Offenders, especially prisoners, have a high prevalence of common mental health problems (e.g. depression, anxiety). Rates of 50–90% for all mental health problems have been found in UK [1, 2] and international [3] prison populations, with high levels of unemployment, relationship problems and past trauma [4, 5]. Prison has been identified as an opportunity for treatment but release into chaotic community environments poses a challenge for treatment and also research [6]. Critical Time Intervention has been developed and evaluated for prisoners with severe mental illness and shown to have a significant impact at 6 weeks and 6 months post release from prison in terms of increasing engagement with services [7]. No studies have attempted to evaluate such ‘through the gate’ mental health interventions for prisoners with common mental health problems.

In 1996 ‘Patient or Prisoner’, a Discussion Paper by Her Majesty’s Inspectorate of Prisons, highlighted the inadequacies in prison healthcare in England and Wales and argued for equivalence, stating explicitly that prisoners should be entitled to the same level of healthcare as that provided to the general public in the community [8]. Prison mental health in-reach teams were established in England and Wales over the last decade for prisoners with severe and enduring mental illness. However, mental health in-reach services fall short of community equivalence, with wide variation in service provision and that services for common mental health problems in prison are even more limited [9–11]. Although the UK government developed the Improving Access to Psychological Therapies (IAPT) for depression and anxiety, these services have been poorly resourced in prisons, with pharmacological interventions often being the only treatment provided [12]. This is despite some evidence that psychological interventions, even ‘low intensity’ treatments, may be as effective for offenders as they are for the general population [13]. An observational, prospective cohort study evaluating IAPT for prisoners found clinical recovery being achieved in 55% of depression and 52% of anxiety cases [12]. However, while this is similar to results in community IAPT demonstration sites in Newham and Doncaster [14] effect sizes in observational studies do not account for regression to the mean.

Discontinuity of care on release from prison is the norm [4]. Once released, ex-prisoners with common mental health problems are, in theory, provided for by mainstream statutory services including general practice and IAPT services. In reality few access these services [6]. Many prisoners with common mental health problems have substantial co-morbidity with personality disorders and substance misuse [2, 6] and these offenders may fall between the cracks in service provision between general practice, IAPT and substance misuse services [15–18]. Offenders are further disadvantaged by their resistance to seeking help and to accepting mental health diagnoses, with lower levels of GP registration and high rates of personal and social problems such as homelessness and relationship difficulties [4–6, 19–22].

To address these multiple problems, we have developed a complex collaborative care intervention aimed at supporting male prisoners with common mental health problems near to and following release (ENGAGER), working with criminal justice providers (prison and community), third sector social inclusion services, health services and people with lived experiences (peer researchers), using a range of methodologies [23]. The underpinning principles and practices of ENGAGER are (a) to develop trust and engagement through showing respect and giving practical support; (b) to support mental health through a psychological therapy informed ‘shared understanding and action plan’ and mentalisation-based approaches which are not disorder specific or based on the ability to turn up for weekly appointments; and (c) to supporting individuals to achieve their personal goals through alignment of personal strengths, family and community resources and joint work with criminal justice, third sector and other health providers.

Mentalising is a natural human ability. It is the capacity to think about our own mind and the minds of others and understand how emotions, thoughts, wishes and impulses lie behind and influence our behaviour and the behaviour of others. Good mentalising involves being able to acknowledge that often we do not accurately know what people are thinking and feeling but that often we can make more or less accurate guesses. Good mentalising also involves having an authentic interest in other peoples’ emotions and thoughts and not making quick assumptions about why a person may have behaved as they did. For example, rather than assuming why your client has started to drink again, instead being open and curious to exploring with them what was happening for them.

Prisons are complex and difficult environments in which to conduct research, and simply getting access can be difficult for researchers. Rigorous evaluation designs such as randomised controlled trials (RCT) can be challenging to implement without sufficient piloting. In addition, studies following recently released prisoners can be affected by low retention rates [24]. Therefore, for all these reasons, it is essential that trial methods are tested prior to a definitive trial.

This pilot trial aimed to investigate whether it was possible to recruit and retain prisoners with common mental health problems in a RCT of the ENGAGER intervention. The specific objectives of the pilot trial were to address feasibility issues and uncertainties about undertaking a definitive RCT of the ENGAGER intervention: inclusion and exclusion criteria, adequate rates of eligibility, consent and randomisation, acceptable rates of participation in the intervention, ability to maintain researcher blinding to trial arm allocation, ability to retain participants at 1 and 3 months post release from prison; and to provide sufficient levels of completion of outcome measures. Success criteria for this pilot trial were to recruit 60 people to the study (sample size sufficient to test trial procedures), to deliver the ENGAGER intervention, and to follow up at least 50% of all participants 3 months after release.

In line with the Medical Research Council framework [25] for the development and evaluation of complex interventions, this paper describes the process of the pilot trial and lessons for trial science.

This external pilot trial is reported in accord with the CONSORT 2010 statement: extension to randomised pilot and feasibility trials (Additional file 1).

This pilot trial sought to assess whether it was possible to demonstrate acceptable levels of recruitment and retention amongst prison leavers with or likely to have common mental health problems, in order to inform a future randomised controlled trial. It is important to perform pilot feasibility trials when the logistics of a large-scale trial are unclear [31]. There are a limited number of intervention trials within criminal justice settings, and studies conducted in prisons and requiring community follow-ups after release are both rare and particularly problematic.

The required number of participants (n = 60) were recruited and randomised within the 9 month timescale set for the pilot trial. Of the people who were assessed for eligibility, the majority were leaving prison to a destination outside the geographical area of where the intervention was delivered. The main trial is addressing this to some an extent, as both the North West and South West sites have extended the release area. In addition, following the introduction of resettlement prisons, prisoners are increasingly likely to be moved to a prison closer to home within the last 6 months of their sentence [32]. These changes should increase the numbers being released to the local area of each prison, and therefore increase eligibility rates.

Participation of individuals eligible to be approached and screened was high, with 71% consenting. Of those who declined to take part, the majority were initially approached via the Clinical Studies Officer and not directly by a member of the research team. Feedback from participants approached by the Clinical Studies Officer indicated that they did not really understand why they were being approached by someone not linked to the research. Also being approached directly by the research team enabled questions about the study to be more thoroughly discussed and hence they were better informed about the project. Face-to-face initial approach from the research team is the chosen method of recruitment for the main trial, and recruitment has been shown to be improved by face-to-face consultations [33–35].

Of the 110 individuals consented, 55% screened in (60 out of 110) as having or likely to have common mental health problems on release. Twenty-eight percent screened in on the basis of the historic common mental health question but not the validated questionnaires (and one of these participants later declined the intervention), raising questions about the appropriateness of this element of the selection process. We have now made changes to the historical common mental health screen by also assessing if participants believe symptoms of common mental health problems are likely to recur following release from prison.

Retention of 73% at 1 month demonstrated our ability to follow up after release. The 47% at 3 months was more disappointing, and we have developed further protocols to deal with changes to plans after release. We found that a flexible follow-up window of 4–12 weeks allows researchers to contact and re-engage more participants and we are using this method in the main trial. Researchers will continue to take multiple contact details at baseline for people at high risk of homelessness or transiency, sending several reminders by text message or phone call and offering alternative ways of engaging (phone interview rather than face-to-face). Losses to follow-up are likely to be reduced further as researchers will follow up participants out of area and approval has been granted to provide ‘thank-you’ vouchers at community follow-up sessions. In addition, we are building closer relationships with other involved agencies (e.g. National Probation Service, Community Rehabilitation Companies, substance misuse and accommodation services), including developing information sharing agreements. All of these being methods likely to improve follow-up rates and obtaining consent to contact.

Blinding of researchers was a key problem. Participants were keen to share their experiences with the researchers and very often experienced the research and practitioner teams as both representing ‘ENGAGER’. To maintain blinding would have required rejecting participants’ very practical requests to share contact information and their commitment to ‘ENGAGER’. We considered a range of possible solutions to maintain blinding, such as using a paper-based self-complete outcome measure for participants but decided against this strategy due to literacy problems and the likely increase in incomplete data. In the main trial the researchers will know trial arm allocation. This is seen very much as a positive, as it allows for the continued building of rapport between the researcher and participant to facilitate follow-up rates and allows the participant to openly share their experiences. The researchers will deliver the primary outcome measure in a protocolised way to minimise bias. All other measures will be delivered in the more conversational style, which had been developed in order to reduce stress within the research process.

In terms of other pilot trial findings: randomisation was accepted and the data completion of outcomes at baseline and follow-up were excellent. Engagement with the intervention, a crucial indicator of viability, was also good exceeding attendance for therapy in IAPT service evaluations [14]. Limitations of the pilot included the use of an outcome dataset which was still being changed and finalised throughout the duration of the pilot. While complete outcome data was collected from all participants successfully followed up, not all participants completed the same outcome measures. The results of the testing of these outcome measures and decisions for inclusion in the definitive trial will be reported elsewhere. Strengths of this study included the use of peer researchers to refine procedures and testing all the key procedures in both prisons proposed for the main trial.

In conclusion, the pilot demonstrated the potential to successfully run a definitive trial. Recruitment is feasible but takes time due to many prison leavers going back to other areas; follow-up is feasible and requires organisation, tenacity and flexibility from researchers; randomisation and collection of outcomes are far less problematic than ensuring researchers are blind to allocation; the ENGAGER intervention was feasible to deliver and acceptable as demonstrated by high levels of engagement. The full trial started in January 2016 (ISRCTN: 11707331).