Critical time Intervention for Severely mentally ill Prisoners (CrISP): a randomised controlled trial

Prevalence of mental illness

In a large-scale study conducted for the Office for National Statistics (ONS) in 1997,1 > 90% of prisoners had one or more of the five psychiatric disorders studied [psychosis, neurosis, personality disorder (PD), hazardous drinking and drug dependence], with remand prisoners having higher rates of disorder than sentenced prisoners. More than half (59%) of men remanded, and 40% of men sentenced, had a neurotic disorder, with the corresponding figures for women being 76% and 63%, respectively. Rates of psychosis varied from 7% in the male sentenced population to 14% in female sentenced population. In addition, 78% of male remand prisoners and 50% of female remand prisoners had a PD. Levels of substance misuse were also high, with 51% of male remand, 43% of male sentenced, 41% of female sentenced and 54% of female remand prisoners being drug dependent in the year before prison. Over 50% of the men in the sample screened positive for hazardous drinking in the year before coming into prison; the analogous figure for female prisoners was 38%. 1

The ONS survey and other studies in England and Wales have shown that psychiatric comorbidity is the norm. 2–5 Between 12% and 15% of sentenced prisoners in the ONS study1 had four or five psychiatric disorders, and many prisoners present with complex psychiatric treatment needs,5 often confounded by issues of dual diagnosis, especially of personality or substance misuse disorders.

The high prevalence of mental disorder in prisons is not confined to England and Wales. In a large-scale systematic review of serious mental disorder in 23,000 prisoners in Western countries,6 approximately one in seven prisoners had either a psychotic illness or major depression, with approximately half of male prisoners and one-fifth of female prisoners having antisocial PD.

Compounding high levels of psychiatric morbidity, prison populations have high levels of suicidal and deliberate self-harming behaviours, with prisoners at a far greater risk of suicide than the general population. In the ONS study,1 around 24% of male and 40% of female prisoners had attempted suicide at some time in their lives. Twelve per cent of male remand and 23% of female remand prisoners reported having experienced suicidal thoughts in the week before interview (rates for sentenced prisoners were considerably lower); these figures rose to 35% and 50%, respectively, when measured over the past year. After a decade of the prison suicide rate in England and Wales reducing year on year, there has been a more recent trend of increasing numbers of suicide from around 2011, attributed by commentators on penal matters to increased population pressures and overcrowding, smaller budgets and significant deliberate reductions in staff numbers. 7

Diagnosis, treatment and care: the role of the prison in-reach team

Historically, Her Majesty’s Prison Service (HMPS), through the existence of the Prison Medical Service, latterly renamed the Prison Health Service, was responsible for the provision of the majority of health-care services for prisoners. Almost all services were provided in house, ranging from primary care for everyday physical complaints through to inpatient care for those with severe mental health problems. Staff, including doctors, prison health-care officers and qualified nurses were directly employed by HMPS. For decades, the development of multidisciplinary care in prisons lagged behind such initiatives in the NHS; for example, at a time when much mental health care in the wider community was being delivered by multidisciplinary community mental health teams (CMHTs), most care in prisons was dependent on input from visiting forensic psychiatrists, with little contribution from wider clinical disciplines. 8,9

In 1999, the NHS and HMPS formed a partnership to modernise the delivery of health care in prisons, acknowledging that the then current arrangements varied considerably in terms of organisation, delivery, quality, clinical effectiveness and links with the NHS. 10 One of the early areas targeted for reform was mental health provision and, in 2001, a specific strategy for mental health, Changing the Outlook, was published. 11

The strategy document reaffirmed that the existing delivery model did not meet prisoners’ needs and was ineffective and inflexible. It was acknowledged that most prisoners with mental health problems were not so ill as to require detention under mental health legislation and, if they were not in prison, would be receiving treatment in the community rather than as an inpatient. The strategy suggested a move away from the historically held assumption that prisoners with mental health problems should be located in prison health-care centres, towards supporting prisoners with mental health problems on ‘normal’ prison wings through the establishment of multidisciplinary mental health in-reach teams. Such teams were to be funded by local primary care trusts and provide specialist mental health services analogous to those provided by CMHTs. Although it was expected that all prisoners would eventually benefit from the introduction of in-reach teams, the early focus of the teams’ work was on those with severe mental illness (SMI), utilising the principles of the care programme approach (CPA),12 in particular to help ensure continuity of care between prison and community on release.

A national evaluation of prison in-reach services reported that, in spite of the new model of care, major challenges remained. 13 In particular, in-reach services struggled to effectively target their priority client group, those with SMI. Research demonstrated that in the month following reception into custody only one-quarter (25%) of those with current SMI were assessed by in-reach services and only 13% accepted into caseload. 14 An examination of the composition of in-reach caseloads identified that only 40% had SMI. Of the 60% with no diagnosis of SMI, 42% had PD, 32% had a common mental illness and 42% had neither. Those with no current diagnosis of SMI, PD or common mental illness exhibited high rates of previous contact with mental health services (lifetime 93%) and substance misuse (69%) before custody. 13

In a parallel study of the management of prisoners with SMI on in-reach caseloads, widespread disengagement from mental health services at the point of release from custody was identified. On examination of the prison in-reach case notes for 53 service users, the authors found evidence of discharge planning for only 27 (51%) individuals; fewer still, 20 (38%), had direct contact with their CMHT before release and, of those, only four (20%) had made contact with the CMHT at follow-up 1 month later. 15 Dyer and Biddle16 highlighted that functioning within the operational constraints of a prison often makes the task of planning and preparation for discharge challenging for in-reach services because of limited resources, functioning overcapacity in the delivery of treatment programmes and/or the sudden transfer of prisoners at short notice.

Transition: reoffending and post-release mortality, health and socioeconomic factors

The transition from institutional to community living is a vulnerable period, associated with a range of negative outcomes. Reoffending by released prisoners in England and Wales continues to be a challenge for the criminal justice system (CJS). In 2013, the reoffending rate of adults released from custody was 45.8%, a rate that has remained relatively static (45–49%) since 2004. Notably, adults who served sentences of < 12 months reoffended at a rate of 59.3%, compared with 34.7% for those who served determinate sentences of ≥ 12 months. 17

Several studies have shown that released prisoners with a history of mental illness and/or comorbid substance misuse, many of whom return to unstable environments characterised by socioeconomic disadvantage, are particularly vulnerable to relapse and reoffending. A large retrospective study in the USA found that prisoners with a diagnosis of SMI were more likely than those without to have experienced multiple prison terms in the 6 years before their index offence. For example, prisoners with a diagnosis of bipolar disorder were 3.3 times more likely to have had four or more previous prison terms than prisoners with no major psychiatric disorder. 18 In a further study of the same cohort, prisoners with comorbid SMI and substance misuse were more likely to have experienced multiple prison terms than those with a single such diagnosis. 19 Similar findings were reported in a study of parole violation in which a diagnosis of SMI was found to increase the likelihood of an individual breaking the terms of their parole; a comorbid diagnosis of substance use disorder increased that risk further. 20

Prison provides an opportunity for individuals with SMI to receive mental health treatment, including participation in programmes designed to reduce and control substance misuse. However, improvement is likely to be quickly jeopardised if, on release, the person does not engage with community services to enable treatment started during imprisonment to be continued. A failure to connect with community providers of mental health and substance misuse services has been linked to the high incidence of mortality in recently released prisoners. Farrell and Marsden21 found that newly released prisoners in England and Wales were at an acute risk of drug-related death in the first 2 weeks of leaving prison, with male and female prisoners, respectively, 29 and 69 times more likely to die of drug-related causes, relative to the general population, during their first week of release. This finding was confirmed in a later meta-analysis of drug-related deaths after release, which reported that drug-using prisoners had a three- to eightfold increased risk of drug-related death in the first 2 weeks post release compared with the subsequent 10 weeks, with risks remaining elevated through weeks 3 and 4. 22 Factors such as reduced tolerance to specific drugs, the high and variable potency of street drugs, and the temptation to engage in ‘celebratory’ drug-taking behaviour on release, were suggested explanations for these findings. Similar findings were reported in a systematic review and meta-analysis of all-cause and external mortality in released prisoners, identifying that released prisoners were at a substantially increased risk of death from all causes but from drugs, suicide and homicide in particular. 23

Studies of suicide by prisoners following release bring most sharply into focus how overwhelming transition can be for some. 24,25 A study of completed suicide by released prisoners found that age-adjusted standardised mortality rates were, compared with the general population, 8.3 times higher for men in the 12 months post release; the risk was even greater for women, at 35.8 times higher. Furthermore, 21% of those who completed suicide did so within 28 days post release, with just over half (51%) of deaths occurring within 4 months of release. 24 The authors conducted a subsequent study, identifying key risk factors for suicide by released prisoners. These included increasing age > 25 years, having a psychiatric diagnosis, previous contact with psychiatric services before custody, a history of alcohol misuse and a history of self-harm. Contact with prison mental health in-reach services while in custody and being recognised as ‘at risk’ of suicide while in prison were also relevant factors. The study authors reiterated the importance of improved release planning for those prisoners most at risk, to ensure immediate engagement with community mental health services, assertive follow-up and intensive post-release support. 25

Planning to prevent discontinuity of mental health and substance misuse treatment on release is critical, but addressing socioeconomic and psychological factors that might adversely impact on service engagement is also vital. Williamson26 listed multiple health and social care needs and factors that reflected the lifelong social disadvantage experienced by many prisoners, all or any of which may negatively influence the establishment of a stable routine lifestyle outside prison. For example, prisoners were more likely to have spent time as a child in local authority care, have a poor education and have a family history of CJS contact than those who had never been imprisoned. On release from prison, 42% had no fixed abode, 50% were not registered with a local general practitioner (GP) and 60% were unemployed. Despite this, it has also been reported that prisoners’ expectations of what life will be like following release may be unrealistically high or, equally damaging, the individual may be overwhelmed by concerns about how they are going to cope. In a longitudinal qualitative study of 40 prisoners, over one-quarter of whom self-reported a mental ill health problem, many prisoners’ aspirations before release were high with regard to finding work or going back into education, overcoming their drug and alcohol misuse and/or generally regaining some stability in the community. However, in the absence of co-ordinated service planning and advice, goals were often unrealistic and, therefore, difficult for the individual to achieve. Having family or peer support helped, but individuals without this support found their plans quickly fell through, negatively impacting on their ability to manage their housing arrangements and control drug and alcohol use, with reoffending once again becoming a coping mechanism. 27

A number of other studies have reported similar findings, noting that having a safe place to live, finding employment and maintaining mental and physical well-being are high priorities on release for prisoners, particularly as good mental and physical health is viewed as very important for securing employment. The prospect of relying on hostel accommodation and entering an environment in which the misuse of drugs and alcohol by others may be commonplace was a source of anxiety for soon to be released prisoners. In addition, the stigma attached to being identified as an ‘ex-offender’, with the associated disadvantages in terms of access to employment and other opportunities, provoked high levels of anxiety. 28,29

Integrating health and social care services to meet prisoners’ needs in a holistic way on release is therefore vital to successful community reintegration. Thus, effective release planning and resettlement requires not only continuity of health care but also measures designed to meet the economic and social needs of the prisoner.

Forerunners to critical time intervention: the case management and assertive community treatment models

Critical time intervention

Critical time intervention (CTI), a variant of ACT, was developed in the USA in the 1990s. CTI was designed as a structured but, unlike ACT, specifically time-limited (to a maximum of 9 months in original trials) intervention to prevent recurrent homelessness in transient individuals with SMI moving from hospital care to the community. 46–48 The intervention had two key components: first, to strengthen ties with service providers, family and friends; and, second, to provide practical and emotional support during transition from institution to the community.

To realise the first component, case managers made appointments with key service providers and accompanied clients to those appointments following discharge from hospital. The case manager ensured that clients had a named contact at each service and facilitated the formation of a relationship between the client and provider to better ensure continued engagement. The case manager also supported the client and his family in re-establishing their relationship; if the client’s family wished to be involved in providing care, the case manager helped them to better understand their relative’s illness, the difficulties they might encounter in their role as carer and ways to resolve those situations. To achieve the second component of the intervention, the case manager maintained close contact with the client, observing how they were adapting to living in the community, stepping in, if necessary, to provide practical help with the development of skills necessary to function independently. The case manager reviewed the extent to which their input was needed throughout the intervention period to the point when they judged they could withdraw without any disruption to engagement.

In an early trial, those in receipt of CTI had significantly fewer nights’ homelessness than the TAU group: 30 compared with 91 homeless nights, respectively. In addition, the strong ties with service providers that CTI put into place persisted after the intervention was withdrawn, with survival curves showing that, after 9 months of the intervention, the differences between the groups did not diminish. The authors noted that CTI could be used in any transition scenario, for example from prison to community, to better co-ordinate and augment existing processes in place to link individuals to other important services. 46

To examine this assertion, as well as the research described in this report, the CTI model is currently being trialled in the transition of two discrete populations leaving shelters for supported or independent housing: (1) individuals previously homeless and (2) women who have experienced domestic violence. 49 In addition, a trial using CTI at the point people first make contact with mental health services in order to put in place a comprehensive and enduring network from the start is under way. 50 These studies will, in due course, add to the body of evidence with respect to the transferability of the model to other scenarios.

Ethics approval

Ethics approval for the study was granted by the research ethics committee (REC) for Wales in January 2012 (reference number 11/WA/0328). The National Offender Management Service research approval was given in February 2012 (reference number 184-11). The trial was registered with the International Standard Randomised Controlled Trial Number (reference number ISRCTN98067793). In addition, all required site-specific permissions and research governance approvals, that is the research and development (R&D) approvals, were obtained from the relevant NHS trusts.

Changes to protocol

The progress of this trial was severely impacted by recruitment shortfalls at the original sites and significant delays in obtaining a number of R&D approvals at new sites. We increased the number of sites involved from three to eight. A summary of the changes to the original protocol notified to the REC is given in Table 1. Approvals were also sought for another three prison establishments but, because of significant delays in obtaining these approvals, recruitment never commenced.

Increase of study sites from three to eight

The number of study sites was increased because of slow recruitment rates in the original three sites. This was mainly caused by potential participants failing to meet one or more of the eligibility criteria. The main reasons people were ineligible were (1) not likely to be released within the lifetime of the study and (2) not likely to be discharged to the local geographical area, thus unable to be followed up. One site, which had been the site with the largest recruitment in our feasibility study, was rerolled during this study, changing it from a category B remand local prison to a category C/D resettlement establishment. This meant that the prison started to take prisoners with longer sentences, so many were not likely to be released within the lifetime of the study. In one site, the original CTI manager left and the NHS trust was unable to recruit a replacement.

Research and development approval

For three of the trusts involved, it took approximately 6 months to obtain R&D approval and, in two cases, a complaint was lodged with the concerned trusts’ medical director in order to expedite matters. Complex commissioning and provider arrangements resulted in a lack of transparency as to where responsibilities lay; this required seeking multiple permissions from several provider organisations at single sites. In addition, retendering processes resulting in the award of contracts to new provider organisations impacted negatively on recruitment. In addition, two sites that had originally agreed to take part in the research pulled out before recruitment could begin, citing staffing shortages.

Use of Operational Criteria Checklist for Psychotic and Affective Illness rather than Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition Axis 1 disorders

Before data collection commenced, the use of the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition Axis 1 disorders (SCID-I)55 by non-clinically trained researchers was reassessed because of concerns about the specialist knowledge required for its accurate completion. The use of the Operational Criteria Checklist for Psychotic and Affective Illness (OPCRIT)56 was agreed to solve this issue, as its completion is not as dependent on expert clinical assessment skills.

Other changes to protocol not requiring research ethics committee approval

Inclusion criteria

Participants were considered for inclusion if they met all the following criteria:

• were male

• had SMI

• were a service user of the prison mental health in-reach team

• were able to give informed consent

• were to be released from prison within the lifetime of the study

• release would be to an agreed geographical area local to the prison.

Severe mental illness was defined as major depressive disorder, hypomania, bipolar disorder and/or any form of psychosis including schizophrenia, schizoaffective disorder and any other non-affective non-organic psychosis.

Exclusion criteria

Participants were excluded if they:

• did not have SMI

• were to be released outside the agreed geographical discharge area

• posed security/safety issues that would compromise researcher/practitioner safety in prison or the community

• were unable to give informed consent

• had previously participated in the trial during an earlier period in custody.

Recruitment procedure

In all sites, the mental health in-reach team identified existing (at the start of the study in each site) and new (as the study continued) service users who fulfilled the inclusion criteria.

The in-reach team informed the service users of the proposed study and asked if they wished to learn more about it. If the service user expressed an interest, the in-reach team member, with the person’s permission, passed their name on to a member of the research team. A researcher then arranged a time to meet with the service user to describe the study. The potential participant was provided with all relevant clearly written information about the study and its implications. They had the opportunity to ask questions about the research and were given a minimum of 24 hours to decide if they would like to take part. Given the unique problems of gaining consent in custodial environments, careful emphasis was given to their rights to consent/not consent, including the right to withdraw at any time, without the need to give a reason for doing so, and free of any coercion or negative consequences to their mental health care or their progress in custody in general. If any concerns regarding capacity to give informed consent because of mental illness were raised, the researcher sought the opinion of the mental health in-reach team. The original signed and dated consent forms were held securely as part of the trial site file, with a copy held in the participant’s clinical records.

The likely release dates for unconvicted prisoners were predicted using the Sentencing Council Guidelines,58 based on the person’s index offence. Geographical discharge area for each prison was based on NHS R&D approval areas.

Treatment as usual

Individuals in the control group received TAU. While still in prison, they were able to access primary care, secondary mental health and substance misuse services as would usually be the case. They also received support from criminal justice and any other third-sector organisations in the standard way.

Critical time intervention: adaptation for current randomised controlled trial

Critical time intervention is intensive CM at times of transition between the prison and community. CTI managers provide direct care where and when needed, for a limited time period. They commence their involvement with the service user in prison. For sentenced prisoners, this starts 4 weeks before discharge. For remand prisoners, or those with unpredictable dates of release, this work commences as soon as the person is on the caseload of the mental health team. The length of their involvement pre release is, therefore, ideally 4 weeks but may be shorter or longer in those with unpredictable release dates. In this adaptation of CTI, the period of contact post discharge was set at 6 weeks. The 6-week period of intervention was adopted because (1) the pilot study indicated that, by this stage, the service user would be engaged with the CMHT if that was going to occur at all; (2) it allowed a reasonable period post discharge in which adjustments to vital support systems, including accommodation and/or benefit entitlement/employment are most likely to be required; (3) to keep community caseloads low and workable for the CTI managers, some of whom were part-time; and (4) the adapted version of the intervention was heavily frontloaded with most of the vital liaison work being completed while the service user was still in prison.

The holistic intervention involved work with clients and clients’ families (when possible), as well as active liaison and joint working with relevant prison and community services. Five key areas were prioritised: (1) psychiatric treatment and medication management, (2) money management, (3) substance abuse treatment, (4) housing crisis management and (5) life-skills training. CTI is not prescriptive; it responds to the needs of each individual client. The intervention comprised four phases.

Phase 1 is conducted while the person is in prison. The CTI manager engages with the individual and develops a tailor-made discharge package based on a comprehensive assessment of the individual’s needs. This typically includes plans for engagement with community mental health treatment and addressing accommodation, financial and social support needs. The CTI manager and prisoner meet as often as required to make the discharge arrangements; pre-release contact is routinely twice weekly. In addition, the CTI manager liaises closely with community services to ensure their availability and suitability.

Phase 2 occurs immediately after release and focuses on providing very intensive personal support. In the first few weeks post discharge, the CTI manager maintains a high level of contact, including accompanying people to appointments to promote engagement and to help them establish relationships with community providers in order to facilitate the development of durable ties. The number of meetings/visits involved is directly influenced by the complexity of each person’s needs, but routinely involves up to 15 meetings per week for the first 2 weeks following discharge.

In phase 3, community services assume primary responsibility for the provision of support and services, and the CTI manager focuses on assessing whether or not the support system is adequate and functioning as planned. During this phase, the CTI manager encourages the individual to start to handle problems on their own. They meet less frequently but maintain regular contact in order to judge how the plan is working. The CTI manager remains ready to intervene if a crisis or potentially destabilising event arises. Again, the frequency of meetings is individually determined, but is typically at least weekly for 3 weeks.

In phase 4, care is fully transferred to community services in order to provide long-term support, thus work focuses on completing the transfer of care. This phase may typically consist of a meeting with the community care co-ordinator, service user and CTI manager, reviewing progress and agreeing future care. Throughout, the CTI manager will gradually reduce their role in delivering direct services to the individual. Their main function in this phase is to ensure that the most significant members of the ‘receiving’ support system meet together and, along with the individual, reach a consensus about the components of the ongoing system of support.

A manual for use by the CTI case manager was also developed to support their delivery of the intervention (see Chapter 5 and reproduced in full in Appendix 1).

Each CTI case manager received 2 days’ training on the manual, undertaken by the study’s principal investigator, a consultant forensic psychiatrist. The CTI case managers then received weekly CTI-specific supervision locally in addition to their normal clinical and/or line management supervision. In addition, group supervision was held by telephone, every 3 months, with the study principal investigator. The aim of the CTI supervision was to correct CM that was inconsistent with CTI principles and practices, provide guidance to assure that the approach was consistent with CTI principles and practices, and schedule case presentations for all new clients within a few weeks of enrolment into CTI.

All professionals who took part in this study in the role of CTI manager were qualified and experienced mental health clinicians; most were mental health nurses and one was a clinical psychologist. All had previous experience working in both prisons and either forensic community mental health services or intensive home treatment/assertive outreach teams. As such, they brought with them an extensive skill set, including existing motivational interviewing training; thus, the majority of training focused on describing the content of CTI as distinct from other ways of working in previous roles rather than clinical skills training per se.

Baseline assessment

At baseline all participants were seen by a member of the research team and the following data were collected:

• OPCRIT+56 – the OPCRIT+ was used to obtain an Axis 1 diagnosis (Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition55). OPCRIT+ is an electronic checklist of psychopathology items with algorithms for objective diagnosis of psychotic and affective disorders. Participants were asked about a range of mental health symptoms and responses entered into the OPCRIT database to produce a diagnosis. In its original format, OPCRIT data are designed to be gathered from case notes alone. 55 However, in a small pilot, it became apparent that there were frequently insufficient data in the notes alone to make a reliable diagnosis and, therefore, the case note data collection was supplemented by direct inquiry with the participant.

• Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition Personality Disorders (SCID-II) – the SCID-II is a semistructured interview for the assessment of PD. The first part consists of eight open questions on the patient’s general behaviour, interpersonal relationships and self-reflective abilities. The second part has 140 items to be scored as 1 (absent), 2 (subthreshold) or 3 (threshold). 60 The full SCID-II interview was administered to all participants and any resulting diagnoses recorded.

• The Michigan Alcohol Screening Test (MAST) – the MAST consists of 24 yes/no questions pertaining to lifetime use of alcohol. Each item is scored 0 or 1, with scores of ≥ 10 indicating evidence of having had a lifetime alcohol problem. 61

• The Drug Abuse Screening Test (DAST) – the DAST is similar in design to the MAST. It consists of 20 yes/no questions, each scored 0 or 1. Scores of ≥ 11 indicate substantial problems with drug abuse. 62

• Adapted Client Services Receipt Inventory – developed from the Client Services Receipt Inventory. 63 A pro forma was developed that enabled data on a specific range of services to be collected from health-care records by the research team.

All baseline assessments were conducted between October 2012 and July 2015.

Follow-up

Follow-up data collection was scheduled to take place at three time points: 6 weeks and 6 and 12 months post release from prison. The 6-week follow-up coincided approximately with the end of the intervention delivery phase and the 12-month follow-up was designed to inform the investigation of any longer-term effects of the intervention on study outcomes. All follow-up data were collected via file information from each participant’s care team in the community. All follow-up data were collected between November 2012 and October 2015.

Primary outcome

The primary outcome measure was engagement with mental health services at 6 weeks post release from prison. Engagement was defined as currently being in receipt of an appropriate level of mental health care, by virtue of (1) having an allocated care co-ordinator, (2) having a current care plan and (3) receiving medical treatment for mental health problems. To create the binary engagement variable, a score of 1 was assigned if all three of these were true, and a score of 0 assigned if any or all of these were not true (i.e. they did not have an allocated care co-ordinator, they did not have a current care plan or they were not receiving medical treatment for mental health problems).

Secondary outcomes

The secondary outcome measures were engagement, as defined above, with mental health services at 6 and 12 months.

Original sample size justification

The original calculation for the research proposal, taking into account the attrition rate in the feasibility trial of 15%, required 100 participants randomised to each arm (CTI and TAU) to give 90% power to detect a difference at 6-week follow-up of 50% in the treatment group compared with 25% in the control group (or greater), at the conventional 5% significance level. Thus, 85 participants were required in each group at 6-week follow-up.

Revised sample size justification

Owing to slow recruitment rates, we checked our earlier assumptions after 120 participants had been randomised and found that the attrition rate was 9%. In addition, we also proposed reducing the statistical power available to detect a significant difference from 90% to 80%. The revised calculations are shown in Table 3. The number required for the primary outcome was 132.

Definitions

Relationship

The expression ‘reasonable causal relationship’ means to convey, in general, that there is evidence or argument to suggest a causal relationship. The research team assessed the causal relationship between reported events and trial participation according to CONSORT65 guidance (see Appendix 3).

Reporting serious adverse events

In this study, SAEs were reported to the chief investigator (JS) regardless of relatedness within 24 hours of the principal investigator (or authorised delegate) becoming aware of the event. All SAEs deemed to have a causal relationship were reported to the Trial Steering Committee. Any non-SAEs (regardless of relatedness) were not reported in this study.

Recruitment

The NHS ethics and National Offender Management Service approvals were received by February 2012 but first recruitment did not commence until October 2012. Table 4 shows the recruitment issues at each site that contributed to delays and the number of participants randomised at each site.

Flow of participants in the trial

In total, 150 individuals were recruited to the trial, with 72 allocated to the intervention group and 78 to the TAU group. Figure 1 presents the CONSORT65 flow diagram for the trial and summarises participant throughput from eligibility screening and randomisation to completion of the 6-week, and 6- and 12-month follow-ups, as appropriate. The diagram also reports numbers of participants who declined, did not meet inclusion criteria, were excluded from the study, withdrew following randomisation or were lost to follow-up at the 6-week, or 6- and 12-month follow-up points.

FIGURE 1.

The CONSORT flow diagram.

Baseline comparability

Table 5 presents a summary of the baseline demographics to describe the sample and demonstrate the baseline comparability of the randomised groups.

The sample were all male, with broadly half (48%) from a white ethnic background and half (52%) from a black or minority ethnic background. The majority of the participants were single (89%), unemployed (89%) and living alone (67%). Nearly 20% of the sample said that they were homeless or had no fixed abode on arrival at prison. Proportionally more participants randomised to the CTI arm had been homeless. In relation to offending, half of the sample had committed a violent index offence. On average, participants had spent 1 year in prison at the point of baseline assessment and the majority (83%) had been in prison previously, with an average of seven previous prison terms.

In relation to mental health, all participants had an Axis I diagnosis, as determined by OPCRIT. 56 The most common primary diagnosis was schizophrenia, affecting 71% and 73% of the CTI and TAU groups, respectively. Overall, 13% of the sample was experiencing a major depressive disorder and 8% a schizoaffective/schizophreniform disorder. There were no significant differences across the intervention and TAU groups in terms of Axis I diagnoses.

In relation to Axis II diagnoses, 53% of the sample overall had at least one PD, as determined by the SCID-II assessment tool. The most common diagnosis was antisocial PD, identified in 49% of the sample overall. Thirteen per cent of the sample was identified as having borderline and 10% a paranoid PD.

In relation to drug and alcohol misuse, nearly 30% of the sample scored ≥ 10 on the MAST, indicating that they had had a severe drinking problem at some point in their life; more participants scoring over > 10 were randomised to the CTI group. In total, 64% of participants scored ≥ 11 on the DAST, indicating substantial problems with drug abuse. Table 6 presents a summary of the key service contact of the randomised groups.

With regard to dual diagnosis, less than half (42%; n = 63) of the sample had a diagnosis of a single Axis I condition alone. Eight (5%) had dual SMI and substance misuse issues, 56 (37%) had dual SMI and PD diagnoses and 23 (15%) had SMI, substance misuse and PD diagnoses.

In relation to contact with services at baseline, the vast majority (97%) had previously received a mental health intervention, with 81% receiving that from a CMHT. For the majority of participants (85%), their first contact with mental health services was > 12 months prior to baseline assessment, and their most recent contact was within the last month (81%). Around half (55%) of participants were in current contact with community mental health services on reception into custody and 32% were receiving support/treatment for mental health problems from their GP on reception. Few participants were receiving drug (15%) or alcohol (3%) treatment on reception. Very few participants were receiving any psychological interventions (8%) while in prison.

Primary outcome

The primary outcome measure was engagement with mental health services at 6 weeks post release from prison. Engagement was defined as a composite variable of (1) evidence of having an allocated care co-ordinator, (2) evidence of having a current care plan and (3) receiving medical treatment for mental health problems.

There were significant differences between the two groups with participants receiving CTI significantly more likely to be engaged with mental health services at 6 weeks; significantly more were likely to have a care co-ordinator in place, have evidence of a current care plan and be in receipt of medication (Table 7).

Secondary outcomes

Issues with fidelity

The contact forms and progress notes of the CTI managers during the engagement and early linking phase provided little written evidence that there was communication with, and/or visits to, housing providers, family or CMHT at two time points (across different CTI managers) (Table 11). This activity may have happened, but was not being recorded adequately.

Overall, intensive outreach was well implemented. However, one CTI manager had very limited contact with clients in the community.

Care plans were well implemented for many, but there was limited evidence of creating care plans for clients at one site. Encouragingly, the rating for this item improved over time.

At one site, monitoring was rated as 1, based on the criterion that contact with the participants at the relevant phase should have been reducing but, because of the complex nature of the participants’ needs, contact with the CTI manager remained frequent.

Cases closed was rated low for some CTI managers, as many participants were still on the caseload after the 6-week (standard deviation ± 2 days) cut-off point.

Post-release contact at 6 weeks post release was rated low for one CTI manager because they had either lost contact with the client or the client had refused to continue to engage, before the 6-week point.

Closing cases, defined as making sure there was a clear transfer of care meeting with the client and services, was limited for some CTI managers and at different time points. Some CTI managers struggled to access other services’ information technology systems to ensure that notes were recorded or they were not part of a specific team and, therefore, again, had no access at certain clinical information technology systems.

When participants were engaged there was evidence that CTI managers were clear that care was ending but, for participants who struggled to engage, evidence of a clear ending was difficult to find.

Overall, clinical supervision was well implemented. However, for one CTI manager at one time point, no clinical supervision had been received and, for another, clinical supervision did not appear to be successful in correcting their use of inconsistent CTI principles, for example limited community contact.

Methods

Theme 1: uncertainty

Conceptually, the ‘uncertainty’ theme reflected participants’ lack of clarity about post-release plans. Their accounts characterised the pre-release period as one during which they experienced increasing levels of stress, tension, worry and anxiety. Uncertainty about release dates is evident from this participant’s response to a specific query regarding his release date:

. . . about a month. But I’m going to court next week and it could be then. I don’t really know.

CTI, prison

This level of uncertainty made it virtually impossible to implement meaningful discharge planning:

I don’t know when I’ll actually be leaving and so I can’t tell people that and that means they can’t see where I can go because they [accommodation provider] won’t keep somewhere open waiting for me if I’m not actually going to be there.

TAU, prison

Among participants with previous experience of incarceration, the spectre of ‘release day’ loomed large. Instead of being a cause for celebration, their return to society was associated with anxiety about the likelihood of negative outcomes, as illustrated by this participant’s account of a previous release day, about how a lack of appropriate pre-release planning resulted in a ‘bad’ release day and ‘things going off track’ from the outset:

My brother picked me up and we had to go to bail hostel but we said we’d go for breakfast first. I felt agitated when I left, but my brother wasn’t there so if I’m honest I went for a beer, yeah. Just wanted to see what it was like.

CTI, prison

Indeed, some participants found their experiences so stressful that they considered reoffending in order to return to prison:

People get nervous when they’re going to go. You leave and you want to turn around and walk back in. Or do something to get back in. I don’t want to be like that though. You know, prison is tough, and the outside is tough too really. But I want to move on.

CTI, prison

In contrast, CTI appeared to have alleviated some participants’ uncertainty and associated anxiety for their upcoming release:

Most things were sorted and everyone knew when I was leaving so yeah I guess you could say ‘integrating’.

CTI, community

I found it easier with the CTI manager. I was getting help that didn’t get before and found everything less stressful. It’s a big help.

CTI, community

Theme 2: support

This theme also contained three related, but distinct, subthemes (‘advocacy’, ‘lack of continuity’ and ‘reliance on others’), each of which is explored in subsequent subsections.

In general, participants in the TAU group were more likely to characterise their current experiences as lacking in support. Those with previous experience of prison and expecting their upcoming release spoke candidly about services that, from their perspective, seemed to lack coherence or focus on the needs of prisoners and recently released individuals:

As far as I’m concerned I’m not in touch with any services. They come and ask questions then just go off and you don’t know who they’re speaking to or who you’re meant to speak to when you get released. That’s how it is, you might get a bit of information when you leave but it all seems to happen between them and not with us.

TAU, prison

No one helps with that [benefits] it’s down to you. You get the forms when you go and you sort that out when you can. Well like I’ve been saying, it’s rubbish isn’t it. There isn’t really any support. You just have to sort things yourself and try and work out whether people know you’re coming out or where you need to go. And it’s always like that.

TAU, prison

This contrasted sharply with participants in the intervention arm. CTI recipients predominantly reported receiving positive support from their CTI manager. Significantly, this support was available at various stages, including pre release:

I’ve been a bit stressed actually, and that support is good. With [CTI manager] I feel like I get seen more. Which I think I need before I go. Just to make sure.

CTI, prison

On the day of release:

[CTI manager] was there waiting outside the gate because I’d told her when I would be released. And she’d checked where we need to go so we could get the train and then the bus over to this hostel together. That was something I was worried about because I get so anxious about being on buses, being with other people in public. I think it would have been too overwhelming but [CTI manager] was there saying it’s OK.

CTI, community

And in the community:

I would have struggled to sort everything and go to appointments on my own as I’d forget.

CTI, community

Critical time intervention managers were also credited with trying to help people reintegrate into their communities:

I’ve been saying I want to get back into doing good activities and [CTI manager] gave me the details of some football teams for people who have problems like me. Like mental health problems or they’ve been in prison. So I want to start doing that. And he showed me how I can join a gym too.

CTI, community

According to this participant, immediate and ongoing support from the CTI manager both reduced the risk of reoffending and provided a realistic hope of a future outside prison:

Well just to say that this thing has been pretty good for me and it definitely has helped. [CTI manager] has helped me a lot and I reckon now I’ve been out this long I can keep it up. You know don’t you that I was in 8 years because of that one thing which was a long time and now I’m looking forward to staying out.

CTI, community

Theme 3: accommodation

Accommodation (in particular uncertainty about accommodation on/post release) was an issue not only in its own right, but also one that impacted on other areas, as exemplified in earlier reports about support and reliance on others. Participants in both arms of the trial voiced concerns about both the availability and suitability of accommodation:

They [hostels] aren’t even that much better than prison. You can go out during the day I guess, but that’d happen on day release anyway. But the rooms aren’t good and you know the people here. They take drugs, they drink. Which I do too. But we shouldn’t be all together should we. Load of mad men all together. It’s not a good idea.

CTI, community

This was especially the case when accommodation had to be shared with others with similar problems, chiefly mental health and substance misuse:

I don’t like being around people with symptoms because it makes me worse. Like in the hostel, if you have someone who’s got symptoms it makes you worse and I don’t want to be around that.

CTI, prison

Participants felt that securing accommodation was a vital part of the resettlement process as lack of suitable accommodation was regarded as a key aspect of being able to access services:

Well because you leave and you need somewhere to sleep that first night out don’t you. And then that gives you a base to plan everything. You can’t be homeless and getting benefits and getting letters from people and your probation knowing how to see you. If you haven’t got that then you’ve got no hope.

TAU, prison

In addition, experience had taught them that a lack of accommodation increased their risk of reoffending and incarceration:

I’ve done this so many times and if you don’t have a hostel then that’s it you’ll be back in a week. That’s why I keep coming back, you don’t get somewhere to live you don’t have any chance.

TAU, prison

However, some participants reported that living in hostels had increased their access to services, some of which were available on site:

That’s the reason they put me here because the CMHT workers come here and they see everyone who needs to be seen rather than you having to go to them. It’s better. It means I don’t have to be going to their building and waiting around I can just come down from my room and see them.

CTI, community

Theme 4: mental health and medication

All participants talked about their mental health problems and symptoms. In particular, they talked about how poor mental health disrupted function and prevented them from working:

Because of my mental health, I don’t have to work. That is fair I think. I mean you can’t be working and thinking you’re hearing voices or something. That doesn’t work does it?

CTI, community

I can only really work when things aren’t too bad. There are days when I can’t get out of the house so then I can’t be in a job where they expect me in at the same time every day.

TAU, prison

Participants in both intervention and control groups talked about medication in relation to their mental health. Some individuals reflected on the positive impact of medication on their lives:

Well just medication to keep me on the level. Keep me grounded like. And that’s all really. I can cope pretty well with that.

CTI, community

Others felt that, without high levels of professional support, they would not be able to continue to take medication on a regular basis. Reinstitutionalisation, to either hospital or prison, appeared inevitable:

It’s going to start going downhill because you either don’t take your medication or when it runs out you stop and then there’s no going back. For me when that happens I’ll end up in prison or hospital.

TAU, prison

A more common theme in participants’ narratives was a sense of coercion in relation to medication:

They always think they know best. They’ll tell you ‘you need this or that drug’ and everything you think is wrong but I can sort myself out.

TAU, prison

Tablets for what they say is schizophrenia but I don’t always take them.

TAU, prison

Participants also protested about the unavailability of other treatment options, in particular a lack of psychological therapies:

They push me to keep taking these drugs and saying I can’t have this or that therapy or services. I don’t want that, drugs and psychiatrists, I want them to get me into counselling. I need counselling, that’s what would help me. Because I’ve been through so much, I don’t just need the drugs, I need someone to sit and listen to me and help me through this.

TAU, prison

Participants seemed to suggest that CTI was providing them with more support around mental health than previously:

All it was, yeah, was an injection. I’d go there, they’d do the injection then I’d go. No psychiatrist or no talking or anything. This time they reckon I’ll have a CPA in first 2 weeks so that’ll be better. I’m hoping everything goes right with the CPA. I needed support. For mental health mainly. And that’s been there this time. I feel very positive about the future now. It might not seem big to you or anyone else but [CTI manager] she’s been really helpful.

CTI, community

Theme 5: stigma

Participants’ accounts highlighted their feelings that the stigma of being an (ex)-offender permeated every aspect of their lives. They reported that stigma negatively impacted quality of life, hindered their job prospects and increased the risk of reincarceration:

There’s no chance of anyone employing me, is there? Why would they [employ me]? They’d find out I’ve been in prison and that’s it. Not even a chance.

TAU, prison

Almost 10 years in prison, I don’t have any chance of a job, do I?

CTI, community

A job isn’t an option with me being like I am and the history I’ve got.

TAU, prison

Participants reported that the stigma associated with being an offender not only impeded job prospects, it also had serious implications for access to their children:

I’m pretty gutted about this but I don’t know whether I can see my lad. That gets to me and I’m sad about that but maybe I just have to accept it. I don’t mean not think about him, I don’t want you to think that. But that I won’t be able to see him like I hoped. I’d have to go to court but I don’t think I can do that. I don’t know how I’d do that. And they’d see me wouldn’t they? Ex con[vict]. They aren’t going to like me are they?

CTI, community

Participants also reported self-imposed stigma in relation to mental illness, often keeping them from disclosing difficulties and from seeking mental health treatment in prison. This participant’s account typified the views of others. When asked about now receiving mental health support in prison, he replied:

Yeah, first time, in here, I always just kept it quiet, the voices and that, and self-medicated, because my idea of mental health and shrinks was that they’d put you in a straitjacket in a hospital and that would be it.

CTI, prison

It is noteworthy that the support this participant had been receiving from the CTI manager appeared to be changing his views of mental health treatment:

Yeah, they’re alright yeah. That guy who’s left seemed like he knew what he was doing, and that tall guy, dresses smart, he’s OK.

CTI, prison

Summary

There were commonalities and differences between participants in the CTI and TAU arms of the study. Commonalities were most apparent in participants’ reports of their reliance on others for practical help, particularly in terms of accommodation and financial support. The former was regarded as a potential source of tension within families, which had the potential to trigger reoffending. Financial reliance on families reinforced self-reported perceptions of being seen as ‘other’ and deviant. Embarrassment at needing financial help, particularly when received from siblings, increased the risk of reoffending. Both CTI and TAU participants stated that a lack of suitable accommodation had serious implications for reoffending. Similarly, not having a permanent address restricted access to benefits and services. Sharing accommodation with other (ex)-offenders and people with mental health and substance misuse problems increased participants’ vulnerability to resuming involvement with drugs and alcohol as well as putting their mental health at risk more generally. In both cases, participants indicated that this increased the risk of reoffending. Both groups of participants reported feeling coerced into taking psychotropic medication and expressed disappointment that there was a lack of access to psychological interventions. This, together with stigma, caused some participants not to disclose their mental health problems to professionals.

There were, however, differences between groups. Most notably, members of the CTI group who had experienced previous incarcerations reported less uncertainty and a sense that on this release care would be more integrated; this was linked to reduction in stress, anxiety and potential for reoffending. The CTI group also reported better continuity of care and improved access to services attributed, at least in part, to CTI managers advocating on their behalf. From these participants’ perspectives, there was a direct correlation between improved discharge planning, increased levels of support, greater continuity of care provided by CTI managers and a reduction in the likelihood of reoffending.

Methods

Theme 1: liaison

Health and justice professionals’ interviews highlighted the importance of liaison between individuals and agencies if interdisciplinary approaches were to prove effective. Within this overarching theme, there were four subthemes: collaborative care, information sharing, gate keeping and resource.

Theme 2: transition

Summary

Perhaps unsurprisingly, the emergent themes from professionals’ narratives indicated a preoccupation with liaison, specifically the extent to which collaborative care was being realised in everyday practice. Participants reported barriers to effective planning and delivery of services as linked to increasingly limited resources. This had led to the raising of thresholds for gaining access to services and more robust gate-keeping on the part of individuals, rather than necessarily being indicative of service culture.

Perceptions and experiences of CTI were positive. However, interviewees raised concerns about the availability of funding to roll out services, highlighting the considerable variation not only across the country, but also within localities such as London. Given that best practice suggests that service users should receive individualised care that is tailored to meet their discrete needs, participants wondered whether or not this was possible in the context of a manualised ‘Improving Access to Psychological Therapies-like’ approach to service, characterised by the delivery of a predetermined number of sessions, compared with a needs-led model, incorporating time to build and appropriately disengage from therapeutic relationships with CTI managers.

Supportive relationships, such as those provided by CTI managers alongside family and friends, were regarded as vital for effective transition from prison to the community. However, participants reported that these could not make up for shortcomings in more practical and essential aspects of care, specifically with regard to accommodation. In common with service users, professionals frequently complained about the lack of suitable accommodation, highlighting the increased risk of reoffending and exacerbation of mental illness within this vulnerable group caused by unsuitable housing.

Results

In the 4-week period before release, about two-thirds of both groups had contacts with care co-ordinators, and around one-third had contacts with psychiatrists and mental health nurses (Table 16). People having contact with ‘other nurses’ and ‘mental health nurses’ were slightly higher in the TAU group than in the CTI group, with frequency of ‘other nurse’ being particularly high in the TAU group. Overall, total contacts per person using services were similar between the groups, at around five per person.

The highest mean cost in both groups, after the cost of CTI, was for psychiatrist contacts, followed by care co-ordinators and other nurses (Table 17). Together, these services made up 76% of costs for both the CTI and TAU groups. The total for the CTI group was 12% higher than for the TAU group.

Table 18 shows the service use in the 6 weeks after release. The CTI group made more use of care co-ordinators and psychiatrists than the TAU group. The overall average contact (excluding inpatient services) was higher for the CTI group. When looking at inpatient care it can be seen that few were admitted. However, the number of days in hospital differed. CTI participants had more days admitted to medical wards and TAU participants had more days admitted to psychiatric wards.

Even though very few participants were admitted to hospital, this care accounted for 77% of costs for the CTI group and 84% for the TAU group, excluding CTI costs (Table 19). Psychiatrist and care co-ordinator costs were around twice as much for the CTI group than for the TAU group.

Critical time intervention contacts account for majority of the costs from release to the 6-week follow-up (see Table 19). This is because the majority of the intervention, in terms of contacts, is carried out in this time period. However, in reality, this cost would likely be more spread out across the follow-ups, as the work would be not be so concentrated in this period. Preparation work that does not involve actual patient contact would shift some of the cost before this period. Similarly, monitoring of patients in the final few months that does not typically result in regular face-to-face contact would also shift some of the costs to later in time. In addition, the single person with a 31-day stay in hospital is driving up the cost per person in the CTI arm as well as the intervention costs.

The period before the 6-month follow-up covers a period of 20 weeks. The number of people in the CTI arm having contacts with care co-ordinators was higher than at 6 weeks and the frequency of these contacts was also higher (Table 20). The TAU group had increased contact with mental health nurses compared with the 6-week follow-up point and also compared with the CTI group. The use of care co-ordinators was slightly higher in the CTI group at 6 months than in the TAU group. The average total service costs were again higher for the CTI group (Table 21). There is not a significant difference in the cost of mental health nurse contacts between the two groups because of the lower frequency of contacts.

At the 12-month follow-up, the CTI group’s use of care co-ordinators remained similar to the previous follow-up and noticeably higher than for the TAU group (Table 22). The mean number of contacts with care co-ordinators was also far higher in the CTI group. Contacts with psychiatrists are slightly higher and more frequent in the CTI group. A higher number of the TAU group have contacts with mental health nurses; however, the frequency of these contacts is lower. Overall, the number of service contacts in the CTI group was about three times higher than in the TAU group.

Table 23 shows that psychiatrists and care co-ordinators are the drivers of costs in the CTI arm. For the TAU group, inpatient care is the main cost driver. Overall, the costs are almost identical including mental health nurse contacts.

At each time point, excluding the 12-month follow-up, the mean costs are higher for the CTI group (Table 24). The cost per week is helpful when comparing between time periods. It can be seen that costs increase initially for both groups and then fall after the 6-week follow-up.

Without the imputed CTI intervention costs, the average service use costs in the intervention arm for the 4 weeks before release and the period from release to the 6-week follow-up are £195.35 (down from £717.19) and £440.21 (down from £2592.80), respectively. Both the service costs at 4 weeks before release and at the 6-week follow-up are not significantly different once CTI contacts are taken out, possibly because of the small sample size and high number of dropouts.

Service costs over the whole follow-up period in relation to age show large differences. For those participants aged 20–29 years the mean cost was £2432.94, for those aged 30–39 years it was £4521.58, for those aged 40–49 years it was £1647.20, and for those aged ≥ 50 years it was £2087.87. Participants with schizophrenia had mean costs of £3217.07, whereas for those with depression the cost was £1867.48. Participants who were diagnosed with schizophrenia complicated by a PD had average service costs of £3528.85. The remaining had an average cost of £2445.49. Those with schizophrenia complicated by alcohol or drug misuse had average costs of £3382.34 compared with £2735.07 for those without. Participants with an Axis II diagnosis had average costs of £3071.89 compared with £2546.88 for those without.

White participants’ costs were higher (£3523.04) than black (£2272.75) or Asian (£1953.12) participants’ costs. Perhaps not surprisingly, costs were substantially higher for participants who were unemployed (£3539.55) than for those in work (£1385.40).

Regression analysis performed on the cost data showed that service costs for unemployed participants were 1.5 times higher than for those who were in employment, and that costs for those not at work due to sickness (and those absent from work due to illness) were, similarly, 1.5 times higher than for those in employment. However, this relationship becomes insignificant when you exclude inpatient costs, suggesting that this relationship is driven by the inpatient costs, a service that relatively few people used. The cost breakdowns previously are for only the patients who had full service use information for the year, which totalled 84 participants. These costs also include the imputed cost of the intervention.

Combining the clinical effectiveness from the primary outcome measure, a difference of proportion of people engaged of 0.161 at 1-year follow-up and, costs, a difference of £2485.10 in favour of the CTI arm, gives a cost per additional person engaged of £15,426. To determine if this is cost-effective depends on society’s willingness to pay for an additional person to be engaged with mental health services at 1 year following release.

Achievement of the objectives of the study

The primary objective at the start of the trial was to establish whether or not CTI was clinically effective for released adult prisoners with SMI in:

• improving engagement with health and social care services

• reducing mental health hospital admissions

• reducing reoffending

• increasing community tenure through reducing time in prison.

The primary objective was partially met.

• Those allocated to the CTI arm of the trial were more likely to be engaged with mental health services at the 6-week follow-up stage of the trial.

• Those who received CTI were also more likely to be registered with a GP.

• Continuing engagement with services was demonstrated for the CTI group at 6 months following release.

• This advantage was not observed at the 12-month follow-up, although CTI participants were still more likely to be currently under CPA at this point.

As noted earlier, delays in recruitment led to us being unable to analyse the impact of CTI on hospital admission, reoffending and overall community tenure. This is considered further in Limitations.

The secondary objectives at the start of the trial were:

• to establish the cost-effectiveness of CTI for this population

• to develop a service delivery manual and training materials to support the implementation of CTI in vivo

• to facilitate and promote active service user, CJS and NHS staff participation in the research programme, thus encouraging greater engagement between the academic and practice communities.

The study’s secondary objectives were achieved.

The health economic analysis consisted of a cost comparison between the two groups. The analysis indicated that:

• participants receiving CTI incurred higher costs in relation to their service use than the TAU group before, and after, release

• at each time point (except at the 12-month follow-up), the associated cost of service use by CTI participants was higher than those for the TAU group. Largely these costs were associated with psychiatrist and care co-ordinator contact

• at no point during the course of the trial were costs incurred because of psychiatric inpatient hospitalisation for the CTI group. This positive outcome for the CTI group may reflect their increased contact with community services. Given that hospitalisation is by far the most expensive form of service provision, the increased service cost incurred by the CTI group may, over a longer time period, be cost-effective.

The main contributor to cost was that of the CTI managers. These costs were mostly incurred during the 6-week follow-up period. However, the use of other services was also generally increased. That costs were increased for the intervention group does not imply that the intervention is not cost-effective. If outcomes are improved by a sufficient amount then these extra costs would be acceptable. The outcomes described show that engagement was improved. We can assume that better engagement leads to more clinically effective treatment and, ultimately, improved well-being. We would argue that the increased costs would be justified. Ultimately, however, it is a value judgement whether or not this is a good use of funds.

A treatment intervention manual, with training materials included, was achieved. The manual was developed through a working group comprising prison- and community-based staff, experts by experience and researchers. The manual details the origins of CTI, outlines results of previous and current research, details the stages and content of the intervention to allow replication and includes skills training suggestions for staff. The establishment of the working group to participate in the development of the case manager manual also achieved our final secondary objective, that of encouraging engagement between researchers, practitioners and users of mental health services.

Qualitative study: service user and professionals’ view of the intervention

The quantitative results summarised were augmented by a strand of concurrent qualitative enquiry running throughout the life of the trial.

The comments received by the service users allocated to the CTI group were overwhelmingly positive. CTI participants reported:

• reduced anxiety with respect to their impending release from prison

• on the value of the CTI manager’s role linking with community services, which made contact and engagement easier and less daunting

• on the CTI manager’s advocacy role, that is having someone who understood their needs, and how community providers operate to ensure appropriate transfer of care

• receiving emotional support from the CTI manager, that is having someone they knew and trusted on both sides of the gate

• an increased ability/wish to persevere with attending mental health/drug and alcohol service appointments, and maintaining treatment programmes that previously have fallen by the wayside

• a continuing increase in self-confidence and self-reliance over time with regard to getting what they want from service providers

• higher personal aspirations in relation to finding work, vocational training and/or education

• improved relationships with peers and family members.

Staff who delivered CTI or staff who had observed the running of CTI within their service reflected on the barriers to, and facilitators of, successful delivery of the intervention, including the following:

• Information sharing across service boundaries was seen as a barrier, taking a long time for CTI managers to establish; however, it could also facilitate care when it was functioning properly, contributing to averting potentially serious consequences.

• The ability to spend more time with service users enabled CTI managers to establish rapport and facilitated continued engagement with them and other services.

• Increasingly limited resources led to raising of thresholds for gaining access to community services, resulting in CTI managers spending increasing amounts of time negotiating access.

• The time-limited nature of CTI was seen as a positive, but some felt that the 6-week cut-off point was too rigid and needed some flexibility to account for individual need.

• There was a lack of suitable accommodation options for clients.

Strengths

This research is the first RCT, worldwide, of CTI as a model for service delivery to support the transition of male prisoners with SMI back to community living. Prisoners have complex and comorbid mental health problems at rates significantly higher than the general population, and they frequently find themselves socially disadvantaged when resettling into the community life. Existing research has identified the problems that they face around gaining suitable accommodation and employment, and in repairing fractured family relationships. Their complex mental health needs make them unpopular patients in many community services both in terms of their clinical needs and their patterns of engagement, which can be chaotic and sporadic, influenced negatively by other lifestyle factors.

The study is also the first UK economic evaluation of delivering CTI as a high-intensity time-limited through-the-gate intervention for prisoners with SMI.

Limitations

Throughout the study we encountered numerous obstacles that challenged study set-up and recruitment. Our experiences provide evidence on the difficulties of obtaining research and ethics permissions and, when these systems fail, with implications for research in general. In addition, it also needs to be acknowledged that retendering and changes of provider, which are common throughout the NHS, including prison-based services, can effectively halt, or at least severely delay, ongoing research.