There are still no simple answers
The ONS analysis completely ignores the statistics on coronavirus deaths that we hear about all the time, and instead simply counts the total number of deaths from whatever cause. It then calculates the excess over the average for the past five years, adjusting for each country’s age patterns. Any excess could be due to Covid-19, or the effects of lockdown, or another reason.
When ranking countries according to the accumulated excess mortality since the start of the year England comes top, with nearly 8% extra deaths over the five-year average. Spain is second (7% extra), then Scotland (5%) and Belgium (4%). Wales and Northern Ireland are fifth and eighth respectively. The UK has done badly.
The UK has not, however, experienced the highest temporary peak of excess mortality: in mid-March, Bergamo in Italy had more than nine times the normal number of deaths, which produced the kind of desperate stories that drove the campaign to protect the NHS. The peak major city in Europe was Madrid, over a week in mid-March. A month later, Birmingham had more than three times its normal death rate, and some local areas were even higher: in one week in April, Brent, in north London, had more than four times its normal number of deaths. These statistics disguise what must have been a devastating time for communities.
To those of us who obsess about these things, a small surprise was Belgium having only half the relative excess mortality of England, since the websites that make international comparisons all show it as being the country (ignoring San Marino) with the highest number of coronavirus deaths per million population. But Belgium has been very generous in labelling deaths as down to Covid-19, which has perhaps given it an even poorer record than it deserves.
It is worth noting that the problems of counting Covid-19 deaths are vividly illustrated every day, when the Public Health England dashboard releases a count for the UK; for example, 119 and 83 additional coronavirus deaths were reported last Tuesday and Wednesday. NHS England is currently experiencing fewer than 15 Covid-19 deaths a day in hospitals, but the implausibly high PHE figures for England apparently also include any of the 250,000-plus people who have ever tested positive and have gone on to die of any cause, even if completely unrelated to coronavirus.
The Department of Health and Social Care has suspended these daily figures, but they are still going on all the international sites, and presumably are being used by others to judge how things are developing in the UK. They may be giving an inappropriately negative picture, as the ONS recently reported that the total number of deaths in the UK has shown no overall excess for the past five weeks.
But when we look at where the deaths are happening it is clear that we are not back to normal: people are still staying away from hospitals and dying at home. In England and Wales there were 766 excess deaths that occurred at home in the week ending 17 July, only 29 of which were with coronavirus, whereas in hospitals 862 fewer deaths than normal were registered. So more than 100 deaths a day were happening in people’s homes that would normally happen in hospital – although this is at least a reduction from the peak of the epidemic, when there were 2,000 additional home deaths a week.
Most people would prefer to die at home, but we seem to have no idea about the quality of these deaths, and whether some of them might have been delayed if they had gone to hospital.
Why has the UK done so badly? One hint is given by the interactive map provided by the ONS report, which shows the development of excess mortality at a local level across Europe. We see strong hotspots in northern Italy and central Spain, which stay fairly localised – for example, Rome has seen no excess mortality. But it is genuinely chilling to see these extra deaths erupt fairly evenly across the whole of the UK, as the thousands of people returning from winter holidays in Spain and Italy seeded hundreds of separate outbreaks across the country. The epidemic in the UK was more widespread and went on longer than in other countries, which saw their mortality return to normal levels by May, while the UK’s excess continued well into June.
But as I said previously, it is misguided to try to attribute good or bad performance to individual causes. Sweden has done badly and ranks sixth in the league table, just behind Wales. How much of this is due to its liberal measures, avoiding a strict lockdown? And how much is due to the fact that a huge number of Swedes take winter holidays in Spain and Italy, and returned and set off outbreaks, or that (like the UK) its care homes were not properly protected? There are no simple answers.
My original comments still hold: we will need years to properly assess the effect of the epidemic and the measures taken against it. We’ve now got a league table, but as to why the UK has done so badly, the arguments will go on.
• David Spiegelhalter is chair of the Winton Centre for Risk and Evidence Communication at Cambridge University. He is the author of The Art of Statistics