Yesterday a great deal of attention was paid to a Financial Times report estimating that the total number of deaths from COVID-19 is 41,000 (double the official figure of 17,337 released on Tuesday).
Editors reached the number by assessing new data from the Office for National Statistics, including deaths outside of hospitals.
The FT says that in “the week ending April 10, deaths in care homes reached 4,927, almost double the figure of 2,471 a month earlier.”
All of this will undoubtedly spark panic and confusion as to the scale of the Coronavirus pandemic – and whether it is much worse than we are being told.
As of today, the UK cumulative total of COVID-19 associated deaths in hospital is 18,100 – but many will wonder how much bigger it will get with non-hospital additions. Hence why the FT has done its own calculation.
There’s nothing wrong with the maths (although if every publication starts releasing their own calculations, it could get rather messy). The problem is that data like this is being used to push premature conclusions about the crisis, which has barely begun. It fits into wider media narrative – that the UK is the most disastrous country at managing the virus.
As ConservativeHome has recently argued, death rates – as well as other Coronavirus statistics – are much more complex than they are currently being presented, and it is much too soon to make assessments on our situation – especially in a global context.
Indeed, differences between countries are not being taken into account in analyses. The UK, for instance, has a higher population density than Germany, France and the Republic of Ireland (where a higher percentage live in rural areas).
London is also the largest city in Europe, with almost nine million people, so this is part of why it will look worse. There are other big variations, such as the average age in different countries – the older the population, the more deaths.
When we hopefully get on top of the Coronavirus crisis, what will help analysts compare death rates is “all cause mortality” – in other words, “all of the deaths that occur in a population, regardless of the cause” over a period of time – which Chris Whitty brought up in April 16’s daily briefing.
The reason this metric will become important is that it is recorded the same way across countries. The issue with current statistics on the Coronavirus is that they have been tracked differently across regions, so it’s difficult to judge them against each other.
Germany, for instance, follows the Robert Koch Institute’s guidance, which instructs medical officials to record: “Both people who died directly from the disease (“died from”) and people with previous illnesses who were infected with SARS-CoV-2 and for whom it cannot be conclusively proven what the cause of death was (“died with”)”.
Other countries, conversely, may simply record COVID-19 as cause of death if the patient has it – never mind underlying causes. Therefore, the only way we can remove these differences is to take the total amount of deaths at the end for each country.
The next complicated part will involve determining what percentage of these deaths was caused by the Coronavirus. This is yet another headache for analysts, not least because a spike in deaths can be caused by other factors than the disease.
Whitty also touched on this on April 16th, saying that: “people die for many reasons in epidemics. They include the direct cause of death, but they can also include people who died, for example, if they stayed at home because they were worried that going into hospital was dangerous for them”.
Another aspect that makes analysing death rates hard is variations in testing regimes.
Some countries, such as South Korea and Germany, have much better testing programmes than others. So they can better assess what percentage of the population died from the Coronavirus – then use this when looking at all-cause mortality. The more data, the more accurate readings on case-fatality ratio and so forth.
In essence, understanding deaths rates is an extremely difficult process. Statisticians have to standardise data between different countries, and that involves untangling lots of different variables, from demographics to environmental factors, to when the pandemic hit.
The effectiveness of Government strategies, such as lockdown, can only be looked at when these variables are “controlled” for – in other words, taken out of the equation.
It may even be the case that one would need another UK (what scientists would call a “control group”) to ever properly assess the impact of our interventions.
In the meantime, we must not jump to conclusions as quickly as much of the media has…