WintonCentre
Published in

WintonCentre

COVID and ‘excess deaths’ in the week ending April 10th

The Office for National Statistics (ONS) released the latest data on death registrations at 9.30am on April 21st, covering the week ending April 10th (week 15). Headline statistics include

  • 18,516 deaths recorded, which is an excess of 7,996 deaths over the 5-year average of 10,520 for that week.
  • 6,213 (78%) of these excess deaths have COVID-19, which means that (22%) did not — much lower than the proportion (41%) in the previous week.
  • 34% of all registered deaths had COVID-19 mentioned on their death certificate. This proportion was highest in London (53%) and lowest in the South-West (19%)
  • 58% of deaths occurring in hospitals had COVID-19 on their death certificate, compared to 17% in care homes , and 8% of deaths at home.

When confronting data, it’s good to stand back and look at the bigger context, before plunging into more details. The excess deaths can be put into perspective by examining the trend starting in October 2019.

We see there has been around 5,000 fewer deaths than normal registered up to March 16th 2020, although there was a similar excess in the first part of the winter. In general the mild winter and limited seasonal flu has, up to recently, produced an undramatic profile. The upward trend appears to have started after 6th March, before deaths from COVID-19 started to be registered, suggesting an early impact of the virus.

Around ~1800 extra non-COVID deaths were registered in the week ending April 10th, although this was fewer than the ~2500 extra non-COVID deaths in the week before. This raises many important questions.

What is causing all these non-COVID excess deaths?

There are three broad categories, which I shall label A B C, that could be contributing to excess deaths that did not get recognised as having COVID-19 as a confirmed or suspected cause.

A. Highly vulnerable people, who had a mild infection, but which was sufficient to lead to their death without any obvious symptoms of COVID-19.

B. Those in which some symptoms had been apparent, but the certifying doctor was reluctant to put COVID-19 on the certificate without further evidence — few patients outside hospital will have been tested, and new regulations mean the certifying doctor does not have to have seen the patient recently. But there has been recent encouragement to put ‘suspected COVID-19’ on the death certificate, and this could have led to a reduction in the number in this category.

C. Deaths of people who have not been infected, but whose normal medical treatment has been disrupted, for example by reluctance to attend hospital in spite of illness that would normally warrant a referral or attendance at A&E.

The relative magnitude of these categories is currently unknown, although there are many strong opinions being voiced in response to a tweet I put out about this last week.

I am not going to try and guess the proportions of groups A, B and C, but will look at what is suggested by looking at the deaths. A similar analysis has been carried out for Scotland by @rogue_wee, and in a recent BMJ data briefing by John Appleby, but with no clear resolution.

Where are these deaths occurring?

ONS now provide a breakdown of the place of death for weeks 11 to 15 (March 7th to April 10th): the figure below uses week 11 as a baseline, as 5-year averages are not available.

This clearly shows that the bulk of extra deaths are not being registered as COVID-19 in care homes (67%) and at home (77%): the details are provided in the Table below.

Excess deaths in England and Wales in Week 15, broken down by place of death.

In summary, there were 1,630 extra non-COVID deaths in care homes, and 1,062 at home. But there was a deficit of 1,349 in hospitals. Some of these may be deaths that normally would have occurred, but since the patient has the virus then they were coded as COVID — deaths with the virus rather than from the virus. But the strong suggestion is that there has been a substantial transfer of non-COVID deaths from hospitals into the community — there has been a 29% drop in hospital attendances in March. If these are just people who would have died anyway, but are staying at home or care homes, then that’s not necessarily a bad thing, as they can be in familiar circumstances and have better access to known health-care workers and family. But presumably some of these could have had their lives extended if they had gone to hospital.

Are there particular categories experiencing excess deaths?

We can now dive into more detail, looking at the excess deaths shown in the final column of Figure 1, broken down by age and sex. Five-year average deaths for age-groups are supplied by ONS , which I have averaged over Weeks 13,14,15. Only the 45+ age groups are drawn: these comprise the overwhelming majority 74/6213 (99%) of the COVID deaths, and excess deaths are unreliably estimated for younger age-groups.

The overall proportion of the excess that is made up of non-COVID-labelled deaths is 22%, but the following figure shows systematic deviations from this average: a formal statistical test rejects the hypothesis of equal proportions.

There is a clear tendency for more of the excess mortality to be labelled as non-COVID for both younger, and older, people. For older deaths, this could be since they are more likely to have been residents of care homes, while the figures for younger people are based on low numbers. But I am sure there are other explanations.

John Appleby has identified that deaths from ischaemic heart disease, cerebrovascular and chronic lower respiratory diseases were lower than usual, suggesting some patients may have had COVID used as an alternative cause of death.

Conclusion

It is impossible from the currently available data to determine the reasons for the substantial spike in excess deaths that do not have COVID on the death certificate. More detailed data on care home deaths, and changes in hospital admissions, might go some way to resolving the magnitude of groups A, B and C. But the major shift of non-COVID deaths from hospital to community deaths points to a substantial impact of the current lockdown on the outcomes for elderly people.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
David Spiegelhalter

David Spiegelhalter

4.2K Followers

Statistician, communicator about evidence, risk, probability, chance, uncertainty, etc. Chair, Winton Centre for Risk and Evidence Communication, Cambridge.