Methodology and methods

Introduction

In this chapter, we describe the methodology and methods used in both reviews. The literature search was conducted for both reviews at the same time (i.e. as one step). However, the other stages were conducted sequentially (i.e. we undertook all of the stages of review 1 first, followed by those of review 2). Given that the methods followed were exactly the same, the methods used in each stage are described in this chapter only once.

The research aims and objectives were constructed to identify insights about lay/health professional experiences of FGM/C-related health care and perceived appropriateness and acceptability of services. These are questions best answered by qualitative research;100,101 indeed, it is increasingly recognised that qualitative evidence syntheses have an essential role to play in understanding barriers to and facilitators of service initiatives.102,103 The overall design and synthesis methodology for each review was the same. Both have been reported as per Enhancing Transparency in Reporting the Synthesis of Qualitative Research (ENTREQ) guidelines.104

Methodology

There are many possible approaches to qualitative evidence synthesis, with most discussions in this area characterising the different types along a continuum between aggregation and interpretation.105 If the purpose of a synthesis is to generate new theoretical insights, a highly interpretive approach such as meta-ethnography may be most suitable, informed by an idealist epistemological stance. However, if the purpose is to inform policy or practice, a more aggregative or thematic approach informed by a realist epistemology is often advocated.106 The latter is also suggested in cases in which the existing evidence is likely to be descriptive (as in much health services research) rather than highly theoretical or conceptual.107 An initial scoping of the literature suggested that this was the case for the proposed syntheses. A thematic synthesis approach involves using thematic analysis techniques to identify key themes from primary research studies.108,109 Synthesis involves an iterative and inductive process of grouping themes into overarching categories and exploring the similarities, differences and relationships between them. Thematic synthesis explicitly aims to move beyond generating a list of descriptive themes (as would be the case in meta-aggregation110) in order to identify new, higher-order, analytical insights that can contribute to new understandings of a phenomenon. Review recommendations, however, are clearly formulated to inform policy and practice. As such, thematic synthesis was considered the most suitable approach for the two systematic reviews.111 Both reviews were conducted following the guidance set out by the Cochrane Qualitative and Implementation Methods Group.83,103,110–114

Protocol and PROSPERO registration

Both review proposals were registered separately with the PROSPERO database.115,116 A combined protocol for both reviews has been published in BMJ Open.117

Search strategy and study selection

An exhaustive and sensitive search strategy was developed by an experienced information scientist (JE). The search strategy was designed to identify papers for both reviews; thus, it is reported here as one search.

Search inclusion and exclusion criteria

The review inclusion and exclusion criteria are set out in Table 1.

List of Organisation for Economic Co-operation and Development countries

• Australia.
• Austria.
• Belgium.
• Canada.
• Chile.
• Czech Republic.
• Denmark.
• Estonia.
• Finland.
• France.
• Germany.
• Greece.
• Hungary.
• Iceland.
• Ireland.
• Israel.
• Italy.
• Japan.
• Korea.
• Latvia.
• Luxembourg.
• Mexico.
• The Netherlands.
• New Zealand.
• Norway.
• Poland.
• Portugal.
• Slovak Republic.
• Slovenia.
• Spain.
• Sweden.
• Switzerland.
• Turkey.
• UK.
• USA.

Search strategy

The searching process was conducted in three phases.

The first phase consisted of searching 10 electronic literature resources [both individual databases and hosted multifile collections (see Table 2 for full details)] that covered the relevant disciplinary areas, so that empirical research published in peer-reviewed journals or book chapters could be identified. The searching used a combination of index terms and text-based queries. Several of the resources (e.g. MEDLINE^® In-Process & Other Non-Indexed Citations) provide access to literature not yet assigned to a medical subject heading (MeSH) term or other indexing, so it was important to include text-based queries for comprehensive and current retrieval.

Key index terms included:

• exp Circumcision, Female/ (MeSH and MEDLINE)
• exp female genital mutilation/ (Emtree and EMBASE)
• exp circumcision/ (Ovid PsycINFO).

The different terminologies reflect different indexing principles and practices in the individual literature databases, and were incorporated into the overall strategies for completeness. Following the initial searches, the same strategies were set up as monthly alerts for four main literature databases (see Table 3 for details) until 31 December 2017 (i.e. the searching was not completed as a one-off process). Example full strategies for Ovid MEDLINE and CINAHL (Cumulative Index to Nursing and Allied Health Literature) on EBSCOhost are provided in Appendices 1 and 2.

The second phase comprised an extensive search for relevant grey literature, particularly to identify research reports or theses not formally published but still available in the public domain through institutional websites or thesis repositories.120–122 This part of the searching process included five resources to help identify this type of grey literature (see Table 2). Google (Google Inc., Mountain View, CA, USA) and Google Scholar were also interrogated and key experts in the field were contacted to elicit additional suggestions of relevant documents.123

The third phase of the search involved hand-searching the reference lists of related systematic reviews and of all the included studies.

Table 2 provides a list of the databases searched and the search dates. Box 2 provides a list of the grey literature sources.

• The British Library EThOS (ethos.bl.uk).
• Networked Digital Library of Theses and Dissertations (www.ndltd.org).
• National Institute for Health and Care Excellence (www.nice.org.uk).
• Trove – National Library of Australia (trove.nla.gov.au).
• Open Grey (www.opengrey.eu/).
• Google (www.google.com).
• Google Scholar (https://scholar.google.co.uk).
• Experts in the field.

Balancing sensitivity versus specificity within the search strategy

In developing the original search strategies for the reviews, search terms around FGM/C seemed to generate suspiciously small answer sets in the main literature databases (MEDLINE and EMBASE) despite the use of a large range of synonyms for FGM/C. To address this, the librarian devised additional search statements that brought in terms relating to the possible physical or psychological complications that are identified as health impacts of FGM/C, making use of the descriptors in a UK Department of Health and Social Care FGM/C guidance document.124 This aimed to retrieve relevant documents, based on symptoms or impacts described, without necessarily mentioning FGM/C or other synonyms. These search statements can then be seen as a proxy for FGM/C in the strategy. In addition, we identified from hand-searching that some relevant studies were not indexed in the databases using FGM/C-related terminology. This appeared to be attributable to the study’s main focus being on a different issue, for example pregnancy experiences or care for a particular migrant group (e.g. Somali). Therefore, the search strategy was broadened by using additional ‘proxies’ for FGM/C (originating countries, e.g. “Somali*”) plus pregnancy terms OR physical/mental complications/disorders associated with FGM/C to draw in such papers, which did not overtly capture FGM/C in the title, abstract or MeSH/Emtree indexing. Because these proxy search statements are necessarily less specific, the high sensitivity was tempered by focusing the strategies to limit retrieval to studies relating to OECD countries and specifically excluding RCTs.

Management of identified records

All retrieved data sets were downloaded into group sets within an EndNote library (Clarivate Analytics, Philadelphia, PA, USA). Using the EndNote grouping functionality, individual group folders were created for each of the database search results sets. Each record was annotated to identify from which database search it had originated. The ‘Find Duplicates’ feature was used to identify duplicate articles from across the different database searches. The duplicates were then moved to a separate ‘Duplicates’ group, leaving one unique representative of each article in the ‘Search Results’ group, usually selecting the MEDLINE representative to be retained, if available.

Screening and selection

Two members of the project team independently screened all potential studies based on a review of titles and abstracts, grouping them into yes, no or unsure using the screening tool in Table 3. Areas of disagreement were resolved through discussion and by consultation with other team members at team meetings.

Full texts of all studies with an initial assessment of yes or unsure were then obtained. These were retrieved via online searches and using the interlibrary loan scheme. If papers and reports could not be retrieved through these channels, authors were contacted by e-mail or via ResearchGate (www.researchgate.net) with full-text requests. All full-text papers were then independently assessed by two team members, alongside discussion with the wider team in cases of uncertainty. Reasons for exclusion were documented (see Appendix 3 for a table of all excluded studies with reasons).

Management of foreign-language papers

Non-English-language papers found to be relevant on the basis of their English-language abstract were sent for complete academic translation. This included a summary of the study purpose or aims and the complete methods and findings sections of the studies. A total of 13 papers were translated by outsourced translators who had experience and training in research. After translation, these texts were then assessed for inclusion and quality appraised and data were extracted if selected for inclusion.

Search results

The search results are presented in a flow diagram in Figure 1 as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.125

FIGURE 1 The PRISMA flow diagram. (Figure not reproduced here; see the original article.)

Quality assessment

The role of critical appraisal in qualitative evidence synthesis is contested and there is a lack of agreement on (1) the appropriateness of excluding studies, (2) the potential impact (or not) of excluding eligible papers on review outcomes and (3) the criteria on which quality should be established.126–129 For these reasons, the team adopted an inclusive approach to critical appraisal, using the appraisal process to enable an in-depth understanding of each paper and to facilitate a critical, questioning approach to the study findings.130 Studies were not excluded on the basis of quality; rather, the quality assessment was used (1) to judge the relative contribution of each study to the overall synthesis and (2) to assess the methodological rigour of each study as part of a process of assessing confidence in the review findings.131–135

Following the guidance of the Cochrane Qualitative Methods and Implementation Group136 and the JBI,84 reports and theses from the grey literature were appraised in the same way as papers that were published in peer-reviewed journal articles.120,122,123

Study quality was assessed by two reviewers (RT and CE) using the Joanna Briggs Institute Qualitative Assessment and Review Instrument (JBI-QARI).137,138 The domains examined in this tool have been found to be more coherent and more sensitive to assessment of validity than other commonly used tools (see Appendix 4).128 Team meetings were used to achieve a shared and consistent approach in operationalising the domains in the tool. A third reviewer (GH or JM) was asked to comment on any papers for which there was uncertainty or disagreement and discussions were held until a final consensus on assessment for all the papers was achieved.

The JBI-QARI has 10 questions. These were applied to each individual paper and an aggregate score was calculated (Table 4). In addition, reviewers’ comments were recorded, providing an explanatory rationale for why questions had been answered in a particular way. There is some debate over the use of critical appraisal tools to ‘score’ papers, especially if arbitrary scores are used to exclude papers from a review.110,135,140 However, as explained above, we did not use the score to exclude papers. Rather, we adopted this approach primarily to enable us to determine an overall, if somewhat crude, picture of the quality of the whole body of evidence within each review, and to assist with the assessment of methodological limitations as a key part of the process of establishing the level of confidence in each of the review findings (see Assessment of confidence in the review findings: CERQual).112 A criticism of ‘scoring’ qualitative critical appraisals is that it can be hard to distinguish between the poor conduct of a study and poor reporting, especially where journal word limits constrain the level of detail that can be reported.110 In addition, there is no consensus regarding the relative importance of any one domain within an assessment tool over another, and, hence, whether or not they should all be given an equal weight. In view of these concerns, we chose to adopt a ‘weighting system’ used in previous studies by Higginbottom et al.,141,142 in which papers were grouped into one of three ‘bands’ (high, medium or low) to enable a broad-brush evaluation to be made of their relative quality, as shown in Table 4.

As an additional strategy for overcoming the potential limitations of solely relying on a checklist to assess quality, we also chose to assess the ‘richness’ of the studies. This is an approach outlined by Popay et al.,143 and subsequently operationalised further in Noyes and Popay144 and Higginbottom et al.141,142 This approach defines study ‘richness’ as ‘the extent to which the study findings provide explanatory insights that are transferable to other settings’.144 ‘Thick’ papers create or draw on theory to provide in-depth explanatory insights that can potentially be transferable to other contexts. By contrast, ‘thin’ papers provide limited or superficial description and offer little opportunity for generalising. Each paper was assessed by Ritah Tweheyo and Catrin Evans against the criteria, as set out by Higginbottom et al.141 (Table 5), and categorised as either ‘thick’ or ‘thin’.

Extraction of study characteristics

Details of each of the included papers were extracted into a Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA) file using the domains from the JBI data extraction template (see Appendix 5).137,146 Data extraction of study characteristics was primarily undertaken by one reviewer (RT); however, a second and third reviewer (CE and JM) double-checked the extractions of a subsample of papers for accuracy. In addition, the team had regular meetings to discuss any uncertainties, to ensure consistency of approach and to agree definitions.

Categorisation of study relevance

After data extraction, papers were categorised by Ritah Tweheyo and Catrin Evans in terms of their relevance to the respective review question. This assessment was made in order to gain a better understanding of the nature of the body of evidence, and also to facilitate the coding process, as described further in Thematic analysis and synthesis. Study relevance was defined as high, medium or low, as set out in Table 6.

Extraction of study findings

PDF (Portable Document Format) files of all of the included papers were imported into NVivo 11 (QSR International, Warrington, UK) software and the ‘findings/results’ sections were coded and analysed. If ‘findings’ were located in other parts of the papers (e.g. the discussion sections), these were also coded.147

Thematic analysis and synthesis

Analysis and synthesis of the study findings followed the principles of inductive thematic analysis,108,109 and comprised four stages, as suggested by Thomas and Harden.107 This was followed by a final, fifth step in which the findings from both reviews were combined into an integrated synthesis from which implications for policy, practice, education and research were derived.

The first stage involved intensive and repeated reading of all of the included papers to gain an overall in-depth understanding of their context and key findings. This stage was returned to many times during the analysis and synthesis to ensure that emerging interpretations remained deeply contextualised.

The second stage involved line-by-line coding of the findings of the primary studies. We coded only findings that were directly relevant to the review question. Where there were multiple papers from the same study, we ensured that distinct findings were coded from each (i.e. we did not code duplicate findings). Owing to the large number of papers in each review, we decided to start the development of the coding framework by focusing on studies that were likely to yield the richest and most relevant findings. Hence, we started with the studies that had been categorised as both ‘thick’ and highly relevant. These generated a large number of free codes, assigned line by line to the text according to its meaning and content. Subsequent studies were coded using the initial framework of codes and adding to them as new codes were identified. Ritah Tweheyo and Catrin Evans both independently coded the initial set of studies, conferring until the initial coding framework was agreed. Subsequently, Ritah Tweheyo completed the rest of the coding, with Catrin Evans and Gina Higginbottom regularly reviewing and discussing the evolving framework.

The third stage involved analysing the codes to explore areas of similarity or difference, grouping them together based on shared meanings to create new codes, and then organising these into a set of descriptive themes that captured the meaning of their constituent codes. In NVivo, this was represented as a hierarchical tree structure with several layers. Each theme was formulated to describe a key phenomenon in such a way as to capture its core meanings but also to explain and account for possible differences or variations in the phenomenon. To ensure rigour of the analytical process, the team actively sought to identify and understand possible ‘disconfirming’ cases that might challenge emerging interpretations,148 and to explore possible subgroup or contextual differences. These processes were aided by the creation of a theme matrix (see Appendices 8 and 12), in which each theme was mapped to its constituent studies. This helped the team to clearly see how common the theme was among the studies and what kind of study contexts or samples the theme related to, and to explore why it may have been present in some studies but not in others. During the analysis, the team referred to a list of questions to help them to develop the descriptive themes, as set out in Box 3.

• Define, explain and contextualise the core concept.
• Explore key groups it may (or may not) apply to.
• Explore key settings/circumstances it may manifest in or be affected by.
• Explore causes, manifestations, variations and consequences in terms of perceived quality of care and potential clinical outcomes or clinical trajectories.
• Identify ‘gaps’ within the theme.
• Explore any key variations between the studies and the study contexts.
• Explore implications of the date ranges of the included studies.
• Explore and understand examples where the theme does not seem to apply.

Descriptive themes are derived from an interpretative process of constructing codes based on underlying meanings and considering how and in what ways these codes are related to each other. Thus, they represent a highly rigorous process that combines the findings of each study into a whole via a listing of themes. Nonetheless, Thomas and Harden107 argue that descriptive themes represent only the first step of synthesis, as, at this point, they generally do not yet ‘go beyond’ the findings of the primary studies to identify new concepts, understandings or hypotheses. This is achieved in the fourth stage, comprising the development of higher-order or ‘analytical’ themes, which constitute the key findings of the reviews.

For the fourth stage of synthesis, analytical themes were evolved through an in-depth process of comparing and contrasting the meanings of the descriptive themes, analysing these in relation to how they were, or were not, able to illuminate the review questions, and inferring broader phenomena, categories of meaning or social processes that they related to. This was a cyclical process that involved re-reading the papers, re-reading the codes and in-depth discussion among team members. Using this analytical process, the descriptive themes were organised into five analytical themes for review 1 and six analytical themes for review 2.

As far as possible, the descriptive and analytical themes were formulated as ‘directive’ findings indicating clear messages and/or suggesting clear lines of action for policy and practice.137

The fifth and last stage of the synthesis involved bringing the findings from both reviews together to generate an overarching novel synthesis. This was done by creating short statements of findings for all of the themes (descriptive and analytical) in each review, listing them out, juxtaposing review 1 themes against review 2 themes and comparing and contrasting these. This process enabled the team to identify key phenomena that could explain and illuminate aspects of FGM/C-related health care through understanding both women’s and professionals’ views and experiences together. An overarching synthesis comprising four ‘synthesised findings’ was then drafted and extensively consulted and revised by the whole team until a final version was agreed.

Assessment of confidence in the review findings: CERQual

Assessment of confidence in the findings of each review was undertaken using the Confidence in the Evidence from Reviews of Qualitative Research (CERQual) approach.110,112,131,136,149–153 Ritah Tweheyo and Catrin Evans led this process, working together and discussing the final assessments with the research team. CERQual is a relatively new, transparent method for assessing confidence in the findings in qualitative evidence syntheses; it is akin to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach for establishing confidence in evidence of effectiveness for each outcome in a quantitative review.154 The assessment of confidence in the evidence is made for each individual review finding and considers four elements: (1) methodological limitations (the extent to which there are problems in the design or conduct of primary studies that contributed to evidence of a review finding),136 (2) relevance (the extent to which the body of evidence from the primary studies supporting a review finding is applicable to the context specified in the review question),149 (3) coherence (whether or not the finding is well grounded in data from the primary studies and can provide a convincing explanation for patterns found in the data)152 and (4) adequacy of data (an overall determination of the degree of richness and number of data supporting a review finding).151 Based on assessments made of any concerns across these four individual domains, each review finding is assigned one of four levels of confidence: high, moderate, low and very low.

In this project, the CERQual assessment was made for each of the descriptive review findings. Each descriptive review theme was summarised into a succinct ‘summary of a review finding’. These were then presented in a ‘summary of qualitative findings table’ containing several columns, including the summary of the review finding, the studies that contributed to the review finding, the assessments for each of the CERQual domains, the overall CERQual assessment of confidence in the evidence and an explanation of this assessment (see Appendices 9 and 13).

The CERQual assessment was made for each of the descriptive review findings (themes), rather than for the analytical themes or the themes in the overarching integrated synthesis, as the latter comprise an aggregation of descriptive themes as well as descriptions of patterns or explanations that are inferred as part of the interpretative process. There is, as yet, relatively little guidance for, or experience with, applying CERQual to higher-level analytical themes or theories.150 The process of applying CERQual to our higher-order themes would, therefore, have required an extended period of methodological development and testing, which was beyond the scope of this project, and which is recognised as being part of the ongoing research agenda for CERQual development.150

As of yet, unlike GRADE assessments, ‘dissemination bias’ within qualitative evidence synthesis is rarely discussed and needs further methodological consideration.153 However, in this review, we sought to minimise possible dissemination bias by undertaking an exhaustive literature search that (1) was highly sensitive rather than specific, (2) included a range of grey literature (research reports, evaluations and theses) as well as published papers, (3) did not limit the date, (4) did not limit the language and, (5) included the original thesis or report in addition to the published paper when the former contained findings that were not reported in the latter. In addition, as noted above, we maintained an inclusive approach and did not exclude any studies on the basis of quality.

Reflexivity, rigour and quality of the synthesis

As with primary qualitative research, there are a number of strategies that can be adopted to enhance the rigour and trustworthiness of qualitative evidence synthesis.148,155 We utilised a range of different approaches, as outlined in this section.

First, we were aware that our own theoretical, cultural and political positions might influence the ways in which we engaged with the texts and developed the interpretations.156,157 The initial analysis and synthesis process was undertaken primarily by two reviewers (RT and CE), aided by the wider project team. Ritah Tweheyo is a black African public health researcher, originally from Uganda, and Catrin Evans is a white nurse and health services researcher from the UK. Neither has direct experience of FGM/C, and neither has worked clinically with women/girls who have experienced FGM/C. Both have had experiences of ‘being migrants’ and having to adjust to foreign health systems, but in different contexts. Both have experienced childbirth. Both consider themselves to be feminists and anti-racist (but again, these standpoints have emerged from different subjectivities). Both are anti-FGM/C but recognise the inherent political and epistemological tensions that exist when individuals located ‘outside’ cultures and communities try to explain a phenomenon that is deeply culturally and socially shaped. During the analysis, we continually challenged ourselves by asking if and how our emerging interpretations were being influenced by our own perspectives and repeatedly returned to the original papers to re-embed coded text within its wider study context and to check our interpretations. As an example, there were many accounts in the papers in review 1 in which women described extremely humiliating, distressing and stigmatising experiences in their health-care encounters, often linked to their race or ethnicity. Although we knew from prior reading that such experiences had previously been reported, reading so many accounts at first hand proved to be a shocking and distressing experience for us as reviewers. We had not expected this to be such a pervasive theme and we felt strongly that we wanted this aspect of women’s experience to be heard in the review. However, we occasionally wondered if we were perhaps giving too much prominence to women’s negative experiences and failing to give sufficient attention to instances where women had reported positive care experiences. This led us back to reading and, in some instances, re-coding the original papers, and this prompted us to develop a more nuanced approach to understanding women’s care experiences.

Second, the analysis process also explicitly drew on the expertise within, and challenge from, the project team. Team members were sent exemplar papers to read before meetings, so that our discussions were grounded in the papers and hence enabled challenge and conceptual development of the initial interpretations of Ritah Tweheyo and Catrin Evans. The same approach was used during consultations with the PAG. However, this consultative process was not always straightforward, as all involved had vast experience and strong views. At times, we had to step back and remind ourselves to base our interpretations on the data from the included papers and not on our own wider experiences.

Third, once the descriptive and analytical themes had been finalised by the project team, we organised a national research stakeholder engagement event in which the draft review findings were presented. Over 65 people attended this event, representing FGM/C survivors, FGM/C activists, community organisations with FGM/C as a remit, academics, researchers, midwives and general practitioners (GPs). Their inputs contributed to the report in four important ways: (1) to explore the credibility of the key findings, (2) to put these into the contemporary UK context, (3) to identify gaps in evidence/understanding and (4) to put forward key implications/recommendations for service development and future research.

The nature of stakeholder validation in the context of a review is controversial and poorly described.158 Generally, non-author stakeholders have not necessarily read the included papers and thus are not directly familiar with the underlying research findings; they may or may not have direct experience of the issues to which they pertain, hence there is some debate over their ability to ‘validate’ the emergent themes. Therefore, as suggested by Booth et al.,148 we used our stakeholder event to challenge our interpretations as well as to explore the credibility of our findings. After each of the review presentations, stakeholders were asked to comment on the questions in Box 4.

• Based on your own experience and perspectives, do the themes seem to capture the main issues?
• Were any review themes surprising for you? Why?
• Were there any review themes that you do not agree with? Why?
• Do you think that there are any issues missing from the review findings that you would have expected to see?

The feedback from the event was reassuring in that there was a very strong sense of validation of the findings. When participants indicated that they had found a theme surprising, we returned to that theme and re-examined it. For example, two of the descriptive themes in review 2 (around professionals’ experiences of deinfibulation and reinfibulation) were rewritten in a more nuanced manner as a result of this process. Participants had felt that further detail and country-specific exploration was required in these themes, as they felt that in the UK context, professionals would know the relevant laws and would not experience the ambivalence described. Likewise, participants noted a number of perceived omissions (e.g. nothing on the impact of mandatory reporting and little on mental health issues), which, again, prompted us to go back and check that these were actual omissions in the evidence and did not reflect errors in the review process. Some comments helped us to analyse the findings in more detail (e.g. one participant asked how the policy and legal context around FGM/C in different countries might have influenced the findings). The event was also used to elicit suggestions for the implications of the review (see Appendix 14 for a full description of the stakeholder event).

Finally, in order to capture feedback from medical practitioners (a group that had been largely absent from the national event), the review findings were then also presented to a group of 30 obstetric and gynaecology registrars during one of their training days. This group included doctors working in cities with a relatively high prevalence of FGM/C as well as some working in settings with a very low prevalence. The subsequent discussions of their own experiences showed a very high level of congruence with the findings of review 2 (at least within the English context), and further helped to shape the implications from a medical perspective (see Appendix 17 for detailed feedback from this group).

Summary of the review process

Figure 2 provides a clear depiction of all of the steps in the review process.

FIGURE 2 Flow chart of the review process. (Figure not reproduced here; see the original article.)