A content analysis of practice and experience

Focus Group Discussions (FGDs) are a common way of gathering qualitative data in Aboriginal health services research; however there have been no studies on the question of whether they are appropriate research tools in such contexts, nor are there are specific guidelines available to ensure that FGDs are delivered to collect data in ways that are consistent with Aboriginal approaches to consultation, ownership and ways of knowing. Furthermore, there is a lack of clarity concerning the theoretical and methodological perspectives that could be operationalised by FGDs to gather data, guide analysis and interpretation in ways that are culturally appropriate, ethically sound and rigorous. We undertook a content analysis of Aboriginal health services research studies using FGDs to determine their use and elements that may provide insight into good practice. A framework is proposed to help guide future FGD research with Aboriginal people.


Introduction
Focus group discussion (FGD) is a recognised qualitative research tool (Morgan et al., 2008) first documented in 1926 (Bogardus, 1926) for gathering large amounts of narrative data arising from informal, moderated interaction between a small group of six to 12 individuals.FGDs can be broadly defined as involving various degrees of structured or semi structured question formats and semidirected interviewing styles, also known as 'focused interviews' (Merton and Kendall, 1946) and 'focus groups' (Morgan, 1996), and the 'group discussion' (Willis, 1990), that has been convened for the purposes of gathering research data.
In response to the consumer movement in health care, organisations have increasingly drawn upon FGDs as a means of gathering direct feedback from clients and community members on quality and acceptability (Dagger et al., 2007) for identifying salient issues for policy development (Taylor et al., 2011) and as forums for change (Race et al., 1994).FGDs provide an important opportunity for participants to collaborate with researchers (Goss and Leinbach, 1996) and can be empowering by enabling participants to explore solutions to issues (Pini, 2002).FGDs can provide important information to assist in the design and evaluation of health interventions for Aboriginal1 Australians who have not always been in a position to ensure their input into planning health services (Mayo and Tsey, 2009).
Despite the importance of the FGD method, few published materials guide the practical application of its use with Aboriginal Australian people.Some authors have considered related matters in research whose primary purpose concerned subjects other than an analysis of FGD methods (Barnett and Kendall, 2011), FGD research with culturally diverse groups other than Aboriginal people (Huer and Saenz, 2003) or generic tools to guide the quality delivery and reporting of FGD research (Onwuegbuzie et al., 2009, Tong et al., 2007).The only investigation of FGDs in Aboriginal Australian settings examines the use of Aboriginal-led adaptations of focus group methods (Willis, 2005).While this is important, there remains a significant gap in our understanding of current approaches to FGDs in Aboriginal health settings, calling into question the substantial investments in time and funding made to not only carry out the research, but to translate the findings into improved health service delivery.
FGDs have been considered in other Indigenous contexts, including in Canadian and American First Nations' settings containing unique aspects of such research and the cultural adaptations necessary.Strickland (1999), for example, identifies various aspects that set FGD processes with First Nations people apart from other cultural groups.This includes the incorporation of cultural rituals, incentives, additional time for discussion, characteristics of participants and seating arrangements.Bartlett and others (2007) describe how audio recording or note-taking can appear inappropriate and intrusive in some Indigenous contexts.To overcome this, data collection in their study was based on the recall of the facilitator and a community elder following the focus group session.
A synthesis of current experience of the use of FDGs in Aboriginal Australian health services research is a useful starting point to provide insight into best practice approaches to conducting FGD research to ensure efficacious, rigorous, culturally appropriate and relevant research.This paper reports the findings of a review of primary research studies to examine evidence related to the factors that affect the use of FGDs in Aboriginal health service contexts.It involved an examination of FGD delivery methods, data collection analysis and interpretation in Aboriginal health service studies.Accordingly, the review sought to answer the following questions:

•
What types of Aboriginal health services research studies have used FGDs, how have they been conducted, with whom and for what purpose?
• What methodology underpins the use of FGD in such studies, how was data analysed, what inferences were drawn and how are these presented?
• How is FGD data treated in relation to other data gathered from Aboriginal people and other sources?
• How have the authors disseminated or shared the findings with Aboriginal people?

Method
An integrative literature review was undertaken using content analysis, a well-established method that allows study findings to be synthesised using a systematic and replicable method of categorising and counting themes (Dixon-Woods et al., 2005, Evans andFitzGerald, 2002).

Search Protocol
A Retrieved records were screened for their focus on service delivery with Aboriginal people as the study population.Papers that were duplicates, not data-based, older than 10 years, whose focus was not health service delivery or reported service delivery from the perspective of service providers were removed from the sample.The references of the screened search results were imported into EndNote (x7) and duplicate items removed.The first author screened all titles and abstracts (n = 110) to identify papers reporting empirical research findings.The inclusion or exclusion criteria developed by all authors to screen these papers (Table 1) 1 were applied.The database source of these papers is identified at Table 2.The PRISMA guidelines (Moher, 2009) were used to report process as shown below in Figure 1.

Quality Appraisal
All 23 papers were appraised to establish whether the research aim and the methodology used were aligned and to evaluate the recruitment, settings, data analysis, ethics, findings and contribution to knowledge.The qualitative papers were assessed for quality using the Critical Appraisal Skills Programme (CASP) tool for qualitative research (NHS, 2006) and the mixed methods papers were assessed using a scoring system designed for this purpose (Pluye et al., 2009).Two items were discarded as quality was deemed low with no mention of ethical processes and insufficient discussion of focus group analysis.The paper by Anjou et al. (2013a) was excluded as the method and data presented is included in another paper by these authors (Anjou et al., 2013b).The paper by Hayman et al. was also excluded as it lacked sufficient explanation of methods (Hayman et al., 2009).

Data Abstraction and Synthesis
A summative form of qualitative content analysis was employed to achieve synthesis.This involved describing and quantifying the data (Bryman, 2001) and carrying out further analysis to discover any underlying meaning (Hsieh and Shannon, 2005).This approach to content analysis has been used in other research involving textual analysis (Evans, 2013).The characteristics of the papers were first described.Keywords for the analysis were developed based on approaches to undertaking FGD, analysing the data described in the literature (Krueger, 2009, Tong et al., 2007) and through discussion with all authors.A template was drawn up to assist with describing data and identifying patterns.A structured search of the methods and findings sections of the studies included in the review was undertaken to identify what data could be categorised according to the a priori keywords and to determine the frequencies of data assigned under these keywords.The analysis was conducted by the first author using tables, then discussed with the second and third authors to critique and reach agreement.The tables were then examined to establish if there were patterns across the categories.Finally, the authors reflected on the approach to the FGDs research taken by the researchers in relation to the characteristics within and across all the studies and drew conclusions about what this means for practice.

Characteristics of FGD Research
Twenty-one papers were included in this this review.These are summarised at Table 3. Fifteen papers employed qualitative methods and six studies used mixed methods.Some studies employed FGDs together with other qualitative methods.Table 4 outlines the methods of the studies in the review.Nine studies employed two qualitative approaches to gathering data from Aboriginal participants.In Whiteside et al. (2012), a table is provided, as well as a description of the number, community and gender of participants in FGDs and face-to-face interviews.The number of FGD participants is recorded in Wong et al (2005), but the authors do not report data on subjects involved in face-to-face interviews.In other studies (Emden et al., 2005, Kildea et al., 2012, Manderson and Hoban, 2006, Nelson et al., 2007, Prior, 2009), it is not clear what Aboriginal participants were interviewed or who was involved in FGDs or small group interviews (Williamson et al., 2010).

Approaches Used to Undertaking FGD and Analysing the Data
The findings are described according to the key areas of the template that was used to guide the content analysis.This is outlined in Table 6.

FGD Participant Sampling and Recruitment
A variety of sampling and recruitment methods were employed to select and recruit Aboriginal participants.Six studies employed purposive sampling (Anjou et al., 2013b, Artuso et al., 2013, Homer et al., 2012, Jan et al., 2004, Manderson and Hoban, 2006, Whiteside et al., 2012), one selected participants on the basis of convenience, while the other studies used random sampling (Wong et al., 2005).Purposive sampling was used to gain relevant, appropriate, diverse and rich participant responses.Consultative processes, such as discussion with peak bodies (Anjou et al., 2013b, Manderson andHoban, 2006) and study reference groups (Kelly and Luxford, 2007), were used to select and recruit participants.Health workers also advertised the studies and invited participants through their networks (DiGiacomo et al., 2013, Kildea et al., 2012, Whiteside et al., 2012).Media, including flyers, posters and local radio, was employed to invite participants (DiGiacomo et al., 2013, Kildea et al., 2012, Prior, 2009, van der Sterren et al., 2006).Seven papers did not provide any details of the selection approach or the procedures used to recruit Aboriginal participants to the FGDs.

FGD Question Development
Authors described how FGD questions were developed from previous research experience (Boudville et al., 2013), from literature in the field (Jamieson et al., 2008) or using a matrix (Wong et al., 2005).
Two papers mentioned separate FGDs for men and women ( Artuso et al., 2013, Wong et al., 2005) and in two other papers, it was not clear if the FGD included Aboriginal participants together with other study participants (Kelly andLuxford, 2007, Nelson et al., 2007).Health information and education was provided as part of the FGD in two studies (Schoen et al., 2010, Wong et al., 2005).No information concerning the FGD facilitation process was provided in four papers.
Many authors described their approach to data analysis as thematics (Anjou et al., 2013b, Boudville et al., 2013, Esler et al., 2007, Homer et al., 2012, Kelly and Luxford, 2007, Kildea et al., 2012, Manderson and Hoban, 2006, Nelson et al., 2007, Prior, 2009, van der Sterren et al., 2006, Wong et al., 2005).An inductive and directed analysis using a health promotion planning framework is described in Artuso et al. (2013), while, in Whiteside et al. (2012), it is less clear if a thematics analysis was undertaken.Other approaches to data analysis included the constant comparative method (Williamson et al., 2010), framework analysis using the access framework (DiGiacomo et al., 2013), content analysis (Emden et al., 2005), open coding (Jamieson et al., 2008) and manual coding (Jan et al., 2004).A 'Plan, Do, Study, Act Cycle' is named in Bloch et al. ( 2005), but it is unclear how this was actually undertaken.There is no mention of how FGDs were analysed in two studies in the review (Schoen et al., 2010, Smith et al., 2011a).
Approaches are described to achieve rigour, including member checking to verify or check the accuracy of the data collected and its interpretation (Artuso et al., 2013, DiGiacomo et al., 2013, Homer et al., 2012, Jan et al., 2004, Kelly and Luxford, 2007, Nelson et al., 2007, Prior, 2009).Member checking was also undertaken to expand the data set in specific areas via discussion (Kelly andLuxford, 2007, Manderson andHoban, 2006) and to promote cultural safety and respect (Homer et al., 2012).Some papers described the use of an independent researcher carrying out coding for the purpose of achieving consistency of coding and verification (DiGiacomo et al., 2013, Emden et al., 2005, Kildea et al., 2012).In one study, descriptors were written for each code in order to assist with this process (Kildea et al., 2012).Collaborative researcher coding is described for the purpose of achieving consensus on coding and multiple perspectives on the analysis process (Homer et al., 2012, Jan et al., 2004).Triangulation is described in Jan et al. for checking the findings of FGDs against other data sources (2004).

FGD Findings
As shown in Table 4, the studies included in the review comprised qualitative studies with FGD data only, qualitative studies with data from FGDs, and qualitative interviews and mixed methods studies with qualitative data from FGDs only, or from FGDs and interviews.It is difficult to establish the results of the FGDs from an examination of the findings in each paper.Aboriginal client and community member FGD data were integrated with interview data collected from other participants (Artuso et al., 2013, Emden et al., 2005, Manderson and Hoban, 2006, Nelson et al., 2007, Prior, 2009).In the mixed method studies, qualitative data from FGDs and semi structured interview data appears to be merged (Kildea et al., 2012, Manderson andHoban, 2006).Client and community FGD data was not separated from other qualitative data collected from service providers and stakeholders in several studies (Anjou et al., 2013b, Schoen et al., 2010, Smith et al., 2011a, van der Sterren et al., 2006, Whiteside et al., 2012, Williamson et al., 2010, Wong et al., 2005).The findings of FGD data is clearly reported in the papers where FGDs are the sole qualitative tool used on data collected from Aboriginal clients and/or community members (see Table 4).
In several studies, Aboriginal community members, including health service board members and study reference group members, were given an opportunity to comment on overall findings (DiGiacomo et al., 2013, Homer et al., 2012, Kelly and Luxford, 2007, Manderson and Hoban, 2006, Prior, 2009).Homer et al. (2012) noted that the review of the data by the Aboriginal Women's Evaluation Group resulted in clarification of some of the terminology and phases, but no themes were altered.In other studies, it is not apparent what comments were made and how they were incorporated into the analysis, as there is no mention of this in the findings, discussion and conclusion section of the papers.

Discussion
This integrative literature review sought to identify the characteristics of FGD studies with clients and community members in Australian Aboriginal health service research and how these and contextual factors affected the way in which FGD research is carried out.Studies included in the review were concerned with a range of Aboriginal health service areas and issues, Aboriginal populations and geographical contexts.The review provides an insight into the variety of approaches taken and identifies gaps in methodological practices.
We found a lack of information concerning the sampling and recruitment of FGD Aboriginal health service users and community participants, with a third of the studies including no information.This gap makes it difficult to conduct a meaningful assessment of the strengths and limitations of the affected studies and introduces uncertainties about the ultimate rigour of their findings.
In some publications, Aboriginal people were said to have been consulted to identify potential study participants.While such consultation is essential, some authors have consistency of coding and verification (DiGiacomo et al., 2013, Emden et al., 2005, Kildea et al., 2012).In one study, descriptors were written for each code in order to assist with this process (Kildea et al., 2012).Collaborative researcher coding is described for the purpose of achieving consensus on coding and multiple perspectives on the analysis process (Homer et al., 2012, Jan et al., 2004).Triangulation is described in Jan et al. for checking the findings of FGDs against other data sources ( 2004).
identified situations where the recruitment of Aboriginal participants was limited to small circles of peers, family and friends (Meadows et al., 2003) and lacked commentary on the extent to which the relevant focus groups may have been more broadly representative.
Iterative data collection is suggested as one solution to the above, where transcription and feedback occurs followed by further recruitment, so that any issues can be identified (Meadows et al., 2003).Such an approach was not noted in the studies within this review.Meadows et al. (2003) also suggest flexible timeframes for community consultation, recruitment and/or training of researcher(s) within the community and multi-stage processes of consent.Three stages of prepare, contact and follow-up are outlined in MacDougall et al. (2001) with respect to planning and recruiting the sample for FGDs and interviews.This approach may provide a useful guide to the design and reporting of such procedures in Aboriginal health services research studies.
The review findings show that some studies gathered qualitative data from both FGD and interviews.However, it is not clear if FGD data were gathered at the same time as interview data.Also confusing is the apparent integration of both types of data in the findings.Meadows et al. (2003) suggest a phased approach for gathering data from Aboriginal women first through the more open approach afforded by FGDs which can provide insight into the design of the more structured individual interview question guide.
Consultation throughout the research process, as well as input into the study findings, were described in some studies through mechanisms such as reference groups or steering committees, however, the level of community acceptance is not clear.Community acceptance of the research process and the health intervention itself is vital (Turner and Sanders, 2007) to ensure not only appropriate recruitment, but to maximise Aboriginal participation and engagement in the development of FGD questions, the running of the FGD and engagement in data collection, analysis and dissemination processes (Dudgeon et al., 2010, Ka'opua et al., 2004).
The use of the focus group has been described as a one-way process for gathering data from Aboriginal people and questioned by authors who highlight the need for participatory processes such as those based on 'yarning' and storytelling.Yarning has been described as a more culturally appropriate approach to accessing data and privileging Aboriginal knowledge (Bessarab and Ng'andu, 2010, Bishop et al., 2006, 2011, Jennings et al., 2013, Nagel et al.).
FGDs have been employed in a participatory manner to empower Indigenous people, such as in a study by Ljunggren et al.where Tibetan indigenous women, through the process of the FGD, built the confidence to make their own health care decisions (Ljunggren et al., 2009).FGDs themselves may be seen to be part of a health intervention.A study with Pakapuka people from the Cook Islands demonstrated that FGDs were part of a process that enabled people to build knowledge and change their behaviour in relation to diabetes (Oliver et al., 2007).FGDs have a number of strengths, including enhancing Aboriginal community control over the context, consensus building and problem solving (Willis, 2005).However, in the Australian Aboriginal context, none of the papers in this review provided details concerning the ways in which Aboriginal people engaged in the FGDs or the effect that these processes may have had on the findings.Multiple factors influence FGDs and, in turn, the resultant data.Some variables include the characteristics of the FGD participants, their contributions and the effectiveness of the moderator.In some studies in the review, such as Nelson et al. (2007), it is not always clear if non-Aboriginal participants were included in the FGDs with Aboriginal health service users and community members.The domination of focus groups by non-Indigenous participants, particularly those with technical knowledge, has been noted in by Willis (2005).In some studies in the review, Aboriginal health service users were in FGDs together with Aboriginal health workers (Kelly and Luxford, 2007) and it is not known what effect this would have had on the discussion and research.
Separate FGDs for Aboriginal men and women has been described elsewhere and identified as an important factor in enabling male and female participants to discuss issues that may not be freely aired in mixed gender settings (Cuneen, 2008).In this review, gender issues were noted in only two of the studies where separate FGDs for men and women were held.
None of the reviewed publications considered the possible effect of the relationships between non-Aboriginal researchers and Aboriginal participants on the FGD processes.In many of the studies, the moderators were non-Aboriginal.While it has been asserted that trust can be built between non-Aboriginal researchers and Aboriginal participants in FGD settings (Kingsley et al., 2010), the use of Aboriginal peers as moderators may be necessary to address issues of power and the need for relationship building (Mooney-Somers and Maher, 2009).Collaboration and consultation were described in one study in the review to ensure cultural safety and, hence, the appropriateness of the data collection and process, where non-Aboriginal people had facilitated the FGD (Homer et al., 2012).Efforts to ensure that FGDs do not promote culturally unsafe practices, such as those that "diminish, demean or disempower" the cultural identity and wellbeing of an individual (NCNZ, 2002), are important to the conduct of research with Indigenous people (Martin, 2008).
In social research generally, there is little literature examining how differences between group and individual discourses affect the analysis and interpretation of focus group data (Kidd and Parshall, 2000).Vicsek (2010) presents a scheme for the analysis of FGD data in non-Aboriginal contexts that may provide useful insights for Indigenous research.This includes suggestions for presenting quotations and explaining contexts that help to address issues of generalisability or representativeness often disputed in FGD research (McQuame and Mclntyre, 1987).
This review did not identify any in-depth discussion of methodology with respect to the FGD data collection, analysis or interpretation.There is no existing research that investigates whether the processes set into motion by FGDs and the data generated is underpinned by rigorous methodologies that are informed by an Aboriginal interpretative framework.For example, although it tends to be presented as a method, the focus group can be used to generate data to operationalise different methodologies.The way a focus group is conceived and implemented, and the way the data generated from it is treated, should reflect the method it was selected to operationalise (Sandelowski, 2010).However, the studies in the review do not point to an Indigenous paradigm to operationalise the FGD research underpinned by protocols such as those outlined by Martin to direct culturally safe and respectful researcher behaviour, and to ensure researcher responsibilities and accountabilities to Aboriginal people (Martin, 2008).

Limitations
There are disadvantages of content analysis that limit its usefulness as a means of synthesis in this review.The reductive nature of the approach may have resulted in a loss of understanding of the complexity and context that is important to the way in which FGDs were carried out in each study.Frequency-counting may not best reflect the structure or importance of the FGD processes described in the papers reviewed.The lack of evidence (nonreporting) could be treated as evidence of absence (not important).The results may be oversimplified and count what is easy to classify and count, rather than what is important.As a result, there is a need for caution in terms of the conclusions reached in this research.However, the gaps identified in the papers were the result of input from all authors, who spent much time discussing the papers in order to reach consensus on what was to be included in the study.The knowledge gaps identified as part of this process raise important questions about the FGD research in Aboriginal health service contexts that must be considered.

Recommendations for Practice and Research
In order to address many of the limitations identified in the FGD research included in this literature review, we propose a framework for improving FGD efficacy, rigour and relevance in Aboriginal health service research contexts (see Figure 2).The framework prioritises Aboriginal narratives, experiences and knowledge, with 'expert' opinion forming part of research problem identification and research question formulation.The framework proposes that the research purpose can then be established with Aboriginal input, allowing the articulation of a focus on: the generation of new ideas and theory (exploratory); and/or sharing lived experience and confirming existing theory (experiential); and/or uncovering motivation in order to explain beliefs or reveal reasons for health service preferences (clinical) (Fern, 2001).Clarification of the research purpose will direct the necessary use of appropriate theoretical perspectives and the methodological stance for carrying out the research, analysing and interpreting the data.In turn, this will provide decision-making guidance concerning the selection of the moderator, the discussion process, where the focus group will be undertaken and how reciprocal capacity-building and relationship negotiation will be achieved (Bartlett et al., 2007).At every step, community members and participants should be involved.

Conclusion
FGDs are an important qualitative tool to collect data in Aboriginal Australian health service research contexts.This literature review, involving a content analysis of the findings of 21 studies across a variety of settings, provided insight into the approaches taken to carrying out FGDs and some methodological gaps.The findings highlight the need for a clearer articulation of processes and procedures undertaken as part of FGD research.The development of a conceptual framework to better guide FGD research may be useful to, not only describe the process of structuring a study involving FGDs, but also to provide an Indigenous knowledge-centred approach and engage Aboriginal communities and partner organisations in collaborative, relevant applied research.
systematic search of the literature published between 2002 and 2013 was undertaken of eight bibliographic databases MEDLINE, CINAHL, Web of Science, PubMed, Scopus, ProQuest (Health & Medical), ATSI Health (Aboriginal and Torres Strait Islander Health), and Medline.In addition, Google Scholar was used to locate literature, as well as hand searching of the reference lists of useful research papers.The following search terms were used: 'Health services research', 'qualitative research', 'focus group discussion' and 'Indigenous Australian', 'Aboriginal' or 'Aboriginal and Torres Strait Islander'.