expert reaction to ONS data on deaths registered weekly in England and Wales, provisional: week ending 11 June 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 11 June 2021.


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

“The latest provisional figures from ONS on death registrations in England and Wales take the data forward to the week 5-11 June.  That’s the week after the Spring Bank Holiday, and that will have caused some minor disruption to the figures.  The reason is that registrations are lower than average in Bank Holiday weeks, for example because registry offices aren’t open so much, but the deaths need to be registered eventually, and so registration figures tend to be slightly higher than average in weeks after Bank Holidays.  It would be better to use figures classified by the date the person died (‘by occurrence’ in ONS language) rather than the date the death was registered, but for the most recent weeks, those figures will be incomplete because not all the deaths are registered yet.  The upshot is that comparing the latest week’s figures, by date of registration, with the figures from the previous week won’t give a very good picture of the real trend, in weeks after bank holiday weeks.

“Because of these Bank Holiday complications, I thought that the number of deaths involving Covid-19 for the most recent week might have been higher than for the week before, whatever the real trend is.  But the excellent news in this bulletin is that number of deaths involving Covid-19 actually fell again.  There were 84 Covid-related deaths in the week ending 11 June, compared to 98 in the previous (Bank Holiday) week and 95 in the week before that (which would not have been affected by the Bank Holiday).  The numbers are down in most age groups and most regions, comparing the latest week with either the previous week or the week before that.  (There are a few exceptions to that pattern, but they aren’t concerning – when the numbers are this low you’d expect a bit of statistical irregularity just by chance.)  Deaths involving Covid-19 made up just 8 out of every thousand total deaths, in the most recent week.  Not all those Covid-related deaths were classified as having Covid-19 as the underlying cause (as always).  For the latest week, about four in five of them were (66 of them).  The number where Covid-19 was the underlying cause is slightly up on the previous week (66 compared to 57), but that could be a Bank Holiday effect, and the latest figure is lower than the number for the week before the Bank Holiday (70).

“A little caution is needed about these figures for Covid-19 deaths, encouraging as they are.  The daily figures for deaths within 28 days of a positive test, as shown (for example) on the dashboard at, have been showing some signs of a slow increase, which has continued after the period covered by the ONS registrations data.  But that increase isn’t entirely clear at present, and the definition used on the ONS data – Covid-19 mentioned on the death certificate (so involving the judgment of a doctor) – is more appropriate than the 28-day definition, which could include the death of someone who had tested positive but wasn’t ill with Covid-19 and died of something unrelated.  But the numbers of newly infected people have been growing in England and in Wales since mid-May, as shown by the counts of new confirmed cases on the dashboard and by results from the ONS infection survey.  It’s not good news that the number of infections is increasing, even though is a little encouraging that the increase in cases hasn’t yet led to any substantial increase in deaths involving Covid (and, judging by these ONS figures, if anything there’s a decrease in deaths).  If, sadly, someone dies from Covid-19, that will on average happen somewhere round three weeks after infection.  So people who died in the most recent week in the ONS registration data would probably have been infected around the middle of May, allowing for a short delay in death registration.  The rise in new cases driven by the Delta variant didn’t really get going until about then, so it is still rather early to use these registration figures to assess what its effect on deaths might be.  Similarly any increase linked to step 3 of the England roadmap on 17 May also wouldn’t have happened in time to show up in the latest registration figures.  Overall, the position is positive so far, but that could change in the next week or two (by which time the effect of the Bank Holiday will also have finished).  I don’t think a large increase in deaths is inevitable, given the very encouraging recent data from PHE on the effectiveness of vaccines against hospitalisations involving the Delta variant.  But I’d be surprised if there is not some sort of increase, sadly – and we mustn’t forget that death is not the only very unpleasant possible consequence of a Covid-19 infection.  The disease itself can be very unpleasant and distressing, and in some people the consequences (long Covid) can last a very long time.

“A negative feature of these death registration figures is an increase in the number of death registrations from all causes.  The number for the latest week (ending 11 June) is 31% higher than the week before.  That large increase is probably another Bank Holiday effect – but the latest number is also 6% higher than two weeks previously, before the holiday.  It is also a small amount (2%) higher than the average for the corresponding weeks in the five years 2015-19, and all those corresponding weeks would also have been the week after the Spring bank holiday.  It’s difficult to say what the reason for this increase might be.  It’s certainly not because of deaths directly involving Covid-19, because there were so few of those in the latest week.  To investigate further would require more detailed data on other causes of death, and that’s not yet available.  But it may be relevant that the change in deaths from all causes, compared to two weeks previously (before the Bank Holiday) seems to be concentrated most in older age groups.  Deaths in people aged under 50 were down by 3% over that two-week period, while deaths in those aged 50 and over were up by 7%.  Maybe this is just some kind of short-term blip, though.  We’ll maybe know more next week.”



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



Declared interests

None received.