Drug-related deaths and suicide in offenders in the community, England and Wales: 2011 to 2021

• We identified 8,385 deaths of offenders supervised in the community by the probation service between 2011 and 2021 in our death registrations database; this is equivalent to 762 deaths per year.

• Of the total identified deaths, 1,267 were suicides.

• The risk of suicide was six times higher in offenders in the community compared with the general population; the risk of suicide was over four times greater in male offenders and 11 times greater in female offenders than in the general population.

• Of the total identified deaths, 2,801 were drug-related deaths, 219 of which were drug-related suicide.

• The risk of drug-related death in offenders in the community was over 16 times greater than in the general population; the risk of drug-related death in male offenders was around 11 times greater than in the general population, and around 36 times greater in female offenders.

If you are a journalist covering a suicide-related issue, consider following the Samaritans' media guidelines on the reporting of suicide. The guidelines advise on terminology and include links to sources of support for anyone affected by the themes in the article.

If you are struggling to cope, please call Samaritans for free on 116 123 (UK and the Republic of Ireland) or contact other sources of support, such as those listed on the NHS help for suicidal thoughts webpage. Support is available around the clock, every day of the year, providing a safe place for you, whoever you are and however you are feeling.

This article reports Experimental Statistics and insights on deaths of offenders in the community by linking our mortality data and HM Prison and Probation Service (HMPPS) data together (see Section 4: Data linkage).

"Offenders in the community" refers to offenders who are supervised in the community by the probation service; around 0.4% of the general population were offenders in the community between 2011 and 2021.

HMPPS provides official estimates of Deaths of offenders in the community in England and Wales (anyone who died under the supervision of the probation service at the time of death). Deaths are then classified into "apparent cause of death" categories, assigned prior to possible inquest. These categories include "self-inflicted deaths", defined as any death of a person who has taken their own life irrespective of intent.

We hold data for all deaths registered in England and Wales which are assigned a final cause of death code based on a coroner's conclusion. However, it is not possible to identify whether the individual was under supervision by the probation service in this data alone. By linking our death registrations database with records held by the Ministry of Justice (MoJ) in the HMPPS data, we identified deaths occurring in this cohort, using National Statistics definitions of suicide and drug-related deaths (see Section 6: Glossary).

The analysis is based on deaths occurring in the period 2011 to 2021. Date of death occurrence was used to improve comparability with deaths of offenders in the community data. This analysis period coincides with the coronavirus (COVID-19) pandemic. While there is no evidence to suggest that the rate of suicide or drug-related death increased in the general population as a result of the pandemic, it should be considered in the context of this analysis.

The introduction of the Offender Rehabilitation Act in 2014 required all offenders on a custodial sentence to be subject to a minimum 12-month supervision under the probation service upon release. This created an increase in offenders in the community after 2014. Note that offenders under supervision in the community are not in the care of HMPPS in the same way as offenders in prison custody are. Therefore, the level of responsibility of the Probation Service for the well-being of offenders is substantially different to that of the Prison Service regarding deaths in custody.

Our previous article investigated suicide and drug-related deaths in prison custody in England and Wales. However, because of low numbers of deaths, analysis was conducted for males only. As the probation community contains a wider demographic who are likely to be exposed to different risk factors than the prison community, and levels of monitoring by Prison and Probation services are different, the risk between these cohorts is not directly comparable.

We identified 8,385 deaths occurring in offenders in the community from 2011 to 2021. The majority of these were male deaths (7,321, 87%) compared with female deaths (1,064, 13%).

Of the total deaths in offenders in the community during this period, 1,267 were suicides (15%) and 2,801 were drug-related (33%). Because of the overlap in our definitions (see Section 6: Glossary), 219 of the 2,801 drug-related deaths were also classified as suicides. Most other deaths were categorised as diseases of the circulatory system (International Classification of Diseases (ICD-10) codes I00 to I99; 1,030 deaths) and neoplasms (ICD-10 codes C00 to D48; 616 deaths). A full breakdown of cause of death by ICD chapter can be found in our accompanying dataset.

Suicide in offenders in the community

Of the 1,267 suicides in offenders in the community, 1,136 were male (90%) and 131 were female (10%). This is approximately 103 suicides each year for males and 12 for females.

The risk of suicide was six times higher between 2011 and 2021 in all offenders in the community compared with the general population, which remained constant over time. This is consistent with previous research, and is likely to be related to risk factors which are disproportionately present in offenders, such as substance misuse and mental health problems, as highlighted in Sirdifield et al's literature review published in 2020.

The risk of suicide in females was 11 times higher than the female general population, consistent with the National study of suicide in all people with a criminal justice history. In male offenders, the risk of suicide was around 4 times higher than the general population (Figure 1).

Figure 1: Offenders in the community are at a significantly higher risk of dying by suicide compared with the general population

Suicide death occurrences in offenders in the community in 2011 to 2021, standardised mortality ratios, males and females, England and Wales

Notes:

1. Estimates for 2019 to 2021 may be an underestimate because of registration delays and should be interpreted with caution..
2. Figures are based on the National Statistics definition of suicide for the year the death occurred. Figures are for persons aged 18 years and over.
3. The dashed line represents the level of suicide among the general population in England and Wales (SMR equals 100). When the value of the SMR per period is above 100, the risk of suicide is higher in the offenders in the community population compared with peers of the same age group and sex in the population.
4. Error bars show the upper and lower confidence limits, which is a margin of error around the SMR estimates. When the range of the upper and lower confidence intervals exclude 100, the risk of suicide for a given period is statistically different to that observed in the general population of England and Wales.

Download the data

Suicide rates also remained statistically significantly higher in offenders in the community compared with the general population between 2011 and 2021. Rates in male offenders peaked between 2012 and 2014, decreased in 2015, and have remained constant over time (see Figure 2). Female offenders had lower rates of suicide across the period compared with male offenders, though they remain statistically significantly higher than in the general population (see Figure 3). For details on how we calculate these rates, see Section 7: Measuring the data.

The most common method of suicide among all offenders in the community was hanging, strangulation and suffocation, which accounted for 64% (808) of all suicide deaths between 2011 and 2021. This is similar to the general population.

For males, the most common method was also hanging, strangulation and suffocation (66%, 755 deaths). However, in females, the most common method of suicide was poisoning (54 deaths, 41%).

Drug-related deaths in offenders in the community

Of the total drug-related deaths in offenders the community between 2011 to 2021 (2,801 deaths), 2,345 were male (84%) and 456 were female (16%). This equates to 255 drug-related deaths in offenders in the community each year.

The risk of drug-related death in offenders in the community was over 16 times greater than the general population (Figure 4). In male offenders, the risk of drug-related death was around 11 times greater than in the general population, while the risk for female offenders was around 36 times greater. This corresponds to Merrall et al's meta-analysis of drug-related deaths, in which offenders in the community are shown to be at significantly greater risk of drug-related death.

Age-standardised rates indicated that, between 2011 and 2021, rates of drug-related deaths have more than doubled in offenders in the community in both males and females, with a particular increase occurring after 2015 (Figures 5 and 6). This increase in recent years is reflected in our Deaths related to drug poisoning in England and Wales bulletins.

Figure 4: Offenders in the community are at a significantly higher risk of dying by drug-related death compared with the general population

Drug-related death occurrences in offenders in the community 2011 to 2021, standardised mortality ratios, females and males, England and Wales

Notes:

1. Estimates for 2019 to 2021 may be an underestimate because of registration delays and should be interpreted with caution. 
2. Figures are based on the National Statistics definition of drug-related deaths for the year the death occurred. Figures are for persons aged 18 years and over.
3. The standard mortality ratio (SMR) is a common measure of mortality. The dashed line represents the level of drug-related deaths among the general population in England and Wales (SMR equals 100). When the value of the SMR per period is above 100, the risk of drug-related death is higher in the offender population compared with peers of the same age group and sex in the population.
4. Error bars show the upper and lower confidence limits, which is a margin of error around the SMR estimates. When the range of the upper and lower confidence intervals exclude 100, the risk of drug-related death for a given period is statistically different to that observed in the general population of England and Wales.

Download the data

Drug-related deaths by substance

Figure 7 shows the number of times a substance was mentioned on the death certificate, with or without mentions of other drugs. Therefore, there are more mentions of drugs (6,841 mentions) than the overall number of individual drug-related deaths (2,801 deaths) for males and females. It is not possible in some cases to tell which substance was primarily responsible for the death.

Between 2011 and 2021, opiates were the most common substance mentioned (1,722, 25% of all mentions) in deaths of offenders in the community. Heroin and morphine (1,126 mentions) were the most mentioned form of opiate. Opiates were also the most common substance mentioned in drug-related deaths in the general population during this period.

The second most mentioned substance on the death certificate was cocaine (525, 8% of all mentions), which was three times higher in 2021 compared with 2015. Benzodiazepines were also commonly mentioned on the death certificate (465, 7% of all mentions), with Diazepam being the most common form (319 mentions).

A more detailed breakdown of drug-related deaths by substance type can be found in our accompanying dataset.

Characteristics of suicide and drug-related deaths in offenders in the community

Most deaths from all causes in offenders between 2011 and 2021 occurred in those with a court-ordered sentence (56%), with 42% of deaths occurring in post-release cases, and 3% unclassified (type of supervision unknown). See Section 6: Glossary for the definition of each sentence type. A greater proportion of suicides occurred in those on a post-release sentence type (67%), compared with 30% with a court-ordered sentence. For drug-related deaths, a greater proportion occurred in those with a court-ordered sentence (51%) compared with those on post-release (47%) on average. However, the number of drug-related deaths in those with a post-release sentence was greater between 2018 to 2021. Where latest release date for those with a post-release sentence was available, most suicide and drug-related deaths occurred between 101 and 365 days after release.

Out of all offender deaths:

• 92% were recorded as being White

• 7% were classified as "all other ethnic groups combined" (recorded as Asian, Mixed, Black, or Other ethnicity by HM Prison and Probation Service (HMPPS))

• 2% were unknown or not stated ethnicity

For offenders with an ethnicity of White:

• 15% of all deaths were classified as suicide

• 34% were classified as drug-related death

For offenders included in the "all other ethnic groups" category:

• 13% of deaths were classified as suicide

• 23% were classified as drug-related death

For additional information please see our accompanying dataset.

Comparing the cause of death recorded by HM Prison and Probation Service with the Office for National Statistics definitions of suicide and drug-related deaths

The system used by the Ministry of Justice (MoJ) for recording deaths under probation supervision provides a classification of the cause of death for administrative and statistical purposes. This is amended when the final underlying cause of death has been determined by the coroner at the inquest. Further information on HMPPS classification of death and how this has changed over time can be found in the Guide to deaths of offenders in the community statistics (PDF, 203KB).

We receive information on deaths when they are registered after a coroner's inquest, which are coded using the International Classification of Diseases (ICD-10).

In the linked data, 2,693 deaths were classified as self-inflicted deaths by HMPPS; most of these deaths (1,012 deaths, or 38%) were defined as suicide and 1,318 additional deaths were defined as drug-related deaths, excluding suicide (49%) (Table 1).

Of the remaining HMPPS self-inflicted deaths which were not included in our suicide or drug-related deaths definitions (363):

• 118 were classified as "external causes of morbidity and mortality", including accidents and assaults

• 69 were classified as "disease of the digestive system" including diseases of the liver

• 58 were classified as "diseases of the circulatory system"

• 118 were classified as infectious, respiratory, blood, metabolic, musculoskeletal or nervous system disease, mental or behavioural disorder, neoplasm, and unknown or uncertain cause of mortality

For drug-related deaths (2,801), most (51%) were classified by HMPPS as "self-inflicted". A further 30% were classified by HMPPS as "unclassified" and 12% as "natural causes".

Registration delays

Deaths with an unknown cause or suspected suicide are subject to a coroner's inquest. Therefore, the investigation of many deaths included in this analysis can take additional time to establish the cause of death.

The median registration delay for suicide deaths in offenders in the community between 2011 and 2021 was 166 days, similar to 158 days for the general population.

For drug-related deaths, the median registration delay was 180 days for both offenders in the community and the general population.

In this article, we identified those who had died in offenders in the community by confidentially linking records in our death registrations database with records held by the Ministry of Justice (MoJ). The data were processed securely, and linkage was approved by the National Statistician's Data Ethics Advisory Committee.

We were provided with a record-level extract of deaths from the HM Prison and Probation Service (HMPPS) for all deaths that occurred under probation supervision from April 2010 to March 2022, inclusive. The extract contained a total of 10,539 individual records and included unique identifiers (for example, names and date of birth), the date of death and the cause of death category assigned to each death (see Section 7: Measuring the data.

Pre-processing

Before linkage, both datasets underwent a basic normalisation process to achieve consistency. We removed any records where the date of death was not present in the HMPPS data (168 records), and any records where the combination of date of birth, first name and surname were not present were excluded (90 cases).

Data for 2022 were excluded from the linkage, as records were only present from January to March and, therefore, did not constitute a full calendar year (284 records). This resulted in 9,997 valid records being taken forward for linkage.

One record was removed from the data where age at death was recorded as 17 years. The HMPPS data does include a small proportion of individuals who have been recorded as being under the age of 18 years; this is likely because the date of birth was incorrectly recorded.

The linkage

We linked the two datasets using available unique identifiers (forename, surname, date of birth and date of death). The linkage process consisted of direct and approximate join types.

We directly joined 5,506 records out of 9,997 records over 2010 to 2021 between the two datasets (55%).

For the records we were unable to match directly, we used the "Jaro-Winkler" approximate joining method. This method calculates the similarity between two strings from each dataset, returning a "similarity distance" between zero and one (where zero indicates no similarity between strings and one indicates an exact match). Of the total records, 4,492 (45%) were approximately joined and assigned a similarity distance. The minimum distance we accepted was 0.9; this high threshold was set to correct for the uncertainty being introduced by shorter forenames and surnames. Therefore, we excluded any approximately joined records which did not meet this threshold (1,234 records excluded).

Following direct and approximate joining, a total of 8,764 valid cases were manually checked to ensure records matched. Any remaining cases where all four identifiers (first name, surname, date of birth and date of death) did not match the Office for National Statistics (ONS) record were removed. Any cases where dates matched, but first name did not match were manually checked against ONS records to confirm identity (for example, checked against middle name). During this checking, a further 123 cases were removed.

Linkage results

This provided us with a total of 8,641 out of 9,997 valid linked records over 2010 to 2021 (86%), leaving 1,356 valid records we were unable to link (14%). Given the much larger proportion of non-linked records for 2010 (255 out of 446 total non-linked records for 2010, 57%), we did not include these records in the analysis. This resulted in the final linkage rate of 8,386 out of 9,641 valid records (87%) between 2011 and 2021.

The percentage of linked records was lowest in the earliest (2011) or latest (2021) years (see Table 2). For 2021, this is likely because the ONS does not yet have the death registrations in the most recently available data at the time of analysis, because of incomplete coroner inquests affecting registration delays (see Section 3: Deaths in offenders in the community). For 2011, this is likely to be because of a small number of inaccuracies or disparities in HMPPS deaths data (for example, recording of first name).

Age-standardised mortality rate

Age standardised morality rate (ASMR) refers to a weighted average of the age-specific mortality rates per 100,000 people and standardised to the 2013 European Standard Population. ASMRs allow for differences in the age structure of populations and, therefore, allow valid comparisons to be made between geographic areas, between the sexes, and over time.

Suicide

The definition of suicide used in this article includes all deaths from intentional self-harm for people aged 10 years and over, and deaths where the intent was undetermined for those aged 15 years and over.

The International Classification of Diseases 10 (ICD-10) codes used to define suicides are as follows:

• intentional self-harm (people aged 10 years and above) (ICD-10 codes X60 to X84)

• injury or poisoning of undetermined intent (people aged 15 years and above) (ICD-10 codes Y10 to Y34)

Drug-related deaths

Drug-related deaths involve a broad spectrum of substances. As well as deaths from drug abuse and dependence, figures also include accidents and suicide involving drug poisonings as well as complications of drug use. They do not include other adverse effects of drugs (for example, anaphylactic shock), or accidents caused by an individual being under the influence of drugs. Further information can be found in our Deaths related to drug poisoning in England and Wales bulletin.

There is an overlap between the National Statistics definitions of suicides and drug-related deaths, whereby deaths caused by intentional self-poisoning and undetermined poisoning by drugs will be included in both suicide and drug-related analyses.

The ICD-10 codes used to define drug-related deaths are as follows:

• mental and behavioural disorders because of drug use (excluding alcohol and tobacco) (ICD-10 codes F11 to F16, F18 to F19)

• accidental poisoning by drugs, medicaments, and biological substances (ICD-10 codes X40- to X44)

• intentional self-poisoning by drugs, medicaments, and biological substances (ICD-10 codes X60 to X64)

• assault by drugs, medicaments, and biological substances (ICD-10 codes X85)

• poisoning by drugs, medicaments and biological substances with undetermined intent (ICD-10 codes Y10 to Y14)

Offenders in the community

Offenders in England and Wales who are under the supervision of the probation service. This includes offenders who are serving court order sentences in the community (that is, not in prison custody), or offenders who are being supervised by the probation service following a custodial sentence. This excludes any offenders currently in prison custody.

Sentence type

Offenders in the community are supervised by the probation service and can hold one of the following sentence types:

• "post-release", where the offender has entered the community following a custodial sentence

• "court order", where the offender has remained in the community supervised by the service as part of a court ordered period

• "unclassified", where the sentence type is not disclosed for the offender

Standardised mortality ratio

Standardised mortality ratio (SMR), in this bulletin, refers to ratios that allow the calculation of the expected mortality risk for a population, given age specific mortality rates from a reference population. 

Ministry of Justice data

HM Prison and Probation Service (HMPPS) uses a classification for apparent causes of death:

• natural causes: any death because of a naturally occurring disease

• self-inflicted: any death of a person who has apparently taken their own life irrespective of intent

• homicides: any death at the hands of another

• accident: any death arising from external causes

• other: any death that cannot be easily classified as natural causes, self-inflicted, accidental or homicide

• unclassified: any death where there is insufficient information to make a judgement about the cause at the time of reporting

Each death is classified based on information at the time of death and reported to HMPPS by the probation provider. Cause of death classification is revised following inquest where necessary. Further information on the data source can be found in the MoJ's guide to deaths of offenders in the community (PDF, 144KB).

Death registrations data

Following coroner inquest and death registration, the Office for National Statistics (ONS) assigns each death with a code from the International Classification of Diseases (ICD-10), based on the coroner's conclusion.

Population data

To define the offenders in the community population at risk, we used probation population data, produced by the MoJ. Because of a technical error in the population data archive, data for offenders in community in 2016 were not provided. To address this, we took the median of 2015 and 2017 estimates to fill population data for 2016.

Population data for offenders in the community provides a snapshot in time for each year of the analysis, and as such may under or overestimate risk because of movement of offenders into and out of the supervision of the Probation Service.

To calculate rates in the general population of England and Wales, we used our mid-year population estimates for England and Wales.

Standardised mortality ratio

Most of the findings in this article are based on the standardised mortality ratio (SMR) (see Section 6: Glossary). This method allows us to calculate how many suicide or drug-related deaths would be expected in offenders in the community, should the overall pattern for these causes of death in the general population be the same. Further information on SMR calculation can be found in our previous article, Drug-related deaths and suicide in prison custody.

We report SMRs using three-year rolling periods. This helps to smooth out fluctuations over time. The SMR should not be used to understand how the level of risk changes over time; for this purpose, we have also included age-standardised rates. Further information on age-standardised rates can be found in our Suicide rates in the UK QMI.

Limitations of the analysis

Official estimates of self-inflicted deaths in the offenders in the community should be taken from the MoJ's deaths of offenders in the community statistics. It should be noted that the total numbers of deaths included in this analysis are not equal to those in offenders in the community statistics, because of both the linkage process and the differences in HMPPS classification and our definitions.

The analysis is based on linked deaths that occurred during the period 2011 to 2021 and excluded cases from 2010 where the proportion of non-matched cases was higher. It may be likely that these findings underestimate the level of risk in more recent years because of registration delays.