1. Overview of work to improve annual suicide statistics
The Office for National Statistics (ONS) has been exploring ways to improve our annual suicide statistics to ensure clarity for all our users. We recently ran a user feedback survey to understand how best to present data on suicide registrations and occurrences in our annual suicides release.
In England and Wales, all deaths by suicide are certified by a coroner and cannot be registered until an inquest is complete. This results in a delay between the date the suicide occurred and the date it is registered and received by the ONS. Currently, ONS suicide releases are published on a registration basis, that is, the year in which the suicide was registered. Death registrations also include the date that the death occurred; we use this to analyse occurrence data, which are currently published in our separate Suicide occurrences, England and Wales dataset. Registration data relate to the previous year and the occurrence data have a lag of an additional year to allow for inquest times. For example, in October 2025, we published 2024 suicide registrations and 2023 suicide occurrences.
Because of lengthening registration delays, a smaller proportion of suicides registered in a given year are occurring within the same year. For example, only 38.7% of suicides registered in 2024 had a date of death that was also in 2024. This is substantially lower than in 2002, when 67% of suicides registered in 2002 also occurred in 2002.
Figure 1: Percentage of suicide deaths registered by year and time taken to register, England and Wales, 2002 to 2024
Percentage of suicides, England and Wales, registered between 2002 and 2024
Source: Suicides in England and Wales from the Office for National Statistics
Notes:
- The accredited official statistics definition of suicide is available in Section 10: Glossary, in our Suicides in England and Wales: 1981 to 2024 bulletin.
- Figures are for persons aged 10 years and over.
- Includes deaths of non-residents.
- Percentages may not sum to 100 because of rounding.
Download this chart Figure 1: Percentage of suicide deaths registered by year and time taken to register, England and Wales, 2002 to 2024
Image .csv .xlsBecause of the increase in registration delays, the ONS has considered how best to present suicide registration and occurrence data in our annual suicides release.
Each proposed option has its own advantages and disadvantages, which were outlined to all respondents in our user feedback survey.
Registration data can be difficult to interpret, where users often interpret the data as occurrences, whereas occurrence data are more intuitive.
Registration data can be released on a timelier basis, as they are released on an annual basis, approximately eight months after all annual registrations are processed; occurrence data can also be released on an annual basis, but this incurs a time lag of 18 to 24 months after the annual occurrence data are processed, which allows us to account for approximately 90% of deaths by suicide.
Registration data are more complete and the data for the previous year do not require revision; occurrence data require repeated revision as new registrations are processed.
Registration data are more comparable with suicide statistics produced by National Records of Scotland (NRS) and the Northern Ireland Statistics and Research Agency (NISRA), whereas using occurrence data would be more comparable to suicide statistics published by Republic of Ireland.
Changing how we report our suicide statistics could affect the interpretability, timeliness, completeness and coherence of those statistics. To better understand the views of our users, and to make evidence-led decisions, we conducted a brief user survey.
Nôl i'r tabl cynnwys2. User feedback survey
In April 2025, we at the Office for National Statistics (ONS) launched an eight-week user feedback survey seeking views from our users on their preferences. We proposed three options and asked for feedback on how their preferred option would best meet their needs.
Option 1 (registrations-first approach)
Continue to publish the annual suicides statistics with all the commentary related to suicides registration data and occurrence data only appearing in a separate dataset, but with clearer commentary on the impact registration delays would have on the annual figures. More signposting to the near to real-time suspected suicide surveillance (nRTSSS) data for England and the nRTSSS data for Wales could also be provided.
Option 2 (combined approach)
Continue to focus the annual suicides statistics on registration data but also include commentary on the most recent occurrence data received. This would provide a more complete picture of trends in suicides but switching between registrations and occurrences could be confusing for users. We will consider the best way to present the registration and occurrence data to reduce confusion.
Option 3 (occurrences-first approach)
Instead of the current approach of providing the registration data only in our annual bulletin, the bulletin would be based on the occurrence data only. This would be more intuitive for users as it would give them annual figures for suicides that occur in a given year. However, because of the registration delays, there would be a longer time lag of approximately 18 months to two years to publish these data. Additionally, the annual occurrence figures would never be complete and would have to be repeatedly revised as new registrations for suicides are received, making year-on-year comparisons more difficult.
With this option, we will also consider whether it is possible to adjust occurrence figures for suicides that have not yet been registered at a given point in time. This would enable more timely suicide occurrence figures to be produced.
We provided free-text boxes to enable our users to provide their feedback on why this was their preferred option, how we could present registration and occurrence data, and any improvements we could make to our annual suicide release.
Analysis of responses
We received a total of 93 responses to the user feedback survey, showing a clear preference for the combined approach, which was selected as the preferred option for 46 respondents. The occurrences-first approach received support from 26 respondents, while the registrations-first approach was the least favoured, with 21 respondents expressing their support.
Figure 2: Number of responses received by selected option
Source: User engagement with suicide statistics: registration and occurrence data from the Office for National Statistics
Download this chart Figure 2: Number of responses received by selected option
Image .csv .xlsWe received responses from a range of stakeholders across multiple organisational sectors including:
arm's length bodies and other public sector
central government and devolved nations
local government
academia and research
charities
business and industry
members of the public
We have broken down the number of responses we received by sector.
Organisational sector | Number of respondents | |
---|---|---|
Arm’s length body and other public sector | 12 | |
Central government and devolved nations | 6 | |
Local government | 28 | |
Academia and research | 9 | |
Charity and voluntary organisation | 12 | |
Mental health organisation | 1 | |
Members of the public and support groups for affected communities | 3 | |
Business and industry with an interest in employee mental health | 2 | |
Citizen | 1 | |
Not Answered | 19 |
Download this table Table 1: Number of responses received by organisational sector
.xls .csvAnalysis showed that there was little variation in preferred approach based on the sector the respondent represented.
Respondents expressed a clear desire for the continued inclusion of the registration data, but with the additional occurrence data, both supported by narrative commentary. Respondents saw this approach as essential for providing a comprehensive overview of suicide trends in England and Wales.
A thematic analysis was conducted on all responses to identify the important themes regarding what respondents would like to see in our annual suicide statistical release.
Identified themes
Improving user understanding
Respondents emphasised the importance of understandable data and provided separate arguments for why both date-of-registration and date-of-occurrence data would be easier to understand. While registrations were recognised as being consistent with past releases, occurrences were viewed as more intuitive and would enable users to better understand and interpret suicide trends.
Respondents acknowledged the complexities involved in reporting suicide statistics and emphasised the importance of addressing these directly within the release to support user understanding. Rather than limiting commentary to either registrations or occurrences, they recommended providing a narrative commentary for both, to ensure transparency and prevent misinterpretation.
Although some concern was expressed about the potential for confusion, respondents felt that failing to clarify the differences between registrations and occurrences could be more misleading. They suggested that the Office for National statistics (ONS) could mitigate this risk by presenting a complete picture, supported by clear interpretation and effective visualisation. Respondents also requested that this was available at local and national levels. This approach was seen as being consistent with best practice in statistical reporting.
Accuracy
Respondents recognised date-of-registration data as a consistent, complete and reliable source for long-term comparisons and public health research. However, they also recognised limitations including regional variations and delays that can obscure when the death occurred. In contrast, date-of-occurrence data were valued for offering a more immediate picture, making it particularly useful for responsive work, such as suicide prevention.
While it was acknowledged that occurrence data is subject to time lags, it was considered more accurate for identifying trends within a given year and assessing the impact of specific events on mental health. This accuracy would support evaluation of local prevention strategies and identification of national or regional patterns.
Respondents highlighted the importance of being able to determine whether suicide rates are increasing or decreasing on a year-on-year basis, and felt that having access to both date-of-registration and date-of-occurrence data would provide the most complete and accurate picture.
Continuity and comparability
The continued inclusion of date-of-registration data was seen as essential for maintaining consistency with previously published figures, enabling users to conduct long-term trend analysis. Respondents also felt that, as registration delays have been observed in previous years, including registration data would allow consistent recording and comparison of these delays over time.
The continued publishing by date of registration would allow the comparison of ONS data with other published data sources that are also based on registration data, such as all-cause mortality statistics, allowing like-for-like comparisons.
On the other hand, using date of occurrence was seen to enhance comparability with other data sources. Some respondents highlighted that occurrence data align more closely to their published data, improving the ability to compare suicide statistics across different geographies. Occurrence data would also align more closely with the near to real-time suspected suicide surveillance (nRTSSS) data (produced by the Office for Health Improvement and Disparities and Public Health Wales), enabling better comparability with suspected suicides at the local level.
By offering both continuity with past releases and enhanced comparability with other sources, the combined approach was viewed as the most effective and pragmatic approach to reporting suicide statistics within the annual release.
Timeliness
Respondents raised concerns about the timeliness of date-of-occurrence data because of the 18-to-24-month time lag for publication. This lag would limit its relevance and usefulness, particularly when informing suicide prevention measures and local decision-making. In contrast, date-of-registration data were valued for providing an earlier view of confirmed suicides, supporting faster identification of emerging trends and more timely policy decisions, particularly at the local level.
Despite the delay, many respondents supported the inclusion of occurrence data, recognising its value in accurately reflecting when suicides happened. This was particularly relevant at the local level and within specific populations, where it could be linked to specific events or policy changes to support strategic planning.
Discussing the need for timely suicide statistics, respondents also supported the signposting of the near to real-time suspected suicide surveillance (nRTSSS) data. These data were recognised for the capacity to improve understanding of suspected suicides at a local level through providing near real-time data, identifying possible clusters early, and informing strategic suicide prevention work. It was seen as a valuable complement to the accredited official statistics provided by the ONS.
The inclusion of both registration and occurrence data was also seen as an opportunity to improve transparency around delays in the suicide registration process. Respondents noted that these delays have increased over time and vary by region. Presenting both data types together would help users better understand and monitor registration delays and would support efforts to improve coronial processes and the timeliness of suicide registrations.
Favoured option: combined approach
Respondents expressed a clear preference for including both date-of-registration and date-of-occurrence data within the annual release. Users felt that this option would provide a better balance for the advantages and limitations of each dataset. It would also offer greater contextual depth, clarity and transparency for suicide statistics across England and Wales.
An approach that integrated all available suicide data would provide users with a more informative picture of suicide trends:
occurrence data, to reflect the actual time of deaths
registration data, for a timelier picture of confirmed deaths and to understand the time it takes to process inquests
near real-time data on suspected suicides (nRTSSS), to enable timely responses to emerging trends
Respondents felt that interpretation of all three datasets would be most beneficial for identifying trends and at-risk groups to inform targeted interventions, local prevention strategies and wider policy development.
Respondents viewed the combined approach as a pragmatic response that could be implemented while exploring alternative data ways to improve the timeliness of occurrence data (such as modelling occurrences that were not yet registered; see Section 3: Future developments).
Considerations
While respondents showed a preference for a combined approach within the annual release, they also shared suggestions for optimal implementation:
clear commentary for both date-of-registration and date-of-occurrence data and trends, including around registration delays
clear caveats for both data types
clear guidance on when to use each type of data and why
analysis of trends, including visual explanations of the differences and narrative commentaries
use of modelling to improve the timeliness of occurrences and to reliably estimate unregistered suicides
Respondents recommended that both date of registration and date of occurrence be presented with equal prominence, supported by clear, narrative commentaries. This approach, along with the necessary caveats, would allow the ONS to provide a more transparent picture of trends and enable users to interpret the data correctly. By clearly outlining the strengths and limitations of each, it would enable users to select the most appropriate data for their specific needs.
Some respondents suggested developing a modelling technique to adjust occurrence data for registration delays, making it more timely and improving its usefulness and relevance. They did acknowledge, however, that such models would need to account for varying delay lengths across regions and changes over time.
Nôl i'r tabl cynnwys3. Future developments
Following the feedback received through our engagement exercise, we at the Office for National Statistics (ONS) have begun to improve the way we present our suicide statistics in our annual Suicides in England and Wales bulletin.
We have adopted the combined approach, where both registration and occurrence data are presented in dedicated sections within the annual suicides release. Each dataset is accompanied by commentary to ensure a comprehensive overview.
The important figures and trends for the occurrence data are included within the main points section. The caveats for each data type are clearly outlined to help users understand how to interpret each data type and determine which is most appropriate for their specific needs.
In addition, we have clarified the impact of registration delays by providing further explanation of the process and including a visualisation, which shows how registration delays have changed over time.
To support the use of more real-time data, there is prominent signposting to the nRTSSS data, which provide more timely information on potential emerging trends within regions.
To improve this combined approach further, the ONS are exploring methods to adjust suicide occurrence figures for suicides that have not yet been registered. This will enable us to report more timely suicide occurrence figures in our annual suicides release. The ONS plans to implement this for our next annual suicides release for 2025 registrations, which will be published in Autumn 2026.
This approach aims to offer a more comprehensive and balanced view of suicide rates and trends, supporting clearer interpretation. We continue to welcome user feedback to improve our suicide statistics.
Nôl i'r tabl cynnwys5. Cite this article
Office for National Statistics (ONS), released 3 October 2025, ONS website, article, User engagement with suicide statistics: registration and occurrence data