30th April 2020
On The Guardian
David Spiegelhalter is a statistician and professor of the public understanding of risk
It’s tempting to try to construct a league table, but we’ll have to wait months, if not years, for the true picture
At prime minister’s questions on Wednesday, Keir Starmer said he had added up a total of 27,241 coronavirus deaths so far, leaving the UK “possibly on track to have the worst death rate in Europe”.
Is he right? Unfortunately, measuring the impact of the virus is a fiendishly complex task. It’s nothing like keeping score in a game. Starmer, as a lawyer, would know that we have to define our terms carefully. And so, assuming we want to make a comparison based on death rates, we first need to decide what a death rate is.
You would think it would be easy for a bean-counting statistician to count deaths – the one certain thing (apart from taxes). But it is remarkably difficult. I have stopped taking much notice of the number given out at the daily press conferences, as it is only based on reports from hospitals, oscillates wildly around weekends, and recently included deaths that occurred a month ago. And this week the number of UK deaths jumped up by nearly 5,000 to 26,097 in one day – rather close to Starmer’s count – by retrospectively including non-hospital deaths that had tested positive for the virus.
But even this is too low, as it does not include the many deaths of people who were not tested. The Office for National Statistics data on death registrations is the last word, although inevitably delayed by around 10 days, and these figures would be expected to take the current total to significantly more than 30,000. But we should be very cautious in comparing even this uncertain total with those of other countries.