select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
The Office for National Statistics (ONS) have released provisional counts of the number of deaths registered in England and Wales in the week ending 29 January 2021.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“There is a glimmer of hope in the provisional data on death registrations in England and Wales for the week ending 29 January. The total number of deaths from all causes registered in that week fell, compared to the previous week. That’s the first time that there’s been a fall in the weekly registration figures this year. While this is encouraging, most other aspects continue to look grim. The number of deaths fell by only 228 out of over 18,000 – that’s a fall of just 1.2%. The total number of registered deaths that week, 18,448, is the fifth highest weekly total since the start of the pandemic. (The previous week’s figure was slightly higher, and the last three weeks in April last year, at the height of the first wave when the pandemic was out of control in care homes, were higher.) Registered deaths are 45% higher than the average number in that week for 2015-2019, and 56% higher than the number for 2020 (before the pandemic had taken hold here). That’s a very large number of excess deaths. And it’s the average over the whole of England and Wales. The position varies from one region or nation to another, with the position being particularly bad in London, where deaths from all causes in the latest week were very close to double the five-year average figure. In the East of England and the South-East, the numbers were about two-thirds higher than the five-year average, and deaths were substantially above average in all the other English regions except the South-West, and in Wales. Even in South-West England they were 8% above average. If deaths really began to fall in the week ending 29 January, they have a very long way to go.
“Really the whole position feels to me a bit like the experience of climbing a mountain. You get to the top of a ridge and you can see the summit in the distance – or is it really the summit and not just some other higher ridge looming out of the mist? And if nothing else, you can see that the summit is still a long way above where you are now. We have a long, long way still to go on Covid deaths. Vaccination will help a lot, and is going well, but that takes some time to have an effect. It’s not like we’re just coming upon a chairlift that will take us to where we want to be quickly and without effort.
“As always, a caveat about these figures is that they are based on the dates when deaths were registered, not when they actually occurred. However, we’re now well clear of the Christmas holiday period when the pattern and timing of registrations is disrupted, so that the trends in the registration counts probably reflect the trends in the numbers of deaths that actually occurred reasonably closely. ONS also provide figures classified by the date of occurrence of death, but these counts are always too low for the latest week available, because of delays in registering deaths. They will be revised upwards in future bulletins – that’s an important reason why ONS describe these numbers as ‘provisional’. ONS estimate that the number of deaths that actually occurred in the week ending 29 January was somewhere between 16,039 and 20,000 – if it turns out to be towards the upper end of that range, then total deaths will have gone up in the latest week, not down. Because of these uncertainties and the very small size of the decrease in registrations, it’s probably most accurate to say that the rising total number of deaths that we’ve seen for the past few weeks has levelled off, rather than being confident that the number is falling.
“What’s also depressing is that the number of registered deaths for which Covid-19 was mentioned on the death certificate was (very slightly) higher in the week ending 29 January than the previous week, at 8,433, or about 1,200 a day. ONS report that 46% of all deaths, from all causes, in that week involved Covid-19. For some of the deaths with Covid-19 on the certificate, the coronavirus was not the underlying cause – but it was the underlying cause for more than 9 out of every 10 of them. That means that 41% – just over 2 in every 5 – of all deaths registered in England and Wales that week had Covid-19 as their underlying cause. It’s true that most of the deaths involving Covid-19 were of older people, but 1 in 8 of the victims were under 65 years old. Everyone is at some risk.
“Is this the sort of trend we might expect to see, given what else is known about trends in the pandemic? After all, these new ONS registration figures are for a week whose midpoint was two weeks ago now, and we’ve had more data on deaths for more recent dates on the dashboard at coronavirus.data.gov.uk and in daily news reports. (Not much more recent though – the deaths classified by the date they occurred, on the dashboard, are shown as reliable only up to 2 February, because the numbers for very recent days are not yet complete.) There are some features of today’s ONS numbers that do match what we know from other sources, but there are other features that don’t entirely match. If we have indeed now reached the peak numbers of registered deaths, with total deaths and deaths involving Covid-19 both levelling off, that would match the pattern on the dashboard. The peak on the dashboard was on about 18 or 19 January, which is in the ONS week 16-22 January, the week before the latest one in the ONS data. Since then the death counts on the dashboard have been decreasing. Given that the peak in new cases and new infections was at around the turn of the year, this all makes sense. If, sadly, someone dies from Covid-19, that would typically be around two or three weeks after they first had symptoms, so the peak in deaths would be around two or three weeks after the peak in new cases, and that’s what we see.
“But what’s different is that the decline in numbers of deaths on the dashboard, since the peak at 18-19 January, is quite rapid. Using the main definition of a Covid-19 death used on the dashboard, Covid deaths in England and Wales fell by about 13% between the week ending 22 January and the week ending 29 January. However, death registrations involving Covid-19 did not fall at all over that period – they remained roughly level, and the same is true for deaths where Covid-19 was given as the underlying cause. (Both of those numbers in fact rose very, very slightly, by 11 deaths for the total number involving Covid-19, and by 18 deaths for those with Covid as the underlying cause.) I can’t yet be clear on the reason for this difference between the trends from the two different data sources. Covid-19 deaths on the dashboard are actually deaths of anyone who died within 28 days of a positive test, even if they died of a reason unrelated to Covid, but they exclude anyone who lived 29 days or more after their positive test. In my view, the counts of registrations are a more appropriate and accurate way of counting deaths from Covid-19 (or indeed any other cause), because they depend on the careful judgement of the doctor who fills in the certificate of cause of death, who can take into account all the relevant information and not simply the fact of a positive test and a somewhat arbitrary number of days between the test and the death. But delays in registration and in reports of death can muddy the waters when we’re looking at these very short-term trends – the position may become clearer when we have more data, and revisions to this week’s data, in the ONS bulletins next week and in later weeks.
“I would very much hope, and indeed expect, the numbers of registered deaths, in total and from Covid-19, to fall in next week’s ONS bulletin compared to this one. But the position is still tricky and I can’t be sure.”
Prof Sir David Spiegelhalter, Chair, Winton Centre for Risk and Evidence Communication, University of Cambridge, said:
“Deaths registered in the week ending January 29th were 45% higher than the five-year average. But this is combined with extraordinarily low levels of deaths that are not caused by COVID-19, the lowest for at least 5 years. There is almost no flu circulating, and sadly many vulnerable older people, who would have survived until now, have already had their lives shortened.
“Deaths occurring in people’s homes are 50% above their normal levels, with the minority involving COVID-19. This is a major shift in the pattern of death in this country.
“Up to 29th January, ONS report that over 126,000 deaths involving COVID-19 in the UK, more than half of which have occurred in the second wave since September. So by now the total will be over 135,000.
“The daily dashboard shows a clear and rapid decline in deaths involving COVID-19 since January 19th, and so next week we should see this reflected in death registrations.”
Prof Rowland Kao, the Sir Timothy O’Shea Professor of Veterinary Epidemiology and Data Science, University of Edinburgh, said:
“The majority of COVID-19 related deaths occur within two weeks of a positive test, however there remain a considerable number that occur for some weeks after. Thus the deaths we are seeing are at least in part due to the very large numbers of infections we saw that continued into January. The fact that the absolute numbers are on the decline mirrors the decline in cases we have seen over the last few weeks, and is good news in that it is further evidence that the epidemic of COVID-19 in England and Wales appears to be on the decline under the current measures. Deaths will decline more slowly than might have been expected prior to the dominance of the “Kent” B.1.1.7 variant, for which the evidence is that it has a higher rate of fatalities for those infected. This, and the still high numbers of deaths means that continued vigilance needs to applied. However, especially as the current vaccines appear to have continued protection against severe disease, we can expect the numbers of deaths to continue to decline, so long as other non-pharmaceutical restrictions remains sufficient.”
All our previous output on this subject can be seen at this weblink:
www.sciencemediacentre.org/tag/covid-19
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.”
Prof Sir David Spiegelhalter: “DJS is a paid non-executive director of the UK Statistics Authority, that oversees the work of the ONS.”
None others received.