Effectiveness of Social Prescribing in the management of people living with physical noncommunicable diseases (NCDs): a rapid review protocol on health and cost-related outcomes

The global challenge of Non-Communicable Diseases

Non-communicable diseases (NCDs), also known as chronic or long-term diseases, were responsible for approximately 45 million deaths worldwide in 2023 and more than 767 million years lived with disability (YLD)¹, representing a major global challenge and placing high demands on health systems worldwide².

NCDs such as diabetes, cancer, musculoskeletal diseases, cardiovascular diseases and chronic respiratory diseases result from a complex interaction of multiple risk factors, including metabolic (e.g. high blood pressure, high body mass index (BMI), high cholesterol levels), behavioural (e.g. tobacco and alcohol use, dietary risks and low physical activity) and environmental factors (e.g. air pollution)³. Most NCDs do not have a cure, requiring long-term pharmacological and/or non-pharmacological management, and sustained self-management, which requires health behaviour changes, including diet and physical activity².

In the European Union, diabetes, cardiovascular diseases, cancer and chronic respiratory diseases represent at least 25% of total health care expenditure and a 2% loss in Gross Domestic Product (GPD)⁴. Musculoskeletal diseases and diabetes alone may account for up to 1.4% and 1.2% of GDP losses worldwide^5,6. Indirect costs due to work absenteeism and loss of work productivity, premature death and early retirement may be comparable to direct costs, impacting households and economies globally^4,7.

Social determinants such as education, income, access to healthcare, gender, and ethnicity significantly influence the risk and outcomes of people living with NCDs. These factors contribute to a social gradient in health, where people from disadvantaged socioeconomic backgrounds have reduced access to high-quality healthcare, greater exposure to risk factors, and poorer health behaviours and outcomes⁸.

While low middle- and middle-income countries, which are undergoing a rapid transition from communicable diseases to NCDs, struggle to deliver effective interventions, high income countries face the burden of NCDs driven by their ageing population and age-specific rates for Years Lived with Disability (YLDs), causing high demands on national health systems⁹. This epidemiological transition has shifted healthcare demands from acute to chronic care, requiring systems to manage both clinical complexity and social needs while ensuring sustainability¹⁰. Strengthening NCD prevention across all levels of care and developing innovative models of care delivery have become major priorities for health systems worldwide¹¹. Integrating community services with healthcare through Social Prescribing (SP) has been described as a promising strategy to address complex needs of people with NCDs, including social determinants of health, leading to improvement in health and well-being outcomes¹².

Social Prescribing as a response to complex health and social needs

SP involves connecting people to activities delivered in the community and voluntary sector that may include activities to support mental health, social inclusion, physical activity, and access to practical resources such as financial or housing advice. There is no universally accepted definition of SP. Several definitions and different models of SP have emerged over the years, across and within countries. In 2023, Muhl et al. constituted a panel with 48 experts, including members of several organizations of SP across the globe (e.g. Social Prescribing Network, Global Social Prescribing Alliance, Canadian Institute for Social Prescribing) and stakeholders (e.g. patients, link workers, healthcare providers), for a Delphi study to reach consensus on a definition of SP (>80% agreement)¹³. In this study SP was conceptually defined as “a means for trusted individuals in clinical and community settings to identify that a person has non-medical, health-related social needs and to subsequently connect them to non-clinical supports and services within the community by co-producing a social prescription—a non-medical prescription, to improve health and well-being and to strengthen community connections”. To avoid ambiguity, this review will use the term “models of SP” to refer to the referral pathways that connect people with NCDs to community-based services. The term “SP activities” will be used to refer to the specific community-based interventions or supports provided by community services in which individuals enrolled in SP programmes participate.

Typically, SP models involve the referral of patients from primary care providers to dedicated social prescriber link workers, who identify health and social needs and connect individuals with appropriate community resources. “Using a definition aligned with that proposed by Muhl et al. (2023), the recent scoping review by Wilson et al. (2025)¹³ identified 37 SP studies involving people with NCDs. These authors concluded that different models of SP have been implemented worldwide, the most common being: signposting (health professionals inform patients about activities), direct referral (health professionals formally refer patients to an activity), link worker (health professionals refer patients to a link worker, who connects individuals referred with one or more activities), and holistic (health professionals refer patients to a link worker hub, which provides follow-up and ongoing support). Additionally, these authors also found that SP referrals were originated not only from primary care but also from community and secondary care, showing that SP is implemented across different levels of the health system, addressing the complex needs of people living with NCDs across healthcare settings¹⁴

SP aims to address the wider social determinants of health and may promote more integrated and comprehensive management of people living with NCDs¹⁵. Over the past decade, SP has been implemented in more than 30 countries and is increasingly recognised as a promising strategy to improve both clinical and non-clinical outcomes relevant to people with NCDs, including health-related quality of life (HRQoL), loneliness, physical activity and mental health¹⁴. The World Health Organization has also published guidance to support the global implementation of SP¹⁵.

Current evidence, evidence gaps and rationale for this rapid review

As SP continues to expand and health systems struggle to implement long-term strategies for NCD management, it is essential to assess the effectiveness and economic implications of SP. Understanding whether SP can be optimised, sustainably delivered, and integrated as a viable approach for continuity of care and chronic disease management is essential. Current evidence on the effectiveness of SP for improving health-related outcomes in people with NCDs remains inconsistent^14,16–18, and there are no synthesised data on its cost-related outcomes in people with NCDs.

The systematic review by Htun et al. (2023) analysed the effectiveness of SP in people attending primary and community care (regardless of health condition) on risk factors for chronic diseases. These authors found that all included studies (n= 7 Randomised Controlled Trials (RCTs)) focused on exercise-based programmes. The meta-analysis showed that SP is likely to slightly improve physical activity and energy expenditure¹⁶. Given that most people with NCDs are physically inactive¹⁹, and that social support is one of the key determinants of adherence to physical activity programs in people living with NCDs²⁰, these findings reinforce the potential of SP to address behavioural risk factors and support self-management.

The systematic review by Kiely et al. (2022) assessed the effectiveness of link worker–based SP in adults with multimorbidity attending primary care, regardless of diagnosis. They included eight RCTs, and concluded that SP may have little to no effect on HRQoL, physical function, social connectedness, mental health, or self-rated health, but evidence was limited. Among people with multimorbidity, SP was associated with improved perception of primary care quality and reduced hospitalizations. No cost-effectiveness studies were identified¹⁷. The systematic review by O’Sullivan et al. (2024) focused on the effectiveness of link worker-based SP for people with long-term conditions and included 12 RCTs or quasi-experimental studies. These authors found that SP interventions improved HRQoL, but reported mixed evidence for physical activity and psychological wellbeing outcomes, and non-significant pooled effects on HbA1c in diabetes. However, their search strategy did not include specific disease terms (e.g., asthma, musculoskeletal disorders, diabetes), which may have excluded relevant studies. Economic outcomes were not evaluated in this review¹⁸. Both reviews focused exclusively on SP models involving link workers, thereby excluding other referral-based SP models previously mentioned in this paper¹⁴.

Moreover, the economic impact of SP remains poorly understood. The Social Prescribing Academy has recently published on grey literature findings from a rapid scoping review that aimed to map the existing evidence on the economic impact of SP on the general population. Among the 19 studies included, 13 addressed Social Return On Investment (SROI), seven used cost description analysis, two employed cost benefit analysis, one used regression modelling and one conducted cost-effectiveness analysis. The authors concluded that SP may have a positive economic impact, particularly among high users of primary care, people from ethnic minority groups, people with only one NCD and older adults²¹. Moreover, a mapping review by Sonke et al. (2023) reported that financial or economic outcomes of SP were commonly assessed in studies from 13 countries²². Taken together, the existing literature shows emerging evidence on the economic impact of SP. However, no systematic review has yet specifically addressed health economic outcomes in people with NCDs.

As SP is expanding globally²³, there is an urgent need to synthesise evidence to optimise SP models for people with NCDs, and to inform policy on their effectiveness, cost-effectiveness, and return on investment. Given the gap in existing evidence, we aim to conduct a rapid systematic review of the evidence on the effectiveness of SP on clinical and cost-related outcomes in people living with physical health-related NCDs. While physical and mental health are closely linked and share behavioural risk factors such as unhealthy lifestyles, mental illnesses (e.g., depression, mood disorders, substance use disorders) often require distinct referral pathways, outcome measures, and approaches to address issues like stigma, discrimination, and suicide prevention. To enhance homogeneity, this review will focus on physical health NCDs as index conditions, while not excluding studies that report mental health comorbidities within these populations²⁴.

Furthermore, this rapid systematic review will extend existing literature SP for people with NCDs by adopting the broader definition of SP by Muhl et al (2023), previously presented, and capturing different models of SP implemented worldwide¹³.

Eligibility criteria

The eligibility criteria are described according to PICO (P-Population, I-Intervention, C-Comparison, O-Outcome).

Population. We will include studies involving community-dwelling adults (>18 years old), in which at least 50% of study participants present with one or more physical NCDs, including but not limited to diabetes, cardiovascular diseases, chronic respiratory diseases (e.g. asthma, chronic respiratory disease (COPD), chronic neurological diseases (e.g. Parkinson’s disease, stroke or multiple sclerosis) chronic pain (e.g. rheumatic and musculoskeletal diseases), even if participants also have mental health comorbidities. As previously stated, studies primarily targeting mental illnesses (e.g., depression, mood disorders, substance use disorders) as index conditions will be excluded.

Intervention. We will include studies that analyse the effects of SP as defined by Muhl et al. (2023): “a means for trusted individuals in clinical and community settings to identify that a person has non-medical, health-related social needs and to subsequently connect them to non-clinical supports and services within the community by co-producing a social prescription—a non-medical prescription, to improve health and well-being and to strengthen community connections”. Accordingly, eligible studies must include participants referred to community-based activities or support services to address social determinants, or other non-medical factors, that influence health and wellbeing, through health professionals or link workers/social “connector”, considering the models of SP referral described previously by Wilson et al. (2025). We will consider SP models integrated in primary, community or secondary care. Since SP may be referred in the literature as community referral, social referral, art prescription, nature prescription or exercise referral schemes, we will include these terms in our search strategy¹⁴.

Studies will be excluded if the SP activities were delivered directly by healthcare professionals, took place in non-community settings such as healthcare units or workplaces, or consisted only of general health advice, education, messaging, or motivational interviewing without referral to community-based services.

Comparison. We will include studies with control groups receiving non-social prescribing interventions, standard care, waitlist, or no intervention.

Outcomes. The primary outcomes of this rapid review are HRQoL, physical function, and physical activity. Secondary outcomes will be considered if measured through valid instruments or measures, across three domains: 1) clinical and wellbeing outcomes, including both physical and mental health outcomes, disease-specific outcomes; 2) social outcomes, including loneliness, social connectedness; and 3) economic and health service utilisation outcomes, which include health service use (e.g., number of general practitioner visits, emergency department attendances, hospital admissions), direct and indirect costs, cost-utility, cost-effectiveness, SROI.

Types of study design

We will include quantitative studies with a comparison group, such as RCTs, quasi-experimental studies, cohort studies, and case-control studies. For economic evaluation studies, we will consider all study designs commonly used in economic evaluations, including cost-effectiveness, cost-utility, cost-benefit, cost-description, as well as SROI analyses, regardless of whether a comparison group is present. SROI studies have been used to capture wider social and economic benefits of a program, in a singular monetary ratio, providing an indication of the efficiency of an investment by comparing the value of its benefit to the value of the resources invested, although it relies on subjective assumptions aiming to an oversimplification of data.

Search strategy

Searches will be conducted in MEDLINE (via PubMed), EMBASE, CENTRAL, and CINAHL databases. The search strategy will be developed using the PICO terms with support from an experienced librarian and peer-reviewed by an independent librarian following the Peer Review of Electronic Search Strategies (PRESS) checklist²⁸. We undertook a preliminary search with the PICO terms “social prescribing” and “noncommunicable diseases,” along with synonyms and related terms, adding the most prevalent physical health NCDs – Supplemental Material. This search strategy is openly available (see Data Availability). To ensure sensitivity, we will not restrict searches by outcomes or study design. To streamline the strategy, we will include studies published in the last 10 years (2015–2025), to filter SP models aligned with current knowledge and recommendations; in English, and in peer-reviewed journals. Reference lists of relevant studies and reviews will be screened through forward and backward citation searching.

Study screening

All references will be uploaded to Covidence systematic review management software, where duplicates will be removed. A standardised form for inclusion and exclusion criteria will be developed, agreed by the research team, and piloted before screening begins. We will select eight records to pilot screening and refine eligibility criteria, including those that are definitely ineligible, and doubtful. For both title/abstract and full-text screening, two reviewers (DAC and HPF) will screen all records. Discrepancies will be solved by consensus, or with the discussion of a third reviewer (BK).

Data extraction

Data extraction will be managed in Covidence using a data collection form, that will be developed according to review objectives. This will be agreed upon by the review team and piloted prior to use. One reviewer (DAC) will extract all data, and a second reviewer (HPF) will verify critical data that may affect the results or conclusions to ensure accuracy and completeness (e.g. outcomes of each study). Extracted data will include study characteristics (authors, year, country, design, baseline measures, duration, timing of data collection, and setting), participant characteristics (sample size, age, gender distribution, inclusion and exclusion criteria, social characteristics of the population if reported – e.g. social disadvantage), and SP characteristics (sources of referral, model of SP referral, professionals involved during the pathway, activities delivered, duration, follow-up, comparator or control group, and number of participants per group). Study results will also be extracted, covering all pre-specified outcomes, the measurement instruments used, units of measurement, timing of assessment, and effect and uncertainty measures (e.g. mean and standard deviations for continuous data, risk ratios and 95% CI for dichotomous outcomes). SP activities characteristics will be reported following the Template for Intervention Description and Replication (TIDieR) checklist²⁹, to enhance comparison between studies. Information on SP referral and SP activities integrity (e.g., adherence and fidelity) will be collected, if reported. In case of multiple records from the same study, data will first be extracted separately for each record and then combined during analysis to ensure completeness of information and avoid duplication.

Risk of bias

We will assess risk of bias using Cochrane RoB-2 for RCTs³⁰, ROBINS-I for non-randomised studies, including observational studies³¹, and ECOBIAS for economic evaluations studies³². One reviewer (DAC) will perform the RoB assessments and a second reviewer (HPF) will verify critical judgements.

Data synthesis

We will perform a narrative synthesis and present in a tabular form, presenting the following study characteristics: study design, SP model, SP activity/activities description, comparison, sample size, follow-up times, outcome measures used and effects. A separate table will synthesize studies that included economic outcomes.

If studies are sufficiently comparable, we will conduct a random-effects meta-analysis for the primary outcomes, as between-study variability is expected. For dichotomous outcomes we will calculate risk ratios, odds ratios, or risk differences. For continuous outcomes we will calculate mean differences when the same measurement tool is used, or standardised mean differences when different instruments are applied. All effect estimates will be reported with 95% confidence intervals. Outcomes reported at multiple timepoints will be grouped a priori (e.g., short-term <6 months vs long-term ≥6 months).

Statistical heterogeneity will be assessed using the χ² test (p<0.1) and quantified with I², interpreted according to Cochrane thresholds (0–40%: might not be important; 30–60%: moderate; 50–90%: substantial; 75–100%: considerable). If there is high heterogeneity (I² = 80%) or findings are inconsistent, we will not pool results. For outcomes not possible to pool, we will present a narrative synthesis instead. For continuous outcomes, study weights will be calculated using reported standard deviations or derived from confidence intervals or standard errors where required. Where outcome data are highly skewed (e.g., costs or service use), log transformation will be applied prior to analysis to improve validity of pooled estimates³³. Sensitivity analyses (where feasible) will exclude studies at high risk of bias to test robustness. Subgroup analyses will be conducted if data permit, by disease condition (e.g. diabetes, arthritis), SP model (e.g. link worker based, direct referral) and by SP activity (e.g., physical activity). Meta-regression is not planned.

Secondary outcomes will not be quantitatively synthesised. These and all other outcomes for which pooling is not feasible will be summarised narratively. All quantitative analyses will be conducted in Stata 19. RevMan Web will be used to generate forest plots for presentation, if applicable.

Certainty of evidence

We will assess the certainty of evidence using the Grading of Recommendations Assessment, Development and Evaluation (GRADE), and all assessments will be managed and documented using GRADEpro software, for each of the primary outcomes separately. Certainty of evidence will be judged across the five GRADE domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. One reviewer (DC) will perform the assessments, and a second reviewer (HPF) will validate the judgements. Certainty of evidence will be rated as high, moderate, low, or very low. If meta-analysis is not feasible, we will apply GRADE narratively by summarizing the direction and magnitude of effects across studies and judging the overall certainty of the body of evidence across the five domains, in line with Cochrane recommendations.

Knowledge user involvement

In accordance with Cochrane Rapid Reviews guidance²⁶, a knowledge user (BK), a general practitioner and researcher experienced in social prescribing, is a co-author of this rapid review and contributed to developing and refining the review question, eligibility criteria, and outcomes, and will assist in interpreting results from a knowledge user perspective to ensure clinical applicability.