select search filters
roundups & rapid reactions
before the headlines
Fiona fox's blog

expert reaction to new ONS stats on deaths registered weekly in England and Wales, provisional: 24 September 2021

The Office for National Statistics (ONS) have released provisional counts of the number of deaths registered in England and Wales in the week ending 24 September 2021.


Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This week’s release of provisional data on death registrations from ONS takes the information forward to the week 18-24 September. As always, the bulletin concentrates on detailed data from England and Wales, because those are the countries where ONS is responsible for processing and analysing the registration data, but (also as usual) it includes some more limited information from Scotland and Northern Ireland, where the main responsibility for the data is with the devolved administrations.

“This week I’ll start with death registrations from all causes of death (and come onto Covid-related deaths later). For the whole UK, the number of deaths from any cause registered in the most recent week was 12,247, which is 16% above the average figure for the corresponding week in the five years 2015-19. So there were excess deaths that week, as has been the position for many weeks now. In England and Wales, deaths from all causes have been above the five-year average for three months now, since early July.

“For England and Wales, the number of registered all-cause deaths fell slightly in the latest week compared to the week before, from 11,009 to 10,684. For the whole UK, the number also fell slightly. But, given the fact that these figures are based on when deaths were registered and not when the people actually died, it’s probably safest not to interpret small trends and instead to say that the picture hasn’t changed much in recent weeks. (ONS do publish figures on the basis of when the deaths occurred rather than when they were registered, but these are incomplete for the most recent weeks because some deaths won’t have been registered yet.) These excess deaths aren’t directly because of Covid. For the whole UK, there were 1,688 more deaths registered in the latest week than the five-year average, but the number of registered deaths involving Covid-19 that week was 1,108, so almost 600 of those excess deaths didn’t directly involve Covid-19. However, we don’t yet know what caused them instead, because these weekly bulletins give little information on causes of death other than Covid-19.

“It’s still the case, as it has been since May last year, that the number of registered deaths in people’s own homes in England and Wales is considerably above the 2015-19 five-year average. For the latest week, there were 840 more registered deaths at home than the five-year average, but only 74 of those deaths involved Covid-19. Roughly speaking, in England and Wales there have been about 100 excess deaths a day at home, not involving Covid-19, for well over a year. Despite some analyses by ONS and others, it’s still not clear (to me at any rate) exactly why this is occurring, or what the quality of the end of life care for the people involved actually is.

“Turning now to deaths where Covid-19 is mentioned on the death certificate, the total number for the UK for the most recent week was about 6% higher than the previous week (1,108 for 18-24 September compared to 1,049 the previous week). The numbers were up on the previous week in three of the four UK countries, but down by 13% in Northern Ireland. The change in England was very small (2% increase), but the changes in Scotland (22% increase) and particularly Wales (up by a third, 33% on the week before) were proportionally larger. I should say that it’s important not to over-interpret these short-term trends, particularly in the smaller UK countries. The numbers I’ve given are counts of registrations. In Wales, we’re talking about just 10 or 12 deaths a day, and in Northern Ireland, about 8 a day. A few day’s difference in when someone sadly dies, or even a couple of days difference in when a death was registered, could make a noticeable difference to the trends there, without really telling us about the underlying patterns.

“But it’s still not encouraging that the numbers rose for the UK as a whole and for three of the four UK nations, and it’s also not encouraging that there were considerably more than 1,000 registered deaths involving Covid-19 in the UK as a whole. It’s true that Covid-19 was not the underlying cause for all those deaths, though it would not be mentioned on the death certificate unless the registering doctor believed that if had an influence on the death (for example, that it might have speeded death up, and/or made the deceased person’s last days more unpleasant or painful). The ONS bulletin does point out that, for England and Wales where more detailed numbers are available. Covid-19 was the underlying cause of death for about 6 in every 7 of the deaths where it was mentioned on the death certificate. Obviously numbers of deaths from Covid-19 are well below where they were back in January and February. But they are currently at around the level from the last half of March this year, which isn’t a comfortable place to be, in my view.

“On the dashboard at, the numbers of deaths within 28 days of a positive Covid-19 test, for the whole UK, were roughly level for the first half of September but have been generally falling slowly since then. (Here I’m looking at the figures analysed by date the person died, not when the death was reported to the dashboard system.)  That does, very roughly, match the trends in these ONS death registration figures. Since the ONS figures only go up to the week ending 24 September, and since they are for registrations (that do not occur immediately on death), I wouldn’t expect them yet to have picked up the falling trend in the dashboard figures from mid-September, assuming that trend does really reflect what’s going on. Maybe the ONS registration figures for Covid-related deaths will start to fall next week, but we’ll have to wait and see.

“Because of the statistical variabilities in counts of Covid-related deaths in the smaller UK countries, I wouldn’t necessarily have expected the short-term trends in the dashboard numbers for Wales or Northern Ireland to match the trends in the ONS registration figures anyway, but very roughly they do match. In Scotland, they do also match – the dashboard figures were rising until late September, and though there are signs of a fall on the dashboard after that, it’s too early for that to have been picked up in the ONS registration numbers.

“In England, where the numbers are higher because of the much larger population than the other countries, one might expect the trends to look more similar. However, the dashboard 7-day averages have been falling since, roughly, the week 2-10 September, and that isn’t really reflected in the ONS registration figures, which have broadly shown continuing increases since then, taking into account disruption to registrations from the August bank holiday (which affect the following week too). Maybe this is because of delays in registration – and in any case these upward or downward trends are pretty slow in both directions. The figure from ONS for death registrations involving Covid-19 in the most recent available week was 2% higher than the previous week, a pretty small change. For the same two weeks, the dashboard count of deaths within 28 days of a positive test fell by about 9%. Give all the uncertainties, I certainly don’t think a discrepancy of this size is any cause for concern – really it just emphasises that we can’t know exactly what’s going on in the very recent past. Things should become clearer in subsequent weeks.”



All our previous output on this subject can be seen at this weblink:



Declared interests

Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee.  I am also a member of the Public Data Advisory Group, which provides expert advice to the Cabinet Office on aspects of public understanding of data during the pandemic.  My quote above is in my capacity as an independent professional statistician.”

in this section

filter RoundUps by year

search by tag