FOI reference: FOI-2026-3649
You asked
I am making a request under the Freedom of Information Act 2000 concerning the ONS publication Suicides in UK armed forces veterans, England and Wales: 2021 and the linked use of death registration records, Census 2021 and the Ministry of Defence Service Leavers Database.
Please provide the following recorded information held by ONS.
1. Methodology and dataset specification
Please provide the final version of any methodology specification, data dictionary, variable list, linkage specification or quality-assurance document used to produce the 2021 veteran suicide statistics.
2. Cohort construction and identification of veteran status
Please provide the final recorded specification showing:
a. the populations eligible for inclusion within the analytical dataset;
b. the populations excluded from it;
c. the data source or linkage route by which veteran status was assigned;
d. how deaths occurring before Census Day were assessed for veteran status;
e. how people leaving the Armed Forces after Census Day were assessed for veteran status; and
f. how conflicting or incomplete veteran-status information across Census 2021 and the Service Leavers Database was resolved.
3. Linkage failure and data quality
Please provide any final or current recorded estimate, table, quality note or methodological assessment showing:
a. the number or proportion of relevant Census or Service Leavers Database records that could not be linked to ONS death registration records;
b. the recorded reasons for non-linkage;
c. any recorded assessment of whether non-linkage was concentrated within particular service-leaver or demographic groups; and
d. any recorded assessment of the effect of non-linkage on completeness, representativeness or bias.
I am not asking ONS to undertake a new analysis where no recorded assessment exists.
4. Available disaggregation variables
Please provide recorded information showing whether the analytical dataset used for the 2021 publication contained variables capable of identifying or deriving:
a. Regular or Reserve service;
b. service branch;
c. discharge year;
d. length of service;
e. rank;
f. nationality;
g. nationality at enlistment;
h. country of birth;
i. ethnicity;
j. a marker identifying personnel recruited as Commonwealth citizens or from Commonwealth countries; and
k. Gurkha status.
For each characteristic, please indicate whether the relevant variable was:
a. included within the analytical dataset;
b. present in a source dataset but not included within the analytical dataset;
c. not supplied to ONS;
d. not held by ONS;
e. not reliably derivable; or
f. unavailable for another recorded reason.
A table or existing data dictionary identifying these positions will be sufficient. I am not requesting counts, rates or record-level extracts for any subgroup.
5. Commonwealth and Gurkha identification
Please provide any final or current recorded methodological note, internal assessment or data-quality consideration addressing whether the linked data could reliably identify:
a. veterans recruited as Commonwealth citizens or from Commonwealth countries; and
b. Gurkha veterans.
Where ONS recorded that either population could not be reliably identified, please provide the recorded reason.
6. Definition and classification
Please provide any recorded operational definition or classification rule used within the analysis to distinguish between:
a. serving personnel;
b. service leavers;
c. veterans;
d. former Regular personnel;
e. former Reserve personnel; and
f. people whose service history could not be classified.
7. Future publication
For the period from 5 April 2024 to the date of this request, please provide any current publication plan, work programme, timetable or recorded decision concerning veteran suicide statistics for years after 2021.
This should include, where recorded:
a. the intended coverage period of the next publication;
b. whether further annual or multi-year publications are planned;
c. any dependency preventing confirmation of a publication date; and
d. any methodological or data-access issue delaying further publication.
We said
Thank you for your request concerning our Suicides in UK armed forces veterans, England and Wales: 2021 bulletin, released on 5 April 2024.
1. Methodology and dataset specification
The methodology used for the release is described in detail in Section 8: Measuring the data of the Suicides in UK armed forces veterans, England and Wales: 2021 bulletin.
As described in the publication, veteran suicides were identified through linkage of Office for National Statistics (ONS) death registration data with Census 2021 and the Ministry of Defence (MoD) Service Leavers Database. Census 2021 provided veteran status at Census Day (21 March 2021), while the Service Leavers Database was used to identify veterans not fully captured through census alone, including those becoming veterans after Census Day.
We record information on data assets within the organisation's Information Asset Register (IAR). The description of the three assets used to produce the veteran suicide release is as follows.
Death registrations:
description – all deaths registered in England and Wales
asset type – statistical data, which means any data used for statistical production, research, or feasibility study (including statistical sources, products and outputs)
data format – raw data
Government Security Classification – Official-sensitive
geographic coverage – England and Wales
lowest geography available – postcode
temporal coverage – 1981 to present
data frequency delivery – daily
format in which the information is stored – electronic; systems and shared drives
data supplier – General Register Office (GRO)
2021 Census – England and Wales outputs:
description – this is a record-level dataset created from the census-processed data including imputed records, SDC – Statistical Disclosure Control
asset type – statistical data, which means any data used for statistical production, research, or feasibility study (including statistical sources, products and outputs)
data format – none of the above
Government Security Classification – Official-sensitive
geographic coverage – England and Wales
lowest geography available – Unique Property Reference Number (UPRN) and postcode
temporal coverage – 21 March 2021
data frequency delivery – other; one-off
format in which the information is stored – electronic, systems and shared drives
data supplier – Office for National Statistics (ONS)
Condensed Ministry of Defence (MoD) Service Leavers Database:
description – MoD Service Leavers Database with forename, surname, initials, date of birth and postcode removed; NHS number will remain and some validation variables will also be available
asset type – statistical data, which means any data used for statistical production, research, or feasibility study (including statistical sources, products and outputs)
data format – unhashed
Government Security Classification – Official-sensitive
format in which the information is stored – electronic; systems and shared drives
data supplier – MoD
2. Cohort construction and identification of veteran status
a. Populations eligible for inclusion
The analytical population consisted of individuals identified as veterans through either Census 2021 or the MoD Service Leavers Database, linked to ONS mortality records. The veteran definition used was persons who had previously served in the UK armed forces, including regular and reserve personnel.
b. Populations excluded
As stated in the publication, the analysis excludes:
people still serving in the armed forces
individuals living outside England and Wales
individuals who could not be linked sufficiently to support the analysis
veterans missing information needed for linkage, for example, not having an NHS number on census-based sources
c. Assignment of veteran status
Veteran status was assigned using:
Census 2021 responses, indicating previous service
the Service Leavers Database maintained by the MoD
These sources were linked to mortality records.
d. Deaths occurring before Census Day
We recognised that Census 2021 alone could not identify veteran deaths occurring between 1 January 2021 and Census Day (21 March 2021). The Service Leavers Database was therefore used to identify such deaths.
e. Personnel leaving the armed forces after Census Day
As stated in the publication, Census 2021 does not identify individuals who became veterans after 21 March 2021. These individuals were identified through the Service Leavers Database.
f. Conflicting or incomplete veteran-status information
There was little conflict between the two data sources used for the publication (Census 2021 and the Service Leavers Database) as they were used to identify different types of leavers. Census 2021 captured veterans who were alive on Census Day (21 March 2021), while the Service Leavers Database was used primarily to identify individuals who left service after Census Day, as well as those who had left service before Census Day but died between 1 January 2021 and 21 March 2021.
However, we did include veterans in the analysis if the Service Leavers Database indicated they left service before Census Day (21 March 2021) but they did not record themselves as a veteran on the census question.
3. Linkage failure and data quality
a. Numbers or proportions not linked
The publication records that:
1,853,112 veterans were identified in Census 2021 as intending to remain in England and Wales
1,740,281 (93.9%) were successfully linked to the Personal Demographic Service (PDS) to obtain NHS numbers
consequently, 112,831 veterans (6.1%) could not be linked to the PDS and were therefore excluded from analyses requiring NHS number linkage
within the Service Leavers Database, 73.2% (10,592) of service leavers in 2021 had a valid NHS number recorded, while 26.8% (3,876) did not, and therefore could not be linked to ONS death registrations
This level of linkage is high; however, the inability to link 6.1% of identified veterans may result in some undercoverage of the veteran population. As veteran status in the mortality analysis relied on successful linkage through NHS numbers, some veteran deaths may not have been identified.
b. Reasons for non-linkage
The reasons for non-linkage include:
missing NHS numbers in the Service Leavers Database
failure to link census records to the PDS
c. Assessment of whether non-linkage varies by subgroup
We have not conducted an assessment addressing this question.
d. Assessment of impact on completeness, representativeness or bias
The publication records that veteran suicides are likely to be underestimated because some veterans cannot be identified because of missing NHS numbers and linkage limitations.
There is also undercoverage arising from missing administrative records, unsuccessful linkage and incomplete veteran identification.
4. Available disaggregation variables
We hold information describing variables supplied through Census 2021 and the Service Leavers Database.
The publication states that:
Census 2021 identifies service type (regular, reserve or both)
the Service Leavers Database used for the publication did not enable reliable distinction between regular and reserve service histories
The full list of variables that the ONS has on the Service Leavers Database data supplied by the MoD is:
name (forename, middle name, surname, previous surname and/or initials)
date of birth
gender
postcode
service exit date
NHS number
early service leaver marker
branch of armed forces served
rank at exit
assignment type
ever regular
Information on the variables available for analysis on Census 2021 data can be found on our Census 2021 Variable by topic web page and in the Census 2021 questionnaire.
5. Commonwealth and Gurkha identification
We have no documentation related to the identification of Commonwealth veterans or Gurkha veterans within the suicide publication.
6. Definition and classification
The publication defined a UK armed forces veteran as:
People who have previously served in the UK armed forces. This covers everyone who has served for at least one day in HM's armed forces, either regular or reserves, or merchant mariners who have seen duty on legally defined military operations. It does not include those who are still serving, even if they have left and rejoined, those who have only served in foreign armed forces, or for this study, those who have served in the UK armed forces and are currently living outside of England and Wales.
7. Future publication
We have publicly committed to future publications of veteran suicide statistics. We plan to include these statistics as part of our annual suicides release. The next release of annual suicide statistics is scheduled for October.