The comparability and coherence of UK homeless deaths statistics
People experiencing homelessness are especially vulnerable to poor physical and mental health. They are also more likely to die early or avoidably. Having quality data on the numbers and nature of homeless deaths is an important part of enabling UK governments to make informed interventions, and to assess the effectiveness of new policy.
Producers of homeless deaths statistics across the Government Statistical Service (GSS) are working together to:
- improve their methodologies
- share learning and new ideas
- find the best ways to communicate the differences between their outputs
The Office for National Statistics (ONS) have produced experimental statistics on homeless deaths for England and Wales since 2018 and The National Records of Scotland (NRS) have produced experimental statistics on homeless deaths for Scotland since 2020.
The Northern Ireland Statistics and Research Agency (NISRA) is exploring the possibility of publishing similar statistics for Northern Ireland for the first time. They are aiming to publish the first set of statistics in late 2023. This means that next year could be the first year for which similar data on homeless deaths are available across the UK. Previous statistics provided by the Northern Ireland Housing Executive (NIHE) to the Museum of Homelessness use fundamentally different methods.
Different public bodies produce these statistics for different countries because both civil registrations and homelessness are devolved matters in the UK. You can find out more about devolution on GOV.UK.
While the methodologies are consistent in principle across Great Britain, differences in legislation and data sources have always meant that the comparability of the statistics has been limited. However, the comparability is negatively affected to a greater extent when considering deaths that occurred during the COVID-19 pandemic.
Comparability of methodologies
The releases for England and Wales and the releases for Scotland use location data collected from death certificates. This could be data about place of residence or place of death. This information can be used to work out whether a person was homeless at the time of their death.
In some cases, these data directly indicate that the death was of a homeless person. For example, this could be where the place of residence is recorded as “no fixed abode” or “homeless”. In other cases, more work is needed to match the death certificate locations against lists of accommodations used by homeless people. For example, we may need to work out whether the location corresponds with places such as temporary accommodation, night shelters, or emergency hostels. The numbers of identified homeless deaths will still be undercounts of the true figures, so statistical modelling is used to estimate the additional number of homeless deaths not captured by these methods.
While this description of the methodologies is consistent for England and Wales and for Scotland, there is a difference in how their lists of known homeless accommodations are compiled. ONS updates their homeless accommodation lists for England and Wales by searching information that is publicly available online. NRS requests their lists each year, from each local authority in Scotland individually.
Because there are only 32 local authorities in Scotland, it is more manageable to speak to them individually to collect information. This contact also allows NRS to check their identified homeless deaths with those councils, who can verify them case by case. There are over 350 local authorities in England and Wales, so it is impractical for ONS to contact them all individually.
Although there are further nuances and differences between the methodologies, these are thought to be minor. You can find more information in:
The effect of COVID-19 on comparability
Many new homeless accommodations were arranged to protect people during the COVID-19 pandemic. This has increased the methodological gap that existed between the England and Wales statistics and the Scotland statistics.
Three days into the first national lockdown, the UK Government wrote to local authority leaders in England asking them to ensure that homeless people were safely accommodated to guard them from the risks of contracting and spreading COVID-19. This included inviting homeless people to stay in hotels around the country. This scheme was known as the “Everyone In” scheme.
Similar schemes were used in Wales and Northern Ireland. No new policy was needed in Scotland because all homeless people already had the right to temporary accommodation, regardless of priority need.
Scotland’s methodology accounts for all their homeless accommodations. This means that it accounts for the full increased use of temporary accommodation during the pandemic. The methodology for England cannot do so because the new homeless accommodations used in the “Everyone In” scheme are not discoverable through ONS web-scraping. Although the new accommodations are discoverable for Wales, they were also used for other purposes, such as providing healthcare workers with another place to stay if they lived with clinically vulnerable people. This means that it is impossible to identify who was homeless in these accommodations, and so they are excluded from ONS analysis.
The difference in methods used to identify homeless deaths, were affected by policies during the COVID-19 pandemic and may have contributed to differences in death rates across the UK. This is in addition to the effect of legislative differences, as there are fewer criteria in Scotland for statutory homelessness applicants to qualify for temporary accommodation. So, although published homeless death rates appear higher in Scotland than in England and Wales in each year where data are available, the data should not truly be compared. It is likely that the different methods have led to the higher estimate in Scotland.
What happens next
The producers of UK homeless deaths outputs will continue to meet regularly. They will work together to consider further developments and improve the coherence of their outputs. Current initiatives include:
- exploring the possibility of having a homelessness indicator added to the death certificate in England and Wales – this would increase the number of homeless deaths that can be identified
- investigating the additional data sources that NRS could bring in to improve the robustness of their statistics for Scotland – this includes benefits data, data from Scotland’s 2022 Census, and case notes from the Procurators Fiscal, which is Scotland’s equivalent of coroners
- exploring the possibility of developing the first release of official homeless deaths statistics for Northern Ireland – NISRA are working on this with the Department for Communities and the NIHE, with the aim of coinciding with the 2023 ONS and NRS publications
- working towards National Statistics accreditation – NRS and ONS are both working towards this accreditation
With the continuing developments of the homeless deaths outputs, the producers are keen to hear from users regarding the value of the statistics. Please contact the team at GSS.Housing@ons.gov.uk if you have any feedback.
You can find more information about the wider work going on across the GSS, to improve housing and homelessness statistics in our 2022 to 2023 cross-government housing and planning statistics work programme.