Background
Personality disorder is highly prevalent among men and women with an offending history. In the United Kingdom, 64-78% of the adult male prison population and 50% of females meet diagnostic criteria for at least one personality disorder [
1]. Personality disorder is the most common mental disorder in the probation population, affecting up to half of probationers [
2]. Within high secure psychiatric facilities, personality disorder is definitely diagnosable in 57-77% of male patients [
3]. People with an offending history and personality disorder (personality disordered offenders: PDOs) are a group whose difficulties come at a considerable cost to themselves, potential victims, the communities in which they live and return, and to society as whole who must meet the costs of service provision. PDOs experience worse physical and mental health, poorer quality of life, reoffend at higher rates and are overrepresented in the commission of serious further offences [
4‐
6], indicating existing approaches may be overlooking important factors.
PDOs are supported by health, criminal justice, social care and third sector services. Whilst approaches have varied internationally and over time, a consistent theme is the requirement of services to reduce risk, improve health and improve social outcomes [
7,
8].
Social outcomes are those that result from functioning effectively in society, for example participation through employment, family roles and independent living. In this paper, social outcomes are conceptualised as
participation, as defined by the World Health Organization [
9] as ‘involvement in a life situation’. Attention to participation among PDOs is vital for two reasons. Firstly participation in personally meaningful
and socially valued (prosocial) activities is integral to functioning, health and social outcomes [
9]. Secondly, in offender populations, participation is also associated with desistance and reduced risk of reoffending [
10‐
12]. Conversely, offenders who do not participate in prosocial activities (e.g. remain unemployed or lack prosocial relationships) or participate in antisocial activities (gang affiliation, substance use) are at higher risk of reoffending [
13].
Social outcomes for ex-offenders are poor. For example in the UK the employment rate is only 27% on leaving prison [
14], and of those referred to support agencies only 16% found and kept employment for 6 months or more [
15]. Research into interventions to facilitate participation and improve social outcomes amongst PDOs specifically is limited, despite this important contributor to health, quality of life and desistance often being mentioned as an aim of service providers.
Objectives
The objective of the review was to determine the effectiveness of interventions to improve social outcomes among offenders with personality disorder.
Methods
We conducted the review according to the stages outlined in the Cochrane Collaboration handbook for systematic reviews [
16]. Review methods and inclusion criteria were pre-specified in a protocol and registered on PROSPERO: ID = CRD42016042304 [
17].
Eligibility criteria
We included English language studies reporting research where participants were adult offenders with personality disorder, reporting any intervention (e.g. psychological, pharmacological, occupational, social) and a social outcome, i.e. participation in a community setting. No limitations were placed on date or quality of research papers. Opinion pieces, commentaries or service descriptions, editorials, and publications addressing laws, policies and/or media reports were excluded.
Offender status was defined as having committed at least one criminal offence as reported from an official source or self-report. Personality disorder or psychopathy was considered present where participants had a formal diagnosis indicated by use of structured tool or justified method. Social outcome was defined as participating in any prosocial activity or engaging in a social role in a community setting (not prison/inpatient hospital) after encountering criminal justice services. For example, employment, volunteering, running a household, caring for children or being in an intimate relationship.
We searched databases for criminal justice, psychological, social, allied health and psychiatric research (Web of Science, SCOPUS, PubMed, EMBASE, AMED, CINAHL, ASSIA, PsycINFO, National Criminal Justice Reference Service (NCJRS) Abstracts Database, Cochrane collaboration, Campbell collaboration) and grey literature (theses, relevant reports, UK government documents). Original database searches were completed in July 2016. Automatic database searches were used to maintain an up-to-date review until September 2017. We reviewed reference lists of included studies and key papers. Relevant journals were surveyed on a regular basis.
Search strategy
The search strategy was tailored to the requirements of each database with input from a search strategist to include terms pertaining to 1) personality disorder or psychopathy, and 2) offenders, and 3) participation in a community setting.
Study selection
We imported database results into Endnote reference management software [
18] and removed duplicates. All titles and abstracts were screened to determine if a citation met inclusion criteria by CC. A random selection of citations (
n = 400, 22%) was independently reviewed by the second reviewer (VF). Inter-rater reliability reached substantial agreement, calculated using Cohen’s Kappa [
19,
20]. Where agreement was not reached on inclusion by discussion the third reviewer (EAM) reviewed the material and gave a definitive judgement. Where insufficient information was available from the abstract the full text was obtained to determine if it met inclusion criteria.
Data collection
A data extraction tool was refined after piloting to include: year of data collection, country of origin, the aim/hypothesis of the study, study design, inclusion criteria, participant demographics, personality disorder diagnosis method and prevalence within sample, offender status, participation outcome of interest, description of intervention, analysis method, and results. Due to the small number of included studies, CC extracted all the data which was checked by the second and third reviewers (VF, EAM).
Risk of bias in individual studies
We appraised study quality using validated structured tools appropriate to study type [
21‐
23]. Studies were not excluded based on quality due to the limited evidence in this area. This is considered in interpretation of the review findings.
Summary measures
As social outcome and participation are rarely discussed in the literature, it was unlikely that these terms would be used as study outcomes. To identify relevant outcomes, a range of terms were anticipated, e.g. employment, education, intimate relationship, community roles and leisure activities. Terms were derived from the WHO International Classification of Functioning chapters on activity and participation [
24].
Data synthesis
The Cochrane Collaboration four-step method of narrative synthesis of effectiveness studies [
25] was applied with consideration to the inclusion of non-RCT designs. The steps are as follows: 1) Develop a theory of how interventions operate, 2) Preliminary synthesis of findings, 3) Exploring relationships in the data within and between studies, 4) Assessing the robustness of the synthesis.
Discussion
This review evaluated the effectiveness of interventions to improve social outcomes among PDOs. It conceptualised social outcomes as analogous to the World Health Organisation concept of participation [
9]. Five studies were included in the review. Narrative synthesis was conducted due to high heterogeneity. Reported outcomes could be grouped into employment and social functioning. There were three potential mechanisms identified in the interventions that may improve participation. There was no evidence for significant change in social functioning and although rates of employment were noted to increase, the quality and designs of the studies prevent attribution to the interventions.
To maximise effectiveness, interventions should be developed based on evidence of what the relevant influencing factors are, and a theory of the mechanisms by which they operate to bring about a desired outcome [
42]. However, interventions identified in this review lacked theoretical explanations of how they may work to improve participation and there was variation in the social outcomes reported.
Attention to employment as a relevant outcome is consistent with the literature that identifies employment as an important factor in desistance from crime in offender populations [
43,
44] and in protecting against serious reoffending among mentally disordered offenders [
10]. Within the health literature, employment is identified as an indicator and facilitator of mental health and wellbeing [
45]. However, employment only reflects a small component of participation, which includes many activities essential for survival, health and social inclusion such as leisure, domestic responsibilities and civic involvement. Social functioning appears more closely related to participation defined in this way. Consensus around measures of participation in mental health is lacking, complicated by ongoing debates on operationalising participation that have resulted in multiple measures being developed [
46]. Until a stronger consensus is achieved on the outcome of interest and its measurement, there is a risk of continued heterogeneity in intervention research that prevents synthesis of trial results.
Three potential mechanisms of change were identified from synthesis of the study descriptions. The first of these was skill development, which may be based on the hypothesis that lack of social, emotional or practical skills impede participation. Skills training is well established in criminal justice programming, including specific programmes for PDOs (e.g. [
47]). However, interventions are institutionally based and research is required to determine if any skills learned are transferred and applied to participation in the community, or ‘real-world’, on release. The second mechanism was facilitating change in values and identity through supported prosocial goal attainment and validation of efforts. This approach is increasingly adopted in forensic practice to address motivation and engagement for risk-focused intervention, by framing offending as a barrier to achieving prosocial goals [
48]. ‘Volitional realignment’ towards prosocial goals and identity change are argued to occur through mastery of new prosocial activities by those practicing from an occupational perspective [
49]. Interventions using this approach are yet to be proven effective. The final mechanism was practical assistance given to compensate for participants’ difficulties, for example taking someone to a leisure centre. Whilst this has an immediate effect, it does not impart a change in the individual him or herself, and thus may not support continued participation on a long-term basis. This approach is consistent with the Individual Placement and Support model, which has been shown to be effective in supporting individuals with serious mental illness into employment [
50]. However, whether employment is then sustained is less clear from the literature. Similarly, in employment interventions for ex-offenders, only 16% retained any employment achieved for longer than 6 months [
15]. This is an important consideration in providing interventions of long-term effectiveness and when working with individuals with personality disorder, whose difficulties may never ‘resolve’, as can be the case for people with psychotic disorders.
The heterogeneity of the studies in this review indicate that a theoretically-informed exploration of the factors influencing participation is required, before systematic development and evaluation of interventions that are likely to be effective can be conducted [
51]. Identifying the influencing factors can be approached from multiple perspectives. For example, identifying the features of personality disorder, such as traits or severity, that influence social outcomes and thus targeting treatment at the modifiable traits or symptoms. An alternative approach would be to identify which components of participation influence social outcomes among PDOs. Intervention would then be focused on modifying these components of participation, rather than attempting to ameliorate signs and symptoms of disorder/s. This approach is more familiar to rehabilitation professionals, who advocate the WHO position that health and functioning are achievable irrespective of disability, disorder or disease [
9].
The WHO International Classification of Functioning (ICF [
9]) provides an internationally recognised framework for describing and classifying strengths and difficulties in participation in great detail. However, it does not explain how different factors interact to produce participation, and as discussed, operationalisation of participation remains contested. The Theory of Human Occupation and its related conceptual practice model [
52] explains of how participation is achieved, experienced, maintained and changed, and has valid and reliable measures for associated and influencing factors. Although it has not been tested specifically with PDOs, the Theory of Human Occupation, like the ICF, is based on universal principles and its utility is evident in its use in international forensic research and practice [
31,
49,
53]. This may present a starting point for identifying relevant factors and the mechanisms of change that can be facilitated through intervention.
Limitations
The review was conducted according to a pre-specified protocol informed by Cochrane guidelines for conducting systematic reviews [
16]. In a clarification to the protocol, studies were included where results were not differentiated for PDOs only where at least 60% of the sample had personality disorder/psychopathy and at least 60% had committed an offence. Previous systematic reviews have set a level of 70% when taking outcomes from a mixed sample [
54]. Because of the known high prevalence of undiagnosed personality disorder among offenders, a slightly lower percentage was considered acceptable. These criteria permitted inclusion of the study by Öhlin et al. [
32].
The feasibility of RCTs is limited in testing complex interventions involving prolonged therapy, such as psychotherapy or occupational therapy. This informed the decision not to exclude studies on the grounds of quality. Due to the inclusion of low quality studies, conclusions drawn from this review must be interpreted cautiously.
Conclusions
No interventions identified were designed to specifically improve social outcomes in the community among PDOs. There is some evidence that employment can be achieved although changes cannot be attributed to interventions due to the study designs used. There was no evidence for interventions aiming to improve social functioning.
There is a sizeable gap in the literature reporting interventions to improve social outcomes, describing the mechanisms by which they are proposed to work, and testing effectiveness. This is further constrained by the focus on different outcomes and variation in how they are measured. Consequently, services for PDOs are unable to apply evidence-based interventions that are likely to increase social outcomes among offenders with personality disorder in the community.
Implications for practice
Services and practitioners working with PDOs in the community currently lack evidence on which to base interventions that specifically target social outcomes. Service providers may consider interventions that are orientated towards achievement of a personally meaningful prosocial identity; target skill deficits that impact on successfully maintaining participation in employment and social relationships; or provide practical assistance to access prosocial roles that have previously been unfamiliar.
Implications for research
Evidence for effective interventions to improve social outcomes is lacking. Further research is required to identify the factors that influence participation, develop interventions to target these, and to test their effectiveness.