Methods of connecting primary care patients with community-based physical activity opportunities: A realist scoping review

Deemed a global public health problem by the World Health Organization, physical inactivity is estimated to be responsible for one in six deaths in the United Kingdom (UK) and to cost the nation's economy £7.4 billion per year. A response to the problem receiving increasing attention is connecting primary care patients with community-based physical activity opportunities. We aimed to explore what is known about the effectiveness of different methods of connecting primary care patients with commu-nity-based physical activity opportunities in the United Kingdom by answering three research questions: 1) What methods of connection from primary care to community-based physical activity opportunities have been evaluated?; 2) What processes of physical activity promotion incorporating such methods of connection are (or are not) effective or acceptable, for whom, to what extent and under what circumstances; 3) How and why are (or are not) those processes effective or acceptable? We conducted a realist scoping review in which we searched Cochrane, Medline, PsycNET, Google Advanced Search, National Health Service (NHS) three refuted for eight Comparisons of the processes and theories of several helpful for future Our base in for well-designed theory-based


| Background
Lack of (physical) activity destroys the good condition of every human being. (Plato) Deemed a global public health problem by the World Health Organization (WHO) (World Health Organisation, 2020), physical inactivity is a leading risk factor for noncommunicable diseases (World Health Organisation, 2018). Levels of physical inactivity in the United Kingdom (UK) are among the highest in the world (Savill et al., 2015) with 40% of adults (aged 18+ years not reaching the WHO-recommended physical activity levels) (World Health Organisation, n.d.). Physical inactivity is estimated to be responsible for one in six deaths in the United Kingdom and a £7.4 billion cost to the nation's annual economy (Public Health England, 2019). In addition to protecting against noncommunicable disease morbidity and premature mortality, and reducing financial burden, physical activity holds a host of other benefits, including reduced anxiety, depression and stress and increased mood, self-esteem, sleep quality, cognitive function and energy levels (Rhodes et al., 2017;White et al., 2017). Physical activity promotion is therefore a priority for public health improvement (Lion et al., 2019;Sparling et al., 2000;World Health Organisation, 2014).
Primary care professionals (PCP) are viewed as instrumental in physical activity promotion (Peckham et al., 2011). In the United Kingdom, the National Institute for Health and Care Excellence (NICE) recommends that such professionals promote physical activity (NICE, 2013). Health promotion is also included in the Royal College of General Practitioners (2019a) Curriculum, and undergraduate medical curriculums are increasingly including education in the importance, and promotion, of physical activity (Gates et al., 2019;Milton et al., 2020). PCP are well-placed to help address the problem of physical inactivity (Douglas et al., 2006;Williams, 2011) (Rowley et al., 2018)-remain popular (Morgan et al., 2016;Sidford, 2006), and other methods of connecting primary care patients with community-based physical activity opportunities have been introduced. These include signposting by PCP (Bull & Milton, 2010) and connection by PCP to an intermediary (e.g. link worker or physical activity professional) who then connects patients with physical activity opportunities (Pescheny et al., 2019)-increasingly referred to in the United Kingdom as 'social prescribing' (Husk et al., 2020).

| Review objectives
The aim of our review was to explore what is known about the effectiveness of different methods of connecting primary care patients with community-based physical activity opportunities in the United Kingdom. It was conducted as part of a larger project aiming to design and evaluate methods of connecting primary care patients to jogscotland-an established community-based physical activity opportunity (https://jogsc otland.org.uk).
It is important to note a fundamental complexity in conducting the review: for a method of connection to be effective, that is for it to achieve a high percentage of eligible and willing patients taking up a physical activity opportunity, two sets of actors are required to undertake particular behaviours-(a) PCP must connect eligible and willing patients with the physical activity opportunity, (b) connected patients must take up the opportunity, that is enrol for and attend the first session. Both these behaviours were therefore of interest in our review. It is also important to note the inclusion of acceptability as an outcome in the review. Acceptability refers to how well an intervention-in this case a method of connection-is received by the target What is known about this topic population and the extent to which it meets the needs of that population and the organisational settings (Ayala & Elder, 2011) and is a necessary condition for effective interventions (Sekhon et al., 2017).
Where no effectiveness evidence is available, acceptability evidence can provide insight into an intervention's potential effectiveness, and where evidence regarding both effectiveness and acceptability is available, acceptability can help to explain effectiveness. We therefore had four outcomes in our review: (a) effectiveness for PCP; (b) acceptability for PCP; (c) effectiveness for patients; (d) acceptability for patients. For the purposes of this review we employed the following outcome indicators: PCP connection rates (the percentage of eligible and willing patients connected with a physical activity opportunity) for effectiveness for PCP; PCP receptiveness and views regarding whether or not personal and organisational-setting needs were met for acceptability for PCP; patient uptake rates (the percentage of connected patients enrolling for and attending the first session of the physical activity opportunity) for effectiveness for patients; patient receptiveness and views regarding whether or not personal needs were met for acceptability for patients.
The development of the specific research questions was initially influenced by two factors: (a) our aim to evaluate all methods of connection from primary care to community-based physical activity opportunities; (b) the value of understanding how and why interventions succeed or fail in different contexts (Craig et al., 2018). A third factor became apparent during data extraction: as each method of connection occurred as part of a multi-stage process of physical activity promotion, evidence was not available regarding the effectiveness or acceptability of methods of connection per se, but rather regarding the processes of physical activity promotion as a whole.
These processes included the identification of eligible and willing patients who would benefit from increasing their physical activity levels and the delivery of behaviour change strategies aiming to enhance the likelihood of those patients increasing their physical activity levels, as well as connection of patients with community-based physical activity opportunities. We therefore refined questions 2 and 3 to broaden their focus from 'methods of connection' to 'processes of physical activity promotion incorporating methods of connection'. We maintained a focus on methods of connection as much as possible as these are not married to the processes of physical activity promotion in which they are embedded. It is therefore useful to know as much as possible about their effectiveness and/or acceptability in order to inform development of future physical activity promotion processes incorporating connection from primary care to community-based physical activity opportunities. The final research questions were: 1. What methods of connection from primary care to community-based physical activity opportunities have been evaluated? 2. What processes of physical activity promotion incorporating such methods of connection are (or are not) effective or acceptable, for whom, to what extent and under what circumstances?
3. How and why are (or are not) those processes effective or acceptable? 2 | ME THODS

| Study design
We undertook a realist scoping review which allowed a complementary combination of the broad focus of a scoping review (Arksey & O'Malley, 2005) and the rich data synthesis of a realist review (Pawson, 2002;. This enabled us to answer our research questions by: (a) identifying and mapping the extent, range and nature of the evidence (Arksey & O'Malley, 2005); (b) exploring the relationships between context, mechanisms and outcomes for each intervention or class of intervention through establishing initial programme theories and testing each component of those theories using relevant empirical evidence to enable refinement of programme theories (Pawson, 2002;. Despite increasing use of this type of evidence synthesis over the last decade (Brydges et al., 2017;Haynes et al., 2018;Kirst et al., 2012;Toohey & Rock, 2011), we could not identify any methodological guidance regarding realist scoping reviews. We therefore followed both scoping (Arksey & O'Malley, 2005;Levac et al., 2010;Peters et al., 2015) and realist review guidance (Pawson et al., 2005;, making modifications where necessary to combine the two. This enabled us to develop methodological guidance concerning realist scoping reviews (manuscript in preparation). In line with realist review guidance (Pawson et al., 2005) we did not pre-publish our review protocol. To report our findings, we follow the Enhancing the QUAlity and Transparency Of health Research (EQUATOR) network guidelines for reporting both scoping (Tricco et al., 2018) and realist  reviews.

| Search strategy
We developed our search strategy in an iterative manner, holding multiple discussions among the research team and testing several potential search terms such as 'social prescribing', 'signposting', 'gym' before finalising the strategy. All types of research design had potential to contribute to the answering of our research questions, and both academic and grey literature was relevant. We did not set date limits on any searches, therefore all databases were searched from inception until October 2018, with an updated search conducted in August 2020. See Appendix S1 for final search strategy.

| Eligibility criteria
We considered documents for inclusion if they: • Provided details of one or more methods of connection from primary care to community-based physical activity opportunities; • Reported an evaluation (quantitative and/or qualitative), or provided data enabling an evaluation, of one or more methods of connection from primary care to community-based physical activity opportunities; • Reported evaluations undertaken in the United Kingdom; • Reported evaluations fully or mainly undertaken in an adult (18+ years old) patient population; • Reported evaluations in which the connecting health professionals were fully or mainly PCP; • Were written in English.
We excluded documents if they: • Focussed on connection to condition-specific physical activity opportunities.

| Document selection
Two authors (RHR, KBC) independently determined eligibility of documents by applying the above eligibility criteria in a two-stage process: (a) screening of titles, abstracts, summaries, lists of contents; (b) screening of full texts. This process was assisted by the use of Covidence systematic review software (www.covid ence.org). Any discrepancies were resolved by discussion, with a third author (SAC) consulted when necessary.

| Document appraisal
In accordance with guidance on conducting realist reviews (Pawson et al., 2005; two authors (RHR, KBC) independently appraised the relevance and rigour of the included documents. Relevance was determined by the number of the four outcomes of interest (PCP effectiveness, PCP acceptability, patient effectiveness and patient acceptability) addressed in the document and was classified as 'high', 'medium' or 'low' (See Appendix S2 for details of classifications). Rigour of documents reporting outcomes concerning acceptability was appraised using the five-item, methodologically eclectic 'rigour' tool of Dixon-Woods et al. (2006) and was classified as 'high', 'medium' or 'low' (See Appendix S2 for details of classifications). Such a tool was not appropriate for appraising rigour of documents addressing effectiveness outcomes as none of those documents reported findings regarding effectiveness. Rather they provided data that enabled us to undertake an evaluation of effectiveness. Appraisal of the methods and findings was therefore not useful-relevant appraisal of rigour related instead to the completeness of the data required to evaluate effectiveness. This was appraised using the question 'Were the data necessary to calculate effectiveness provided?' and was classified as 'high', 'medium' or 'low' (See Appendix S2 for details of classifications). Any discrepancies were resolved by discussion, with the option to consult a third author (SAC) if necessary. As per guidance concerning scoping reviews (Arksey & O'Malley, 2005;Levac et al., 2010) we took the decision not to exclude any documents based on their rigour. We rather took their relevance and rigour into account in the data synthesis as advocated by Dixon-Woods et al. (2006).

| Data extraction
Data extraction involved identifying key components of the methods of connection and the physical activity promotion processes in which they were embedded, along with the expected impacts/ outcomes pertinent to the review. This required us to infer some elements of the processes based on reported information. Data extraction also involved eliciting theories of change (sequences of events leading to a desired outcome, together with underlying assumptions about mechanisms, where mechanisms refers to how and why the sequence might generate that outcome; Vogel, 2012). We inferred the underlying assumptions about how and why the processes of physical activity promotion incorporating methods of connection might generate the outcomes of interest in the review, as the included evaluations did not explicate these assumptions. To do so we employed the Capability, Opportunity, Motivation -Behaviour (COM-B) model (Michie, van Stralen, et al., 2011)-a framework for understanding behaviour/behaviour change. In this model behaviour is viewed as the result of individual decision-making, and as part of an interacting system involving the three components of capability to perform a behaviour (determined by psychological and physical capacity/incapacity), opportunity to perform a behaviour (determined by enabling/disabling external social and physical factors) and motivation to perform a behaviour (determined by energising/discouraging automatic and reflective mental processes). Data extraction also involved establishing evidence regarding effectiveness and acceptability. We calculated PCP connection rates and patient uptake rates using the data provided-effectively conducting primary evaluations-as these were not a focus of the included evaluations.
Establishing acceptability required us to identify relevant findings employing different terms, as the included evaluations did not use the term 'acceptability'. Finally, data extraction involved identifying relevant information regarding context and mechanisms. For the purpose of our review 'context' included: (a) the actors, that is, the type of PCP and patients and their characteristics-specifically gender, age and physical activity level; (b) the circumstances, including the workload associated with the processes of physical activity promotion incorporating methods of connection for both sets of actors, as well as any other factors relating to capability, opportunity or motivation (Michie, van Stralen, et al., 2011) to engage with the processes. Identifying relevant information regarding mechanisms required us to analyse how and why the methods of connection and the physical activity processes in which they were embedded generated or did not generate the desired outcome(s), as again this was not a focus of the included evaluations. To do so we firstly developed initial programme theories, comprising key components of the methods of connection and the physical activity promotion processes in which they were embedded, expected outcomes/impacts as relevant for the review and theories of change. We then established context-mechanism-outcome configurations to test and refine those theories. Where effectiveness evidence was not available, we employed acceptability evidence as the outcome. For processes where evidence regarding both was available, we employed acceptability evidence to help explain why the process was or was not effective.
To facilitate the process of data extraction we developed forms based on our research questions-one for PCP and one for patients.
The data were extracted by KBC and reviewed by RHR. Any disagreements were resolved by discussion, with third and fourth authors (SAC and GO) consulted when necessary.

| Data synthesis
Data synthesis involved classifying evidence regarding workload, effectiveness and acceptability as 'high', 'medium' or 'low' (see Table 1 for details of classification schemes). It also involved categorising behaviour change strategies using the Refined Taxonomy of Behaviour Change Techniques to Help People Change Their Physical Activity and Healthy Eating Behaviours (CALO-RE Taxonomy) (Michie, Ashford, et al., 2011) to aid description and facilitate comparison, of the processes of physical activity promotion incorporating methods of connection. We developed separate forms to conduct the data syntheses for each of the three research questions.
For question 1 we identified the ways in which PCP connected patients with community-based physical activity opportunities. We classified these into overarching methods of connection and included detail regarding the different modes of implementation of each.
For question 2 we identified the processes of physical activity promotion in which the methods identified in question 1 were embedded. We then undertook comparisons of those processes to identify any syntheses that could occur as per standard realist (systematic) reviews within the realist scoping review. In other words, we sought opportunities to compare findings for a process or class of process implemented in multiple contexts. Each of the processes identified was unique, differing significantly from all the others, therefore such opportunities did not exist and it was necessary to conduct data synthesis separately for each process. Information regarding context was relevant for answering this question. We classified the workload component of context as 'high', 'medium' or 'low' for PCP and patients. See Table 1 for details of the classification schemes. Evidence concerning acceptability and effectiveness was also relevant for answering this question. We classified such evidence-for PCP and patients separately-as 'high', 'medium' or 'low'. See Table 1 for details of the classification schemes.
For question 3 we examined mechanisms underlying the processes that were (or were not) effective or acceptable and compared them with the hypothesised mechanisms in the theories of change contained in the initial programme theories. This enabled us to identify whether the evidence supported or refuted the hypothesised TA B L E 1 Details of the classification schemes for workload, acceptability and effectiveness for PCP and patients  Passive organisation/enrolment: patient must respond to contact initiated by PCP or PA professional/researcher or provider of PA opportunity.
c These cut-offs are in line with those of Pavey et al. (2011). mechanisms, leading to discernment of necessary refinements to initial programme theories.

| Search results and document characteristics
We identified 1,030 records through database searching and 24 records from other sources. Following de-duplication we screened 1,004 titles, abstracts, summaries and lists of contents and 36 full texts for eligibility. Ten documents were included in the review. See Figure  Three of these reported the total number of eligible and willing patients connected, enabling calculation of full connection rates.
Four reported only a partial number of eligible and willing patients, allowing calculation of what we termed a partial connection rate. Two documents provided data enabling calculation of effectiveness for patients (patient uptake rates). We were able to identify acceptability evidence for PCP in four documents. We considered acceptability as an outcome in all four of these. We were able to identify acceptability evidence for patients in four documents-in two of these the evidence related to the whole process of physical activity promotion and in two the evidence concerned specific components of the process. We considered acceptability as an outcome in three of these documents and as a factor helping to explain effectiveness in one document. See Table 2 for details, and relevance and rigour classifications, of the included documents.

| What methods of connection from primary care to community-based physical activity opportunities have been evaluated?
We identified that five methods of connection from primary care to community-based physical activity opportunities had been evaluated-one method was employed in a direct route in which the PCP connected the patient with physical activity opportunities, the other four were employed in an indirect route in which the PCP connected the patient with an intermediary-a physical activity professional or researcher-who then connected the patient with opportunities.
The method employed in the direct route was active signposting, which involved the PCP actively communicating information about, and recommending attendance at, physical activity opportunities. The methods employed within the indirect route were: PCP: • Effectiveness-partial connection rate • Acceptability (outcome) Patients: • Acceptability -whole process (outcome)

Relevance:
High 4. referral/prescription followed by further referral/prescriptioninvolving the PCP facilitating contact with an intermediary who facilitates enrolment in physical activity opportunities.
These methods were implemented in several different ways. See    Table 3. We identified a total of 15 different processes of physical activity promotion. See Table 3 for details.
Evidence concerning PCP behaviour, that is whether or not PCP connected eligible and willing patients with physical activity opportunities, was available for all 15 processes. However, we were only able to calculate full connection rates for two of those processes (A, F). Both processes were highly effective. We were able to calculate the full connection rate for PCP of two further processes (G and H) in combination. These processes were of low effectiveness. We were able to calculate partial connection rates of six processes (B, C, I, J, K, M). The partial effectiveness of processes B, C and J was high. The partial effectiveness of processes I, K and M was low. Acceptability evidence was reported for three processes (L, N, O). The acceptability of process N was medium and the acceptability of processes L and O was low. Acceptability evidence was reported for a further two processes (D, E) in combination. These were of medium acceptability.
Evidence concerning patient behaviour, that is whether or not patients took up the PA opportunity following connection, was available for only seven processes (A, F, M, B, C, D, E) and mainly regarded acceptability. We were able to calculate patient uptake rates for only two of these processes (A, F). Both processes were highly effective. Findings regarding the acceptability of the overall process for patients were available and considered as an outcome for only one process (M). Its acceptability was high. Findings regarding the acceptability of specific components of the process were available for two processes (B, C). The acceptability of the information provided in process B was high whereas the acceptability of that provided in process C was low. Findings regarding the acceptability of specific components of the processes were reported for two further processes (D, E) in combination. The 'Let's Get Moving' (LGM) resource pack was of medium acceptability.
We were able to have at least a moderate level of confidence in the credibility of acceptability evidence. All relevant documents containing this evidence scored 'high' or 'medium' for rigour. Despite the rigour of the documents containing effectiveness evidence ranging from 'high' to 'low', we were able to have a high level of confidence in the credibility of this evidence as the evidence contained in documents scoring 'low' for rigour was explicitly presented as partial connection rates, rather than full connection rates. See Tables 4 and   5 for details of the context, that is for whom and in what circumstances the findings apply, and outcomes for each of the processes for patients and PCP respectively.
Consideration of the PCP and patient outcomes for each process in combination revealed strong positive findings for four of the processes in the contexts in which they were implemented (A, F, B, J).
Those processes had only evidence of 'high' effectiveness or acceptability. Processes A and F were highly effective for both PCP and patients. The partial effectiveness for PCP of process B was high, while the acceptability of information provided was high for patients. The partial effectiveness for PCP of process J was also high, however there was no evidence regarding the effectiveness or acceptability for patients. Consideration of the PCP and patient outcomes for each process in combination also revealed moderately positive findings for three processes (D, E, N

Medium
Practice not using Vision: • Impact on practice manager workload minimal, with majority of work falling to reception staff • Ran smoothly and had no adverse effects on receptionists' workload • Access to IT system at GP practice organised externally and took a couple of weeks, however once this was done accessing the appropriate systems was not problematic Practice using Vision: • Struggled to export the patient lists to Docmail due to type of software used • Practice staff had to do the reports manually, which increased their workload and pressure-became easier with practice

TA B L E 5 (Continued)
process M was low and the acceptability was medium for PCP, despite high acceptability for patients. The acceptability of processes L and O was low for PCP and there was no evidence regarding the effectiveness or acceptability for patients. See Table 6 for the outcomes for PCP and patients.
Despite the uniqueness of each of the processes, and the differ-

| How and why are (or are not) those processes effective or acceptable?
The theories of change of the initial programme theories were highly supported for three of the processes with strong positive findings (A, F, B)-that is all components of those theories were supported by the effectiveness and/or acceptability evidence. The theories of change were supported to an extent for one of the processes with strong positive findings (J) and the three processes with moderately positive findings (D, E, N). In these cases, some TA B L E 6 The outcomes for the 15 processes of physical activity (PA) promotion incorporating methods of connection for PCP and patients

| Key findings
Ten documents were included in our review aiming to explore what is known about the effectiveness of different methods of connecting primary care patients with community-based physical activity opportunities in the United Kingdom.
We identified that five methods of connection from primary care to community-based physical activity opportunities had been evaluated. One method-active signposting-was employed in a direct route in which the PCP connected the patient with physical activity opportunities. The other four methods involved the PCP connecting the patient with an intermediary who then connected the patient with opportunities, and were: (a) active signposting followed   the most influential factors of effectiveness; (e) the workload for patients per se may not be one of the most influential factors of effectiveness, however facilitation of patient action, through referral or prescription, might be a key influential factor; (f) financial discounts for physical activity opportunities for patients may positively influence effectiveness.
The theories of change contained in the initial programme theories were highly supported for three of the processes with strong positive findings. They were supported to an extent for one of the processes with strong positive findings and the three processes with moderately positive findings. They were refuted for the eight processes with negative findings. The findings of comparisons of the theories of change also generated several indications helpful for future development of effective processes of physical activity promotion incorporating methods of connection from primary care to community-based opportunities: (a) the importance of strategies to enhance patient capability and motivation to undertake the behaviours required for connection to, and uptake of, a physical activity opportunity, including a particular need for improved strategies to enhance patient capability and motivation to organise the session with an intermediary; (b) inclusion of an in-person session may be a strategy necessary for patient capability and motivation to uptake a physical activity opportunity; (c) ensuring a low workload appears to be an effective strategy to enhance the capability and motivation of PCP to connect patients; and (d) so long as the process is not too demanding for PCP, providing training is an effective strategy for enhancing the capability and motivation of PCP to connect patients.

| Strengths and limitations
Our review is the first synthesis of the evidence regarding the effectiveness for PCP and patients of methods of connection from primary care to community-based physical activity opportunities in the United Kingdom. In order to answer the research questions the review overcame significant challenges: (a) the use of a relatively novel type of evidence synthesis which lacks methodological guidance; (b) the focus on two sets of actors and two outcomes for each; (c) the need to further interpret findings provided, as well as to conduct secondary data analysis. Our review has several other strengths.
The use of realist methodology facilitated understanding of how and why processes succeeded or failed in different contexts. It also enabled development of realist programme theories relevant not only for the case of physical activity promotion but also for other cases of health promotion involving connection from primary care to community-based opportunities, including social prescribing activities . The inclusion of grey literature was a particular strength as less than half of the documents included were from academic journals, which, along with the types of sources of funding reported, indicates the 'real-world' application of the topic.
The use of behavioural theory-the COM-B behavioural model and the CALO-RE Taxonomy of behaviour change techniques-facilitated secondary analysis and interpretation of primary evaluation data/ findings, thereby enhanced the evidence base in the area.
The main limitation of the review is the dearth of evidence regarding the effectiveness of different methods of connecting primary care patients with community-based physical activity opportunities in the United Kingdom. Limited attention has been paid to the effectiveness and acceptability of such methods, and the processes of physical activity promotion in which they are embedded, in primary evaluations. Despite undertaking of secondary data analysis for the included primary evaluations, evidence regarding effectiveness for both PCP and patients was available for only two of the fifteen identified processes. There was greater emphasis on acceptability than on effectiveness in the included evaluations, however, while a necessary condition for effectiveness, acceptability does not guarantee effectiveness. Such evidence is therefore more useful in combination with effectiveness evidence to enable understanding of how and why a process was or was not effective. The lack of evidence precluded definite conclusions regarding their outcomes, and the underlying theories of change. It is important to note that full evaluations might have led to different conclusions regarding outcomes and theories of change, as well as greater understanding of why theories of change were supported or refuted. A further limitation was the inclusion of only documents written in English, however given the UK context, we do not perceive this to be a significant limitation.

| Comparisons with existing literature
The paucity of evidence specific to the effectiveness of methods of connecting primary care patients to community-based opportunities limits comparisons of our review with existing literature. Our findings were in line with those of the theory-driven qualitative study we conducted as part of the same larger research project (Carstairs et al., 2020). Our qualitative study explored primary care patient and PCP views regarding methods of connection. Patients and PCP discussed three methods of connection from primary care to community-based physical activity opportunities that they believed could be effective: informal passive signposting; informal active signposting; and formal referral/prescription. Similar to our review, the use of the COM-B model to understand perceived barriers and facilitators to their potential effectiveness provided useful insight to the future development of processes of connection with methods embedded.
Of note in the wider social prescribing field is a recent UK-based realist review identifying a dearth of evidence regarding the effectiveness of methods of connection from primary care to communitybased activities (Husk et al., 2020). The need for improvement in the evidence base in this area is corroborated by other social prescribing literature, which also advocates the importance of theorybased evaluation of methods of connection (Hopewell, 2017;Kellezi et al., 2019;Price, Hookway, & King, 2017;Roland et al., 2020;Stevenson, 2019)-a point receiving less attention in the physical activity-specific literature.

| Recommendations
The findings of our review have several implications for both practice and research in the area. Firstly, future development of processes of physical activity promotion-and other health promotion-involving connection from primary care to community-based opportunities, including social prescribing activities, should take into account the useful indications from our review in order to enhance their effectiveness. Secondly, well-designed theory-based evaluations are needed to progress the evidence base in the area. Such evaluations should be built in to the design and implementation of processes of physical activity promotion incorporating methods of connection from the outset, and efforts should be made to capture in practice the data necessary for such evaluations. This includes data relating to the effectiveness of all three stages of the processes: (a) approaches to identifying eligible and willing patients who would benefit from increasing their physical activity levels; (b) behaviour change strategies aiming to enhance the likelihood of patients increasing their physical activity levels; (c) methods of connecting patients with community-based physical activity opportunities. These data should be complemented by data regarding how and why the processes are or are not effective for PCP and patients. Thirdly, regarding the first two stages of the processes, although not a focus of our review, we observed in several of the 15 processes low willingness of patients to actively participate in eligibility checks, and low willingness of eligible patients to participate in processes of physical activity promotion and thus be connected with community-based opportunities. Research into techniques and strategies to improve the willingness of potentially eligible patients to participate in eligibility checks, and the willingness of eligible patients to participate in processes of physical activity promotion, would therefore be beneficial. In particular our review identified a need for a focus on techniques and strategies to enhance patient capability and motivation to organise the session with an intermediary in indirect routes.

| CON CLUS IONS
The evidence base concerning the effectiveness of methods of connection from primary care to community-based physical activity opportunities is lacking. Our review revealed several indications useful for the future development of such methods and the processes of physical activity promotion in which they are embedded. It also highlighted the need for well-designed theory-based evaluations.

CO N FLI C T S O F I NTE R E S T
All the authors declare no competing interests.