David Spiegelhalter's Guardian piece on COVID-19 statistics
I have not been to impressed by much of what has appeared in the UK press in general of late, and the Guardian is one (though for from the only one) of the papers which has published a lot of nonsense.
But I do want to comment them for publishing an article by a statistician and professor of the public understanding of risk called David Spiegelhalter, which is entitled
Coronavirus deaths: how does Britain compare with other countries?
and which gives a well-informed, intelligent and accessible explanation of why we should be careful not to jump to conclusions (or to think we have justification for what we always wanted to believe) based on incomplete or misunderstood data.
It is not an article defending the government: it is an article which looks at some of the questions we would need to ask and to which we need to understand the answers before we can make meaningful comparisons of data from different countries.
After discussing UK estimates of deaths linked to Coronavirus, and why they are probably too low, he says of the ONS data which is probably most complete that
"We should be very cautious in comparing even this uncertain total with those of other countries."
As he points out,
"Every country has different ways of recording Covid-19 deaths: the large number of deaths in care homes have not featured in Spain’s statistics – which, like the UK’s require a positive test result. The numbers may be useful for looking at trends, but they are not reliable indicators for comparing the absolute levels. If we were naive enough to take the counts at face value, the new figures propelled the UK past France and Spain into second place in Europe behind Italy, which is not encouraging because we are behind Italy in terms of what stage of the epidemic we are at.
A more equitable metric might be Covid-19 deaths per million. Ignoring tiny countries, our current score of 388 puts us fourth, behind Belgium (632), Spain (509) and Italy (452)."
"But these are still deeply unreliable numbers, as it is not clear if we should just be looking at Covid-19-labelled deaths anyway.
"Many feel that excess deaths give a truer picture of the impact of an epidemic. The latest ONS figures reveal that in the week ending 17 April, the week after Easter, around 22,500 deaths were registered in England and Wales, compared with an average of around 10,500 a week at this time of year. Nearly 9,000 of these had Covid-19 on their death certificate, but that leaves 3,000 extra deaths – more than 400 a day – essentially unaccounted for. Many of these will be indirect effects from the disrupted health service: routine treatments have been largely abandoned, A and E attendances and non-Covid-19 hospital admissions have plummeted."
"But, of course, people are not so interested in the numbers themselves – they want to say why they are so high, and ascribe blame. But if it’s difficult to rank this country, it’s even trickier to give reasons for our position."
"Covid-19 mainly harms the elderly, with the average age of deaths above 80, and its fatality rate doubles every seven years as a person ages. Italy’s population is elderly (it has a median age of 47), while Ireland’s is much younger (a median age of 37), so we would expect different effects. And Covid-19 is a disease of crowded areas – New York is rather different from Reykjavik. An obsessive comparison is being made between Norway and Sweden: Sweden’s more relaxed social distancing policies may or may not have been instrumental in their current death rate being 233 per million, compared with Norway’s 38."
"Will we ever know the direct and indirect health effects of the epidemic, taking into account reduced road accidents, the benefits of reduced pollution, the effects of recession and so on?"
"My cold, statistical approach is to wait until the end of the year, and the years after that, when we can count the excess deaths. Until then, this grim contest won’t produce any league tables we can rely on."
These are extracts from the piece, you can read the whole article here.
But I do want to comment them for publishing an article by a statistician and professor of the public understanding of risk called David Spiegelhalter, which is entitled
Coronavirus deaths: how does Britain compare with other countries?
and which gives a well-informed, intelligent and accessible explanation of why we should be careful not to jump to conclusions (or to think we have justification for what we always wanted to believe) based on incomplete or misunderstood data.
It is not an article defending the government: it is an article which looks at some of the questions we would need to ask and to which we need to understand the answers before we can make meaningful comparisons of data from different countries.
After discussing UK estimates of deaths linked to Coronavirus, and why they are probably too low, he says of the ONS data which is probably most complete that
"We should be very cautious in comparing even this uncertain total with those of other countries."
As he points out,
"Every country has different ways of recording Covid-19 deaths: the large number of deaths in care homes have not featured in Spain’s statistics – which, like the UK’s require a positive test result. The numbers may be useful for looking at trends, but they are not reliable indicators for comparing the absolute levels. If we were naive enough to take the counts at face value, the new figures propelled the UK past France and Spain into second place in Europe behind Italy, which is not encouraging because we are behind Italy in terms of what stage of the epidemic we are at.
A more equitable metric might be Covid-19 deaths per million. Ignoring tiny countries, our current score of 388 puts us fourth, behind Belgium (632), Spain (509) and Italy (452)."
"But these are still deeply unreliable numbers, as it is not clear if we should just be looking at Covid-19-labelled deaths anyway.
"Many feel that excess deaths give a truer picture of the impact of an epidemic. The latest ONS figures reveal that in the week ending 17 April, the week after Easter, around 22,500 deaths were registered in England and Wales, compared with an average of around 10,500 a week at this time of year. Nearly 9,000 of these had Covid-19 on their death certificate, but that leaves 3,000 extra deaths – more than 400 a day – essentially unaccounted for. Many of these will be indirect effects from the disrupted health service: routine treatments have been largely abandoned, A and E attendances and non-Covid-19 hospital admissions have plummeted."
"But, of course, people are not so interested in the numbers themselves – they want to say why they are so high, and ascribe blame. But if it’s difficult to rank this country, it’s even trickier to give reasons for our position."
"Covid-19 mainly harms the elderly, with the average age of deaths above 80, and its fatality rate doubles every seven years as a person ages. Italy’s population is elderly (it has a median age of 47), while Ireland’s is much younger (a median age of 37), so we would expect different effects. And Covid-19 is a disease of crowded areas – New York is rather different from Reykjavik. An obsessive comparison is being made between Norway and Sweden: Sweden’s more relaxed social distancing policies may or may not have been instrumental in their current death rate being 233 per million, compared with Norway’s 38."
"Will we ever know the direct and indirect health effects of the epidemic, taking into account reduced road accidents, the benefits of reduced pollution, the effects of recession and so on?"
"My cold, statistical approach is to wait until the end of the year, and the years after that, when we can count the excess deaths. Until then, this grim contest won’t produce any league tables we can rely on."
These are extracts from the piece, you can read the whole article here.
Comments