The meta-ethnography has been registered with the International Prospective Register of Systematic Reviews (PROSPERO); registration number CRD42020184343. Protocol development was informed by recent advances in meta-ethnographic theory and practice [33, 34, 36, 38-42]; and with reference to the Preferred Reporting Items for Systematic Review and Meta-analysis Protocols (PRISMA-P) checklist (see Additional File 1). Data from eligible papers will be synthesised following the seven phases (see Figure 1) outlined by Noblit and Hare (1988); however, meta-ethnography is not a linear process and these phases will likely overlap and repeat as the synthesis proceeds [36]. Findings will be evaluated using the Confidence in Evidence from Reviews of Qualitative Research (CERQual) approach [43]; and reported in accordance with the eMERGe reporting guidelines [44]. Ethical approval is not required for a synthesis of published peer-reviewed studies (http://www.hra-decisiontools.org.uk/ethics/).
Figure 1: Noblit and Hare’s (1988) seven phases of meta-ethnography
Theoretical perspective
We (HP, KM and NB) collaborated to prepare a National Institute for Health Research funding application with the intention to build on previous work [45, 46] and this has now been funded. HP has worked as a National Health Service Hospital Pharmacist for the past 15 years, specialising in the field of antimicrobials since 2009, and is now a National Institute for Health Research (NIHR) Clinical Doctoral Research Fellow. JF and NB are Medical Sociologists with extensive qualitative research and meta-ethnography experience. KM is a Professor of Medical Education and has broad research experience including qualitative methods and analysis. SR is an experienced information specialist.
Meta-ethnography is an inductive, highly interpretive approach [36, 40]. We will be endeavouring to develop new interpretations from first order data (primary study participant interpretations); and second order data (author interpretations of participant interpretations) presented in the primary studies. To broaden perspective and to support the development of insightful, practical theory we have formed a stakeholder group who will contribute to the synthesis. Stakeholders will include: a Consultant Surgeon, Anaesthetist and Microbiologist; a Hospital Pharmacist; and several patient representatives. Additionally, technical support will be provided by an Information Specialist (SR) and a Research Fellow (EC) from the NIHR Applied Research Collaboration South West Peninsula Patient and Public Involvement team.
Phase one: getting started
Formulating the research question
Our research question is: how and why do contextual factors act and interact to influence surgical APB in hospital settings? Several qualitative studies have explored the subject. However, a synthesis offers us a tool to understand this body of work more fully, more deeply and more convincingly [47] whereas other approaches might remove context and/or impede explanation [36]. Meta-ethnography is the preferred qualitative evidence synthesis method as it is systematic and has the potential to preserve interpretive properties from the primary studies [48]. Furthermore, it aims to develop conceptual understanding [36], which aligns with our intentions, and has been implemented previously to develop theory about antimicrobial prescribing interventions in general practice [35]. As far as we know, this will be the first meta-ethnography addressing surgical APB.
Phase two: deciding what is relevant
HP and an information specialist (SR) will develop a search strategy (see Appendix 1); and will systematically search eight databases (AMED, CINAHL, EMBASE, MEDLINE, MEDLINE-in-process, Web of Science, Cochrane Library and PsycINFO) from their inception to identify potentially relevant studies. This combination of databases will enable representation from a range of domains including medical and allied health professional research and clinical practice; sociology; psychology; and related disciplines. This is important as qualitative research is frequently catalogued outside the medical domain. The SPIDER tool [49] (see Table 1) has been used to provide structure for the search, although search terms will be individualised for each database. Because qualitative literature can be challenging to find [50], we will employ supplementary search methods [51] to identify additional suitable studies: (1) forwards and backwards citation searching using studies that meet the inclusion criteria for the meta-ethnography; and (2) we will contact experts in the field, including the authors of all included studies, to ask them to suggest any additional studies (including those in-press).
The purpose of the comprehensive search is to identify the relevant body of literature containing information on the contextual factors associated with surgical APB. A comprehensive approach has been chosen as: (1) it will ensure that all relevant work is cited, to facilitate the development of theory and to prevent unhelpful research repetition/waste; and (2) it is more likely to resonate with our target audience (surgical and stewardship teams) who are more familiar with quantitative systematic reviews. Scoping searches suggest that the volume of applicable literature will be manageable. However, should the number of studies uncovered become unwieldy we will use a purposive or theoretical sampling strategy in keeping with the epistemology of meta-ethnography [50].
Table 1: SPIDER table of study inclusion and exclusion criteria
|
Inclusion criteria
|
Exclusion criteria
|
Sample
|
· Surgical teams (any members including surgeons, trainee surgeons, anaesthetists, surgical nurses, surgical pharmacists etc.)
· Secondary care setting including wards; out-patient clinics; theatres etc.
|
· Non-surgical specialities
· Other care settings e.g. primary care; dentists
· Veterinary studies
|
Phenomenon of interest
|
· Antimicrobial/antibiotic prescribing behaviour (treatment and/or prophylaxis)
|
· Prescribing behaviour related to other medication classes
|
Design
|
· Qualitative or mixed method studies reporting primary qualitative data collected using qualitative methods (e.g. through direct observation; focus groups; or interviews)
|
· Studies that report quantitative data only including questionnaire studies with open-ended free text questions
|
Evaluation
|
· Qualitative analysis of antimicrobial prescribing behaviour (using any qualitative evaluation e.g. grounded theory; and framework analysis)
|
· Studies that evaluate using quantitative methods only
· Studies that do not explicitly state the method of analysis
|
Research type
|
· Peer-reviewed journal articles
· Full text available
· English language
|
· Reviews; protocols; theoretical work; editorials; opinion pieces; and grey literature
· Non-English language
|
Inclusion and exclusion decisions
All retrieved studies will be imported into Endnote reference management software and de-duplicated. HP and SR will then independently screen the studies based on the title and abstract.
Studies will be excluded if they do not have a qualitative component or do not describe APB in a surgical context (see Table 1). In the event of uncertainty or disagreement studies will be sought in full, in addition to the studies that definitely appear relevant, to be assessed by HP and one other author (KM or JF). Studies will be included if they use recognised qualitative methods (e.g. interviews, focus groups or observation) and analysis (e.g. framework analysis or thematic analysis); focus on contextual factors associated with surgical (any speciality) APB within a hospital setting; and are available in full in the English language. Any disagreement at the final screening stage will be resolved by consensus between three reviewers (HP, JF and KM).
Quality assessment
No studies will be excluded based on quality alone [52]. However, all studies will be assessed using the qualitative Critical Appraisal Skills Programme (CASP) tool [53] to support careful and systematic reading [40] with consideration of a range of aspects [41]. Lower quality assessment scores – for example, due to poor reporting or abridged methods sections (often the case in medical journals) – does not always reflect the quality of the research; however, it can draw the reviewer’s attention to shortcomings in the interpretation of study findings that may have an impact on the results of the synthesis [38].
Three reviewers (HP, KM and JF) will then use a pragmatic approach, first described by Dixon-Woods et al. (2007) [54], to classify studies based on their perceived utility to the meta-ethnography (see Table 2). Those studies deemed ‘irrelevant’ or ‘fatally flawed’ will be excluded. Remaining papers – key papers; satisfactory papers; and questionable papers – will be included in the synthesis. Any disagreement regarding categorisation of a study will be resolved by consensus between the three reviewers. Additionally, synthesis messages derived from the included studies will be examined against ‘key’ papers (only) to test their contributions and promote further discussion and insight, consistent with previous work [41]. A Microsoft Excel spreadsheet will be used to collate study demographics; appraisal scores; and inclusion/exclusion decisions.
Table 2: Study classifications [54]
Category
|
Study characteristics
|
Key papers
|
Conceptually rich with the potential to make an important contribution to the synthesis
|
Satisfactory papers
|
Less valuable than key papers but still relevant
|
Questionable papers
|
Uncertain contribution
|
Irrelevant
|
Not relevant to the review question
|
Fatally flawed
|
Study data not presented in a usable format
|
|
Phase three: reading the studies
This stage of the meta-ethnography involves repeated careful reading of the studies to gain familiarity and to identify the main concepts described i.e. what is each study telling us. Contextual information, such as study setting; participants (e.g. sub-speciality, grade and number included); research design; and aim, will be recorded in a Microsoft Excel spreadsheet. Following repeated close reading (HP, KM and JF), the studies will be imported into NVivo qualitative data analysis software. Key concepts (potentially explanatory ideas) from each study’s results and/or discussion section will then be independently coded by two reviewers (HP and JF or KM). The coded data will include quotations from participants (first-order data); and quotations from the original study’s authors (second-order data). Reviewers (HP, JF and KM) will then discuss and agree the key concepts, recording them in a Microsoft Excel spreadsheet.
Phase four: determining how the studies are related
The next phase requires us to identify the relationship between the primary studies, and their key concepts, through a process of close comparison. We will begin to determine whether the synthesis is ‘reciprocal’ (primary studies’ concepts are directly complementary), ‘refutational’ (primary studies’ concepts oppose each other) or ‘lines-of-argument’. In the latter case, primary studies identify different aspects of a larger phenomenon which when taken together offer a new interpretation; a ‘whole’ is discovered from a set of parts [36]. We are aware that meta-ethnographies frequently produce reciprocal or lines-of-argument translations [55]. Lines of argument syntheses often bring together interrelated concepts, but may also represent lack of attention to conflicting findings. We will actively seek disconfirming or contradictory findings and concepts.
Phase five: translating the studies into one another
Translation involves sorting the key concepts (from primary studies) into conceptual categories or ‘piles’ [42]. HP, KM and JF will independently compare key concepts across the primary studies, grouping them all into conceptual categories (third order data) with a definition of what each conceptual category encompasses. The conceptual categories will be developed inductively, through a process of constant comparison (of key concepts), rather than according to any a priori theory, although we recognise each reviewer’s interpretations will be influenced by their backgrounds. The reviewers will then compare interpretations and, with input from stakeholders, collaboratively develop a final list of conceptual categories which will be tested against the primary studies to ensure a good fit. The multi-disciplinary input will enable us to challenge our own understandings and will support the identification of a range of possible analytic interpretations.
Phase six: synthesising translations
Synthesising translations is the on-going process whereby findings are further abstracted to form a conceptual framework [42] which explains the phenomena of interest. It cannot be reduced to a set of mechanistic tasks [39] but will involve three reviewers (HP, KM and JF) working collaboratively to ‘make sense’ of the conceptual categories, with the aim of developing a new theory that explains how and why contextual factors act and interact to influence APB among surgical teams. If appropriate, a visual way of representing the findings will be developed iteratively, to convey the theory.
Emerging interpretations will be discussed with: (1) the authors of the primary studies (where possible) to test the validity of our third-order interpretations; (2) academic and surgical/stewardship audiences to receive feedback, for example at departmental seminars, conferences and methodological discussion fora; and (3) our wider stakeholder group to ensure that the knowledge is applicable and meaningful. Additionally, we will assess the synthesis findings using the CERQual approach [43] to transparently determine how much confidence can be placed in them.
Phase seven: expressing the synthesis
Findings from the meta-ethnography will be published in a peer-reviewed journal; presented at suitable fora (including surgical and infection conferences and relevant teaching); and made available to patients and members of the public. We will work with key stakeholders and patient representatives to ensure that the media are acceptable to their target audience and disseminated via the most effective channels.