FOI Ref: ​FOI/2021/3218

You asked

I would like to know how many registered Covid-19 deaths are NOT caused by the virus for each month since April 2021.

Please provide the total registered Covid-19 deaths and the percentage of these not caused by Covid?

We said

​Thank you for your request.

Our mortality data is derived from information collected from the death certificate.

When a person dies, in most cases a doctor writes a medical certificate of cause of death (MCCD) which is then recorded in the death registration (at a local authority registration office).

The details are printed out as the official 'death certificate' for the next of kin. The same information is sent electronically from the registration office to ONS for us to produce statistics about causes of death.  The doctor or coroner certifying a death can record more than one health condition or event on the form. The medical certificate of cause of death has two parts, Part 1 contains the sequence of health conditions or events leading directly to death, while Part 2 can contain other health conditions that contributed to the death but were not part of the direct sequence. For statistical purposes one of the health conditions on the certificate is chosen as the 'underlying cause of death'.

The underlying cause of death is defined as the health condition or event that started the train of events leading to death and is worked out according to rules from the World Health Organisation (WHO).

If someone dies in circumstances involving an accident, violence or suspicious circumstances, the case is referred to a coroner for investigation. A post-mortem examination is carried out and usually an inquest is held. The Coroner's Court hears all the evidence and follows legal rules of evidence when deciding the causes of death.

It is extremely unlikely that a coroner would find that someone was involved in a traffic accident, or was the victim of violence, because of having COVID-19 or a positive COVID-19 test -- so they would not mention COVID-19 on the death certificate. This applies to any death caused by an accident, violence, poisoning, or other external causes. ONS Mortality data will not show deaths involving an external cause as a COVD-19 death.

We do report COVID-19 deaths where the doctor suspects COVID-19 but a test has not been completed. These are recorded using ICD-10 code U07.2.

We report these in the following monhtly publication: Monthly Mortality Analysis.

As of October 2021, analysis of the 134,375 deaths due to COVID-19, 4,139 (3.1%) were classified as "suspected" COVID-19. Including all 151,574 deaths involving COVID-19, "suspected" COVID-19 was recorded on 3.1% (4,763 deaths) of all deaths involving COVID-19 in England and Wales.

We do not hold information regarding deaths listed as a COVID-19 death that have not been caused or contributed to by the COVID-19 virus. The UK Health Security Agency (formerly PHE) have a different methodology and they do publish all deaths, including those involving external causes as a COVID-19 death, if there is a positive COVID test within 28 days of death.

With regards to the COVID-19 "within 28 days of a COVID-19 test" methodology for reporting COVID-19 deaths, ONS and UK Health Security Agency COVID-19 deaths data are produced in different ways and have different purposes:

  • The daily COVID deaths counts reported on the government's COVID-19 dashboard (produced by UKHSA) shows deaths within 28 days of a first positive laboratory-confirmed test.

  • ONS weekly death registrations data for England and Wales, relate to the week that ended 11 days prior and are based on the cause of death reported on the death certificate, thereby not restricted to only deaths that showed a positive test.

More information can be found in a  blog by Professor John Newton of Public Health England about the complexities of counting COVID-19 deaths and the different methods used.

For more information on how the UK Health Security Agency publish COVID-19 mortality statistics please contact InformationRights@UKHSA.gov.uk.