Official Statistics

Drug and alcohol treatment for victims and suspects of homicide: report

Published 28 March 2024

Applies to England

Summary

This report presents the results of the first data linkage between the Home Office Homicide Index (HI) and the National Drug Treatment Monitoring System (NDTMS), which is owned by the Department of Health and Social Care (DHSC). DHSC’s Office for Health Improvement and Disparities (OHID) carried out this linkage as a proof of concept and to better understand the relationship between homicide and substance misuse treatment. In English law, the offence of homicide is an act resulting in the death of another person. It includes murder and manslaughter.

The data linkage explores the use of substance misuse services by victims and suspects of homicide. We intend that this report’s findings will contribute to part of the evidence base for homicide prevention policy.

The report is official statistics in development and is part of the Better Outcomes through Linked Data (BOLD) cross-government programme led by the Ministry of Justice (MOJ).

The NDTMS extract extended to March 2022, a year beyond the HI extract, to enable us to analyse suspects after the homicide.

Methodology

Between April 2019 and March 2021, there were 1,245 homicide victims and 789 suspects recorded in the HI. Compared with most other crimes, the homicide rate remains very low, with 9.9 homicides recorded per million population during the year ending March 2023. This was the lowest rate per population since the year ending March 2016. Between August 2018 and March 2022, NDTMS had records of 442,242 people in community treatment and 53,482 people receiving treatment in secure settings, using standard NDTMS methodology to identify an individual person.

We linked the HI extract covering the period April 2019 to March 2021 to the NDTMS extract covering the period August 2018 to March 2022 using deterministic and probabilistic methods. The linkage processes identified 36 victims and 90 suspects of homicides occurring between April 2019 and March 2021 who had at least one treatment record between August 2018 and March 2022. We undertook the analysis separately for victims and suspects.

Main findings

This report demonstrates the feasibility of linking the HI with NDTMS to explore how many homicide victims and suspects have been engaged in substance misuse treatment. These results will contribute to the evidence base to inform policy and commissioning of services. The main findings from our analysis include the following.

We identified 3% (2% to 4%, 95% confidence interval (CI)) of victims and 11% (9% to 14%, 95% CI) of suspects in the HI as having a treatment record in NDTMS.

The CI represents the range that we are 95% confident that the true rate lies within.

There were 118 homicide incidents in which a victim or suspect was identified as having at least one treatment record for either drugs or alcohol. This corresponds to 10% (8% to 12%, 95% CI) of all homicide incidents in the HI for the period of the analysis. The number of homicide incidents is lower than the number of individuals because some homicide incidents involved multiple victims or suspects that were identified with a treatment record.

Most suspects (68% of those with a treatment record) had a treatment record only after the homicide, most of which was for treatment in prison.

Almost two-thirds (62% of those with a treatment record) of all suspects in the HI were involved with a homicide that was recorded as drug-related. For suspects with a treatment record, this proportion was 80%.

When compared to the overall population in treatment, both victims and suspects were less likely to be in treatment for opiates. Most suspects in treatment after the homicide (68% of those with a treatment record) were in treatment for alcohol and non-opiates or non-opiates only.

Limitations

This publication is the first time OHID and the Home Office have collaborated on data linkage to better understand the relationship between homicide and substance misuse.

The linkage rate (the proportion of people in the HI that were matched to NDTMS data) is likely to be an underestimate where insufficient or conflicting data meant we could not find a link. Our findings will also not include people who have substance misuse problems but have not accessed treatment in this period or ever.

Still, our analysis shows the feasibility of the linkage and the initial findings suggest that ongoing linkage and analysis can provide valuable insights for evidence, policy and delivery.

Introduction

Better Outcomes through Linked Data

Better Outcomes through Linked Data (BOLD) is a 3-year cross-government programme led by MOJ. It was created to demonstrate how people with complex needs could be better supported by linking and improving the cross-government data held on them, in a safe and secure way.

There are 4 main BOLD demonstrator pilots. These are focused on:

• reducing homelessness (led by the Department for Levelling Up, Housing and Communities)
• supporting victims of crime (led by MOJ)
• reducing reoffending (led by MOJ)
• substance misuse (led in England by OHID and in Wales by Public Health Wales)

Homicide victims and suspects’ engagement with substance misuse services

This report presents results from a collaboration between OHID and the Home Office to explore the use of specialist substance misuse services by victims and suspects of homicide in England.

Research on drugs and violence (Goldstein 1985) showed that people who use illicit drugs are at greater risk of being a victim or suspect in a homicide. The Home Office report Trends and drivers of homicide sets out 3 ways in which illicit drugs might cause homicide. These are:

• psychopharmacological (related to an immediate loss of inhibition, cognitive dissonance and aggression)
• economic (related to the basic need to finance drug use or addiction and craving, which may contribute to high-risk taking behaviour to secure any means of funding)
• systemic (related to prohibition, illegal drug markets and territories, and associated gang warfare and debt collection enforcement)

This report uses 2 sources of data:

• homicide data from the Home Office
• substance misuse treatment data from OHID

Homicide data

The Home Office receives 2 sources of information on homicide from police forces in England. These are the: 

• aggregated police recorded crime (PRC) return
• Homicide Index (HI)

The Office for National Statistics publishes information from both sources. Information on homicide from the aggregated PRC return is published quarterly. More detailed information from the HI is published annually at Homicide in England and Wales articles.

This published HI data showed there were 661 victims of homicide in England in the year ending March 2022, which was 113 (21%) more than the previous year (548). However, the year ending March 2021 covered time periods where COVID-19 restrictions were in place to limit social contact. These restrictions may have affected patterns of various crime types during this period, including homicide offences. The number of homicides in the latest year (ending March 2022) was more in line with the 4 pre-pandemic years, from the year ending March 2017 (661 victims) to the year ending March 2020 (638 victims). Overall, the recurring theme through the years is that homicide crimes continue to be rare in England.

The HI includes information on all homicides, including whether they were drug-related. The Home Office’s definition of a drug-related homicide includes any case where the police believe the victim or suspect:

• was an illegal drug user or dealer
• had taken a drug
• had motive to obtain drugs or steal drug proceeds

It also includes cases which were otherwise marked as drug-related.

The proportion of homicides in England and Wales that have involved drug users or dealers, or have been related to drugs in any way, has increased over the last decade. They rose from 43% in the year ending March 2012, to 52% in the year ending March 2022. Figures for England only are not available.

From April 2021 to March 2022 there were 360 homicides that were thought to involve drug users or dealers or were in some way drug-related. This was 58 more than the previous year and the highest number since the Home Office began collecting this information in the year April 2007 to March 2008.

This report uses HI data from April 2019 to March 2022. We chose this period to ensure there was a reasonable period preceding it where individuals had contact with treatment services.

NDTMS treatment data

Publicly funded specialist drug and alcohol treatment services are available in every local authority in England. Each month, data from people accessing these services is collected by the National Drug Treatment Monitoring System (NDTMS).

In NDTMS reporting, people in treatment are divided into 4 substance groups:

1. opiate: people who are dependent on or have problems with opiates, mainly heroin
2. non-opiate: people who have problems with non-opiate drugs only, such as cannabis, crack and ecstasy
3. non-opiate and alcohol: people who have problems with both non-opiate drugs and alcohol
4. alcohol only: people who have problems with alcohol but do not have problems with any other substances

You can find more details on these substance groups in the annual report Adult substance misuse treatment statistics 2021 to 2022. You can also find other reports from NDTMS, including about young people (aged under 18) and prisons and secure settings, in OHID’s Alcohol and drug misuse and treatment statistics collection.

NDTMS data showed that there were 289,215 adults in contact with drug and alcohol treatment services in the community between April 2021 and March 2022. This is a rise of 4.8% compared with the previous year (275,896). NDTMS data also showed that there were 11,326 young people (under the age of 18) in contact with drug and alcohol services in the same period.

Drug and alcohol treatment is also available in prisons and secure settings in England, and NDTMS also collects this data. NDTMS data showed that there were 45,096 adults in prisons and secure settings in drug and alcohol treatment between April 2021 and March 2022. NDTMS data also showed that there were 553 young people receiving drug and alcohol treatment in prison between April 2021 and March 2022, the lowest number since reporting began in 2015.

This report uses NDTMS data from August 2018 to March 2022. We gained approval to use NDTMS for data linkage and research in 2018 through updated patient information.

The NDTMS extract extended to March 2022, a year beyond the HI extract, to enable suspect analysis following the homicide.

Aims and objectives

The aim of this project was to carry out an accurate assessment of the association between drug and alcohol dependence and the risk of dying by or committing homicide.

To provide evidence for this assessment, we answered the following research questions:

1. What proportion of homicide victims and suspects have been in drug and alcohol treatment?
2. What is the sociodemographic and clinical profile of homicide victims and suspects (for example, sex, age, ethnicity, drug use and dependence, age of first use, injecting status, referral source, accommodation and employment status)?
3. How many victims and suspects successfully completed their last substance misuse treatment?
4. How many victims and suspects were engaged in substance misuse treatment on the date of the homicide?
5. How much time passed since the victim and suspect previously engaged with substance misuse treatment?
6. For suspects who have engaged with treatment, what types of homicide do they tend to be involved in? For example, was there a relationship to the victim, and if so, what kind of relationship did they have, what was the method of homicide and the location of the homicide?
7. Has the proportion of homicide victims and suspects accessing treatment changed over time?
8. What (if any) is the relationship between police indicators for drug-related homicide and treatment indicators focused on levels of engagement, retention in treatment, planned or unplanned discharge and successful completion?
9. What (if any) is the relationship between police indicators for mental health related homicide and treatment indicators focused on levels of engagement, retention in treatment, planned or unplanned discharge and successful completion?

We intend that this report’s findings will contribute to the evidence base for homicide prevention policy.

This work has not, and will not, involve any follow up with individuals in NDTMS or the HI, or their families.

Methodology

Overview of data linkage

Alongside demographic data, the HI captures details on the location and method of the homicide and the circumstances surrounding the homicide. For suspects, the HI also records:

• whether they had previously committed a homicide
• the relationship with the victim
• the conclusion of the Crown Court trial

NDTMS collects data on all people accessing publicly funded structured drug and alcohol treatment in England. The data collected includes:

• sociodemographic information
• clinical reviews
• physical health status
• mental health status
• the substances people have problems with
• the interventions people receive
• people’s treatment outcomes

Consent and confidentiality

People accessing drug and alcohol treatment are asked for consent to share their information in line with the NDTMS consent and privacy notice. Part of this consent allows their information to be linked with other administrative data sets. There is no unique identifier that enables NDTMS to perfectly link the same person with the HI. However, the 2 systems share information about:

• a person’s forename and surname initials
• a person’s date of birth
• a person’s sex and ethnicity
• what area a person lived in

Statistical approaches used

We used the shared information to link the HI and NDTMS using deterministic and probabilistic approaches.

Deterministic linkage occurs when records from 2 or more data sets agree on a set of identifiers. For the deterministic linkage in this project, a perfect match was required on initials, sex, date of birth and area a person lived.

Probabilistic linkage gives more flexibility to identify a probable match in instances where a person’s information might have changed, for example their surname changing after marriage. It assigns a match probability for 2 records. A threshold can then be chosen to determine what counts as a match for analytical purposes.

Data scientists undertook clerical checks on a random sample of record pairs. This means manually examining each combination of records to determine whether they match, to understand what probability threshold would identify the most true matches, while minimising the number of false matches being misclassified as true matches. We chose the threshold to be 98.7%. The data scientists then viewed all links above this threshold and removed about 10% of the links in which they had the least confidence, to ensure we had confidence in the matched data.

Managing the data

The final linked data set did not contain enough records to be able to be anonymised in line with the Information Commissioner’s Office code of practice on anonymisation (PDF, 1.9MB).

It was not possible to remove all direct identifiers, so the data set is classed as pseudonymised, not fully anonymised. This means there is a theoretical potential for individuals to be identified. To mitigate this risk, access to the final data set was strictly limited to a small number of OHID staff.

Results

Data set linkage

The HI had a total of 2,034 records between April 2019 and March 2021. This included 1,245 homicide victims and 789 suspects, of which 569 were the principal suspect (in homicides involving more than one suspect).

Where there are multiple suspects, one is recorded as the principal suspect. Police forces can use their judgement to select which suspect is the principal suspect. For example, they can consider which suspect was most involved in the homicide such as which one fired the weapon or landed the most blows. However, they may select the suspect with the closest relationship to the victim. Following the court outcome, they should revise the principal suspect to the one who received the longest prison sentence.

In NDTMS, there were 53,482 individuals accessing treatment in secure settings, and 442,242 individuals who started treatment in a community setting between August 2018 and March 2022.

The linkage processes (first deterministic, followed by probabilistic) found 126 individual records with at least one treatment record for either drugs or alcohol. Of these, 36 were victims (3% of all victims in the HI for this period) and 90 were suspects (11% of suspects). We then determined that these corresponded to 118 homicide incidents in which a victim or suspect was identified as having at least one treatment record for either drugs or alcohol. This was around 10% of all homicide incidents in the HI for the period. The number of homicide incidents is lower than the number of individuals, as some homicide incidents involved multiple victims or suspects that were identified with a treatment record.

We conducted the analysis in this report using the results of the probabilistic linkage.

Treatment characteristics of victims and suspects

Timing between homicide incidents and contact with treatment

Of all the homicide incidents involving a victim or suspect with at least one treatment record for either drugs or alcohol (118 incidents), there were:

• 57 where the suspect was only detected in treatment after the homicide
• 61 where the victim or suspect were detected in treatment before the homicide

Where we report people’s level of engagement with treatment as a rate, we give it alongside its 95% confidence interval (CI). The CI represents the range that we are 95% confident that the true rate lies within.

Treatment status of victims

Most victims who had at least one treatment record (36 in total) were actively engaged in treatment at the time of the homicide (26, 72% of linked victims) (see table 1). This represents 3% of all victims (2% to 4%, 95% CI). It is not possible to do further analysis on the remaining 10 victims due to statistical disclosure requirements (breaking these down would mean the numbers are too low and individuals’ details could potentially be disclosed).

Table 1: treatment status of victims linked with NDTMS

Treatment status	Number of victims detected in treatment	% of victims detected in treatment	% of all victims
Victim left treatment before the homicide	10	28%	1%
Victim was actively engaged in treatment at the time of homicide	26	72%	2%
Total victims	36	100%	3%

Treatment status of suspects

Overall, 90 suspect records in the HI had a record in NDTMS between August 2018 and March 2022, representing 11% of suspects (9% to 14%, 95% CI). Most (68%) were only identified in treatment following the homicide (61 suspects) while 32% (29 suspects) had been engaged in treatment before or at the time of the homicide. This equates to 8% (6% to 10%, 95% CI) of all suspects in the HI as having newly engaged in treatment services following the homicide, while 4% (2% to 5%, 95% CI) of suspects had been in treatment on or before the date of homicide (see table 2).

Of the 61 suspects who only engaged in treatment after the homicide, a small number (fewer than 5) received treatment in the community, while the majority received prison treatment. However, as the treatment data only covers a limited period, suspects could have been in treatment before August 2018 or after March 2022 without it being picked up in this report.

Table 2: treatment status of suspects linked with NDTMS

Treatment status	Number of suspects detected in treatment	% of suspects detected in treatment	% of all suspects in HI
Suspect only in treatment after the homicide	61	68%	8%
Suspect engaged in treatment before or at the time of the homicide	29	32%	4%
Total suspects	90	100%	11%

Sociodemographic and clinical profile of victims and suspects engaged in treatment

For the results in this section, we present sub-categories as percentages instead of numbers due to the small number of cases and the need to protect anonymity of individuals. Any that relate to a number less than 5 have been supressed and replaced by a rounded-up figure. We took the figures for the ‘General NDTMS community treatment population’ column in table 3 from the NDTMS report Substance misuse treatment for adults: statistics 2020 to 2021. NDTMS records age, ethnicity and substance groups for all adults in treatment during the reporting period (289,215 individuals). But it records injecting status, referral source and accommodation status only for people who are new to treatment during the reporting period (133,704 individuals).

Table 3 shows the age distribution at the start of treatment for both victims and suspects with a treatment record. Data on suspects is divided into those with a treatment journey with a starting date before the homicide and those with a treatment journey solely after the homicide.

For suspects in treatment at the time of or before the homicide, the largest group was 25 to 39 years old (41%). For suspects only with a treatment journey after homicide, the largest group was under 25 years old (39%). For victims, 50% were aged 40 years and over. Both victims and suspects of homicide are more likely to be under 25 than the general NDTMS community treatment population.

Table 3: age (at start of treatment journey closest to homicide incident) of victims and suspects linked with NDTMS

Age	Victims	Suspects in treatment at time of or before the homicide	Suspects only with a treatment journey after homicide	General NDTMS community treatment population
Under 25	19%	31%	39%	6%
25 to 39	31%	41%	38%	36%
40 and over	50%	28%	23%	58%
Total	100%	100%	100%	100%
Total individuals	36	29	61	289,215

Table 4 reports the ethnicities of victims and suspects with treatment records. The largest ethnic group of both victims and suspects who were linked to NDTMS were recorded as white.

Table 4: ethnicity of victims and suspects linked with NDTMS using NDTMS ethnicity data

Ethnicity	Victims	Suspects in treatment at time of or before the homicide	Suspects with a treatment journey solely after the homicide	General NDTMS community treatment population
White	72%	66%	72%	88%
Ethnicities other than white, not stated or missing	28%	34%	28%	12%
Total	100%	100%	100%	100%
Total individuals	36	29	61	289,215

What substances bring victims and suspects into treatment

Many victims and suspects experience difficulties with drugs and alcohol and receive treatment for both.

Table 5 shows the main substance groups for victims and suspects who received treatment. Suspects who were in treatment on or before the date of the homicide were mainly treated for opiates (41%). Suspects who started treatment after the homicide were mainly treated for alcohol and non-opiate substances (69%).

Table 5: NDTMS substance groups for victims and suspects linked with NDTMS

Substance group	Victims	Suspects in treatment at time of or before the homicide	Suspects only with a treatment journey after homicide	General NDTMS community treatment population
Alcohol and non-opiate and non-opiate only	31%	41%	69%	22%
Alcohol only	36%	17%	16%	29%
Opiates	33%	41%	15%	49%
Total	100%	100%	100%	100%
Total individuals	36	29	61	289,215

Note: percentages may not total 100% due to rounding

Injecting status at start of treatment

Table 6 presents information on self-reported injecting status at the start of treatment. Most victims and suspects were recorded as having never injected, ranging from 53% for victims to 79% for the suspects who were only in treatment after the homicide. When compared to the general NDTMS community treatment population, a high proportion of victims and suspects (between 11% and 17%) had an unknown injecting status.

Table 6: injecting status of victims and suspects linked with NDTMS

Injecting status	Victims	Suspects in treatment at time of or before the homicide	Suspects with a treatment journey solely after homicide	General NDTMS community treatment population
Currently injecting or have injected previously	31%	17%	10%	17%
Never injected	53%	55%	79%	83%
Unknown	17%	28%	11%	1%
Total	100%	100%	100%	100%
Total individuals	36	29	61	133,704

Note: percentages may not total 100% due to rounding

Referral source at the start of treatment

When people start drug and alcohol treatment in the community, they self-report their referral source for treatment. People who start treatment in prison do not report their referral sources.

Table 7 shows that 47% of victims were self-referred or referred by a family member or friend. The most frequent referral route for suspects in treatment before or at the time of the homicide was through the criminal justice system (38%) or by themselves, a family member or a friend (38%). Most suspects who started treatment after the homicide did so in prison.

Table 7: referral source of victims and suspects linked with NDTMS

Referral source	Victims	Suspects in treatment at time of or before the homicide	Suspects with a treatment journey solely after homicide	General NDTMS community treatment population
Criminal justice system	19%	38%	Less than 10%	12%
Health and social care	19%	Less than 20%	0%	18%
Other	14%	Less than 20%	0%	6%
Self, family or friend	47%	38%	0%	59%
Substance misuse service	0%	0%	0%	4%
In treatment in prison	0%	0%	More than 90%	Not applicable
Total	100%	100%	100%	100%
Total individuals	36	29	61	133,704

Note: percentages may not total 100% due to rounding

To stop individuals’ details being disclosed, we have suppressed numbers less than 5 and given a percentage the number is less than. We have also done the same to the next smallest value in the column, because this could be used to identify the value under 5. We have used the same method in other tables in this report.

Accommodation status at start of treatment

When people start treatment in the community, they are asked about whether they have a housing problem. This question is not asked to those starting treatment in prison.

As shown in table 8, most victims and suspects who were in treatment at the time of or before the homicide had no housing problem when they started treatment.

Table 8: accommodation status of victims and suspects linked with NDTMS

Accommodation status	Victims	Suspects in treatment at time of or before the homicide	Suspects with a treatment journey solely after homicide	General NDTMS community treatment population
Housing problem or urgent housing problem	25%	Less than 20%	0%	16%
No housing problem	75%	86%	Less than 20%	84%
Other	0%	Less than 20%	0%	0%
In treatment in prison	0%	0%	Less than 80%	Not applicable
Total	100%	100%	100%	100%
Total individuals	36	29	61	133,704

Note: percentages may not total 100% due to rounding

Employment status at start of treatment

We show the employment status of victims and suspects who had treatment records in table 9. We have not given the employment status numbers for the general NDTMS community treatment population because they are not publicly available.

The question about employment status is not part of the prison data set. So, data is not available for most of the suspects engaged with treatment after the homicide (fewer than 5 suspects engaged in community treatment after the homicide).

Victims and suspects in treatment before the homicide (47% and 41% respectively) were unemployed or economically inactive. While almost 20% of victims were in regular employment, no suspects in treatment before the homicide were in regular employment. A similar proportion of both victims (22%) and suspects (28%) in treatment before the homicide were recorded as long-term sick or disabled.

Table 9: employment status of victims and suspects linked with NDTMS

Employment status	Victims	Suspects in treatment at time of or before the homicide	Suspects only with a treatment journey after homicide
Long-term sick or disabled	22%	28%	0%
Regular employment	Less than 20%	0%	0%
Unemployed or economically inactive	47%	41%	Less than 10%
No record of employment status	Less than 20%	31%	Less than 10%
In treatment in prison	0%	0%	More than 80%
Total	100%	100%	100%
Total individuals	36	29	61

Characteristics of the homicide for victims and suspects

The HI records information about the following categories for the principal suspect only:

• relationship to victim
• mental state of suspect
• domestic homicide

In the HI data set, we found a total of 73 principal suspects recorded that had a treatment record in NDTMS.

Relationship between homicide victim and principal suspect: breakdown for suspects

We show the relationship between victims and principal suspects with and without a treatment record in table 10. We could not break down suspects into being treated before or after homicide due to small numbers.

The proportion of principal suspects that were strangers or unknown to the victim was similar between principal suspects with a treatment record and other principal suspects identified in the HI (42% and 43% respectively). However, a greater proportion of those in treatment were friends or acquaintances with the victim than other principal suspects in the HI (44% compared with 27%).

Table 10: relationship between principal suspect and homicide victim

Relationship to victim	Principal suspects with treatment record	Principal suspects in the HI
Friend or acquaintance	44%	27%
Stranger or unknown (includes ‘other’, ‘relationship not known’ or ‘no suspects’)	42%	43%
Partner or ex-partner	Less than 10%	18%
Family (includes ‘parent’, ‘son or daughter’ or ‘other family’)	Less than 10%	13%
Total	100%	100%
Total principal suspects	73	569

Note: percentages may not total 100% due to rounding

Relationship between homicide victim and principal suspect: breakdown for victims

We show the relationship between homicide victims and principal suspects, for victims who had a treatment record in table 11.

Overall proportions of these were similar to those in the HI overall. The largest category was ‘stranger or unknown’ (61%), followed by ‘friend or acquaintance’ (25%).

Table 11: relationship between homicide victim and principal suspect

Relationship between victim and principal suspect	Victims with treatment record	Victims in the HI
Friend or acquaintance	25%	20%
Stranger or unknown (includes ‘other’, ‘relationship not known’ or ‘no suspects’)	61%	58%
Partner or ex-partner	14%	11%
Family (includes ‘parent’, ‘son or daughter’ or ‘other family’)	0%	11%
Total	100%	100%
Total victims	36	1,245

Note: percentages may not total 100% due to rounding

Apparent homicide method: suspects

We report the apparent homicide method for suspects with a treatment record compared to those in the HI overall in table 12. We could not break down suspects into being treated before or after homicide due to small numbers.

Using sharp instruments was the most common method of homicide both for all suspects in the HI (53%) and for suspects with a treatment record (49%). Compared to all suspects on the HI, those with a treatment record had a greater proportion where the homicide method was hitting or kicking (26% compared to 17%) or blunt instruments (13% compared to 7%).

Table 12: apparent homicide method used by suspects, for suspects detected in NDTMS

Homicide method	All suspects with treatment record	All suspects in the HI
Sharp instrument	49%	53%
Hitting or kicking	26%	17%
Blunt instrument	13%	7%
Other (includes shooting, strangulation, burning, drowning, drug poisoning, motor vehicle and not known)	12%	24%
Total	100%	100%
Total suspects	90	789

Note: percentages may not total 100% due to rounding

Apparent homicide method: victims

We report the apparent homicide method for homicides where victims have a treatment record compared to those in the HI overall in table 13.

Approximately 40% of all homicides in both groups were carried out by suspects using sharp instruments. Another 40% fell under the ‘other’ category (several offences were grouped into this category due to small numbers, see table 12).

Table 13: apparent homicide method used by suspects, for victims detected in NDTMS

Homicide method	All victims linked to NDTMS	All victims in the HI
Sharp instrument	42%	40%
Hitting or kicking	14%	18%
Other (includes shooting, strangulation, burning, drowning, drug poisoning, motor vehicle and not known)	44%	42%
Total	100%	100%
Total suspects	36	1,245

Main apparent circumstance of the homicide: suspects

In a police investigation, the senior investigating officer records the main circumstances of the homicide. This can happen either at the time of the incident or later when circumstances become clearer. The homicide record can also be updated as more information is obtained.

Table 14 shows that most of the homicides were recorded as ‘quarrel, revenge or loss of temper’ (71% for suspects with a treatment record and 66% of all suspects in the HI). We could not break down suspects into being treated before or after homicide due to small numbers.

Table 14: main circumstance of the homicide for suspects

Main circumstance of the homicide	All suspects linked to NDTMS	All suspects in the HI
Quarrel, revenge or loss of temper	71%	66%
Other circumstances	12%	16%
For theft or gain	9%	9%
Irrational act or not known	8%	9%
Other (includes arson, attributed to acts of terrorism and while attempting to restrain or arrest individual)	0%	1%
Total	100%	100%
Total suspects	90	789

Note: percentages may not total 100% due to rounding

Main apparent circumstance of the homicide: victims

We provide the main circumstances of homicides for a victim with a treatment record in table 15.

The proportions of victims with a treatment record were similar to those in the HI overall, with ‘quarrel, revenge or loss of temper’ the largest category (56%). This was followed by ‘other circumstances or irrational act’ and ‘not known’ categories (both 22%).

Table 15: main circumstance of the homicide for victims

Main circumstance of the homicide	Victims with a treatment record	All victims in the HI
Quarrel, revenge or loss of temper	56%	54%
Other circumstances or irrational act	22%	19%
For theft or gain	0%	6%
Not known	22%	20%
Other (includes arson, attributed to acts of terrorism and while attempting to restrain or arrest individual)	0%	1%
Total	100%	100%
Total victims	36	1,245

Mental health of the principal suspect

In a police investigation, the senior investigating officer can record that the homicide might be linked to the mental health of the principal suspect. This includes any information from:

• witnesses
• suspects
• investigating officers
• other third parties (such as informants)

It can also include the judgement of investigating officers that the offence is linked to the mental state of the suspect. This information does not necessarily have to be verified by other sources and may not necessarily be viewed as evidence admissible in court.

We show the proportion of principal suspects with a treatment record and in the HI overall in table 16. We could not break down suspects into being treated before or after homicide due to small numbers.

Overall, homicides were linked to the mental health of around 1 in 6 principal suspects (16%), which was higher than the rate for principal suspects with a treatment record (8%).

Table 16: mental health of the principal suspect

Homicide linked to the mental health of the suspect	Principal suspects with a treatment record	Principal suspects in the HI
No or not known	92%	84%
Yes	8%	16%
Total	100%	100%
Total principal suspects	73	569

Domestic and non-domestic homicides

We show whether the homicide was recorded as domestic in table 17. We could not break down suspects into being treated before or after homicide due to small numbers.

Homicides are recorded as ‘domestic’ when the relationship between a victim aged 16 years and over and the perpetrator falls into one of the following categories:

• spouse
• common-law spouse
• cohabiting partner
• boyfriend or girlfriend
• ex-spouse
• ex-cohabiting partner or ex-boyfriend or ex-girlfriend
• adulterous relationship
• son or daughter (including step and adopted relationships)
• parent (including step and adopted relationships)
• brother or sister
• other relatives

There was a higher proportion of domestic homicides for principal suspects in the HI (26%) compared with principal suspects with a treatment record (14%).

Table 17: domestic and non-domestic homicides

Domestic and non-domestic homicides	Principal suspects with treatment record	Principal suspects in the HI
Non-domestic or no suspect charged or suspect acquitted	86%	74%
Domestic	14%	26%
Total	100%	100%
Total principal suspects	73	569

Drug-related homicides

Table 18 shows whether suspects in the HI were involved with a homicide that was recorded as drug-related in any way.

Almost two-thirds (62%) of all suspects in the HI were involved with a homicide that was recorded as drug-related in any way. For suspects with a treatment record, this proportion was higher (80%), which was broadly consistent whether a suspect had accessed treatment before the homicide or had only accessed treatment afterwards (83% and 79% respectively).

Table 18: drug-related homicides

Homicide is drug-related in any way	Suspect in treatment before or at time of homicide	Suspect in treatment after homicide only	All suspects with a treatment record	All suspects in the HI
Yes	83%	79%	80%	62%
No	17%	21%	20%	38%
Total	100%	100%	100%	100%
Total suspects	29	61	90	789

Main findings, strengths and limitations

Feasibility of linking data sets

This report demonstrates the feasibility of linking the HI with specialist substance misuse data sets to quantify the extent to which homicide victims and suspects have been engaged in drug and alcohol treatment. It also demonstrates that, in this process, probabilistic linkage techniques are better than deterministic linkage approaches, given the lack of a common identifier across both data sets. This approach increased the number of victim and suspect records linked with NDTMS from 99 to 126, which is a relative increase of 27%.

The linkage process allowed us to do an initial analysis that, for the first time, provides evidence on the use of substance misuse services by victims and suspects of homicide in England. This evidence can now inform national and local government, and police and crime commissioners. The findings also contribute to the goals set out in the government’s 10-year drug strategy and Beating crime plan. The beating crime plan estimates that 48% of homicides in the year ending March 2020 were drug-related.

Findings from the analysis

Main findings from the analysis included the following.

Between April 2019 and March 2021, there were 118 incidents recorded in the HI where the victim or suspect had a treatment record, including 90 suspects and 36 victims. This corresponds to 10% of all homicide incidents in that period (8% to 12%, 95% CI).

Most suspects (68%) had a treatment record only after the homicide, most of which was prison-based.

Both victims and suspects with a treatment record tended to be younger than the overall population in treatment, particularly suspects, most of whom were under 40 years old.

Most victims and suspects were in white ethnic groups (between 66 and 72%), but this proportion was lower than the overall population in treatment (88%).

Compared to the overall population in treatment, both victims and suspects were less likely to be in treatment for opiates. Most suspects in treatment after the homicide (69%) were in the alcohol and non-opiate treatment group, or the non-opiate only group.

A quarter of victims (25%) had a record of a housing problem at the start of treatment, a higher proportion than for the overall population in treatment (16%).

Of the 29 suspects in treatment before the homicide, none were in regular employment. Also, less than 20% of victims were in regular employment during their treatment.

When compared to all principal suspects in the HI, suspects with a treatment record were more likely to be a friend or acquaintance of the victim (27% and 44% respectively).

When compared to all suspects in the HI, principal suspects with a treatment record were more likely to kill by:

• hitting or kicking (17% compared with 26%)
• using a blunt instrument (7% compared with 13%)

In the overall HI, the homicide was thought to be linked to the mental health of 16% of principal suspects. However, for principal suspects with a treatment record, this proportion is only 8%.

In the overall HI, the homicide was considered domestic for 26% of principal suspects. However, for principal suspects with a treatment record, this proportion is 14%.

Almost two-thirds (62%) of all suspects in the HI were involved with a homicide that was recorded as drug-related. For those suspects with a treatment record, this proportion was 80%.

Limitations of the analysis

We acknowledge that there are several limitations with the analysis, which means that it’s likely the number of matches we made between the HI and NDTMS is an underestimate.

Matching individuals

While we made all reasonable efforts to link individuals across these 2 datasets, it’s possible that the identifiers recorded on these systems were too dissimilar to generate a probable link. We selected only links with an extremely high probability of being a true match (greater than 98.7%). This makes it likely that we excluded some true matches from the analysis if they fell under the match probability threshold. But this higher threshold increases the likelihood that all matches found were true matches.

At the 98.7% probability threshold we found 145 links, of which we removed 19 (13%) after manual clerical checking. When we used a lower probability level, such as 98%, this increased the number of links to 159. When we used a probability threshold of 90%, there were 200 links. Since we removed 13% of potential matches after manual clerical checking at the 98.7% threshold, we would expect a large proportion of the increased links at a lower probability to be removed through clerical checking. So, we determined that the threshold we selected (98.7%) was the optimal level to ensure the linked cases were correct.

Length of analysis period

The analysis was limited to 2 years of the HI (April 2019 to March 2021) and 3 and a half years of NDTMS (Aug 2018 to March 2022). It is possible that both victims and suspects could have spent time in treatment before or after these dates. So, analysing a longer NDTMS period could have increased the proportion of victims (3%, 95% CI of 2% to 4%) and suspects (4%, 95% CI of 2% to 5%) who had accessed treatment.

The latest Opiate and crack cocaine use prevalence estimates report approximately 341,000 people with problems with these drugs. The latest Alcohol dependence prevalence in England report estimates there are about 608,000 people (18 and over) with alcohol dependence. However, treatment data suggests that at least 40% of people with an opiate or crack problem do not engage with treatment services. And this is double at 80% for people with alcohol problems. Prevalence estimates are not available for other drug use.

So, it’s possible that more victims or suspects have problems with these substances but are not accessing treatment. This would not have shown up through our analysis. We have shown that 68% of suspects only access treatment (in prison) after the homicide, suggesting many needed treatment before but did not access it.

Official statistics in development

These are official statistics in development. The Office for Statistics Regulation (OSR) replaced the designation of ‘experimental statistics’ with ‘official statistics in development’ in September 2023. Official statistics in development are developed under the guidance of the head of profession for statistics and are subject to the UK Statistics Authority’s Code of practice for statistics.

You can find more information about official statistics in development on the OSR website.

Enquiries or feedback

We are keen to hear from people who have read this report and would like to contribute ideas towards future development. If you have any enquiries or feedback about these statistics, contact MOJ at bold@justice.gov.uk.

The purpose of these statistics is to understand the extent to which drug or alcohol dependence may be associated with being either a victim or suspect of homicide in England. This report has established the feasibility of linking these 2 datasets together and highlighted some areas where we could make better quality estimates. These could include:

• expanding the timeframe of people accessing treatment to include as much historical treatment information as possible
• feedback from users of these statistics on whether the probability threshold we used to determine a true match should be lowered, even at the cost of including some false matches
• other information systems we should link with

We would welcome comments, concerns or suggestions by the end of September 2024. At that stage, OHID and the Home Office will decide whether we should make future iterations of this analysis and what (if anything) we should change to the approach we took for this report.

Reference

Goldstein, PJ. The drugs/violence nexus: a tripartite conceptual framework (only accessible with a subscription). Journal of Drug Issues 1985: volume 15, issue 4, pages 493 to 506.