Barriers and enablers to sustaining self-management behaviours after attending a self-management support intervention for type 2 diabetes: a protocol for a systematic review and qualitative evidence synthesis

Review question (Stage 1 – Step 1)

A scoping search was conducted to explore the amount and nature of the evidence and to inform the review aim and question. The following review question was developed using the SPIDER (Sample, Phenomenon of Interest, Design, Evaluation, Research Type) framework³⁴:

• What are the barriers and enablers perceived by adults with type 2 diabetes to sustaining self-management behaviours after attending a diabetes self-management support intervention?

Eligibility criteria

The following eligibility criteria were informed by the SPIDER framework³⁴.

Sample. Studies including adults (aged ≥ 18 years) with a diagnosis of type 2 diabetes who have attended a self-management support intervention will be included. If the study also contains participants with type 1 diabetes, individuals with type 2 diabetes aged under 18 years, women with gestational diabetes or adults with type 2 diabetes who did not attend a self-management support intervention, studies will be included only if it is possible to extract the barriers and enablers perceived by the relevant participants.

Phenomenon of interest. Studies will be included if they focus on barriers and enablers to sustaining self-management behaviours at least three months after attending a self-management support intervention. For the purpose of this review, a barrier is defined as any factor that impedes or obstructs sustaining self-management behaviours. An enabler is defined as any factor that facilitates or helps sustain self-management behaviours. The data collection timing will be verified in the study abstract or full text to ascertain the data reported in the study was collected at least 3-months post-intervention. If the study is focused on barriers and enablers to sustaining self-management behaviours, but the data collection timing is not reported in the abstract, the study will be retrieved for full-text screening. When the time gap between intervention attendance and data collection is not reported in the study, the research team will contact the corresponding author(s) for clarification and further information. If it is not possible to ascertain the time gap between intervention attendance and data collection, the study will be excluded. Consistent with a previous review¹⁴, a self-management support intervention for type 2 diabetes is defined as any intervention that aims to support or facilitate self-management of type 2 diabetes. In accordance with the available literature on self-management support for type 2 diabetes, interventions should explicitly focus on self-management support of type 2 diabetes and target one or more of the following self-management domains to be included in the review^17,35:

Cognitive skills

Behavioural skills

Emotional skills

There will be no restrictions for inclusion based on the intervention setting, mode of delivery, type of facilitator, intensity, duration, and theoretical basis of the intervention. Studies that include carers or family relatives will be included in the review, as long as the intervention is primarily intended for the person with type 2 diabetes. If the study includes both individuals still attending the intervention and individuals who have attended the intervention, studies will only be include if it is possible to extract data from individuals who attended the intervention and have experiences of self-management for at least three months post-intervention at the time of the study. If studies explore sustaining self-management behaviours and other concept(s) in self-management, it must be possible to extract the information specific to sustaining self-management behaviours to be included in the review. Studies examining other aspects of living with type 2 diabetes (e.g., coping, emotional distress) where it is impossible to extract the data on barriers and enablers to sustaining self-management behaviours after attending a self-management support intervention will be excluded.

Design and Research Type. Primary qualitative and mixed-methods research studies will be considered for inclusion. Although grey literature can be difficult to search and retrieve, non-peer reviewed studies will be considered for inclusion as there is a growing consensus that the inclusion of grey literature can widen the scope of reviews, thus providing a more complete picture of the evidence available^36,37. Studies will be included if a qualitative method is used for data collection (e.g., focus groups or interviews) and analysis (e.g., thematic analysis or grounded theory). The data collection and analysis methods should be clearly reported in the studies to be included in the review. Studies that collect data using qualitative methods but analyse it quantitatively (e.g., descriptive statistics) will not be included in the review. Mixed-methods or multi-methods studies will be included in the review if the qualitative data is reported separately and is relevant and of sufficient depth to be synthesised in the review. Both mixed and multi-methods studies that report qualitative findings in either a separate linked publication that was cited in the study retrieved in the search, or as a separate section of a study retrieved in the search will be considered for inclusion. Studies will be included if there is a focus on barriers and enablers to sustaining self-management behaviours after attending a self-management support intervention for type 2 diabetes. Studies will be deemed relevant and of sufficient depth to be synthesised in the review if relevant findings are supported by at least one quote from participants. Only articles published in English will be included in the review to avoid issues with language translation of qualitative studies as meaning may be lost in translation³⁸, as well as pragmatic constraints including a limited timeframe for the review, and limited access to translation services. Quantitative studies, literature reviews, qualitative evidence synthesises, editorials, commentaries, opinion pieces, and abstracts in proceedings will be excluded. Where the full text article is not available online, the corresponding author(s) will be contacted by email with one follow-up. If the corresponding author(s) do not reply within one week after the follow-up, the article will be excluded.

Information sources and search strategy (Stage 1 – Step 2a)

A combination of systematic searching of the literature of electronic databases and supplementary searching techniques will be used to maximise the identification of relevant papers for inclusion in the review³⁴. A comprehensive search will be conducted on the following databases: MEDLINE (Ovid), EMBASE (Elsevier), CINAHL (EBSCO), PsycINFO (Ovid) and SCOPUS from inception to September 2021. An expert librarian provided support on the selection of the databases and development of the search strategy.

The search strategy was informed by the SPIDER framework³⁴ (See Table 1 for further details) in consultation with an information specialist. The search combines free-text terms with index terms (e.g., Medical Subject Headings) for type 2 diabetes, self-management support interventions, sustained behaviour change, and qualitative research. The search strategy was developed iteratively and informed by existing reviews^{14,19,39–42} to ensure the search was as comprehensive as possible. Methodological filters for qualitative research were also used where available in specific databases to enhance the specificity of the search^43,44. Using validated search filters in qualitative research can result in the omission of potentially relevant studies due to the diversity of terms used to index qualitative research within electronic databases³⁷. Therefore, search filters have been combined with other synonymous free-text terms in qualitative research to ensure potentially relevant studies were not missed³⁷. Search terms were truncated where relevant to ensure all spellings are captured (e.g., behavio*). In addition, Boolean terms, such as OR and AND were included in the search to enhance specificity and sensitivity. A sample search strategy for the MEDLINE (Ovid) database is presented in Extended data: Supplementary File 3²⁸.

A search of grey literature will also be undertaken on ProQuest Dissertations and Theses, WorldCat via the Online Computer Library Center (OCLC) and Open Grey in September 2021. To counteract common challenges in identifying qualitative literature through systematic searching of electronic databases alone³⁷, forward, and backward citation searches will be conducted on all included studies.

Screening (Stage 1 – Step 3a)

The lead author will import all references to the electronic reference manager EndNote X20 and remove duplicates. The eligibility criteria will be pilot tested with a random sample of 6 papers by two authors (MC and PD) and the criteria clarified if needed. The lead author will then screen the titles and abstracts and full texts of the identified articles against the eligibility criteria using Rayyan QCRI software⁴⁵. A second author (PD) will independently screen a random sample of 20% of the articles at both stages. A chance-corrected Kappa statistic will be calculated to assess inter-reviewer agreement⁴⁶ at both stages. Disagreements will be discussed between the authors and, if necessary, with a third author until consensus is achieved. When the abstract is not available or does not contain enough information to make an informed decision about the inclusion of an article, the article will be retrieved for full text screening. If necessary, authors of primary studies will be contacted for clarification and further information. All the studies identified as potentially relevant by one or both authors will be retrieved for full text screening. A table listing studies excluded from the review and the main reasons for exclusion will be recorded by the research team and presented in the review using a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram³².

Data extraction and management (Stage 1 – Step 3a)

Full-text articles will be imported to QSR International’s NVivo v12 software⁴⁷. This software will be used to store the data and to assist in the data extraction and synthesis to ensure clarity and transparency⁴⁷. A data extraction form (See Extended data: Supplementary File 4) will be created for the purpose of this synthesis within NVivo^48,49. The following data will be extracted from each study⁴⁸: study information (e.g., authors, year of publication), study characteristics (e.g., aims and objectives, sample size), participants’ characteristics (e.g., age, gender, time living with the diagnosis), intervention characteristics (e.g., intervention components, duration, and mode of delivery), data collection and analysis methods. Findings, including participants’ verbatim quotes and reported interpretations by the primary study authors will be extracted as “best fit” framework synthesis allows the analysis, synthesis and integration of primary and secondary data^31,50. Reported strengths, limitations and implications of the study will also be extracted.

The data extraction form will be initially pre-tested with a random sample of three papers by two authors (MC and PD). One author (MC) will extract data from all included articles and a second author (PD) will cross-check 20% of the articles to ensure consistency and minimise potential bias during the data extraction process. Any disagreement will be discussed between the authors and, if necessary, with a third author until consensus is achieved.

Assessment of methodological limitations of the included studies (Stage 1 – Step 3a)

An adapted version of the Critical Appraisal Skills Programme (CASP) Tool⁵¹ for qualitative studies will be used to assess the methodological limitations of the included studies. The following domains will be considered: context, sampling strategy, data collection, data analysis, support of individual study findings in data, reflexivity, and ethical considerations. Each domain will be judged as yes (i.e., the domain is sufficiently, clearly, and appropriately described in the study), no (i.e., the domain is not described in the study) or unclear (i.e., the study only offers a limited or unclear description of the domain). MC will appraise all the included studies. PD will independently appraise 20% of the included studies. Disagreements will be discussed, and a third author will be consulted if necessary. Studies will not be excluded based on this assessment⁵², in line with recommendations⁵³, but this information will be considered in the analysis of the findings, assessment of confidence in the review findings and the reporting of the review.

Identification and development of the a priori framework (Stage 1 – Steps 2b and 3b)

The a priori framework for this review will be selected through a combination of literature search, expert consultation, and research team consensus²⁹. Potential a priori theories will be identified by the review team opportunistically from within the topic-relevant searches and articles selected for full text screening, and purposively from an independent parallel search in Google Scholar combining the search terms ‘model*’ or ‘framework*’or ‘theoretical’ or ‘theory’ or ‘concept’ or ‘conceptual’³¹ with the terms ‘type 2 diabetes self-management’ and ‘sustained behaviour change’. A list of relevant candidate theories will be developed based on the results of this search. This approach has been previously used to identify an appropriate existing conceptual model in a worked example of “best Fit” framework synthesis³¹. Google Scholar has the advantages of covering mainstream and non-mainstream academic literature and ability to sort the results of the search by relevance. Records retrieved will be read to identify relevant candidate theories. The aim is to build a comprehensive list of candidate theories. For each new record, the theory or theories identified will be matched to ones previously identified and added to the list if not previously identified. The search will be terminated when no new theories are identified after five new records have been read⁵⁴.

As it is anticipated that the number of hits for the initial search will be high with a large number of records of very low relevance, the search results will be ordered by relevance using the Google algorithm ‘sort by relevance’⁵⁴.

When a list of potential theories is identified, the review team will meet to discuss the suitability of the candidate theories, and to determine if a single comprehensive theory can be used, or if it is necessary to develop a meta-framework using concepts or constructs from different theories in the existing literature⁵⁰. The list of candidate theories will be circulated by the research team in advance of the meeting. Additionally, the three senior research team members (DK, MB, JMS) with extensive knowledge of and experience using behavioural theories will be consulted to identify any potentially relevant additional theories they are aware of which are not included in it the list.

The review team will discuss the conceptual fit of each of the candidate theories until consensus is achieved. The three criteria outlined by Damschroder and colleagues⁵⁵ will be considered by the review team when evaluating candidate theories, as suggested by Booth and Carroll⁵⁰ (See Table 2). The assumption by Booth and Carroll⁵⁰ that a priori theoretical framework does not need to be “a perfect match for the question or evidence”, but only offer “a ‘good enough’ starting point as designated by the phrase “best fit” (p. 701) will be taken into consideration by the research team. After identifying the most suitable theory, secondary thematic analysis²⁹ will be employed to create the a priori framework. Thematic analysis will be used to generate a set of explanatory constructs and theoretical propositions, referred to in the “best fit” synthesis method as themes, which represent patterns of theoretical explanations. The theoretical themes identified might be further organised by subthemes if appropriate.

Based on the information available in the primary studies and original papers of the selected theory, definitions will be created for each theme²⁹. As the suitability of a theory or theories also depends on the proportion of the data that can be accommodated within it, the choice of the a priori framework will be revisited during the data analysis and synthesis process to ensure the framework selected is the most appropriate²⁹. The a priori framework will be considered appropriate if it accommodates at least 50% of the data extracted from the included primary studies.

Data analysis and synthesis (Stage 2 – Steps 4 to 8)

The lead author will develop a coding tree on NVivo with the themes and constructs identified to facilitate the coding of the data against the a priori framework. A second author (PD) will cross-check the final list of themes to ensure different authors can consistently code data from primary studies with a sample of three studies²⁹. The findings from the included studies will be coded against the themes generated based on the a priori framework (Step 4). New themes will be generated to code data that cannot be coded against the a priori framework secondary thematic analysis²⁹. New themes will be based on the author’s interpretation of the data and constant comparison of such data across studies²⁹ (Step 5). The new themes resulting from this analysis will be added to the a priori framework.

A new updated framework composed of a priori and new themes supported by the evidence will result from this process^29,50 (Step 6). One author (MC) will conduct all stages of data analysis and synthesis from coding to interpretation with continuous input from the rest of the research team. A second author (PD) will independently analyse and synthesise 20% of the articles. Any discrepancies will be discussed between the two authors and if necessary, with a third author until consensus is achieved.

The potential ways in which themes may relate to each other will then be explored using constant comparison method^56,57, facilitating the generation of a new conceptual model describing the process of sustaining type 2 diabetes self-management behaviours after attending a self-management support intervention^29,50 (Step 7).

Finally, a sensitivity analysis^29,58 will be performed to examine the contribution of studies with methodological limitations to the review findings^29,48,58 (Step 7). Query tools within NVivo, such as matrix-coding queries, will be used to conduct the sensitivity analysis following the guidance provided by Houghton and colleagues (2017)⁴⁸. Matrix-coding queries allow comparing multiple nodes/codes and attributes as a numeric matrix table with different shading colours, which indicates whether studies with methodological limitations skewed the findings in any particular way based on their contribution to the finding⁴⁸. Each study will be assessed for methodological limitations based on the CASP tool criteria to facilitate the analysis⁵¹. Subgroup analyses will be conducted where appropriate and if possible, comparing studies based on intervention characteristics, time gap between attendance at the intervention and data collection, participants’ characteristics, and study context, including continent and income level of country where the study was carried out.

Confidence in the review findings

The Confidence in the Evidence from Reviews of Qualitative Research (GRADE-CERQual)⁵⁹ approach will be used to assess confidence in each theme included in the final model. GRADE-CERQual assesses confidence in review findings, based on the following key components: methodological limitations (i.e., the extent to which there are concerns about the design or conduct of the primary studies that contributed to the review finding); relevance (i.e., the extent to which findings from the primary studies are relevant to the review question), coherence (i.e., the extent to which the review findings are grounded in data from the primary studies), and adequacy of the data (i.e., the extent to which a review findings are supported by rich data and a large number of studies)⁵⁹. After assessing each of the four components, a judgement about the overall confidence in the review finding will be made. The confidence in each review finding will be graded as high (i.e., it is highly likely), moderate (i.e., it is likely), low (i.e., it is possible) or very low (i.e., it is unclear) dependant on whether the review finding is judged to a reasonable representation of the phenomenon of interest⁵⁹. All findings will initially be graded as high confidence and will then be graded down if there are important concerns regarding any of the GRADE-CERQual components⁵⁹. The final assessment will be based on consensus among the two authors (MC and PD) involved in the confidence assessment with discussion with the full review team if needed.

Reflexivity

To ensure rigour and quality, the research team members will maintain a reflexive stance throughout all stages of the review process, from study selection to data synthesis and interpretation of the findings. A team-based reflexive approach will be adopted involving individual critical reflection on assumptions and potential biases and reflexive group discussions⁶⁰. The research team will continuously reflect on their background and how their personal views and beliefs could influence their choices regarding methods to use, data extraction, coding synthesis, and interpretation of the review findings. MC, JMS, DK and MB have a background in Health Psychology and primarily work in research focused on health behaviour change. MB is a professor, JMS a lecturer, DK is a senior research fellow, and MC is a PhD candidate. PD is a PhD candidate with a background in nutrition/dietetics and has extensive clinical experience delivering self-management support interventions for people with Type 2 diabetes. MB and JMS have experience in conducting qualitative evidence synthesis and primary qualitative research focused on type 2 diabetes mellitus. MC and PD have previous experience in conducting primary qualitative research. DK has experience and expertise in theory review and synthesis. All authors believe that sustaining self-management behaviours is challenging, and people might struggle to integrate these behaviours into their daily lives.

During the screening, data extraction, coding and synthesis, and assessment of confidence in the review findings, the team will regularly meet to discuss progress and potential disagreements. The two authors (MC and PD) who will conduct the study screening, data coding, extraction and synthesis, and assessment of confidence in the review findings, will meet regularly to discuss how their background, experiences and presumptions on the review topic may be influencing their assessments and analysis and will both record and reflect on their decisions in memos. Both MC and PD will keep reflexive diaries and fieldnotes to document and critically reflect on the research process, which will be recorded in memos and form the basis for ongoing discussion in research team meetings. The authors will use the seven questions orientating for team-reflexive accounts proposed by Barry et al. (1999)⁶⁰ to guide their reflexive accounts. Examples of questions proposed by Barry et al. (1999)⁶⁰ include: (1) ‘in what way might my experience colour my participation in the project?’, (2) ‘what is my orientation to qualitative research?’, and (3) ‘what results do I expect to come out of this project?’. By keeping this reflexive stance, the research team hopes to reduce potential threats to rigour and any potential negative impact of any personal or professionals’ beliefs or biases on the interpretation of the findings. As suggested by Flemming and Noyes (2017)⁶¹, the following aspects relevant to author reflexivity will be reported on the final manuscript of the review: (a) the funder and their potential involvement and influence on the development and editing of the review findings; (b) the composition of the review team and the key positions or beliefs about the review question and the phenomenon of interest that could influence the interpretation of the findings; (c) potential conflicts of interest both financial and non-financial; (d) team governance procedures and processes to ensure internal validity; (e) procedures for processing evidence when one of the review authors is also an author of a primary qualitative study; and (f) procedures and processes to engage with key stakeholders ensuring no potential undue influence on the review and interpretation of the findings⁶¹.

Public and Patient Involvement

The involvement of key stakeholders in systematic reviews is increasingly recognised as fundamental to the quality, relevance, and impact of the review findings^62,63. Patient and public representatives of people with type 2 diabetes who attended a self-management support intervention for type 2 diabetes will be involved in the review from the point of data synthesis. Patient and public representatives will be asked to review the a priori framework and the new themes that are generated by the research team and contribute to the new updated framework and interpretation of the synthesis findings. In addition, they will also be invited to contribute to the development of dissemination strategies and assist in the preparation of dissemination documents, such as the lay summary, to ensure clarity and readability. The involvement of the members from the advisory panel in the systematic review process will be guided by the ACTIVE (Authors and Consumers Together Impacting on eVidencE) framework, which outlines a range of methods and approaches to guide both the involvement of stakeholders in systematic reviews and the reporting of their involvement in the review process⁶³. The activities and contributions of the advisory panel will be reported in line with the Guidance for Reporting Involvement of Patients and the Public Version 2 (GRIPP2) Checklist⁶⁴ and the ACTIVE framework⁶³.

Study status

The review is currently underway. The database searches, title and abstract screening and full text screening, data analysis and synthesis, and assessment of in confidence in the review findings have been completed.

Underlying data

No data are associated with this article. However, dataset associated with the review will be published Open Access online on the Open Science Framework review page.

Extended data

Open Science Framework: SUSTAIN T2DM: Supporting people with type 2 diabetes to sustain self-management behaviours, https://doi.org/10.17605/OSF.IO/KJVGU²⁸.

The project contains the following extended data (under Work Package 1: Barriers and enablers to sustaining self-management behaviours after attending a self-management support intervention for type 2 diabetes: A systematic review and qualitative evidence synthesis):

Reporting guidelines

Open Science Framework: Supplementary File 1. PRISMA-P checklist, https://doi.org/10.17605/OSF.IO/KJVGU²⁸.

Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).