Deaths of people experiencing homelessness in England and Wales: 2013 to 2025 registrations

• There were an estimated 974 deaths of people experiencing homelessness registered in England and Wales in 2025, with a 95% confidence interval of 908 to 1,041 estimated deaths; this was similar to the 978 estimated deaths in 2024.

• The estimated crude mortality rate was 21.3 deaths per million people aged 15 to 74 years in 2025; this was similar to the 21.5 per million in 2024.

• The estimated number of deaths was 77.1% higher in 2025 than in 2013, increasing from 550 to 974.

• Males accounted for an estimated 833 deaths (85.5%) in 2025, compared with 141 deaths (14.5%) among females.

• London had the highest estimated number of deaths among English regions in 2025 (151 deaths), followed by the North West (146) and the South East (144).

• Accidents were the largest underlying-cause group, with an estimated 380 deaths (39%).

• Among identified deaths of those aged 15 to 74 years in 2025, mean age at death was 47.3 years for males and 43.2 years for females.

Homelessness can be defined in several different ways. In this release, we have identified deaths from information recorded at death registration indicating that the person was sleeping rough or using emergency or temporary accommodation at or around the time of death. The definition does not align with official statistics on homelessness produced by other public bodies.

For example, these statistics do not use the same definition of homelessness as:

• the Ministry of Housing, Communities and Local Government (MHCLG) Rough sleeping counts
• the MHCLG Statutory homelessness statistics
• Welsh Government Homelessness statistics
• Welsh Government Homelessness accommodation provision and rough sleeping statistics

More information on the methodology used in this bulletin is available in our Deaths of people experiencing homelessness in England and Wales QMI.

Annual estimates

In 2025, 834 death registrations of people experiencing homelessness were identified with an estimated total of 974 deaths. This indicates that 140 additional deaths were of people experiencing homelessness beyond those identified directly. The 2025 estimate was four deaths lower than in 2024.

As these are model-based estimates, there is uncertainty around estimated deaths. Therefore small year-to-year differences should be interpreted with caution. For example, the 95% confidence interval around the 974 estimated deaths in 2025 was 908 to 1,041.

Figure 1: The estimated number of deaths was broadly unchanged between 2024 and 2025

Estimated deaths of people experiencing homelessness and 95% confidence intervals, England and Wales, registered between 2013 and 2025

Notes:

1. Estimated numbers include identified deaths and additional deaths estimated using capture-recapture modelling.
2. The shaded area represents the 95% confidence interval.
3. Figures are for deaths registered, rather than deaths occurring, in each calendar year.
4. Figures for England and Wales may include deaths of non-residents.

Identified and estimated deaths

The number of identified deaths in England and Wales doubled from 415 in 2013 to 834 in 2025. Estimated deaths increased by 77.1% from 550 to 974 over the same period. Estimated totals were higher than identified counts in every year because the capture-recapture model includes an allowance for deaths not identified directly in death registration data.

Figure 2: Estimated deaths were higher than identified deaths in every year

Identified and estimated deaths of people experiencing homelessness and 95% confidence intervals, England and Wales, registered between 2013 and 2025

Notes:

1. Estimated numbers include identified deaths and additional deaths estimated using capture-recapture modelling.
2. The light blue boxes represent the 95% confidence intervals around the estimated numbers.
3. Figures are for deaths registered, rather than deaths occurring, in each calendar year.
4. Figures for England and Wales may include deaths of non-residents.

Deaths by sex and age

In 2025, there were an estimated 833 male deaths (85.5%) and 141 female deaths (14.5%) of people experiencing homelessness. Among males, the largest number of deaths was in the 50 to 54 years age group, accounting for 16.9% of male deaths (141 deaths). Among females, the largest number of deaths was in the 40 to 44 years age group, accounting for 18.4% of female deaths (26 deaths).

Among identified deaths of those aged 15 to 74 years in 2025, the mean age at death was 47.3 years for males, up from 44.2 years in 2013. For females, the mean age at death was 43.2 years, which is similar to the 43.3 years in 2013. This is lower than the mean age at death for all males (61.1 years) and all females (62.1 years) aged 15 to 74 years in 2025. Mean age at death is not the same as life expectancy; it describes the average age among people who died in the period.

Geographic variation

In England, an estimated 911 deaths of people experiencing homelessness were registered in 2025. London had the highest deaths among English regions (151), followed by the North West (146) and the South East (144). The North East had the lowest deaths among English regions (40). In Wales, an estimated 63 deaths of people experiencing homelessness were registered in 2025.

Capture-recapture modelling was used to estimate the total number of deaths at the England and Wales level. Estimates for individual regions and local authorities were then calculated by distributing this overall estimate according to the proportion of deaths identified in each area, rather than creating separate models for each area. Smaller-area estimates should therefore be interpreted with caution.

Underlying cause and drug-poisoning deaths

Accidents were the largest underlying-cause group for people experiencing homelessness in every year between 2013 and 2025, with an estimated 380 deaths in 2025 (39%). Further information on the types of deaths included in the definition of "accidents" is available in Section 4: Glossary.

In 2025:

• intentional self-harm accounted for 84 deaths (8.6%)

• diseases of the liver accounted for 67 deaths (6.9%)

• ischaemic heart diseases accounted for 55 deaths (5.6%)

• malignant neoplasms accounted for 45 deaths (4.6%)

There were an estimated 304 drug-poisoning deaths in 2025 (31.2% of all estimated deaths), compared with 335 in 2024.

Local authority results

The accompanying dataset provides identified and estimated deaths for local authorities. Rates are suppressed where the estimated count is below three, or a matching population denominator is unavailable. Estimates are produced from the England and Wales total rather than fitting separate local authority models.

Capture-recapture

A statistical method that uses the overlap between different ways of identifying cases to estimate the number not identified by any of them. This helps improve the completeness of estimates of deaths among people experiencing homelessness.

Crude death rate

The estimated number of deaths per million people aged 15 to 74 years. These rates are not age-standardised and are not rates among the population of people experiencing homelessness.

Deaths of people experiencing homelessness

Deaths identified from information recorded at death registration indicating that the person was sleeping rough or using emergency or temporary accommodation at or around the time of death. This definition does not align directly with definitions used in statutory homelessness or rough-sleeping statistics.

Mean age at death

The arithmetic mean age among people who died in the group and period. It is not a measure of life expectancy.

Identified deaths

Deaths found in death-registration data using homelessness-related evidence and retained following identification and review.

Estimated deaths

The total of identified deaths and additional deaths estimated through capture-recapture modelling. These figures represent our best estimate of the number of deaths of people experiencing homelessness in England and Wales.

Proportional allocation

A method that distributes the annual England and Wales estimate across subgroups according to their shares of identified deaths. The drug-poisoning estimate was produced by applying the observed drug-poisoning share among identified deaths to the annual England and Wales capture-recapture estimate; it was not produced from a separate drug-specific capture-recapture model.

Registration year

Mortality statistics are compiled from information supplied when deaths are certified and registered as part of civil registration, which is a legal requirement. The registration year is the calendar year in which a death was registered, which can be later than the year in which it occurred.

Accidents

Types of death where the underlying cause was any International Classification of Diseases, 10th edition (ICD-10) code ranging from V01 to X59. This category includes fire-related deaths and deaths from:

• road traffic collisions

• falls

• accidental drug poisoning and overdoses

• drowning

• accidental suffocation

Further information on accidents and the definitions used are available in this freedom of information request.

We last published these statistics in our Deaths of homeless people in England and Wales: 2021 registrations bulletin in 2022. Publication of these statistics was paused following the Health and social care statistical outputs consultation, which was completed in 2024.

Data coverage

The dataset covers deaths registered in England and Wales from 2013 to 2025, for those aged 15 to 74 years. The upper age limit reduces the risk of incorrectly including older people in some institutional settings. Rates use population denominators for the same age range. Figures for England and Wales may include deaths of non-residents.

Figures are based on deaths registered in each calendar year rather than deaths occurring in that year. Date of registration of the death is usually later than the date it occurred, particularly where a coroner investigates the death. Official statistics in development are those still undergoing development and in the testing phase.

Within this release, identified deaths are deaths found in death-registration data using homelessness-related evidence and retained following identification and review. Estimated deaths are the total of identified deaths and additional deaths estimated through capture-recapture modelling. These figures represent our best estimates of the number of deaths of people experiencing homelessness in England and Wales.

How the figures were produced

The figures were produced using a two-stage process. First, complete death-registration records for deaths registered in England and Wales between 2013 and 2025 were analysed using multiple search strategies. These searches identified deaths where the available registration information indicated that the deceased was experiencing homelessness at or around the time of death. Cases with ambiguous or conflicting evidence were subject to manual review by an analyst who judged whether the death was of someone experiencing homelessness or not.

Second, the results of these searches were used in a statistical method known as capture-recapture. This uses the overlap between different identification sources to estimate the number of deaths of people experiencing homelessness that may be present in the registration data but were not identified directly.

The capture-recapture model uses three grouped sources of evidence:

• residence or no-fixed-abode information

• coroner information

• place-of-death information

The model is fitted separately for each registration year at England and Wales level.

The Mth Chao lower-bound estimator was selected as the preferred model, based on the nature of the data and the diagnostic and goodness-of-fit information examined during development. The Chao model is a robust but conservative estimator, meaning that the figures should be interpreted as lower-bound estimates and the true number of deaths may be higher.

Estimates for England, Wales, regions and other subgroups are allocated from the annual England and Wales estimate in proportion to identified deaths. They are not produced from separate capture-recapture models.

Strengths

Cross-referencing different indications of homelessness within death-registration records provides a strong dataset of identified deaths. The method uses several types of evidence, including registered-address information, no-fixed-abode indicators, coroner information and place-of-death information. Ambiguous or conflicting cases can also be examined through manual review.

The statistics provide insight into changes in the number and characteristics of deaths of people experiencing homelessness in England and Wales over time.

The identification and estimation methods are applied consistently across geographical areas. This means that figures for one local authority can be compared with figures for other local authorities, subject to the limitations of the small-area estimates.

Limitations

The identification process depends on the information recorded when a death is registered. A person's homelessness may not always be apparent to the informant, registrar or coroner, or may not be recorded in the available fields when a death is registered. Capture-recapture modelling is therefore used to estimate deaths that may not have been identified directly.

The capture sources are obtained from information within the same administrative system and may not be fully independent. The selected Chao model is conservative, so the resulting figures should be treated as lower-bound estimates rather than exact totals.

Estimates for local authorities and other subgroups are allocated according to their proportion of identified deaths. The method cannot allocate estimated additional deaths to a local authority or subgroup with no identified deaths in that year. Therefore, a small number of deaths may have occurred in areas shown as having no deaths.

Small-area estimates are particularly sensitive to low identified counts. Therefore, rates and annual changes should be interpreted cautiously, especially where the underlying numbers are small. Rates are crude rates for people aged 15 to 74 years, and are not age-standardised or rates among the population of people experiencing homelessness.

The statistics are based on year of registration rather than year of occurrence. Registration can take place after the death occurred, particularly where a coroner's investigation is required.

More quality and methodology information

More information on strengths, limitations, appropriate uses and how the data were created is available in our Deaths of people experiencing homelessness in England and Wales QMI.