Background
The systematic review is an important technology for the evidence-informed policy and practice movement, which aims to bring research closer to decision-making [
1,
2]. This type of review uses rigorous and explicit methods to bring together the results of primary research in order to provide reliable answers to particular questions [
3‐
6]. The picture that is presented aims to be distorted neither by biases in the review process nor by biases in the primary research which the review contains [
7‐
10]. Systematic review methods are well-developed for certain types of research, such as randomised controlled trials (RCTs). Methods for reviewing qualitative research in a systematic way are still emerging, and there is much ongoing development and debate [
11‐
14].
In this paper we present one approach to the synthesis of findings of qualitative research, which we have called 'thematic synthesis'. We have developed and applied these methods within several systematic reviews that address questions about people's perspectives and experiences [
15‐
18]. The context for this methodological development is a programme of work in health promotion and public health (HP & PH), mostly funded by the English Department of Health, at the EPPI-Centre, in the Social Science Research Unit at the Institute of Education, University of London in the UK. Early systematic reviews at the EPPI-Centre addressed the question 'what works?' and contained research testing the effects of interventions. However, policy makers and other review users also posed questions about intervention need, appropriateness and acceptability, and factors influencing intervention implementation. To address these questions, our reviews began to include a wider range of research, including research often described as 'qualitative'. We began to focus, in particular, on research that aimed to understand the health issue in question from the experiences and point of view of the groups of people targeted by HP&PH interventions (We use the term 'qualitative' research cautiously because it encompasses a multitude of research methods at the same time as an assumed range of epistemological positions. In practice it is often difficult to classify research as being either 'qualitative' or 'quantitative' as much research contains aspects of both [
19‐
22]. Because the term is in common use, however, we will employ it in this paper).
When we started the work for our first series of reviews which included qualitative research in 1999 [
23‐
26], there was very little published material that described methods for synthesising this type of research. We therefore experimented with a variety of techniques borrowed from standard systematic review methods and methods for analysing primary qualitative research [
15]. In later reviews, we were able to refine these methods and began to apply thematic analysis in a more explicit way. The methods for thematic synthesis described in this paper have so far been used explicitly in three systematic reviews [
16‐
18].
The review used as an example in this paper
To illustrate the steps involved in a thematic synthesis we draw on a review of the barriers to, and facilitators of, healthy eating amongst children aged four to 10 years old [
17]. The review was commissioned by the Department of Health, England to inform policy about how to encourage children to eat healthily in the light of recent surveys highlighting that British children are eating less than half the recommended five portions of fruit and vegetables per day. While we focus on the aspects of the review that relate to qualitative studies, the review was broader than this and combined answering traditional questions of effectiveness, through reviewing controlled trials, with questions relating to children's views of healthy eating, which were answered using qualitative studies. The qualitative studies were synthesised using 'thematic synthesis' – the subject of this paper. We compared the effectiveness of interventions which appeared to be in line with recommendations from the thematic synthesis with those that did not. This enabled us to see whether the understandings we had gained from the children's views helped us to explain differences in the effectiveness of different interventions: the thematic synthesis had enabled us to generate hypotheses which could be tested against the findings of the quantitative studies – hypotheses that we could not have generated without the thematic synthesis. The methods of this part of the review are published in Thomas
et al. [
27] and are discussed further in Harden and Thomas [
21].
Qualitative research and systematic reviews
The act of seeking to synthesise qualitative research means stepping into more complex and contested territory than is the case when only RCTs are included in a review. First, methods are much less developed in this area, with fewer completed reviews available from which to learn, and second, the whole enterprise of synthesising qualitative research is itself hotly debated. Qualitative research, it is often proposed, is not generalisable and is specific to a particular context, time and group of participants. Thus, in bringing such research together, reviewers are open to the charge that they de-contextualise findings and wrongly assume that these are commensurable [
11,
13]. These are serious concerns which it is not the purpose of this paper to contest. We note, however, that a strong case has been made for qualitative research to be valued for the potential it has to inform policy and practice [
11,
28‐
30]. In our experience, users of reviews are interested in the answers that only qualitative research can provide, but are not able to handle the deluge of data that would result if they tried to locate, read and interpret all the relevant research themselves. Thus, if we acknowledge the unique importance of qualitative research, we need also to recognise that methods are required to bring its findings together for a wide audience – at the same time as preserving and respecting its essential context and complexity.
The earliest published work that we know of that deals with methods for synthesising qualitative research was written in 1988 by Noblit and Hare [
31]. This book describes the way that ethnographic research might be synthesised, but the method has been shown to be applicable to qualitative research beyond ethnography [
32,
11]. As well as meta-ethnography, other methods have been developed more recently, including 'meta-study' [
33], 'critical interpretive synthesis' [
34] and 'metasynthesis' [
13].
Many of the newer methods being developed have much in common with meta-ethnography, as originally described by Noblit and Hare, and often state explicitly that they are drawing on this work. In essence, this method involves identifying key concepts from studies and translating them into one another. The term 'translating' in this context refers to the process of taking concepts from one study and recognising the same concepts in another study, though they may not be expressed using identical words. Explanations or theories associated with these concepts are also extracted and a 'line of argument' may be developed, pulling corroborating concepts together and, crucially, going beyond the content of the original studies (though 'refutational' concepts might not be amenable to this process). Some have claimed that this notion of 'going beyond' the primary studies is a critical component of synthesis, and is what distinguishes it from the types of summaries of findings that typify traditional literature reviews [e.g. [
32], p209]. In the words of Margarete Sandelowski,
"metasyntheses are integrations that are more than the sum of parts, in that they offer novel interpretations of findings. These interpretations will not be found in any one research report but, rather, are inferences derived from taking all of the reports in a sample as a whole" [[
14], p1358].
Thematic analysis has been identified as one of a range of potential methods for research synthesis alongside meta-ethnography and 'metasynthesis', though precisely what the method involves is unclear, and there are few examples of it being used for synthesising research [
35]. We have adopted the term 'thematic synthesis', as we translated methods for the analysis of primary research – often termed 'thematic' – for use in systematic reviews [
36‐
38]. As Boyatzis [[
36], p4] has observed, thematic analysis is
"not another qualitative method but a process that can be used with most, if not all, qualitative methods...". Our approach concurs with this conceptualisation of thematic analysis, since the method we employed draws on other established methods but uses techniques commonly described as 'thematic analysis' in order to formalise the identification and development of themes.
We now move to a description of the methods we used in our example systematic review. While this paper has the traditional structure for reporting the results of a research project, the detailed methods (e.g. precise terms we used for searching) and results are available online. This paper identifies the particular issues that relate especially to reviewing qualitative research systematically and then to describing the activity of thematic synthesis in detail.
Results
Six main issues emerged from the studies of children's views: (1) children do not see it as their role to be interested in health; (2) children do not see messages about future health as personally relevant or credible; (3) fruit, vegetables and confectionery have very different meanings for children; (4) children actively seek ways to exercise their own choices with regard to food; (5) children value eating as a social occasion; and (6) children see the contradiction between what is promoted in theory and what adults provide in practice. The review found that most interventions were based in school (though frequently with parental involvement) and often combined learning about the health benefits of fruit and vegetables with 'hands-on' experience in the form of food preparation and taste-testing. Interventions targeted at people with particular risk factors worked better than others, and multi-component interventions that combined the promotion of physical activity with healthy eating did not work as well as those that only concentrated on healthy eating. The studies of children's views suggested that fruit and vegetables should be treated in different ways in interventions, and that messages should not focus on health warnings. Interventions that were in line with these suggestions tended to be more effective than those which were not.
Conclusion
This paper has discussed the rationale for reviewing and synthesising qualitative research in a systematic way and has outlined one specific approach for doing this: thematic synthesis. While it is not the only method which might be used – and we have discussed some of the other options available – we present it here as a tested technique that has worked in the systematic reviews in which it has been employed.
We have observed that one of the key tasks in the synthesis of qualitative research is the translation of concepts between studies. While the activity of translating concepts is usually undertaken in the few syntheses of qualitative research that exist, there are few examples that specify the detail of how this translation is actually carried out. The example above shows how we achieved the translation of concepts across studies through the use of line-by-line coding, the organisation of these codes into descriptive themes, and the generation of analytical themes through the application of a higher level theoretical framework. This paper therefore also demonstrates how the methods and process of a thematic synthesis can be written up in a transparent way.
This paper goes some way to addressing concerns regarding the use of thematic analysis in research synthesis raised by Dixon-Woods and colleagues who argue that the approach can lack transparency due to a failure to distinguish between 'data-driven' or 'theory-driven' approaches. Moreover they suggest that,
"if thematic analysis is limited to summarising themes reported in primary studies, it offers little by way of theoretical structure within which to develop higher order thematic categories..." [[
35], p47]. Part of the problem, they observe, is that the precise methods of thematic synthesis are unclear. Our approach contains a clear separation between the 'data-driven' descriptive themes and the 'theory-driven' analytical themes and demonstrates how the review questions provided a theoretical structure within which it became possible to develop higher order thematic categories.
The theme of 'going beyond' the content of the primary studies was discussed earlier. Citing Strike and Posner [
59], Campbell
et al. [[
11], p672] also suggest that synthesis
"involves some degree of conceptual innovation, or employment of concepts not found in the characterisation of the parts and a means of creating the whole". This was certainly true of the example given in this paper. We used a series of questions, derived from the main topic of our review, to focus an examination of our descriptive themes and we do not find our recommendations for interventions contained in the findings of the primary studies: these were new propositions generated by the reviewers in the light of the synthesis. The method also demonstrates that it is possible to synthesise without conceptual innovation. The initial synthesis, involving the translation of concepts between studies, was necessary in order for conceptual innovation to begin. One could argue that the conceptual innovation, in this case, was only necessary because the primary studies did not address our review question directly. In situations in which the primary studies are concerned directly with the review question, it may not be necessary to go beyond the contents of the original studies in order to produce a satisfactory synthesis (see, for example, Marston and King, [
60]). Conceptually, our
analytical themes are similar to the ultimate product of meta-ethnographies:
third order interpretations [
11], since both are explicit mechanisms for going beyond the content of the primary studies and presenting this in a transparent way. The main difference between them lies in their purposes.
Third order interpretations bring together the implications of translating studies into one another in their own terms, whereas
analytical themes are the result of interrogating a descriptive synthesis by placing it within an external theoretical framework (our review question and sub-questions). It may be, therefore, that
analytical themes are more appropriate when a specific review question is being addressed (as often occurs when informing policy and practice), and
third order interpretations should be used when a body of literature is being explored in and of itself, with broader, or emergent, review questions.
This paper is a contribution to the current developmental work taking place in understanding how best to bring together the findings of qualitative research to inform policy and practice. It is by no means the only method on offer but, by drawing on methods and principles from qualitative primary research, it benefits from the years of methodological development that underpins the research it seeks to synthesise.
Acknowledgements
The authors would like to thank Elaine Barnett-Page for her assistance in producing the draft paper, and David Gough, Ann Oakley and Sandy Oliver for their helpful comments. The review used an example in this paper was funded by the Department of Health (England). The methodological development was supported by Department of Health (England) and the ESRC through the Methods for Research Synthesis Node of the National Centre for Research Methods. In addition, Angela Harden held a senior research fellowship funded by the Department of Health (England) December 2003 – November 2007. The views expressed in this paper are those of the authors and are not necessarily those of the funding bodies.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
Both authors contributed equally to the paper and read and approved the final manuscript.