This protocol is compliant with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) guideline recommended for systematic review protocols (23). The PRISMA-P checklist is included in Additional file 1.
Review design
We will use the ‘best fit’ framework approach (19–21) to synthesise evidence for implementing the PCMH in primary care practices. ‘Best fit’ framework synthesis is “a pragmatic, flexible approach to integrating theory with findings from practice” (26). It involves selecting an a priori (starting) framework, so called because it need not be perfect for the research question; just “good enough” (26). The concepts represented by the framework are then used to organise data extracted from the included literature. For data that does not align with the concepts, thematic analysis is used to identify further concepts, which are then added to the framework, modifying it. A final step is to explore the relationships between the concepts in the revised framework.
There are different approaches to selecting the a priori framework for best fit synthesis, including identifying a single framework that may be the ‘obvious choice’ for the question or topic, or a systematic search for candidate published frameworks and synthesis to achieve a single ‘meta-framework’.
As mentioned above, this review will use the CFIR (27) as the a priori framework to synthesise evidence on the implementation of the PCMH. With 39 constructs (theoretical concepts), it is the obvious choice for studying the implementation of a complex intervention like the PCMH, and it is also a meta-framework, incorporating 19 theories of implementation. Its constructs are organised into five domains (Table 1).
Table 1
Consolidated Framework for Implementation Research (CFIR) description of domains
Domain
|
Description
|
Examples
|
Characteristics of the intervention
|
Attributes of the intervention being implemented
|
Strength of evidence, complexity
|
Outer setting
|
Economic, political and social context
|
External policy and incentives
|
Inner setting
|
Organisational structure, politics and culture
|
Implementation readiness, culture
|
Characteristics of individuals
|
Attributes of the individuals adopting the intervention
|
Knowledge and beliefs about the intervention
|
Processes
|
Steps involved in the implementation
|
Planning, engaging
|
Source: Damschroder, Aron (27)
Aims of the review
The review will aim to answer the question: What factors affect the implementation of the PCMH model in primary care practices and how? In answering this question, the review will also identify strategies that primary care practices have used in implementing the PCMH model to overcome barriers and to smooth transition to the model.
The outcome will be a refined theoretical model of implementing organisational change to specifically apply to implementing the PCMH in primary care.
Stakeholder group
We will establish stakeholder group to provide input to the review, comprised of a patient, a general practitioner, a practice manager, and a practice facilitator (a role assisting primary care practices in redesigning business and clinical processes). These are roles that are either likely to use the tool arising from the review or be impacted by its use. The group will meet four times over the course of the review (Table 2).
Table 2
Planned meetings of the stakeholder group
#
|
Timing
|
Purpose
|
1
|
Design phase
|
Receive a briefing on the review and the nature of members’ involvement.
|
2
|
Before data extraction
|
Identify criteria to narrow the selection of included papers, if required.
Review the data to be extracted from the included studies.
|
3
|
During the thematic analysis
|
Sense check the work undertaken by the reviewers.
|
4
|
During data synthesis
|
Provide advice on the representation of the refined theoretical framework, including the diagrammatic representation and wording of the accompanying narrative.
|
Eligibility criteria
Types of studies
The review will include primary research using any study design (qualitative, quantitative, or mixed methods), published in peer-reviewed publications (selected due to having undergone external review). Primary research published in grey literature (including evaluation reports), thesis/dissertation or conference abstract or presentation will be excluded. In these instances, steps will be undertaken to find out if the study has been published in a peer-reviewed publication. Systematic reviews, editorials, commentary and opinion pieces will also be excluded.
Domain
The domain studied is the PCMH model of care. Papers will be accepted if they include a statement that the study is related to a practice's/health organisation’s/health system’s implementation of a practice-level model that includes: a formalised ongoing relationship between a physician and a patient, team-based care, patient-centred care, coordination, prompt access to care, preventative care and continuous quality improvement.
Participants
This review will focus on stand-alone primary care practices (that is, not part of a hospital) led by physicians (that is, not nurse-practitioners or allied health professionals). It will also focus on primary care practices providing generalist primary care services (that is, not focussing on a specific population or disease/condition). In relation to the latter, studies may report on the implementation of elements of the model for a specific population cohort (for example, older people or people with diabetes), but the implementation should involve practices providing services to a range of patients and across disease/condition groups. We will exclude studies involving specialist primary care services, such as paediatric primary care, behavioural/ mental health care, addiction services, cancer care, palliative care and geriatrics.
Intervention
Studies will be included if they report on factors affecting the implementation of the PCMH. The study need not be focussed on implementation, but to be included, it must include amongst its findings factors affecting implementation. The topic of study may be one or more features of the PCMH (such as team-based care or preventative care). However, to be included, the feature(s) must be related to a practice's/health organisation’s/health system’s implementation of the PCMH. Studies reporting on the uptake by a practice(s) of a feature associated with the PCMH, but not in relation to the implementation of the PCMH, will be excluded.
Comparators
There are no comparators for this review.
Outcomes
The primary outcome is factors affecting implementation of the PCMH. Factors are any determinants of implementation of a PCMH feature or the model. These will be classified as barriers and enablers.
Search strategy
Searches for this review will be through:
-
Ovid Medline [1946–present]
-
Ovid Embase [1974–present]
-
Web of Science Core Collection (including Science Citation Index and Social Science Citation Index) [1900–present]
-
CINAHL EBSCOHost [1982–present]
The first two have been selected as general health care databases, the third as a multi-disciplinary database, and the last one is a specialist database relating to nursing and allied health, where additional studies relating to implementation from the perspectives of these disciplines might be found.
Search terms will relate to the setting/population (primary care) and the phenomenon of interest (PCMH). Gaps arising from the database search will be addressed through snowballing (following up references to other primary studies mentioned in included studies), citation tracking (tracking subsequently published studies that cite the included studies, using databases such as Google Scholar and Scopus) and review of reference lists of systematic reviews of the PCMH model (relevant reviews will be set aside for this purpose during the title and abstract screening). Also, any relevant study protocols identified during the title and abstract screen will be followed up to check whether the study has been published subsequently.
Where databases have a controlled vocabulary or subject heading list, this will be used, alongside keywords representing the key concepts.
Sample search terms and a search strategy using Medline (Ovid) are in Additional file 2.
Study selection, data extraction and analysis
Selection
EndNote X9 will be used to manage citations returned from the searches and Covidence will be used for recording the outcomes of the title/ abstract screen and full-text review.
One reviewer will screen the titles and abstracts from papers returned from the search, and a second reviewer will independently screen 20% of the papers. Where there are disagreements, the reviewers will discuss and decide jointly.
All papers that make it through to the full text review will be reviewed independently by the two reviewers. Disagreements will again be resolved through discussion.
Depending on the number of studies selected for synthesis, to manage the review, further criteria may be applied to narrow the selection. For example, included studies may be rated on the 'richness' of the information that they contribute to the review question, and only conceptually rich studies selected for synthesis.
Data extraction and analysis
MaxQDA, a qualitative analysis software package, will be used to code segments of the included papers and enter additional variables related to each study. Data entered/ coded will include:
-
Information about the study. For example, author, year, study design, country, setting, participants.
-
Details about the PCMH model. For example, features of the PCMH model, feature(s) that the study is focusing on, number of primary care practices involved, staff types involved, any special populations such as disadvantaged groups, older people or people with specific health conditions.
-
Information related to quality assessment criteria. See below.
-
CFIR constructs. Factors affecting implementation of the PCMH relating to the constructs in the CFIR, organised into barriers and enablers, how the constructs work together to achieve implementation and strategies that practices have used to overcome barriers within specific constructs.
-
Other data. Other factors affecting the implementation of the PCMH element or model, not captured by the CFIR constructs. These will be analysed in a subsequent stage, described below.
One reviewer will code the data from the primary studies to the appropriate fields in MaxQDA and enter additional information about each study. The same reviewer will code any data relating to implementation that cannot be mapped to the CFIR constructs into a separate ‘miscellaneous’ category. The second reviewer will progressively review the coding of the first reviewer and suggest alternative or additional ideas for consideration, which will be discussed between the reviewers as the coding develops.
Thematic analysis will then be used to develop additional constructs from the data coded to the ‘miscellaneous’ category, using the approach detailed by Braun and Clarke (28). This will be done by one reviewer; however, another reviewer will sense check the work of the first reviewer and raise alternative views for discussion.
The unpopulated CFIR constructs will be used to identify where further primary research may be required and explore the potential for publication bias.
Quality assessment
The mixed methods appraisal tool (MMAT) (24, 25) will be used to assess the methodological quality of the included studies. The MMAT allows appraisal of heterogenous study designs within the single tool.
The results of the quality assessment will be used to test the synthesis (see ‘Testing the synthesis’). Lower quality studies will be ones without a ‘Yes’ response to one of the two screening questions for all study types, or two or less ‘Yes’ responses from the five related to the study design.
Data synthesis
The CFIR will be modified using the additional categories identified during the thematic analysis of the residual data not able to be mapped to the existing CFIR constructs. Narrative of the results of the synthesis will be accompanied by a diagrammatic representation of the revised framework, showing the differences from the CFIR.
Testing the synthesis
Carroll, Booth (25) suggest that a final step in a best fit framework synthesis is to test it by comparing the a priori framework with the framework arising from the synthesis. This means explaining any unpopulated constructs from the CFIR as well as explaining the added constructs/ refinements. Differences between the two may be due to the differences between the PCMH and implementations of other types of interventions (namely the types on which CFIR has been based) or publication bias.
Further testing of the synthesis will be through removing lower quality studies to see if their omission makes a difference to the new framework.