The database searches identified 744 articles, of which 250 were duplicated and removed. After reading the title and abstract, 42 were selected and read in full; ten papers met the inclusion criteria. Subsequently, the bibliographical references of the ten papers were analyzed, and one article was identified and included in the review. A total of 11 papers met the inclusion criteria and were taken under the systematic review. The exclusion criteria were as follows: the wrong population, a person with another type of diabetes mellitus or of pediatric age; the wrong study design, a pilot study protocol; incorrect indication, not referring to social prescribing interventions; and incorrect setting, which took place in a hospital environment (Fig. 1).

Flow diagram of the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines. From: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71. https://doi.org/10.1136/bmj.n71. For more information, visit: http://www.prisma-statement.org/

The eligible studies included two randomized controlled trials (Mercer et al. 2019; Williams et al. 2017), one nonrandomized controlled trial (Kiely et al. 2021), one mixed methods study with a matched control group (Cranes et al. 2017), three cohort studies (retrospective and prospective) (Hamlin et al. 2016; Wildman and Wildman 2021) and (Munford et al. 2020b) and four observational studies (Munford et al. 2020a; Pescheny et al. 2019; Poulos et al. 2019), and (Wakefield et al. 2022) (Table 1).

The studies were from five different countries: England (Carnes et al. 2017; Munford et al. 2020a, 2020b; Pescheny et al. 2019; Wakefield et al. 2022; Wildman and Wildman 2021), Scotland (Mercer et al. 2019), Ireland (Kiely et al. 2021), New Zealand (Hamlin et al. 2016; Williams et al. 2017), and Australia (Poulos et al. 2019). All the studies involved n = 19,202 participants with chronic disease, defined as a long-term disease in which T2DM was present (Table 1). In the cohort study by Wildman and Wildman (2021), with 8086 participants, and in the randomized controlled trial by Williams et al. (2017), with 138 participants, the focus was the person with T2DM. The mean age (SD) of the population was ± 58.5 years, but two of the papers, Munford et al. (2020b) and Poulos et al. (2019), focused on people older than 65 years. The female population was predominant, and five social prescribing programs have been developed in socioeconomically deprived areas (Carnes et al. 2017; Hamlin et al. 2016; Kiely et al. 2021; Wakefield et al. 2022; Wildman and Wildman 2021).

Different approaches to social prescribing programs have been identified. Six studies involved three stages in the development of the programs (Kiely et al. 2021; Mercer et al. 2019; Munford et al. 2020a, 2020b; Wildman and Wildman 2021; and Pescheny et al. 2019) (Table 2). Each had two types of settings, primary care and community, with the first stage involving the recognition of the person’s needs by healthcare professionals. The provider mainly identified in primary care was a general practitioner in ten studies (Carnes et al. 2017; Hamlin et al. 2016; Kiely et al. 2021; Munford et al. 2020a, 2020b; Pescheny et al. 2019; Poulos et al. 2019; Wakefield et al. 2022; Williams et al. 2017) and a practice nurse in five (Hamlin et al. 2016; Mercer et al. 2019; Poulos et al. 2019; Wakefield et al. 2022; Williams et al. 2017).

In the second stage, an articulation between primary healthcare and community organizations was referred to as performed by the linked worker (Kiely et al. 2021; Wakefield et al. 2022), also referred to as the social prescribing coordinator (Carnes et al. 2017), navigator (Pescheny et al. 2019), community health worker (Wildman and Wildman 2021), green prescriptor facilitator (Williams et al. 2017), or community-link practitioner (Mercer et al. 2019), which normally occurs in the community. The third stage involved the development of social prescribing interventions in the community. Despite other reported approaches, healthcare professionals can refer directly to the community, such as network support sports (Williams et al. 2017) or professional arts (Poulos et al. 2019). Different approaches involved an intermediate stage, between stage 1 and stage 2, performed by the health coach (Wakefield et al. 2022) or trained volunteers to assist in the delivery of the service in the community by providing additional support (Carnes et al. 2017).

The delivery methods were divided into individuals and groups and differed in the purpose of the activity in the community (Table 2). The social prescribing programs analyzed combined the two types, individually in the beginning and individually or in a group, according to community activities; only in one study were all activities in groups (Poulos et al. 2019). In terms of delivery mode, four studies reported face-to-face mode and by phone (Hamlin et al. 2016; Kiely et al. 2021; Mercer et al. 2019; Williams et al. 2017), two reported home visits (Mercer et al. 2019; Wildman and Wildman 2021), and four reported accompanying patient to community activities (Carnes et al. 2017; Kiely et al. 2021; Mercer et al. 2019; Wildman and Wildman 2021). Compared with phone mode, face-to-face mode had a small effect on HbA1c and body weight, with a greater effect (Table 3).

The interventions took place for a minimum of 6 weeks (Kiely et al. 2021) or for a maximum of 25 months (Wildman and Wildman 2021) (Table 2). Most of the studies did not detail the intervention duration or frequency. Only one study mentioned the number of sessions (Pescheny et al. 2019), and another study mentioned the number of courses, showing better results for participants who engaged in three or more courses (Poulos et al. 2019). Another three studies reported the appointment frequency with linked workers: six sessions (Carnes et al. 2017), one to five (Mercer et al. 2019), and one session face-to-face per month, with a duration of 15–60 min or four phone conversations per month (Williams et al. 2017). The studies revealed that there was a small to large effect of the linked worker intervention when three or more contacts were made during the program (Table 3).

Social prescribing intervention was developed in the community by carrying out different activities (Table 2). The most common interventions were physical activity (Carnes et al. 2017; Hamlin et al. 2016; Kiely et al. 2021; Munford et al. 2020a, 2020b; Pescheny et al. 2019; Poulos et al. 2019) (Wildman and Wildman 2021; Williams et al.