The number of Qualitative Evidence Syntheses (QES) that aim to systematically search for and synthesise the findings from qualitative research is increasing. In 2011, Campbell and colleagues identified 41 syntheses [1]. Other reviews suggest that the number is much larger than this and that it is increasing dramatically [2, 3]. For example, Hannes and colleagues demonstrate that the number of qualitative syntheses in 2008 had doubled within four years [3]. QES is useful in health research because the proliferation of qualitative studies can sometimes make it difficult for stakeholders to utilise qualitative knowledge to inform practice and policy [4, 5]. Insights from several QES have contributed to a greater understanding of complex processes in healthcare. A few examples are: medicine taking [6], diabetes [7] antidepressants [8], chronic musculoskeletal pain [9] and chronic pelvic pain [10].

In some areas, there is a growing number of QES exploring the same or similar topics. The increasing number of QES [3, 11] with no clear way of identifying existing or planned QES raises the danger of research duplication. It also raises the question for stakeholders, which synthesis do I use? For example, we became aware of the growing number of syntheses, including our own, that explored the experience of living with chronic pain. Framed in a more positive light, an increase in QES might provide an opportunity for useful synthetic products for the purposes of policy and practice. These syntheses of QES would be useful in order to aggregate existing findings or, alternatively, to move our conceptual understanding forward. This reflects a distinction often made between QES that (a) aggregate, or (b) develop, conceptual understandings [2–4, 12–14]. Frost and colleagues explore the possibility of synthesising QES of diabetes [15]. They indicate that there has been a move away from interpretation and theory development in QES towards aggregative forms of synthesis. They caution reviewers not so simply produce ‘more of the same’ [15]. We aimed to explore whether a synthesis of QES had the potential to add a conceptual level that was greater than the sum of its QES parts. We felt it would be valuable to synthesise existing QES into a conceptual output to help us to more fully understand what it is like to live with chronic non-malignant pain.

Meta-ethnography is a method of QES that focuses on conceptualisation and theory development. As such, it is not necessarily recommended for decision-making contexts such as guideline or policy recommendations. Our aims were both substantive and methodological: (1) to bring together QES that explored patients’ experience of living with non-malignant chronic pain and develop conceptual understanding of what it is like to live with chronic non-malignant pain for improved healthcare; (2) to undertake the first mega-ethnography of QES using the methods of meta-ethnography. This is important as we know that many people living with chronic non-malignant pain continue to experience healthcare as an adversarial struggle [9]. The innovation of this study is to bring together eleven QES exploring patients experience of living with chronic non-malignant pain using the methods of meta-ethnography [12, 16]. To our knowledge, this is the first meta-ethnography of QES and we have coined the phrase ‘mega-ethnography’.

We used the methods of meta-ethnography developed, refined and reported in a previous synthesis of patients’ experience of chronic musculoskeletal pain [9] and applied these methods to produce a mega-ethnography of QES.

Stage 4: Determining how studies are related to each other

For meta-ethnography, determining how studies are related involves creating ‘a list of key metaphors, phrases, ideas and/or concepts’ [12] (page 28) from primary research studies and comparing these across studies. Schütz’ concept of first and second order constructs are frequently used in meta-ethnography studies to distinguish the levels of data [21]. Schütz distinguishes (1) first-order constructs (the participants’ ‘common sense’ interpretations in their own words) and (2) second order constructs (the primary researchers’ interpretations based on first order constructs). The ‘data’ of meta-ethnography are the primary research findings which are second order constructs. These findings are further abstracted to develop QES findings which are third order constructs (the reviewers’ interpretation of the primary authors’ interpretations). These QES findings are the data of mega-ethnography which are further abstracted to develop fourth order constructs (the mega-reviewers’ interpretation of reviewers’ interpretations). This is illustrated in Fig. 1. For mega-ethnography, the focus is still on conceptualising rather than describing. Two reviewers challenged each other’s interpretation of the QES findings in order to remain confident that their interpretation remained grounded in that study [16]. Any disagreements were discussed and resolved with a third reviewer. Once we had agreed upon a description of each QES finding, FT wrote a statement of this finding in the first person. For example:

‘This theme described how their whole life and future as they knew it had been disrupted’, becomes:

‘My whole life and future as I know it has been disrupted’.

We have found that writing concepts in the first person is a powerful way for reviewers and QES stakeholders to fully engage in the meaning and sentiment of each concept. It also facilitates the use of accessible language for a diverse audience in both analysis and dissemination.

The final stage of meta-ethnography is to synthesise the conceptual categories into a line of argument that helps us to further understand the phenomenon of interest (Fig. 3). Through our discussions we agreed upon the imperative of the diagnostic ‘holy grail’ in people’s experience of living with chronic non-malignant pain. The quest for the diagnostic ‘holy grail’ is instigated by pain invading all aspects of a person’s life, both now and in the future (life is impoverished and confined). The person struggles against their body in order to try and keep hold of a sense of self for themselves and others (struggling against my body to be me). They thus take up a quest to find out what is wrong with their body so that it can be fixed (a quest for a diagnostic ‘holy grail’). They fall short of this quest and find nothing to prove that pain is real. People no longer believe them and they are robbed of personal credibility (lost personal credibility). They therefore try harder to keep up appearances and hide the changes so that they are not judged harshly (trying to keep up appearances). They need to be treated with dignity and to find someone in healthcare who makes them feel worthy of personal respect and who treats them as an embodied individual (need to be treated with dignity). Even if they realise that they must learn to live alongside pain in the absence of a diagnosis or cure, giving up the quest is easier said than done (deciding to end the quest for the grail is not easy).

Our innovation is to use the methods of meta-ethnography modified by Toye and colleagues [16] to synthesise findings from eleven QES that explore patients’ experience of living with chronic non-malignant pain. To our knowledge this is the first synthesis of QES using the methods of meta-ethnography, and we have coined the phrase, mega-ethnography. An overriding benefit of QES is that it synthesises a large body of qualitative data and makes the findings available for practise, policy and education. Our mega-ethnography synthesises concepts from eleven QES which were drawn from 187 primary qualitative studies (Additional file 2) to provide a conceptual understanding for the benefits of healthcare. The number of QES is increasing exponentially [3, 11] and whilst there is no means of identifiying existing or planned QES, duplication is likely. We felt it would be valuable to synthesise existing QES into a conceptual output to help us understand what it is like to live with chronic non-malignant pain. Meta-ethnography is a method of QES that focuses on theory development and is therefore not necessarily recommended for use in decision-making contexts such as evidence to recommendation guideline processes. More methodological testing is required to demonstrate the application of meta-ethnography in these contexts.

Some methods of QES (including meta-ethnography) abstract their individual findings into a line of argument or model. However, this is not standard amongst QES approaches. For example, some present individual findings and not a line of argument [49, 50, 52, 53, 55–57], whereas others presented a line of argument [9, 10, 51, 54]. We therefore used individual QES findings as the data for this mega-ethnography rather than their lines of argument. Both individual QES findings and lines of argument are third level constructs, although lines of argument may present a higher level of abstraction.

Our focus was on conceptual development and, although we excluded one QES where it was not possible to link QES findings to specific primary studies [46], we did not formally appraise QES quality. Although there have been calls to standardise the reporting of QES [17–19], qualitative analysis is underpinned by an interpretive framework and therefore efforts to regulate are likely to prove challenging. However, we cannot ignore the fact that some QES are indeed better (methodologically and conceptually) than others. For quantitative reviews of the effectiveness of interventions, there is now an established method for evaluating confidence in review (http://gradeworkinggroup.org/). In contrast, attempts to establish confidence in QES findings are in their infancy [5, 20]. There are two proposed method for appraising the confidence in QES which offer overlapping but distinct approaches. For QES undertaken alongside quantitative reviews of interventions, Lewin and colleagues propose four areas that allow us to determine ‘Confidence in the Evidence from Reviews of Qualitative Research (GRADE-CERQual): (1) Methodological limitations are the ‘extent to which there are problems in the design or conduct of the primary studies’; (2) Relevance is the extent to which the evidence is applicable to the QES question; (3) Adequacy of data is the determination of richness or weight of data supporting a finding. (4) Coherence considers the consistency (or inconsistency) of a particular finding across the primary studies. Munn and colleague propose two alternative but overlapping criteria for establishing ‘confidence in the output of qualitative research synthesis’ (ConQual): (1) Dependability aligns with the ‘methodological limitations’ of GRADE-CERQual; (2) Credibility, is a global evaluation of ‘fit’ between the primary data and the reviewers’ interpretations as demonstrated by adequate exemplars. This resonates with concept-indicator fit [58].

One of the challenges of attempts to determine confidence in QES is the limited agreement about what determines a ‘good’ primary qualitative study [24, 59–61]. Some feel that useful studies might be excluded from QES if our primary concern is methodological reporting rather than conceptual insight [1, 24]. It might also mean that studies embedded in other research disciplines, such as anthropology, are overlooked. Thus, a significant number of qualitative reviewers legitimately choose not to appraise for the purpose of QES [1]. This choice does not imply low quality. Indeed, although quality appraisal might help us to recognise methodological flaws, it does not necessarily help us to appraise the conceptual value of findings.

GRADE-CERQual considers adequacy and coherence of data to contribute to confidence in QES findings. Adequacy regards the vertical depth of data supporting each finding and coherence considers consistency (and difference) across the horizontal spread of data. Both constructs contribute to the gravity, or central pull of a developing idea. However, it is important to consider that gravity has a qualitative component, and that a singular idea can exert a strong gravitational pull. We should be cautious about truth-claims based purely on sheer weight or consistency of data; remember that, ‘a single word . . . can jog our memory or spark off insight where we had not expected it’ [24]. The Tale of the Emperor’s New Clothes has taught us about the validity of a small voice in the crowd. In order to demonstrate the depth and breadth of a finding, although it can be useful to provide a tally of ideas or studies supporting the finding, a single idea can stimulate a new way of thinking or highlight possible areas for further research. There is no agreed way to determine how much data provides enough gravity to support the validity of an idea. One idea may be weightier than ten other ideas. However, it is important to consider that CERQual was designed to apply to QES that are going to be used in a decision-making contexts such as a guideline panel. Further research would usefully explore the application of CERQual to conceptual QES approaches such as meta-ethnography. Although you could argue that issues of scale are not applicable to qualitative methodologies, it would be useful to consider these issues. For example, how much qualitative knowledge is adequate to support a policy decision? This issue resonates with the unresolved, even unresolvable, question ‘how many qualitative interviews is enough?’ [62]. From our experience of QES we would argue that there is confidence to be gained from performing systematic searches and incorporating a large number of ideas for comparison. However, qualitative researchers can find themselves caught between a rock and a hard place as they face criticisms for undertaking studies that are ‘too small’ (thus anecdotal) or ‘too large’ (thus not in-depth). We argue that large and in-depth studies are possible for conceptual development. Further thought should be given to the place of small idea provoking studies for policy contexts.

One of the criticisms of large QES is that it is possible to lose sight of the nuances of the primary studies. This criticism is potentially amplified in a mega-QES. We identified 78 QES findings that explored the experience of living with chronic non-malignant pain. At this point, the data has gone through a series of abstractions from first to fourth order construct. Some of the subtle nuances of the original QES may therefore have been lost in abstraction. For example, Toye and colleagues theme, ‘Moving forward alongside my pain’ [9] originally incorporated several sub-themes (integrating my painful body; redefining normal; being part of a community; being open about my pain; realising that there is no cure; becoming expert). These themes became abstracted along with other QES findings into the idea, ‘deciding to end the quest for the diagnostic holy grail’. We did not use either of our studies [9, 10] as an index paper to ‘orient[ate] the synthesis’ [23]. Although it is unrealistic to claim that our interpretations were not influenced by our own ideas, we made every effort to keep an open mind and utilise our existing ideas as sensitising rather than definitive; ‘definitive concepts provide prescriptions of what to see, sensitising concepts merely suggest directions along which to look’ [63] (p. 7). It is useful to consider conceptual qualitative analysis as a dialectic, rather than linear process where tension between existing and new ideas can create innovative ways of thinking [26].

Although qualitative research philosophy emphasises the idiographic nature of knowledge and focuses on the unique contextual experience, we argue that it also has the power to say something valuable that is transferable beyond its context. Thus, we found that further abstraction brought conceptual gains. Importantly, the imperative of ‘the diagnostic holy grail’ helps us to understand that investing in the healing powers of diagnosis can leave patient’s vulnerable [64]; if a diagnosis is not made the person in pain can suffer the consequences of lost personal credibility. This applies in situations where the person has either failed, or given up, the quest for a diagnosis. A decision to end the quest is therefore not taken lightly and patients may need support to make this decision. The way that we explain illness is embedded in cultural understanding and thus underpinned by powerful emotions [65].Our findings demonstrate that biomedical model is a culturally embedded model that is integral to a person’s credibility, and this may explain why it is so difficult to give up the quest for a diagnostic ‘holy grail’.

The innovation of this study is to present the first mega-ethnography of QES to develop our conceptual understanding of what it is like to live in pain. Findings demonstrate that people with chronic pain are fundamentally altered by pain and often feel that health care professionals do not hear or believe them. This finding has implications for practice as lost credibility may encourage a continuing diagnostic quest. Patients may need support to make the decision to cease the continuing search for a diagnosis and cure. It would be useful for health care professionals to take affirmative action to reaffirm patients’ personal credibility through listening, hearing and believing. This part of clinical work is not an adjunct to, but integral to effective health care.