Dear Mr Smalley,

Thank you for your response to our email.

We acknowledge that there are limitations caused by primarily reporting deaths data by the date of registration, as opposed to date of death (most of ONS’s publications report the date the death is registered, but the Monthly Mortality Analysis also reports the date of death). This method of reporting is subject to the effects of registration delays (there are often delays in recording deaths over bank holidays, or deaths recorded by coroners). We do however appreciate the merits of keeping a static record of mortality, as opposed to a dynamic dataset that constantly updates. Because of this, we intend to engage with ONS to explore the ways in which some of these limitations could be mitigated. This will be through the Compliance Check we outlined in our previous email.

We would also like to acknowledge the steps taken by ONS to inform users of the impacts of their focus on the date of registration rather than occurrence, and the impact of registration delays on the data. In addition, ONS conducts their own analysis of the impacts of registration delays on the data, and provides data to allow users to conduct their own analysis.

In response to our email, you outlined three questions. Our answers are as follows:

1. The impact of registration delays on data from 2022: You have conducted your own analysis which you claim shows ‘substantial registration delays for all ages under 75, dating back to the end of 2021’. You ask whether we have conducted our own analysis of Coroners data, and whether that analysis shows registration delays for 2022.

With regards to registration delays, we have based our conclusions on the latest published official statistics. These show that registration delays did increase in 2020. Coroners statistics for 2021 indicate that the number of deaths registered by a coroner dropped to the lowest number since 1995. ONS has not yet published its analysis of registration delays for 2021, but it has pre-announced the release of the analysis, coming out in March or April 2023. The Coroners statistics for 2022 are also not yet published. These are due to be published on 11 May 2023. A delay between the year ending and the production of the annual report allows for the analysis to be conducted and for quality assurance processes to be completed. We have not and would not conduct our own analysis. As a statistics regulator (rather than auditor), it is not appropriate for us to conduct an analysis of data in advance of statistics being published.

2. Reporting deaths by the date of registration allows for the creation of an incomplete record of mortality

As your email states, ONS have published various reports regarding the benefit of using date of registration – primarily because it ensures that data are complete for deaths registered in a given time period and allows for comparisons over time. You are however correct in your assertion that this results in an incomplete record of all deaths that occurred. This is particularly apparent as deaths that are registered more than a year after occurrence are not included in ONS’s weekly, monthly and annual mortality publications. However, as these deaths tend to occur as a result of external influences, they are usually covered in specific ONS publications, such as those on suicide, drug-related deaths and alcohol-specific deaths.

We acknowledge that both methodologies have their limitations. However, we feel it is important here to acknowledge the steps taken by ONS to explain the reasons for their chosen methodology, and to provide users with sufficient data and statistics to mitigate these issues, such as the provision of data on the date of death in the Monthly Mortality analysis, and the publication of statistics assessing the impact of mortality delays.

3. ONS meeting the needs of users: You have asked how we can be satisfied that ONS has met the needs of their users when you, as a user, do not feel that your needs are met.

While we do encourage producers to meet the needs of their users, we recognise that this will not always be possible. As stated in my previous email, ONS does not have access to real-time data and therefore is unable to meet the user need that you have identified.

ONS previously reported the number of deaths occurring in a time period, which meant that a period of time had to lapse before the analysis could take place. This process was impacted by registration delays and as a result, in 2006 ONS changed their methodology to their current method of reporting on death registration. At the time of the change, this resulted in a significant improvement, as it allowed for a much more timely release of the data. ONS conducted a user consultation in 2007 to determine which methodology would best suit their users.

It is not possible for ONS to receive mortality data from the General Register Office at an earlier date. Once a death occurs, a death certificate is issued by either a GP, a doctor or a coroner. The death certificate is then sent to the Local Registrar by email or post.

The Local Registrar manually enters the data into the online registration system. The data are then received by the GRO. From here, the data are transferred to ONS. Neither the GRO nor the ONS are made aware of a death until after the death is formally registered with the Local Registrar. This process works well for deaths registered by doctors, but it does mean that delays are present with coroner registered deaths.

NHS Digital have published a Mortality Data Review, which describes the death registration process in detail. The flowchart on page five of the document details the steps involved in the statutory reporting process, and highlights areas where delays are likely to occur.

Overall, ONS has increased the number of regular publications on deaths, as well as ad hoc publications when needed. We are continuing to conduct 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 2023 and would be happy to send the findings of this review to you when it is published, should that be of interest.

Yours sincerely
Ed Humpherson