Parts of this chapter have been reproduced from Law et al.1,2 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.
Literature searching
We reviewed the existing literature to look for evidence that suggested how and for whom physical activity interventions work to optimise physical function in the primary care setting. As interventions or services based in other areas of literature (e.g. secondary care, social services, the voluntary sector, exercise science) also hold relevant insight for the development of the initial programme theories, searches were not restricted. A systematic search strategy was developed and amended for use with the following databases: Cochrane Library, MEDLINE, Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycInfo® (American Psychological Association, Washington, DC, USA), Sociological Abstracts, Web of Science™ (Clarivate Analytics, Philadelphia, PA, USA), Applied Social Sciences Index and Abstracts (ASSIA; ProQuest®, ProQuest LLC, Ann Arbor, MI, USA), Social Care Online and Social Care Institute for Excellence (see Report Supplementary Material 4).1 Keywords were developed from the early scoping exercise and the key themes underpinning the initial programme theories were adapted for each information source as necessary. The searches were run in March 2019 and updated in 2020.
Our searches included adults of all ages and socioeconomic backgrounds. We translated non-English-language papers where relevant and practical. We did not limit our searches by publication date and there was no restriction on the type of publication or study type. We examined published and unpublished literature including research articles, systematic reviews and documents detailing policy and local and national initiatives. We did not search for, or include, studies that had limited transferability to NHS primary care, such as interventions involving pharmacological agents or very technical, high-cost equipment.
Literature initially was screened for relevance to the initial programme theories and cross-checked by two members of the research team. To assist with this, articles were uploaded to Covidence (Melbourne, VIC, Australia) to complete the initial sift. This software enabled multiple members of the team to participate in the sifting out of irrelevant articles.
The review team then followed predefined inclusion and exclusion criteria to remove irrelevant articles (see Report Supplementary Material 5). Thirty per cent of articles were reviewed by two members of the review team, with conflicts resolved through discussion between the two main reviewers and other members of the review team as necessary. As part of these discussions, study team meetings and following the initial theory-building workshops, we iteratively refined the inclusion and exclusion criteria. After this initial phase of double reviewing, the criteria were deemed sufficiently robust, and the remaining articles were reviewed by a single member of the review team. This resulted in a library of 2083 relevant papers. From that library, a purposive sample of 170 papers were selected for data extraction, chosen on the basis that they were the most relevant to the developing theory areas and gave the clearest and richest examples of evidence of interventions aimed at functional improvements in people with long-term conditions, and evidence of how these interventions are organised and operate with different primary care contexts. We supplemented the systematic search with forwards and backwards citation tracking of key articles. We also drew on the expertise of the project team, external Project Advisory Group, patient and public representatives, and other key researchers (nationally and internationally) and organisations to ensure that we did not miss any relevant evidence that may not have been retrieved by our traditional systematic searching methods.
We also carried out a grey literature search by targeting relevant organisations and programmes:
organisational websites of professional bodies (i.e. Royal College of General Practitioners, Royal College of Physicians, Royal College of Nursing, Chartered Society of Physiotherapy, Royal College of Occupational Therapists, Academy of Medical Royal Colleges, Royal College of Surgeons, British Association of Sport and Exercise Sciences and Royal College of Psychiatrists)
government departments and national centres (i.e. Public Health Wales, Public Health England, National Centre for Sport and Exercise Medicine, UK Faculty of Sport and Exercise Medicine, Sport England, Sport Wales and The King’s Fund)
specific organisations and charities for people with long-term conditions
Natural Resources Wales and National Parks England.
We also searched the Evidence for Policy and Practice Information and Co-ordinating Centre, NHS Evidence, Social Care Online and OpenGrey. In addition, we conducted an open web search for any grey literature (including from commercial leisure services). We used the grey literature sources to ensure that we had captured information about specific campaigns around physical activity for people with long-term conditions.
As previously described,109 unlike a traditional systematic review, a realist synthesis requires an iterative process for identifying literature. In addition to the systematic review of the literature, we also performed additional purposive searches enabling the initial programme theories developed in stage 1 to be expanded. The purposive searches were as follows:
Guidelines: MEDLINE was searched using specific physical function keywords (physical function* OR physical activity OR physical fitness or exercise) with guideline keywords (exp guideline/ OR Guideline$.ti OR (guideline or practice guideline).pt. We also manually checked the websites of major guideline producers: NICE, Centre for Reviews and Dissemination, professional organisations, World Health Organization, Scottish Intercollegiate Guidelines Network and the National Guideline Clearinghouse.
Social prescribing: we carried out a MEDLINE search using the phrase ‘social prescribing’ and filtered for recent reviews. We followed up with forwards and backwards citation searching on those reviews and consulted with the project team for additional references.
Physical literacy: we carried out a MEDLINE search using the phrase ‘physical literacy’ and filtering for adults only. We also followed up with forwards and backwards citation searching on the papers identified and consulted with the project team.
Data extraction
Consistent with the realist synthesis approach,110 the test for inclusion was whether or not the evidence was ‘good and relevant enough’ to be included.79 Relevance was defined as the ability of the data to contribute to the programme theory.81 Assessment of relevance involved seeking any ‘trustworthy nuggets of information to contribute to the overall synthesis’ (p. 90).111 For example, evidence-rich papers included detailed and reflective descriptions of what it was about interventions that worked (or not) (e.g. papers including qualitative or service evaluation elements), whereas less-rich papers provided limited in-depth description of the intervention and the factors influencing whether or not it worked (e.g. randomised controlled trials with mainly objective outcomes). Rigour or whether or not the quality of the evidence is ‘good enough’ was the research team’s judgement of the credibility of the data, including fidelity, trustworthiness and value.82
Owing to the large data set, we adopted the following approach to data extraction:
We identified all of the systematic reviews and conducted a ‘realist critique’ of all those which were relevant to gather an overview of the evidence available. The realist critique consisted of brief notes describing what information the paper provided about what was working or what was not working to improve physical activity or physical function and why, for whom and in what settings. It enabled us to identify the rich systematic reviews and also increased our awareness of broader patterns in the emerging contexts, mechanisms and outcomes.
Using the information from this process and from the stakeholder work, we then developed eight theory areas.
To facilitate data extraction, we designed a bespoke data extraction form to ensure that we captured data informing the developing theory areas, including intervention details and any differences in implementation.
We then identified literature that was specific to primary care and used the bespoke data extraction form to capture relevant data (see
Report Supplementary Material 6).
When discrepancies were encountered, the project team discussed whether or not the evidence provided met the criteria to be included.
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram depicts the flow of information through the different phases of the synthesis, and details the number and type of papers identified, included and excluded (). The list of final included papers is included in Report Supplementary Material 7.
The PRISMA flow diagram detailing the flow of information through the different phases of the review and the purposive searches.
Theory-refining interviews
As part of the iterative process of refining final programme theories, we explored the developing theory areas with stakeholders through 10 qualitative telephone interviews. Purposive sampling of the stakeholders was informed by the stakeholder analysis to provide a range of perspectives, which included three people with long-term conditions, three GPs, two practice nurses, one health-care assistant and one researcher with a background in pedagogy. One of the people with a long-term condition and one of the GPs were also participants in a theory-building workshop. The researcher with a background in pedagogy also participated in a later co-design and knowledge mobilisation workshop.
A semistructured interview topic guide was used to elicit the views of stakeholders on their resonance with the developing theory areas. The approach used in the interviews was a ‘teacher–learner cycle’ whereby the researcher presented the developing theory areas to the stakeholder (‘teaching’) and then verified with the stakeholder where they needed adjusting (‘learning’) to create an improved, refined version and a ‘mutual understanding’ of the developing theories.1,112
With permission, the telephone interviews were audio-recorded and transcribed verbatim for descriptive analysis of the key themes, which contributed to refinement of the theories.1 NVivo version 12 (QSR International, Warrington, UK) was used to organise the data. Coding linked the themes with the developing theories.
Synthesis of evidence from literature and interviews
This analytical stage involved synthesising the evidence to elicit relationships between the contexts, mechanisms and outcomes. Through the research team’s experience of conducting realist syntheses,82,108,113 suggestions from Pawson and Tilley,78 and underpinned by the principles of realist enquiry, we used the following approach:
organisation of extracted information into evidence tables representing the different bodies of evidence
developing themes across evidence tables in relation to emerging patterns among the developing programme theories to seek confirming or refuting evidence
linking patterns to develop hypotheses that support or refute the developing programme theories.
1
Three very early ‘conjectured’ CMO configurations focused primarily on early themes of primary care culture, providing advice adapted to individual circumstances, and the confidence and behaviour of primary care health professionals. We then developed two overarching CMOs with further explanatory subthemes based on the theory areas, covering organisational and system-wide influences, as well as influences affecting people with long-term conditions at an individual level. However, to clearly acknowledge the interactions between systemic and individual factors, and after incorporating additional evidence and discussion, we settled on five ‘final’ CMOs. These provided further nuance and depth to the initial CMOs and expanded on the area around credibility.
Following this process, a set of synthesised statements were written together with a narrative summarising the nature of the links between context, mechanism and outcome (i.e. what works, for whom and in what circumstances). This also summarised the evidence underpinning the statements (see Chapter 5). This process involved ongoing, iterative discussion among the project team members and the Project Advisory Group, which included public contributors.
Taxonomy
Alongside the evidence synthesis process, we developed a taxonomy of primary care physical activity interventions for people with long-term conditions, which categorised and provided examples of interventions in the following categories:
brief interventions
telephone interventions
online/‘eHealth’ interventions
exercise referral schemes
community navigators
referral to exercise specialists
intervention delivery by existing primary care staff
physical activity pathways
practice-wide initiatives
community initiatives
whole-system approaches to embed physical activity promotion in clinical practice
multifaceted interventions.
