select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The Office for National Statistics (ONS) have released the latest statistics for deaths in England and Wales, including deaths from COVID-19 in all settings.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“There are certainly some concerning aspects of this week’s ONS data on death registrations, but they aren’t showing anything like the alarming patterns of excess deaths that we saw in April and May. More deaths were registered in England and Wales in the most recent week (ending 18 September) than the average of the previous five years, but not very many more (259 more deaths out of about 9,500, just 2.8% above the five-year average). Analysing the data by the dates when the deaths occurred rather than the date they were registered, there aren’t more deaths compared to the five-year average in the latest week, and there haven’t been since early June, apart from two single weeks (one in June, one in mid-August). The data analysed by the date of death, for the most recent week, will be affected to some extent by late registrations – they include only deaths registered on or before 26 September – but there’s no sign of a major or rapid increase in excess deaths.
“What’s concerning, though, is the rise in the number of deaths where Covid-19 is mentioned on the death certificate. Measured by the date when the death was registered, that number went up from 99 in the week ending 11 September to 139 in the week ending 18 September, a 40% increase in a week. In fact the position may be rather worse than that – it’s possible that registrations in the week ending 11 September were still slightly affected by late registrations from the late summer bank holiday. Analysing by date of occurrence of death, there were 97 Covid-related deaths in the week ending 11 September, and 154 in the week ending 18 September, a 59% increase in a week, and the number for the week of 18 September will inevitably be revised upwards a certain amount (and more than the 11 September figure) because of late registrations after 26 September. These figures are not too concerning yet, I’d argue, because the weekly numbers of Covid-related deaths were higher than these levels right up to late July or early August, so this is by no means a large spike in deaths. But the recent rise in the numbers of infections, shown by data from the ONS Infection Survey and the REACT-1 study from Imperial College, did not really get started until late August or early September. If the rise in infections is going to lead to a corresponding rise in numbers of deaths – that seems very likely but we can’t be completely certain yet – that rise in deaths mostly won’t have showed up yet. That’s because it takes time, typically several weeks, for an infected person to become ill enough to need hospitalisation and, eventually and sadly, to die. This week’s release of ONS death registration data goes up only to the middle of September. The rise in registered death with Covid on the certificate may well be the first sign of an increase linked to the increase in infections, but we’ll have to wait for a week or two’s more data to see whether and how that rise continues.
“Another concerning feature, which is longer-term, is that the number of deaths taking place at the person’s home (from any cause) continues to run at levels considerably higher than the average of the previous five years. That’s been the position right since the early days of the pandemic. From the start of July up to mid-September, in England and Wales, there have been 4,600 more deaths at home than the five-year average. Over the same period there were about 1,300 fewer deaths in care homes than average, 950 fewer in hospices, and over 9,000 fewer in hospitals. That’s a major change in the pattern of where deaths occur. Maybe this does not indicate anything problematic – maybe, to some extent at least, it’s a good thing for people to be able to spend their last days at home – but I’d like to know what the reason is for this change.”
Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:
“The Office for National Statistics has carried out important statistical work on influences, such as weekends and bank holidays, on the registration-delay for deaths in England. This work should now be applied to achieve nowcasting of COVID-mention deaths in England by occurrence-date and age-group [under 70 years, 70-79 years, 80+ years] so that registration-delays do not “delay” our appreciation of the rate of increase in UK’s second wave of COVID-mention deaths as was the case in March 2020.”
Prof Sir David Spiegelhalter, Chair of the Winton Centre for Risk and Evidence Communication, University of Cambridge, said:
“The extra deaths at home continues – around one-third more than the five-year average. This may be an overall positive development, since most people would prefer to die at home than in hospital – it depends on the care and support the family are receiving. Only 8 out of these 700 extra at-home deaths in England and Wales had Covid on the death certificate.”
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 the Advisory Committee, but my quote above is in my capacity as a professional statistician.”
None others received.