Integrating Non-Clinical Supports into Care: A Systematic Review of Social Prescribing Referral Pathways for Mental Health, Wellbeing, and Psychosocial Improvement

Search strategy

A comprehensive search strategy was developed in conjunction with a medical librarian at Macquarie University, encompassing key terms related to SP (e.g., community referral, non-medical prescribing), mental health (e.g., anxiety, depression), psychosocial health (e.g., loneliness) and wellbeing (e.g., quality of life; see Appendix 1 for the search strategy). Medline, Embase, PsycINFO and Scopus were searched on 7 August 2023 for empirical articles published between January 2010 and August 2023. We searched publications in this date range to focus on contemporary SP approaches and evaluation methods, which are rapidly advancing [1039]. The search was updated on 3 April 2024 covering the period August 2023 to April 2024. During screening, additional articles were also identified through backward citation searching (i.e., snowballing) of articles eligible for full-text review and of review articles.

Study selection

All database search results were uploaded into Rayyan [40], an online data management platform, where duplicate records were identified and removed. Six reviewers (SS, SW, LAE, MS, TS, YZ) screened the articles according to the inclusion and exclusion criteria (Table 1). Studies were eligible if they reported on SP pathways that integrated community supports (inclusive of both participant-directed and pre-structured pathways), that involved referral by a care provider, and targeted mental health, psychosocial and/or wellbeing outcomes. To improve inter-rater reliability, 10% of the abstracts were screened by all reviewers (88% agreement in raw scores). Any disagreements or uncertainties were discussed by the team until consensus was reached. The full text of articles was screened by the same six reviewers using a data extraction form developed using the REDCap electronic data capture software hosted by Macquarie University. Five articles were independently full-text-screened by all reviewers to ensure consistency and usability of the extraction form, with the remaining articles apportioned among the reviewers and separately screened.

Data extraction

Extracted data included details about the study (publication year, country, design, methods), participants (age, referral reason, symptoms), SP pathway (name, components, services, duration), referral provider and linkage process, quantitative outcomes (instruments, timepoints, sample sizes at each timepoint, type of analyses, reported direction of effect), control/comparison group (type, sample size), and qualitative outcomes (key themes and subthemes). Outcome data extracted were limited to mental health, wellbeing and psychosocial outcomes. Qualitative data were extracted using the thematic framework developed by Pescheny and colleagues [22]. This framework was chosen for data extraction and analysis because it was derived from a synthesis of studies closely aligned with the eligibility criteria of the current review.

Data synthesis

Quantitative data were narratively synthesised through textual descriptions, groupings, tabulation and frequency analysis, using the ‘Guidance on the Conduct of Narrative Synthesis in Systematic Reviews’ [41]. Characteristics of included articles were summarised textually and using descriptive statistics. Details of referral pathways, interventions, participants, and outcome measurement were summarised and grouped into common categories, and the frequency of each category was calculated to determine the range and type of groupings across the included studies.

Qualitative data were analysed using the Framework Method approach [42]. Themes and subthemes were analysed using a combined deductive-inductive approach by three reviewers (SS, MS, TS), using the Pescheny et al. framework [22]. Reviewers familiarised themselves with the extracted qualitative data and used a combination of deductive and inductive techniques to identify and code, within each category of the Pescheny et al. framework, SP outcomes and their enablers. A thematic framework was created and iteratively revised as new codes and themes were identified. The stages of the data synthesis process are displayed in Appendix 2.

Quality assessment

The Mixed-Methods Appraisal Tool (MMAT) [43] was utilised to assess the scope and quality of included studies. Two team members (SS, SW) assessed each study according to the appropriate MMAT criteria and resolved uncertainties through discussion, with a third team member available for arbitration if necessary. Discussions were focused on potential bias in study designs, analyses, and reporting, to determine whether each study adequately fulfilled the MMAT criteria. The studies that did not meet all assessment criteria were still included, as they provided valuable information about the quality of SP research.

Search results

The study selection process is summarised in the PRISMA flow chart (Figure 1). Database searches yielded 7578 articles and snowballing yielded 51 articles. From a total of 7629 identified articles, 2004 duplicates articles were removed, and 5625 articles were screened by title and abstract. Following title and abstract screening, 5468 articles were excluded because they did not meet criteria or could not be retrieved, leaving 157 articles eligible for full-text review. Of these 157 articles, 127 were subsequently excluded for failing to meet the eligibility criteria, and the 30 remaining studies were included in the current review.

Figure 1

Characteristics of included articles

Studies were conducted predominantly in the UK (n = 24, 80%), particularly England (n = 21, 70%). Two studies were conducted in Australia (7%), with the remaining four studies carried out in Sweden, Finland, Canada, and The Netherlands. Most studies (n = 18, 60%) had quantitative nonrandomised designs, which were either single-group longitudinal evaluations (n = 15) or matched comparison group evaluations (n = 3). Seven quantitative nonrandomised studies incorporated qualitative methodology (i.e., mixed methods; 23%). Nine studies (30%) were qualitative designs, and three studies (10%) were randomised controlled trials. To measure mental health-related outcomes of SP, studies primarily utilised quantitative surveys (n = 21, 70%) and interviews (n = 15, 50%); methods used less frequently were focus groups (n = 4, 13%), observations (n = 1, 3%), and patient diary entries (n = 1, 3%). The full list of included studies and their characteristics is provided in Appendix 3.

Quality of included articles

Thirteen studies scored ‘yes’ responses across the relevant MMAT criteria, indicating high quality. Articles using quantitative nonrandomised designs (solely or combined with a qualitative design, n = 18) were typically uncontrolled before and after studies examining a single group. The most commonly failed criteria for these nonrandomised designs were appropriately adjusting for confounders in study design and analysis (n = 7), and providing adequate detail on whether the intervention was administered as intended (n = 7). For example, the delivery of some interventions were impacted by COVID-19 pandemic restrictions [4445]. For the randomised controlled trial studies (n = 3), none reported that the outcome assessors were blinded to the intervention. Two studies that used mixed methods did not adequately address the divergences between qualitative and quantitative results [4647]. Qualitative studies were generally of high quality (see Appendix 4).

Target participants and reasons for referral

Most studies (n = 24, 80%) reported on SP that was targeted to adults in general with no specific age range. Four studies targeted older adults, with the minimum age of eligibility ranging from 50–65 years [48495051]. The primary reasons for referral were loneliness, social isolation, or unmet social needs (n = 14, 47%), poor mental health or wellbeing (n = 13, 43%), and the presence of chronic health conditions or risk of developing chronic conditions (n = 12, 40%). Other reasons for referral were bereavement/grief (n = 5, 17%), chronic pain (n = 5, 17%), psychosocial problems (n = 4, 13%), low self-esteem or confidence (n = 2, 7%), distraction from health-related issues (e.g., to support smoking cessation; n = 2, 7%), and cognitive issues (n = 1, 3%).

Referral and linkage pathways

Almost all studies (n = 27, 90%) reported on SP referral through primary care. Primary care referrals were mostly initiated by general practitioners (n = 16, 53%) but also by practice nurses (n = 5, 17%), mental health professionals (n = 5, 17%), assistant practitioners (n = 2, 7%) and allied health professionals (n = 2, 7%). Studies also reported on referrals from the community sector (n = 7, 23%) third or voluntary sector (n = 5, 17%), secondary care (n = 4, 13%), social care (n = 4, 13%), and tertiary care (n = 2, 7%).

The SP referral process varied among studies but was broadly categorised into two types: direct referral or referral through a link worker (Figure 2). The most common process was a direct referral from a care provider to an SP intervention (n = 21, 67%), where individuals were connected with intervention facilitators. In some of these direct referral pathways (n = 8), referral was followed by assessment and support by a link worker, mental health worker, or an expert intervention instructor (e.g., exercise professional) [4651525354555657]. Referral from a care provider to a link worker was reported in 10 studies (33%), where participants were connected with a link worker for consultation and referral to a suitable intervention. Link workers were described with various terms including coordinator, navigator, wellbeing coach, community-links practitioner, or ‘holistic’ link worker, and terms were sometimes used interchangeably. Across both pathways, the SP process often involved ongoing coaching and counselling beyond the initial referral (n = 9, 30%) [485456585960616263].

Figure 2

Intervention types

Across 30 studies, 27 SP interventions targeting mental health, wellbeing, or psychosocial improvement were identified (see Appendix 3). Most common were arts-based interventions, 11 of which followed an Arts on Prescription model [64] (n = 13; 43%). Many arts-based interventions incorporated a broad range of activities including painting, drawing, pottery, photography, museum visits, theatre, music, and dance [444749556566], and others focused on specific activities such as art-and-craft workshops [51526768] and museum visits [5068]. Two studies provided little detail regarding the specific arts activities included in the intervention [6469].

Tailored interventions were the second most common (9 interventions; n = 9, 30%) [48535758596061626370], in which a combination of activities was customised with a link worker to suit individual needs. These activities varied broadly and included arts, exercise, social groups, volunteering, and specific support groups. Intensive link worker support was typically provided as part of this tailoring process, in the form of coaching, motivational interviewing, action planning and progress monitoring [48585960616263].

Five interventions incorporated nature-based activities (n = 5, 17%), which focused on spending time in natural environments and included guided walks, learning about nature, and time for socialising [4546687172]. Two of these interventions offered exercise as an alternative or adjunct to nature-based activities [4546], and two studies described gardening-focused interventions such as creating a community garden [6872].

Four interventions incorporated exercise activities (n = 4, 13%), such as walking, climbing, group sports and group exercise classes [45465456]. Two of these interventions involved intensive coaching and goal setting by a health or exercise professional [5456]. Two interventions also offered nature-based activities as an alternative or adjunct to exercise [4546].

Quantitative evaluations

Qualitative outcomes

Qualitative outcomes of SP were reported in 16 studies (53%). Sample sizes ranged from six to 52 individuals, who were mostly interviewed immediately after completing the intervention, and only one study collected pre and post qualitative data [58]. Across 16 studies, 57 themes and eight subthemes were identified as relevant to the aims of this review, and SP outcomes and their enablers were identified within each category of the Pescheny et al. framework (see Appendix 6 for tabulated qualitative results).

Quantitative outcomes

Consistent with the findings of previous reviews, this review revealed a great deal of heterogeneity in quantitative outcomes measured and tools used for measurement [182237].

The most consistently assessed construct was mental wellbeing, measured using the WEMWBS, aligning with the findings of prior reviews [737]. In eight studies, mental wellbeing improved in SP participants over time, but these studies were uncontrolled, had widely varying and mostly limited sample sizes with minimal a priori power calculations, and measurement timeframes were narrow (as little as six weeks) [6771]. Similarly, studies assessing anxiety, depression, and health-related QoL (which were the three other most consistently measured constructs) showed some improvement on these constructs over time, but these positive effects were not consistently demonstrated in between-groups evaluations [4454], an issue identified in previous research [10]. The remaining 22 constructs were assessed in either one or two studies, most of which were uncontrolled (67%) and had inadequately powered sample sizes. These methodological limitations challenge the generalisability and reliability of the observed effects, which might have occurred for other reasons that were not accounted for (e.g., social desirability, lower scores at baseline) [101736]. In the absence of longer-term follow-up, inferences cannot be made about the clinical meaningfulness of the findings.

Quantitative measurement of SP outcomes needs to be given a greater deal of thought by researchers. Individuals referred to SP often have complex problems related to the wider social determinants of health, and choice of measurement instruments must take this into account. Quantitative measurement of discrete constructs may not adequately capture change as individuals continue to face multifaceted life challenges [10]. While a wide range of quantitative measures could contribute to the evidence base for SP effectiveness, researchers should focus more on the intended effects of interventions and select appropriate measures that are responsive to change in the targeted populations [37]. There have been calls for standardised outcome measures aligned with the core aims of SP to build capacity for comprehensive evidence synthesis and informed decision-making across care contexts and in policy [821]. Frameworks for assessing integrated care performance and outcomes could be leveraged to develop an SP-specific assessment framework that incorporates key domains and indicators, with consumers as partners to ensure local relevance [73].

Qualitative outcomes

Qualitative outcomes of mental health, wellbeing, or psychosocial improvement both reflected and differed from the observed quantitative outcomes captured in the included studies, and enabled greater understanding of how SP impacts mental and psychosocial health and wellbeing [22]. The most commonly reported qualitative outcome across studies was an increased sense of belonging and reduced loneliness, enabled by social interactions. This aligns with prior research which demonstrated that positive impacts of SP were largely driven by socialisation, and conceptualised positive social outcomes as uniquely different from health and wellbeing improvements [28]. Other research has conceptualised positive social outcomes as a component of improved health and wellbeing [74]. Our findings align with the former; different precursors and influencing factors were associated with belonging (e.g., feeling safe and supported [4672]) than with wellbeing (e.g., improvements in mood and anxiety [4772]). It is well recognised that the complex, multifaceted nature of wellbeing can make it difficult to comprehensively measure quantitatively [7576], and qualitative studies can enhance understanding of the drivers of wellbeing, particularly the role of belonging and connectedness [77].

Other common qualitative outcomes reported were increased purpose and meaning, and increased confidence. Purpose and meaning was related to the self-concepts and feelings theme, and was enabled by feeling useful, motivated and accomplished. Increased confidence was both an outcome in its own right, enabled by various factors, and an enabler of outcomes. Although many SP approaches are designed to enhance empowerment and support meaningful, purposeful engagement [639], the quantitative instruments used for measurement appear lacking in their ability to capture these nuanced processes and outcomes [22]. The frequent measurement of wellbeing might be driven by the belief that improvements in wellbeing will then foster other outcomes, including improved confidence, self-esteem, and autonomy [39], but the directionality of this relationship can be called into question. Greater attention should be paid to how evaluations of SP for mental health can capture these nuanced outcomes. Rigorous qualitative methodologies with well-developed conceptual frameworks should be utilised in qualitative studies and as an adjunct to quantitative studies, to ensure that contextual factors influencing outcomes are identified and considered when interpreting results. Realist evaluations can be valuable for highlighting contextual factors that determine intervention success across different populations and settings [1629].

Strengths and limitations

This systematic review comprehensively synthesised and evaluated quantitative and qualitative research on SP referral pathways targeting mental health, psychosocial needs and wellbeing, canvassing the range of outcomes, measures, and evaluative approaches reported. The inclusion of qualitative research enabled a greater understanding of the varying person-centred integration approaches offered by SP, and the impacts these can have on mental health, and on social, emotional, and environmental needs that can influence mental health. Another strength of this review was the focus on research reporting outcomes of SP referral pathways, from the point of referral to intervention outcome measurement, rather than the outcome of SP interventions alone. By examining studies of SP referred by care providers, we can better understand the impact of integrating SP within healthcare systems.

There are limitations of this review worth noting. First, the design and conduct of this review did not involve people with lived experience accessing SP pathways. Nevertheless, the review provides a basis for future engagement with people with lived experience, enabling crucial insights into the measurement of outcomes during the co-design, implementation and evaluation of SP pathways targeting mental health. Second, it was outside the scope of this review to examine implementation evaluation, but we believe it to be the next key step for progressing research on SP as an integrated model of care for mental and psychosocial health. For example, it is essential to synthesise the research on the capacities of link workers and other interdisciplinary professionals required to enable effective SP referral pathways for individuals with mental health, psychosocial, and wellbeing needs [15]. Third, although our search strategy was comprehensively designed, we may have missed relevant studies reporting on similar models of care that leverage community resources to support individuals with mental health problems or psychosocial needs impacting their mental health [4]. Fourth, we did not search grey literature, which may contain relevant information about SP referral pathways that has not been indexed in academic databases. As a result, we may have missed other outcomes relevant to the aims of the current review. We did, however, include a broad range of evaluations (including uncontrolled studies) to ensure an accurate view of how SP is enacted in practice, but additional research that broadens these findings will be important, particularly in such a rapidly evolving field.

Conclusion

Approaches to integrating non-clinical supports into care for people with mental health, wellbeing and psychosocial needs are increasingly being developed and evaluated. High heterogeneity in how SP referral pathways for mental health improvement are being evaluated limits conclusive inferences about their effectiveness. The development of SP referral pathways must be guided by specific goals and be accompanied by the design of rigorous mixed methods evaluation strategies. A standardised set of quantitative outcome measures alongside rigorous qualitative and implementation evaluations should be utilised to progress the evidence base on SP as a model of integrated care for holistic mental health improvement.