The Office for National Statistics (ONS) have released provisional counts of the number of deaths registered in England and Wales in the week ending 22 January 2021.
Dr Jason Oke, Senior Statistician at the Nuffield Department of Primary Care Health Sciences, University of Oxford, said:
“Provisional numbers of deaths registered in England and Wales for the week ending the 22nd January reflect the continuing impact of the pandemic on mortality. At 8,422 deaths, this is the second highest weekly number recorded during the pandemic but it is not the second highest in terms of excess deaths. This is because unlike the first wave when there was also an excess of non-Covid-19 deaths, deaths from other causes in the second wave have been significantly lower than expected for the time of year. In term of excess deaths, there were 5,460 (41%) more deaths registered this week than the five-year average and this is markedly lower than the excess death seen at the height of the first wave (75 – 110%).”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“Today’s provisional data from ONS on death registrations in England and Wales takes the information up to the week ending 22 January. The position for the latest week isn’t as good as my best hope from last week, but nor is it as bad as my worst reasonable fear. Superficially, one depressing part of today’s figures is that, not only did the total number of deaths registered in the latest week go up compared to the previous week (by 634 deaths), but the rise was a little faster than last week’s rise compared to the week before that. The latest week’s figure is 3.5% higher than the previous week – last week’s figure was 1.6% higher than the week before that. In fact this particular issue isn’t as depressing as it sounds, for reasons I’ll explain later. What’s more worrying is that the number of deaths where Covid-19 is mentioned on the death certificate rose again, quite substantially. The number of those deaths was 8,422, which is 1,177 more than the previous week – that’s up by 16% in a week. The only positive I can see in these very large figures is that the rise is a bit slower than the previous week’s figures, when deaths involving Covid-19 were nearly 20% up on the week before that. Over 90% of these deaths involving Covid-19 actually had Covid-19 recorded as the underlying cause of death – these aren’t people who just happened to die of something unrelated to Covid-19 after a positive test. Even for the other 10% or so, where Covid-19 was mentioned on the certificate but not as the underlying cause of death, the certifying doctor would have considered than Covid-19 contributed in some way to their death, though not as the main cause. The number of excess deaths – the difference between the number of deaths in the latest week and the average number for 2015-19 – is up as well.
“With these continuing very large increases in deaths involving Covid-19, why are total deaths from all causes not rising so quickly? That will be largely because the reductions in social contacts, because of the lockdown restrictions, reduce the transmission of a lot of respiratory and other infectious diseases too, and those diseases are usually important causes of death at this time of year.
“So why are deaths from Covid-19 continuing to increase so fast in the registration data, when (for example) coronavirus deaths reported on the dashboard at coronavirus.data.gov.uk have been falling recently? One important point is that the main daily figures on the dashboard and in news reports are for people who died, from any cause, within 28 days of having a positive test for Covid-19. The great majority of those people are likely to have had their death actually caused by Covid-19, but some will not. Fundamentally the counts from death registration are more accurate. But the main reason for an apparent difference in the trends is simply a question of timing. For England and Wales (together), the number of deaths (using the 28-day definition) for the week ending 22 January was 8,091. For the week ending 15 January it was 7,497*. So the number of deaths went up by 8% between those weeks. That’s not as big an increase as the increase in registrations mentioning Covid-19, but it’s still a substantial increase. On the 28-day definition, the peak of numbers of deaths in England and Wales was in the week ending 22 January, the latest week in the ONS registration figures. That corresponds to what one might expect from previous figures on new confirmed cases – the peak of confirmed cases was around New Year’s Day, and, in the sad event that someone dies of the horrible disease, that would on average happen two or three weeks after they were first recorded as a confirmed case. The numbers of deaths on the dashboard on the 28-day definition have been falling since the middle of the week ending 22 January. If the death registrations continue to be broadly in line with those figures, and if the relationship between confirmed cases and deaths two or three weeks later continues to hold, we might expect to see death registrations involving Covid-19 falling in next week’s bulletin. It’s just about possible that vaccinations might begin to have an affect on the figures too. But none of this is certain, and in any case the figures for Covid-19 deaths on all the definitions are still alarmingly high – we have a very long way still to go.
“The regional picture mostly isn’t positive either. Total deaths, and deaths involving Covid-19, were higher in every region of England in the latest week, than the week before, though both of these counts were a little lower in Wales than the previous week. Deaths in every English region, and in Wales, were above the 5-year average – particularly in London where they were over double the five-year average. There is perhaps a sign of hope – in the three regions on the south and east of England (London, South East, East of England) where tight restrictions were imposed before the national lockdown, there are signs that the rate of increase of deaths is slowing, and in Wales, where very tight restriction were also imposed earlier than in much of England, death registrations fell slightly.
“I said I’d explain why the higher rise in total deaths for the latest week compared to the week before (3.5%), than the position in last week’s bulletin (1.6%), isn’t as depressing as it sounds at first, if you dig a bit deeper. The rise in the number of deaths registered, reported last week for the week ending 15 January, compared to the previous week ending 8 January, was considerably lower than the true underlying figure, because the registrations in the week ending 8 January were high, because quite a lot of deaths that happened in the week or two before that were registered late because of the Christmas and New Year holidays. Data are also available classified by the date when the deaths actually occurred, not when they were registered. Those show a much greater rise in deaths from all causes between the week ending 8 January and the week ending 15 January – a rise of nearly 11%, not 1.6%. (The data by the date of occurrence of death are, pretty obviously, much more clearly related to what’s actually going on in the country, because they are affected very much less by things like public holidays and how quickly a death registration can be done.)
“This couldn’t be known from last week’s bulletin – the problem is that deaths do take time to register, at any time of the year, so that the count of deaths the occurred in the latest week available is always quite considerably too low because quite a lot of deaths that happened that week will not have been registered yet. These figures are revised every week as more registrations come in – that’s why ONS are careful to call these numbers ‘provisional’. The number of deaths recorded in this week’s bulletin as having occurred in the weeks ending 8 and 15 January will doubtless be revised upwards again in future bulletins, but the revisions are unlikely to be huge. This issue of incomplete registrations means that I can’t say anything very solid about the count of deaths that occurred in the latest week (ending 22 January) from this week’s bulletin – they are bound to be revised upwards quite a lot next week as more deaths that happened in that week are registered.”
* These are the total daily counts for the days in the weeks in question. Note that these are not what you would get by multiplying the 7-day average counts for those days, as given on the coronavirus.dat.gov.uk dashboard, by 7. That is because those 7-day averages are not for the 7 days ending on the day on question, but for a 7-day period with the day in question in the middle of it. So, for instance, the 7-day average for England for 22 January is given as 1041.3 on the dashboard, and this is the average of the daily totals for the 7 days from 19 to 25 January inclusive. I mention this because I’ve seen it wrongly reported in the media quite frequently, and also, I admit, because I got this wrong myself a few weeks back.
Prof Sir David Spiegelhalter, Chair, Winton Centre for Risk and Evidence Communication, University of Cambridge, said:
“The latest ONS data reveals that over 117,000 deaths involving Covid had occurred by 22nd January in the UK, so by now there must be well over 127,000.
“7,592 deaths with Covid as the primary underlying cause were registered in England and Wales in the week ending January 22nd. And yet only there were only 5,586 excess over the five-year average. So this means there was a substantial deficit of around 2,000 non-Covid deaths, that’s around a sixth of the 12,000 that usually occur. This positive finding will be partly due to social distancing almost eliminating flu, but also because so many vulnerable people, who would have died this winter, have already been part of the 75,000 excess deaths in 2020.
“The major shift towards deaths in private homes continues to grow: normally we would expect around 2,900, but 4,300 were registered, nearly 50% extra. Around 900 of these extra deaths did not involve Covid.
“Over 1,800 deaths with Covid were reported in care homes – even at the peak of the first wave, there were only three weeks in which this figure was exceeded.”
Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:
“Provisional counts for COVID-mention deaths by date of occurrence are key data. As evidence mounts that death-rate may be increased for those infected by the Variant of Concern, it is still too early to know if the peak of COVID-mention deaths in England and Wales has been reached in the week ending 15 January 2021 or if there is worse to come in the subsequent two weeks.
“My table shows COVID-mention deaths in England & Wales (by week of occurrence) from early September 2020 to 22 January 2021. The festive season’s perturbation of registration-delays has now shaken out and I can again use a) the pattern of registration-delays and b) fortnightly rates of change (currently increase) in the occurrence of COVID-mention deaths to estimate likely totals for COVID-mention deaths in the most recent weeks of occurrence.
“In England and Wales, not all deaths which occurred in the week ending 22 January 2021 have yet been registered. Taking registration delay into account and the most recent fortnightly rate of increase, sadly we can expect that 8,900 to 9,900 COVID-mention deaths may have occurred in England & Wales in the week ended 22 January 2020, see Table. If so, then COVID-mention deaths in the week ended 22 January 2021 would have topped the peak weekly count in wave 1 in England & Wales.”
All our previous output on this subject can be seen at this weblink:
www.sciencemediacentre.org/tag/covid-19
Declared interests
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.