Research Brief
Old Victorian prison landing wing with bars and staircase of high security  immigration detention centre with brown cell door in England bright daylight

Photo by Matthew/Adobe Stock

The 2009 Bradley review drew attention to the considerable disparities in mental and physical health between parts of the population who do and do not come into contact with the criminal justice system (CJS). Contact with police and custodial environments can exacerbate poor mental health and other vulnerabilities, create stigmas and preclude access to adequate services to support their needs. Lord Bradley called for more efforts to divert people from the CJS and recommended the creation of a national model for Liaison and Diversion services (L&D). These would identify and support people in the CJS with health issues and vulnerabilities including mental health, physical health and learning disabilities.

In 2014, NHS England launched the national Liaison and Diversion Operating Model for L&D services. It provides 24-hour, seven days per week services for people of all ages in the adult and youth justice pathways. The intended outcomes of L&D services are fourfold:

  1. Improve access to healthcare and support services for vulnerable individuals and a reduction in health inequalities
  2. Reduce reoffending or escalation of offending behaviours
  3. Divert individuals, where appropriate, out of the youth and criminal justice systems into health, social care or other supportive services
  4. Deliver efficiencies within the youth and criminal justice systems.

Following our 2014 study into the early implementation of the National Model for L&D in ten sites, RAND Europe was commissioned by NHS England to conduct an outcome evaluation of the L&D model at 27 sites in England. The evaluation looked at the impact of the National Model for L&D on:

  • Health service utilisation (in particular, use of A&E, mental health services and drug and alcohol treatment services)
  • Reconviction
  • Diversion from the criminal justice system (in particular, being diverted from prosecution, from court or from a custodial sentence)
  • Timeliness of court processes (in particular, the number of adjournments, number of hearings per case, duration of a case and time from arrest to sentence).

The evaluation also addresses a fifth question relating to the economic effects of any identified impacts on the outcomes outlined above.

L&D services succeeded in engaging people with a broad range of vulnerabilities

88% of people referred to L&D services had at least one vulnerability

Overall, 88 per cent of people referred to L&D services had at least one vulnerability identified. Almost three-quarters (71 per cent) of those referred had a mental health need, and approximately half (52 per cent) experienced drug or alcohol misuse. Other vulnerabilities identified included: risk of suicide or self-harm, unstable accommodation, being an abuse victim, financial needs, and needs relating to physical health, communication difficulties, or learning disabilities. Almost 20 per cent of people referred to L&D services were recorded as having more than one mental health need. This may present a challenge for L&D staff if multiple vulnerabilities are present, but none of these meet criteria for referral. A quarter (26 per cent), however, of people referred to L&D services had no prior service contacts recorded, indicating that L&D services were providing their first contact with health and social care services.

L&D services appear to intervene at a point of crisis

We identified increased use of multiple healthcare services in the months leading up to the arrest that led to referral to an L&D service. The six to 12 month period prior to L&D referral is often characterised by a steep increase in contact with Accident & Emergency (A&E) services, specialist mental health services, and declining self-reported health in those attending drug treatment services (Figure 1). L&D service users, however, with and without previous criminal justice system contact, were as likely to go to A&E after involvement with the L&D service. This suggests that for L&D service users it is acute health vulnerabilities that lead to contacts with the criminal justice system, rather than the other way around. We also observed an increase in detentions under the Mental Health Act (1983, amended 2007) in the six months prior to referral to an L&D service, suggesting that referral to L&D was not their first crisis-related contact with police for some service users. We did not observe such a marked increase in offending behaviour, except in the 1–2 months immediately prior to the arrest leading to L&D referral. This pattern of repeated health service contacts over an extended period suggests there may be a window of opportunity for intervention prior to L&D referral.

Figure 1. Pattern of (a) A&E utilisation; (b) drug and alcohol treatment referrals; (c) offending before, during, and after committing an offence in the evaluation and historic control cohorts. Data presented are unadjusted rates.

These tables list the data for Figure 1, which is described in the paragraph preceding the figure. Skip Figure 1 data tables.

A&E rate per person, per month

Months Relative to Index DateEvaluation CohortHistoric Control Cohort
-240.110.13
-230.130.14
-220.120.14
-210.120.16
-200.130.12
-190.130.15
-180.120.13
-170.140.13
-160.130.14
-150.120.13
-140.140.13
-130.140.16
-120.140.16
-110.140.13
-100.150.16
-90.150.16
-80.160.17
-70.160.18
-60.150.15
-50.160.16
-40.160.17
-30.180.20
-20.200.18
-10.290.24
00.290.22
10.180.14
20.170.18
30.150.14
40.160.16
50.150.15
60.160.16
70.150.17
80.140.16
90.150.18
100.150.14
110.160.15
120.160.15
130.180.14

Drug treatment referrals per person, per month

Months Relative to Index DateEvaluation CohortHistoric Control Cohort
-240.0050.009
-230.0060.009
-220.0060.007
-210.0070.007
-200.0050.013
-190.0060.008
-180.0050.009
-170.0050.009
-160.0050.008
-150.0070.009
-140.0080.012
-130.0060.008
-120.0060.012
-110.0050.009
-100.0070.010
-90.0060.013
-80.0050.011
-70.0070.014
-60.0070.012
-50.0070.009
-40.0070.012
-30.0070.009
-20.0080.012
-10.0090.008
00.0220.016
10.0120.016
20.0100.016
30.0060.011
40.0090.010
50.0080.012
60.0050.008
70.0060.009
80.0060.008
90.0060.013
100.0060.007
110.0060.006
120.0040.006
130.0060.012

Percentage of cohort committing an offence

Months Relative to Index DateEvaluation CohortHistoric Control Cohort
-144.3912.18
-134.3212.18
-124.7512.07
-114.9612.48
-104.7913.07
-95.0211.30
-85.1012.89
-75.6413.42
-65.9514.13
-55.8015.60
-45.7914.89
-37.3615.77
-28.2618.13
-127.8125.19
020.34100.00
17.046.83
25.857.18
35.397.95
45.428.77
54.709.12
64.6210.54
74.2510.42
84.2510.54
93.7910.42
103.9710.89
113.9211.07
123.6711.77
133.3212.07

Source: Accident and Emergency data from Hospital Episode Statistics; National Drug Treatment Monitoring System data from Public Health England; Police National Computer data from the Ministry of Justice

L&D services appear to increase diversion from custodial sentences

L&D services may reduce the proportion of offences resulting in custodial sentences and thus increase diversion from the criminal justice system. Considering changes in the proportion of offences resulting in a custodial sentence over time, the likelihood of service users receiving a custodial sentence after involvement with L&D services was almost half that of the historic control group (p = 0.05). There was, however, no evidence for an impact on the length of custodial sentences.

The interventions offered, and their uptake, varies by individual and by L&D site

There is a wide spectrum of interventions offered by L&D services, from advice and brief interventions, to primary care referral, to detention under the Mental Health Act for psychiatric assessment. Referral to healthcare services varied significantly between sites, particularly for the national Improving Access to Psychological Therapies (IAPT) service. Variability between sites is a planned aspect of the L&D programme as it is intended that referrals and other interventions are tailored to individual needs. Yet, insofar as this suggests variation in the delivery of the programme, this variability represents an unplanned aspect of its implementation. Intervention uptake also varied; people with substance use vulnerabilities were more likely to decline L&D referral and interventions overall, despite being more likely to have multiple vulnerabilities. Results from the qualitative interviews with L&D services staff indicated that timeliness of accessing onward referral services may contribute to variation in both provision and uptake; lengthy delays decrease the likelihood that service users engage with services, but interviewees reported that in some instances waiting times could be up to six months.

Following L&D referral, there is a short-term increase in referral to mental health services

Referral to L&D services appears to be followed by an increase in referral to IAPT services and non-L&D specialist mental health services. We did not, however, find evidence that these referrals translated into a substantial increase in face-to-face attendances following L&D referral.

Drug and alcohol treatment referral and attendance may increase following L&D referral

After L&D referral, drug and alcohol treatment service referrals appear to increase. A substantial minority of the drug and alcohol treatment service referrals were for individuals without previous recorded contact with these services. Additionally, referrals to drug and alcohol treatment services appear to translate into increased attendance at appointments, although this increase was not statistically significant.

Referral to L&D services does not appear to reduce offending

Overall, we did not find any evidence for an impact on offending behaviour after L&D referral. This lack of effect was at odds with the perceptions of L&D service staff; interview participants consistently identified that L&D services reduce offending by addressing the unmet needs of offenders through appropriate assessment and referral. This discrepancy may partially be due to heterogeneity in the population targeted by L&D services, and the wide variation in interventions offered both within and between services. We also found, however, that (after accounting for other vulnerabilities) contact with the CJS is not a predictor of healthcare utilisation after L&D referral. This suggests that offending behaviour in this cohort is a symptom of a wider health and/or social problem, which contact with a single L&D service may not be sufficient to address.

View of Crown Court room inside St Georges Hall, Liverpool, UK

Photo by sas/Adobe Stock

Court processes are not significantly affected by L&D services

The duration of court proceedings was reduced for offences committed at or just after referral to L&D services, but this reduction is not statistically significant based on the available data. Following L&D referral, the time from first hearing to completion was almost seven days shorter in the evaluation cohort compared to the historic control, but this difference was not statistically significant. There was no evidence for a statistically significant impact on number of hearings per court case. The L&D service staff interviews, however, did provide additional support for the role L&D staff play in expediting court decisions via drafting on-the-day reports for pre-sentencing.

The L&D programme contributes to savings in the criminal justice system

L&D services may contribute savings in the criminal justice system of between £13.1 million and £41.5 million

L&D services may contribute savings of between £13.1 million and £41.5 million in the criminal justice system through diversion from custody and consequent increases in productivity. The economic analysis hinges on custodial sentence length. When considering the diversion from custody based on average sentence length, L&D is associated with a total saving of £38.1 million, or £858 per L&D referral not declined. If we also consider increases in productivity due to avoidance of custody, then the savings increase to £933 per L&D referral (average sentence length) or £294 per L&D referral (median sentence length). We estimate that the National Model of L&D costs £659 per L&D referral not declined. Based on the average sentence length, the cost of L&D referral is thus more than offset by the £933 of savings that are due to diversion from custody.

Unique data set incorporates information from the criminal justice and healthcare systems

For this evaluation, we created a large-scale linked data set combining information from four separate healthcare sources (Hospital Episode Statistics Accident & Emergency, Mental Health Services Data Set, Improving Access to Psychological Therapies database, National Drug Treatment Monitoring System) and two separate criminal justice sources (Police National Computer, Her Majesty's Courts and Tribunals Service). To our knowledge, this is the first example of such an extensive, cross-sector data linkage study. It enabled us to overcome the key challenges of evaluating the outcome of the national programme, namely:

  • The National Model was implemented before the evaluation was commissioned, so it was not possible to directly gather baseline data from sites before the intervention was deployed. The linked data provide extensive information on people's contact with health and criminal justice services both before and after referral to L&D services.
  • During the evaluation period, the National Model was being rolled out across England, so there was no contemporaneous control group (i.e. those without access to L&D services) and a randomised controlled design could not be used to assess causation. Instead, we used the longitudinal linked administrative data to create a 'historic' control group, using data from an earlier time period before the implementation of the national programme. This enabled us to attribute causation.

Data were available for an evaluation cohort of 8,729 service users referred to L&D services between January and September 2017 who consented to the use of their data and were followed-up for at least 12 months. The evaluation cohort was broadly representative of the wider population referred to L&D services during this period, although people from ethnic minority groups may be under-represented in this cohort. We also used these linked data to create an historic control group. To do this, we identified an historic offence for which 1,699 members of the evaluation cohort were charged during 2013/4, before the National Model of L&D was implemented, and analysed the data before and after this offence. We also conducted interviews with staff from L&D services to understand the implementation of the National Model and how this may affect the outcomes achieved.

Evaluation conclusions

This evaluation used a novel, large-scale linked data set combining nationally collected administrative data from both the healthcare and criminal justice sectors. This information provided valuable insights that could not have been reliably obtained in other ways.

Analyses conducted across these data sources consistently showed that L&D services are successfully engaging with a group of service users with a broad range of vulnerabilities, often at a time of acute crisis when they are most in need of support. There was, however, substantial variation between L&D services in the types of interventions offered, and referrals to healthcare services often did not translate into face-to-face contact with health service providers. This is likely due to a combination of factors including differences in service user needs at each site, variation in availability of services to which L&D staff can refer people, and length of waiting time for face-to-face appointments.

The main impact of the L&D programme appears to be increased referrals to mental health and drug and alcohol treatment services and diversion from custodial sentences. There was no evidence that outcomes became worse due to L&D referral, and we did not identify any unintended consequences of referral.

Some factors of the programme that could be addressed to improve the overall impact of L&D services include: (1) increasing capacity for onward referrals; and (2) developing approaches to support people who have multiple vulnerabilities, but who are not currently eligible for referral because no single vulnerability meets the required therapeutic threshold.

Research conducted by

This report is part of the RAND Corporation research brief series. RAND research briefs present policy-oriented summaries of individual published, peer-reviewed documents or of a body of published work.

Permission is given to duplicate this electronic document for personal use only, as long as it is unaltered and complete. Copies may not be duplicated for commercial purposes. Unauthorized posting of RAND PDFs to a non-RAND Web site is prohibited. RAND PDFs are protected under copyright law. For information on reprint and linking permissions, please visit the RAND Permissions page.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.