Dear Mr Smalley
ONS Mortality Statistics
Thank you for contacting the UK Statistics Authority with your concerns about the methodology used in mortality publications published by the Office for National Statistics. I am writing to you from the Office for Statistics Regulation (OSR), the independent regulatory arm of the UK Statistics Authority. At OSR, we don’t publish statistics ourselves and are separate from the ONS.
We have looked into the methodology used in ONS’s mortality publications. You are correct to say that all ONS mortality publications use the date that the death is registered, not the date that the death has occurred. We have spoken to ONS about the reasons for this, and they are as follows:
Death Registrations
Firstly, by way of background: for a death to be registered, the cause of death must be determined. In the majority of cases, this is done by a doctor, who will have seen the individual in the last two weeks of their life. At present, 74% of deaths in England and Wales are registered by doctors. The remaining deaths are registered with a coroner. Registration by a coroner typically takes place if the cause or the circumstance of death is uncertain, with a post-mortem sometimes being used to determine cause of death. As such, deaths registered by coroners are more likely to be subject to registration delays than those registered by doctors.
You are right to say that registration delays increased in 2020, as a result of the pandemic. This happened for two reasons. Firstly, there were more deaths than average in 2020, caused by COVID-19. Secondly, there were delays in coroner-registered deaths. This meant that in 2020, the median time between death occurring and death registration increased from four days in 2019 to 5 days in 2020. ONS has not yet published its analysis of registration delays for 2021, but Coroners statistics for 2021 indicate that the number of deaths registered by a coroner dropped to the lowest number since 1995.
Deaths resulting from COVID-19 are some of the least likely causes of deaths to be subject to registration delays, with the majority of these deaths being registered within one week in 2020 (86.5%). The rapid reporting of deaths due to COVID-19 occurred because there was a specific need for this data, as an early indicator of deaths during the pandemic. The diagnostic criteria for a death resulting from COVID-19 (a positive test within a number of days of the death) also allowed for a low level of registration delays.
As indicated in an analysis by ONS, deaths that are most often subject to registration delays are those resulting from external causes (ICD-10 chapter 20). Deaths in this category include suicide and drug and alcohol related deaths. Registrations of deaths from these causes may take months or years to be registered, particularly if an inquest is required.
Date of Recording and Date of Occurrence
At present ONS publishes mortality data by the date that the death is registered. When a death is recorded, the General Register Office (GRO) sends information on the date of registration and the date of death. All cases have a date of recording and the majority have a date of occurrence as well. However, in some cases the date of occurrence is missing, such as cases when the exact date of occurrence is unknown.
The primary reason for ONS publishing data using the date the death is registered is that it allows for the creation of a static mortality dataset that is comparable over time. If the date of death was to be used, historic mortality publications would need to be updated whenever a new death is registered. If mortality delays are considered, this would mean that some mortality publications would continue to be updated for potentially over a year after the date of death. Using the date of registration mitigates this issue. It ensures that data are complete for a given time period.
ONS does not receive notification of a death until it is registered with the GRO. Because of this, it would not be possible for ONS to publish real-time data as you suggested in your email.
Our view
Prior to the pandemic, ONS produced a series of mortality publications on a regular basis. ONS has since increased the number of regular publications on deaths, as well as publishing ad hoc publications when needed. We are satisfied that these publications meet the needs of users of these statistics. We have attached a table below with a breakdown of mortality publications.
As a result of the COVID-19 pandemic, we have seen an increase in interest about statistics on mortality. Because of this, we have been conducting a Compliance Check on ONS’s mortality statistics, focusing on compliance with the Quality pillar of our Code of Practice for Statistics. We are aiming to publish the findings of this review in early December 2022. As part of our Compliance Check, we will be asking ONS to make clear some of the reasons for the current methodology, such as its decision to publish data by the date the death was registered, not the date that the death occurred.
Thank you again for contacting us about this issue and also for your patience as we have investigated your concerns.
Yours sincerely,
Ed Humpherson
Director General for Regulation
Related links
Ed Humpherson to Joel Smalley: Response to ONS Mortality Statistics (30 January 2023)