Design

This systematic review follows Cochrane guidance [19] on conducting reviews and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines on reporting review processes [20]. The eligibility criteria for study inclusion were developed with the participants, interventions, comparators, outcomes, study design (known as PICOS) framework.

The review protocol was registered with PROSPERO (CRD42021257078).

Search strategy

Five electronic databases (Embase, Medline, PsycInfo, CINAHL and Web of Science Core Collection) were utilised for the search. All databases were searched from their inception to the 29 May 2021 by the first author (SI). The search was limited to papers written in English. Additional searches were performed through hand-searching and forward citation searching of the included papers to ensure all relevant papers were identified.

An example search strategy for Ovid Embase is shown in S1 Appendix. Search terms were identified for four concepts: “self-management intervention”, “influencing factors”, “secondary care”, and “severe mental illness”. These terms were generated through conducting a limited search of the literature, analysing the keywords of the generated papers from this search, and by using the databases’ thesaurus to extend those keywords. Also, the terms were combined with standard MeSH terms from the PubMed and Cochrane databases and Subject Headings for the PsycINFO database.

All references were imported into EPPI-Reviewer 4 [21] for screening and selection, after de-duplication.

Inclusion and exclusion criteria

Inclusion criteria. Participants: Staff facilitating the self-management intervention in a secondary mental health service, or people with a clinical diagnosis of non-affective psychosis (schizophrenia, schizophreniform disorder, schizoaffective disorder, delusional disorder, psychotic disorder not otherwise defined), bipolar disorder, or major depression who are receiving the intervention.

Interventions: The criteria used to define supported self-management were as in a previous review of effectiveness of supported self-management interventions conducted by members of the same research group [7]. To be included in the review, the intervention was required to include all of the following three domains, which are three of four domains that Mueser and colleagues [6] described as “effective areas of self-management”:

1. Psychoeducation about mental illness and its treatment
2. Recognition of early warning signs of relapse and development of a relapse prevention plan
3. Coping skills for dealing with persistent symptoms

Comparator(s)/control: A comparator or control group was not required.

Exposures: Any primary data on factors which influence the implementation of self-management interventions for people with SMI in secondary mental health services.

Design: Full-text journal articles with primary qualitative or quantitative data (or both). If the primary focus of the study was not an investigation of the barriers or facilitators to the use of self-management interventions but the results section included data relevant to the research question, the study was included.

Screening and selection

The titles and abstracts of the articles from the search were screened by the first author (SI) to identify studies which appear to meet the inclusion criteria. A second reviewer (CHH) screened 25% (1443) of titles and abstracts independently. The second reviewer (CHH) also screened 10% (561) of the studies excluded by the first author (SI). Agreement between reviewers at title and abstract screening was not calculated.

The full text of the included studies was independently reviewed for eligibility by the first author (SI) only. The reference lists of full-text articles selected were also screened by the first author (SI). After the included studies were determined, the second reviewer (CHH) screened all of the full texts of these studies. Any disagreements were resolved through discussion with other team members.

Data extraction

A data extraction form was developed and piloted on five studies. Data extraction was conducted using EPPI-Reviewer 4 [21].

The data extraction form included details of study design, aims, gender, age, ethnicity, SMI diagnosis, study setting and context, intervention details, factors affecting implementation and implementation outcomes. To gather information related to implementation, a brief version of the Consolidated Framework for Implementation Research (CFIR) [23] was used, as well as an established taxonomy of implementation outcomes [24]. The CFIR gathered data related to factors potentially influencing implementation and the taxonomy of implementation outcomes gathered data related to factors ensuring successful implementation, in particular. These frameworks can help identify factors and aspects within implementation processes and create an evidence base around successful implementation. The two frameworks were combined to synthesise the factors which can influence implementation.

The CFIR [25] is a theoretical framework used to identify factors which influence implementation of interventions, within conceptual domains. The five major domains are: intervention characteristics (the perceived internal or external origin, evidence quality and strength, relative advantage compared to other interventions, trialability, adaptability, and complexity); outer setting (external influences such as organisational networks, peer pressure from other services and external policies and incentives); inner setting (characteristics of the implementing service including team culture, implementation climate, leadership engagement and networks and culture); characteristics of individuals, such as staff, involved in implementation (staff’s beliefs, knowledge, self-efficacy, and personal attributes which could affect implementation); and the implementation process (information on any stages of implementation, such as planning, executing, reflecting and evaluating, and any processes put in place to support implementation). The framework was adapted [23] to also include an additional domain on service user needs or resources; this domain covers the extent to which service user needs are known and feedback from service users about the intervention.

The taxonomy of implementation outcomes [24] is used to conceptualise and evaluate successful implementation. An intervention is unlikely to be effective if it is not implemented properly, so these outcomes indicate whether an intervention can be successfully put into practice and whether it is thus likely to achieve its goals in routine practice. The taxonomy includes the following concepts: acceptability, adoption, appropriateness, feasibility, fidelity, cost effectiveness, penetration and sustainability. Acceptability refers to satisfaction with the intervention. Adoption covers intention to try, uptake or utilisation of the intervention. Appropriateness refers to the perceived fit or compatibility of the intervention for the setting or service user. Feasibility is defined as the extent to which a new treatment can be successfully carried out within given setting. Fidelity is the degree to which an intervention was implemented as it was intended. Cost is defined as the cost impact of implementation. Penetration refers to the integration of an intervention within a service setting and its subsystems. Sustainability is the extent to which a newly implemented intervention is maintained within a service’s ongoing operations.

The first author (SI) independently extracted data from all included studies, and the second reviewer (CHH) independently extracted data from 20% of the included studies. Any disagreements were discussed with a third reviewer (RA).

Data synthesis

The narrative synthesis began with an analysis of study characteristics, and further results were structured around the implementation of the interventions and the influencing factors. To guide the synthesis, the CFIR [23] and the taxonomy of implementation outcomes [24] was used. The frameworks provided a table with set categories (the defined domains) to extract relevant data into, and were used for data extraction by deductively coding findings from included studies to the domains. Data within each CFIR domain, and domains from the taxonomy of implementation outcomes, was inductively coded thematically; when conducting the synthesis with the frameworks, we identified any relevant themes across studies. We used a hybrid deductive-inductive approach for this analysis method.

Study quality assessment

The quality of all included studies was assessed by the first author (SI) using the Mixed Methods Appraisal Tool (MMAT) [26], and the second reviewer (CHH) assessed 25% (6) of the included studies. The MMAT was used to inform us about the quality of information provided about implementation barriers and facilitators. The MMAT is a quality assessment tool which enables appraisal of qualitative, quantitative and mixed methods studies. The MMAT was scored by counting the fulfilled criteria. Qualitative and quantitative studies are appraised using five questions, and mixed methods studies using 15. The questions differ based on the study design and have been created to assess the methodological quality. Each study was assigned a percentage based on the score. Studies were not excluded based on the results of quality assessment.

Study selection

Searches of five electronic databases identified 8081 references, of which 23 studies met the inclusion criteria and were included in this review. No further articles were included by reviewing the reference lists of the included studies. Fig 1 shows the PRISMA diagram representing the stages of study selection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA diagram.

https://doi.org/10.1371/journal.pone.0282157.g001

Study characteristics

Of the 23 included studies, three were categorised as qualitative, six as mixed methods, nine quantitative descriptive and five as quantitative randomised controlled trials (RCTs). The studies identified were published between 2006 and 2020.

All of the studies were conducted in community mental health services or outpatients in secondary mental health care settings. This distinction is made based on what the study authors described as the type of service. The authors did not explicitly state the difference between the two, or describe what the service entailed as this was often not relevant to the study being conducted. Four studies also had a service user led/peer support element [27–30]. The characteristics of the included studies are shown in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Study characteristics.

https://doi.org/10.1371/journal.pone.0282157.t001

Six (26%) of the 23 studies were primarily investigations of self-management interventions. The remainder had other primary goals, but reported data relevant to implementation: six (26%) were feasibility and acceptability trials, five (22%) were studies mainly assessing the effectiveness of interventions, four (17%) looked at perspectives of stakeholders and two (9%) were pilot studies.

The majority (18 out of 23) of included studies were conducted in USA, with one study [8] being conducted in both Australia and the USA. The other studies took place in the UK [32,42,44], Ireland [35] or Switzerland [34].

The majority of studies had samples including people with severe mental health problems of a mixture of types (including schizophrenia or schizoaffective disorder, depression and bipolar disorder) [27,28,30–32,38,39,45,48]. Three studies had a sample consisting of schizophrenia or schizoaffective disorder, depression and bipolar disorder, along with an “other” category [29,41,47]. The remaining studies had a sample made up of only people with psychosis [43], only people with schizophrenia or schizoaffective disorder [36,37,46], only people with bipolar disorder [33,35,42,44], people with schizophrenia or schizoaffective disorder and bipolar disorder [34,40] and one study had the former population makeup along with “other” [8].

The most common intervention within the included studies was the Illness Management and Recovery (IMR) programme [28,31,37–39,45,46,48]. The goals of the IMR programme are for people with SMI to learn about their mental illness and different treatment strategies, and to develop skills to alleviate symptoms, reduce relapses and progress towards recovery [8]. Strategies to achieve these goals include learning ways to increase social support, learning skills to discuss medication issues with clinicians, creating an individual relapse prevention plan, and identifying ways to cope with distressing symptoms. Two studies used the standard IMR programme with an additional module dedicated to healthy living lifestyles [40,41], one study delivered the programme individually and in groups [8] and one study employed a peer specialist to deliver the programme [27]. Three studies used the Wellness Recovery Action Planning (WRAP) programme [29,30,32]. The WRAP programme aims to help users develop skills to maintain wellbeing in their everyday life [49]. This involved creating a daily maintenance plan, understanding triggers, identifying early warning signs, creating an action plan, and crisis and post-crisis planning.

Other interventions include psychoeducation groups for people with bipolar disorder [33,34,42,44]; the Graduated Recovery Intervention Program (GRIP) for first episode psychosis [43], a cognitive behaviour therapy (CBT)-based programme for people who have experienced psychosis; the Wellness Enhancement and Recovery Program [47], which uses the topics from IMR in a group setting; an online self-management intervention for bipolar disorder [35] and the Coping with Voices programme for people with psychosis [36], an online CBT-based intervention.

Two of the studies described an implementation plan [45,48] and one study reported on the use of implementation model [39]. The aims of most studies were either evaluating the effectiveness or implementation of an intervention, or gathering service user or staff views of an intervention. Studies evaluating the effectiveness of an intervention were included as, although the focus of the study was not on barriers and facilitators to implementation, the outcomes measured did provide data relevant to our research questions, such as views of service users and evidence on implementation.

Quality assessment

Of the 23 included studies, three were categorized as qualitative, 14 as quantitative and six as mixed methods. The quality of included primary research studies was mainly high, as 18 out of 23 studies met above 80% of quality criteria. Three studies were of moderate quality [8,33,44] and met between 60 and 75% of the criteria. Two low quality studies [37,40] met 40% of the criteria.

Barriers and facilitators to self-management interventions

Findings on the barriers and facilitators to implementation were categorised with an adapted version of the CFIR framework [23]. The key findings have been summarised below.

Implementation outcomes

Findings on the outcomes of implementation were categorised using the Implementation Outcomes Taxonomy [24]. The key findings have been summarised below.

Main findings

In this study, we conducted a systematic narrative review of the barriers and facilitators for implementing self-management interventions within secondary mental health settings. A total of 23 studies were included. To our knowledge, this review is the first of its kind to examine this topic.

The barriers and facilitators identified mainly related to organisational factors, with some being related to the individual. The narrative synthesis found some evidence to demonstrate feasibility of including self-management interventions in routine secondary mental health care, and high fidelity was a facilitator of implementation. Organisational facilitators of implementation included a strong team structure, sufficient number of staff available to support the intervention staffing, and support from colleagues. Factors directly related to the implementation process included the provision of training, supervision and the involvement of a designated implementation champion. Aspects of the intervention itself which could improve implementation include interventions being adaptable, the quality of the materials, and staff and service users perceiving the intervention content as relevant and helpful.

Barriers for implementation appeared to include organisational issues like a high staff turnover, staff shortage, lack of infrastructure for supervision and lack of support for staff delivering the intervention. A pervasive barrier was concerns over staff being expected to provide the intervention alongside their regular duties. Also, a lack of senior clinical leadership for the intervention was identified as a barrier to implementation. As mentioned previously, these barriers are related to organisational aspects, specifically readiness for implementation [25]. Readiness for implementation refers to indicators of organisational commitment to implementing an intervention and consists of leadership engagement, available resources and access to information. The identified barriers are generic ones that would potentially make it challenging to implement any new way of working.

Comparisons to existing literature

The findings of our review, focused on mental health, are similar to systematic reviews examining implementation of self-management interventions in other settings, including physical health care settings. van Beest and colleagues [17] identified facilitating factors which included a tailored intervention that could adapt to organisational changes, and the involvement of multi-disciplinary teams. Whilst involvement of multi-disciplinary teams was not found to be a facilitator in our review, we did find that the lack of involvement from multiple disciplines was a barrier [33]. Additionally, they found that gathering feedback on effectiveness was a facilitator for implementation; similarly, in our review, one study [33] found that a facilitator for implementation was the use of an organisational learning approach whereby, after receiving feedback on implementation, the authors reflected and changed their approach. Furthermore, they reported that having a feasible business model facilitated implementation, and our review also found that having sufficient staffing and organisational resources was key to implementation. van Beest and colleagues [17] also reported that facilitators for implementation included the involvement of stakeholders such as service users, local and business partners, and anticipating required changes within the healthcare system. These were areas that seemed to have been little explored in the studies we included.

Common facilitators between our review and the synthesis conducted by Harvey and colleagues [18] include service user’s knowledge of their condition, adaptability of the intervention, and efficient integration of the intervention within the service. Common barriers identified include time and resource constraints, and a lack of support from senior leadership. In contrast, our review did not identify motivational factors as a facilitator, nor the visible effectiveness of the intervention. Furthermore, we did not identify sustainable funding and resources as a facilitator as there was a lack of data surrounding this topic.

Implications

Our review makes the following recommendations after identifying several factors which could improve implementation. Firstly, developers of self-management interventions should consider adapting interventions to make them more inclusive for participants with less formal education lower education, as it was found that educational level was associated with attendance and therefore, a potential barrier in programmes which are curriculum-based. Other factors found to be associated with attendance included less mental health problem severity, and older age. This provides a basis to consider strategies for engaging service users who do not have the aforementioned attributes. In addition, the content of the intervention should be tailored to the service users and have flexibility to be personalised, since it was found that the intervention did not always fit their needs, and at times, service users were already very familiar with the content. To do this, developers could co-produce the content with potential service users; co-production can be used at each stage of the implementation process, including during the development of the implementation strategy. Alternatively, a novel self-management intervention could be created which is informed by the barriers and facilitators of implementation identified by this review. Also, interventions that require clinical skills from providers were difficult to implement for staff–this suggests that further training focused on clinical skills is required for staff facilitating the intervention.

The barriers and facilitators identified in this review have potential to form part of the basis for development of an implementation strategy for self-management interventions in secondary mental health care; this is likely to be most effective if co-produced and iteratively tested by a collaborative team including researchers, service users and clinicians. In the included studies, drop-out rates were found to range from 11% to 32%, and completion rates ranged from 46.6% to 77%; increasing these rates is a potential focus for implementation-focused interventions.

Based on the finding that several barriers are related to organisational readiness for change (ORC), service managers and stakeholders involved in implementing new self-management interventions could conduct an assessment of ORC [51] before introducing new interventions. This would allow identification of present barriers and provide implementers with the opportunity to address them. Services with less readiness for change may require more flexibility, or engagement with service leaders with the aim of addressing any remediable organisational barriers.

This review also identified current gaps in the literature which could provide focus for future research. For example, future research should explore the cost-effectiveness of implementing self-management interventions, as well as penetration and sustainability, as the studies included in this review did not contain much information on these areas of implementation.

Strengths and limitations

To our knowledge, this is the first systematic review mental health self-management interventions for SMI in secondary care. A strength of this review is the use of the CFIR and established taxonomy to analyse the results. This allowed us to translate the quantitative and qualitative evidence in a broader context for policy development and evidence-informed practice. Furthermore, the majority of the included studies were of high quality.

While the findings have implications for informing care delivery for people with SMI, there are some limitations that must be considered. Firstly, only articles published in the English language were included, therefore, some relevant literature published in other languages may have been omitted. Secondly, although exhaustive search methods were used, some studies may have been missed, especially since grey literature was not searched. However, the likelihood of publication status being a significant source of bias is low, especially in the context of this topic. We were looking at full-text journal articles with primary data, because to accurately gauge influence on implementation of self-management interventions for people with SMI, we believe a rigorous study would need to be conducted. Thirdly, all of the included studies were conducted in high-income countries, which limits the generalisability of the findings. Mental health services may be less developed in low- and middle-income countries [52], due to scarce resources for mental health. Thus implementation barriers may be greater in these countries, although supported self-management may be potentially valuable given the relatively low levels of resource and staff training that tend to be involved [53]. Due to differences in health care service delivery, future research is required to explore the factors which affect implementation of self-management interventions in low- and middle-income countries. Another limitation to consider is that only a minority of the included studies actually focused on implementation. Only three studies were explicitly based on implementation science theories and models [39,45,48]. Whilst the other studies did provide relevant data related to implementation, the richness and completeness of data is likely to be less than for research based on implementation science. A further limitation of this study is that whilst we collected data on the study setting, we did not investigate whether there is variation in secondary care settings as this is beyond the scope of the project. Typically, community mental health services will be multi-disciplinary and based in non-hospital settings [54], whereas outpatient clinics may only involve care from a psychiatrist and are typically hospital-based [55]; however, there may be considerable variation and some overlap within these categories. This should be addressed in future research. Finally, more than one reviewer did not double-screen, or extract data from all of the studies. Only one reviewer coded all of the identified barriers and facilitators, which could potentially cause bias as qualitative data is subjective [56]. An attempt to remedy this was taken by having a second reviewer also review a portion of the included studies at each step.