A change in the standard of proof for how suicides are recorded that occurred in 2018 and increasing delays in the time taken for suicides to be registered have created challenges for ONS in communicating our suicide statistics. In this blog, David Mais, Head of Mortality Statistics, explores how these developments are changing our releases.
We are committed to delivering accurate and impactful suicide statistics to provide people with useful information about this important topic. Today we have published the latest suicide figures with some improvements to better communicate changes in the data. A growing delay in suicide registrations has made the distinction between when a death occurred and when it was registered more important, so the release now provides both to help people better understand the figures. We have also improved our communication following changes to how coroners determine suicides. The importance of these changes was raised by Full Fact, who noted that the lack of commentary in our releases could lead to people misunderstanding suicide statistics.
Looking to the past
Every possible suicide death is referred to a coroner. In July 2018, England and Wales implemented a significant legal change affecting how coroners determine whether a death was caused by suicide. The “standard of proof” – the level of evidence required to reach this conclusion – was lowered from the criminal standard of “beyond all reasonable doubt” to the civil standard of “on the balance of probabilities”. This makes it easier to reach this determination when the evidence suggests it is more likely than not that someone intended to end their own life.
Such a change could lead to more deaths being labelled as a suicide. We carried out analysis a couple of years after the release to see if this was the case, but the results were inconclusive. Since then, the rate of suicide registrations have risen and it has become increasingly important to determine if this is due to an actual change or the result of this lowering of proof, something Full Fact flagged last year.
So, with several more years of data to work with, we re-examined changes to suicide registrations since this lowering of the standard of proof. The results painted a complex picture. Suicide registration rates did increase by a larger than expected amount after the change in the standard of proof, but this increase began before the change and was also seen in Scotland where no such change occurred. We also found fewer cases where the intent of poisoning or injury was undetermined, as coroners seemed more willing to attribute it to intentional self-harm, consistent with the standard of proof being lowered.
Together, these indicate that the rates of registered deaths caused by suicide have increased in recent years, and it is likely that the change in the standard of proof has contributed to this trend alongside other factors – such as an actual rise in the number of suicides.
Refining in the present
This was another issue noted by Full Fact last year, who noted that these figures are not very intuitive. Specifically, we publish figures based on when the death is registered, but not every death registered in a given year occurred in that year. Many conflate registrations and occurrences, which is not just a misunderstanding but could lead to wrong conclusions, reporting the wrong number of suicide deaths in a given year by mixing these numbers.
To make matters worse, this delay between occurrence and registration has worsened, with only 38.7% of suicides registered in 2024 actually occurring in 2024, compared to 67% in 2002. Full Fact explored how this distinction was often missed in discussion of suicide statistics, particularly in media headlines.
The alternative is to publish data based on when the suicide actually occurred. However, as it can take a long time for a year’s occurrence figures to be published this would add an extra year’s lag to our figures as we wait for all the data. With both approaches having their strengths and limitations, we launched a user feedback survey to learn which would be most useful for people: by registrations, occurrences, or a combined approach where the most recent registration data and last year’s occurrence data are published together.
After over 90 responses from across government, charities, academia, and the public, the results showed that the overwhelming preference was for this combined approach. Respondents valued this comprehensive method because it provides both timely information about confirmed deaths and a more accurate picture of when suicides occurred, essential for understanding trends and informing prevention strategies. They emphasised the importance of clear commentary explaining both datasets, proper caveats, and equal prominence for each type of data to prevent misinterpretation There was also an emphasis on the need to keep publishing registration data to lobby for improvements to the delay in registrations. This would not just make the statistics more accurate but would help the families of those involved get answers faster.
Improving in the future
Today we have published our latest suicide figures and are already making improvements based on this work. They include a clear note explaining the changes in standard of proof, with a note on relevant graphs highlighting when it happened so readers can take it into account. The bulletin also features both the 2024 registration data and 2023 occurrence data, launching the combined approach our survey responses recommended, along with prominent discussion of the limitations of each approach.
We are still working to make further improvements to our suicide statistics. We are developing nowcasting techniques – a statistical method that combines current data with predictive modelling, which will allow us to provide a faster estimate of suicide occurrences. We have plans to implement this nowcasting methodology for the annual suicide statistics covering 2025 registrations, which will be published in autumn 2026. When combined with near real-time suspected suicide surveillance by the Office for Health Improvement and Disparities, people will have a range of suicide statistics at varying levels of timeliness and accuracy for different uses. This represents a major step forward in making suicide occurrence data more immediately useful and better able to be understood.
David Mais is Head of Mortality Statistics at the ONS