures, no one should make pronouncements on death rates.
Another tragedy of this pandemic is that it’s caused an increase in non-Covid-19 deaths; The Guardian reports there were an extra 3,312 of these in the week ending April 24.
Some speculate that these were caused by delayed hospital visits for other life-threatening conditions, economic hardship, and/ or mental illness.
We need to fully understand this data to make any assessment of the UK Government’s approach, and others’.
There are even more variables we may not know about now
This is just an initial list in understanding what variables play into death rates.
It may be that critics of the Government are completely right to be angry. Perhaps lockdown should have come sooner; perhaps it did not do enough to shield vulnerable people and so forth.
But no one can make any assessments without doing a proper analysis. And that means stripping back data so you know what you’re measuring. Lockdown and other interventions (or lack of interventions) have to be isolated to see their impact.
As the Government says, all-cause mortality, adjusted by age, will ultimately be the best way to access this – as it’s the only measurement that can be properly compared across countries – all of which are dealing with and recording Covid-19.
Until then, sensational assumptions help no one.