Step 1 –Identifying the research question

The research question, purposes, and objectives have been developed using the Population/Participant–Concept–Context [60] mnemonic (Table 1). This adds clarity to, and alignment between, the research question and protocol purposes and objectives [57].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Population/Participant–Concept–Context parameters with inclusion and exclusion criteria.

https://doi.org/10.1371/journal.pone.0276776.t001

The overall research question is deliberately broad and seeks to identify what online STB talk research has been conducted. There are three purposes associated with this review, with each purpose related to a specific objective. The first is to map the range, nature, and extent of research conducted on online STBs talk. Doing this will allow the field to advance in a systematic, evidence-based manner increasing the potential of review findings to inform suicide prevention and intervention practices, policies, and research.

The second purpose is to disseminate what is known about online STBs talk. This may enhance the training of health practitioners and peer/professional online moderators in suicide prevention strategies. Specifically, this is in terms of how to best talk with people experiencing STBs. Dissemination extends to improving suicide prevention and intervention policies and practices to enable online spaces to be safe places to talk about STBs. Developers of safe language guidelines may also use findings to audit how well current guidelines reflect empirical evidence. The final purpose is to identify gaps in the existing knowledge base that may be used to guide future online STBs talk research in a systematic manner. Findings may also be used to refine understandings of contemporary suicidal theories/models.

Step 2 –Identifying relevant studies

To identify relevant studies, a three-stage approach will be undertaken with the first stage having been conducted as part of the protocol development. Open Science Framework, Prospero, Cochrane, and JBI Evidence Synthesis were searched by reviewer 1 (the first author) on the 18^th of March 2022 to identify any existing systematic or scoping reviews that reflect the proposed review’s research question. No scoping or other type of systematic review was identified. The second stage has already been conducted with reviewer 1 undertaking a preliminary search of the EbscoHost Academic Search Ultimate database to test the search strings on the 18^th of March 2022. Subject terms and keywords from the first 30 returned articles have been scanned to identify if there were any key search terms that were not included in the search string strategy or were not relevant to the search string strategy. No new search terms or potential changes to the search string were identified. Based on reviewer feedback Snapchat and Weibo were added to the second search string. The final stage will be for reviewer 1 and 2 (the second author) to independently search all the databases, key journals, and reference lists of included articles.

The search strategy (see S1 File) and electronic database selection has been developed in consultation with a research librarian who has expertise in systematic review study database and search string identification and development. No language limitation will be imposed during database searching. Should studies in a language other than English be identified, a translated version of the title and abstract will be sought from the corresponding author of the study, if such a translation has not been identified via a search of the selected databases and journal website where the study is published. If a translated version is not available, the study will be included in a supplementary table and noted as a potential limitation of the review’s findings. Translation apps or software will not be used to translate study titles or abstracts. Research suggests that they are useful for simple everyday phrase translation with error rate increasing with sentence complexity [67]. Such complexity is likely to be found in research article titles and abstracts. A date limitation of 1989 onward will be imposed during this search stage that reflects when the internet became publicly available.

A quality assessment limitation will also be implemented during the search stage. Although quality or critical assessment of studies is not a scoping review requirement [56, 57], some published scoping review protocols [68, 69] and reviews [70–73] include such an assessment. Further, as one of the purposes of the proposed review is to identify gaps in the literature, quality assessment has been recommended in such situations [74]. Therefore, a two-phase quality assessment process has been adopted. The first will be to impose a search limitation of only peer reviewed journals during the search phase, where possible [75]. This means that only peer reviewed research studies will be included in this review. The second is outlined at step 4.

A tension for any scoping review occurs between the feasibility of doing the review and its breadth. The need for a comprehensive mapping of the field needs to be balanced against practicalities so that it can be conducted in such a way that it meets the review’s purpose and objectives in a timely manner [57]. The databases have been chosen to reflect this in that they that are health orientated as well as sociological/discursive/language focused databases. This is because the substantive area is both health (STBs) and sociological/discursive (talk, language, conversations, or interactions). The databases have also been chosen with a view to identifying relevant studies whilst minimising study duplication.

Database searching will be undertaken independently by reviewer 1 and reviewer 2 on the same day. Databases will be searched one at a time in the following order Medline, CINAHL with Full Text, Academic Search Ultimate, APA PsycInfo, Communication Source, Health Source: Nursing/Academic Edition, Humanities Source Ultimate, MLA Directory of Periodicals, Psychology and Behavioural Sciences Collection, Sociology Source Ultimate, Scopus, and SAGE Journals: Social Sciences and Humanities to minimise duplicates [76]. Searches will be conducted by title and then by abstract, where possible. For each database search, search string 1 will be run first, following by search string 2, then search string 3, and then search string 4. The outcomes of the three separate searches will be combined using the Boolean operator AND to become the final search.

Both reviewers will note the outcome of each database search in a separate Microsoft Excel file that will act as a review audit trail (see S2 File). After each database title and abstract search, reviewer 1 and reviewer 2, will compare results. Should discrepancy arise, the reviewers will discuss this in order to identify the source of the discrepancy. Discrepancy identification will be noted on the Microsoft Excel review audit trail files with a corrected search then run. If discrepancy identification fails, a third reviewer (the third author) will overview reviewer 1 and 2’s search processes to identify the discrepancy. This outcome will also be noted on the file. This general comparison-discrepancy-consensus process will be used throughout the review, unless otherwise stated.

Besides electronic databases, reference lists of studies identified at the end of stage 3 will be independently searched by reviewer 1 and 2 to ensure that all potentially relevant studies have been considered for this review. Identified studies will be considered for inclusion using the same inclusion-exclusion criteria (see Table 1) as studies identified via database searching. The reviewers will compare results after searching all reference lists with outcomes being noted on the Microsoft Excel review audit trail files.

A separate independent search of key suicide, discursive, and cyberpsychology research journals will also be undertaken by reviewer 1 and 2. This is required because electronic databases have a British and American journal bias [56]. Key suicide journals to be searched are Suicide and Life-Threatening Behaviors, Archives of Suicide Research, Death Studies, and Crisis: The Journal of Crisis Intervention and Suicide Prevention. Key discursive/language research journals are Sociology of Health and Illness, Qualitative Health Research, Discourse Studies, Journal of Pragmatics, and Discourse and Society. The key cyberpsychology journals are Cyberpsychology, Behavior and Social Networking and Computers in Human Behavior. Other key journals may be added depending on search results. The same result comparison process will occur after all journals have been searched with results being noted on the Microsoft Excel review audit trail files.

Given the inclusion of a two-phase quality assessment process, a search of the grey literature will not be undertaken. This is not only related to quality assessment but also due to the difficulty in identifying what grey literature to include and how to do this in a systematic and reproducible manner. For example, mental health or suicide prevention organisations may have commissioned research to investigate online talk. Typically, the websites of such organisations would be searched for such studies with these organisations being identified using an internet search engine. The issue with using an internet search engine is that the same search conducted by two different reviewers will return different results [77]. Indeed, the same search conducted by the same reviewer on different occasions will also return different results. The algorithms that underpin such a search, returns results that are bespoke to the particular reviewer’s preferences which means the algorithms are constantly changing and adapting [77]. Any results from such a search are not systematic nor reproducible and thus not appropriate to be considered for a scoping review.

Whilst including a search of the grey literature has been a notable feature of scoping reviews [56, 58], there are no clear guidelines as to what grey literature must be included in a scoping review. Further, there are no best practice recommendations as to how to search for and thus identify all relevant grey literature sources [78]. Currently it is recommended that overlapping search strategies be adopted in order to identify all relevant pieces of grey literature, however this brings with it the probability of also identifying irrelevant literature [78]. Thus, it is not only replicability that becomes an issue but as Peters and colleagues [58] posit the scope and breadth of the review needs to be balanced alongside the feasibility of completing the review with the resources available. We recognise that the exclusion of grey literature will impose limitations on review findings, and this will be noted in limitation discussions. As per Peters et al. [58, 79] recommendation when deciding to not include particular searches, we have justified, detailed, and made transparent our decision to not include grey literature and recognise this as a limitation.

The results of the above database, reference list, and key journals searches will be managed using EndNote X9 [80]. Each reviewer will manage their own search results thus two independent EndNote libraries will be developed. Each library will have a separate EndNote group folder for each database search and for each search type (journal search, reference list search).

Step 3 –Study selection

Reviewers 1 and 2 will independently engage in the below study selection stages using the inclusion and exclusion criteria presented in Table 1. Team members have met and discussed the appropriateness of the databases and search strings as well as the inclusion and exclusion criteria [57]. Team meetings will be held regularly throughout step 3 to ensure that inclusion and exclusion criteria remain relevant as this allows for issues and challenges to be identified early and acted upon as required. The first meeting will be before study selection begins with the second occurring after the first half of the identified studies have been reviewed. Any changes to the inclusion and exclusion criteria will be through discussion and consensus with changes noted in the Microsoft Excel search review audit trail files. Given the iterative nature of scoping reviews [56], any changes may require reviewer 1 and 2 to restart the study selection process using any new updated inclusion and exclusion criteria. The final meeting will occur at the end of the study selection process to review the study selection process and to identify any additional changes that may be required to step 2 or 3. Any changes are likely to require reviewer 1 and 2 to re-visit steps 2 and 3.

The first stage in study selection will be for reviewer 1 and 2 to independently identify duplicate results using the Find Duplicates function in EndNote. The reviewers will do this following the systematic 13 step process outlined by Falconer [76]. The reviewers will meet after this to confirm remaining study numbers. Any discrepancies will be managed using the aforementioned discrepancy process. Once duplicates have been removed, reviewers 1 and 2 will independently apply the inclusion and exclusion criteria to both title and abstract at the same time. Reviewer 1’s experience with previous reviews [69–71] indicates that a two-phase study title and then study abstract screening has the potential to miss studies that should be included in the final review. Given this, combining title and abstract screening is more feasible and less likely to exclude studies that should be included.

In stage 2, reviewers 1 and 2 will compare results after screening all studies where the first author has a last name starting with A, they then will compare after screening studies where the first author has a last name starting with B, with this continuing through to Z. This way any errors, issues and challenges are identified as they occur and can be acted upon with immediacy. The same consensus and discrepancy process that occurred in previous steps will continue during this screening process. One difference is for any studies where both reviewers are unsure whether to exclude or include. When such instances occur, the study will be included for full text screening.

Once the title and abstract stage screening has finished, the full text of remaining studies will be read independently by each reviewer. Using the inclusion and exclusion criteria, each full text study will be assessed for final review inclusion. The reviewers will meet after reviewing all full-text articles to compare results. After which, studies to be included for final review will have been identified.

Step 4 –Data extraction

A Microsoft Excel data extract template based on the JBI data charting template has already been developed in consultation with all team members (S3 File). It contains both general (e.g., authors, publication year, location of study) and study specific (e.g., type of social media platform or online space, data, moderated/unmoderated online space, suicidality–thoughts and/or behaviours, limitations) information. Extracted information will align with the research question and purpose and objectives of the review. The template will be trialled independently by reviewer 1 and 2 with the first 5 articles for usability [57]. The research team will discuss the outcomes of this, with any changes being agreed to by all team members. As review and refinement of the data extraction template is likely to take place during this step [57]. There is the possibility that study information will be re-extracted as the extraction form evolves.

Reviewer 1 will engage in extracting data from the remaining studies with reviewer 2 undertaking an audit of a random selection of 20% of the remaining extractions to check for errors, omissions, and biases. The 20% figure has been used in a number of reviews as an appropriate study audit percentage [70, 71, 75]. This audit also offers an opportunity for further refinement of the extraction form. Outcomes of the audit will be charted on the Microsoft Excel review audit trail file with the same previously mentioned discussion and adjudication process being adopted. Reviewer 1 will contact corresponding author/s if relevant information has not been reported in the article, prior to extraction and audit taking place.

Stage 4 is where the second quality assessment will occur. Reviewer 1 will undertake a quality assessment of each included study during the data extraction process. Studies that are qualitative will be assessed using a modified JBI Appraisal Checklist for Qualitative Research [81]. Quantitative studies will be assessed using a modification of the appropriate JBI appraisal checklist (e.g., cross sectional studies will be assessed using the JBI Appraisal Checklist Analytical Cross-Sectional Studies [82], case reports the JBI Appraisal Checklist Analytical Case Report [83]). The JBI checklists have been chosen as they have been developed using an evidence-based approach, with the JBI Scientific Committee giving their approval for use. The checklists have been modified by reviewer 1 with changes being agreed to by the research team. These changes are based on Macrynikola et al [46] recommendations for selection, measurement and confounds. In terms of selection: did the study include clear sample descriptions/definitions such as age, gender orientation, nationality, sexual orientation, geographical location; was a participation rate included (if relevant) and was the recruitment criteria applied appropriately and uniformly; and was a justification of the sample size included. Measurement focused on the quality of measurement used in each study with a rating of high to low given (see S4 File) with confounds focusing on whether basic and potential confounds that have been identified in the literature were taken into consideration and discussed.

This review will adopt a 80%, 50%-80% and below 50% rating process that previous reviews have used [68, 70, 71]. For each item on the checklist, 1 point is allocated to a ‘yes’ response and 0 for a no. When more than 80% of the checklist criteria has been met, the study is assessed as being of good methodological quality, moderate quality studies fall between the 50% to 80% range, and below 50% is indicative of poor methodological quality. Quality assessment outcome will be noted on the relevant data extraction form. Once this is completed the results of all the extraction and quality assessment processes will be entered into one Microsoft Excel file so that the key study aspects can be more easily identified, compared, and summarised.

Step 5 –Collating, synthesising, and reporting the results

Stage 5 is where research that has been identified for final inclusion will be collated, synthesised, and reported in such a way that it answers the research question and meets the review’s purposes and objectives. To assist with this, article selection stages and processes will be visually represented using the PRISMA flow chart (S1 Fig) with the PRISMA-ScR checklist (S5 File) being used to guide the overall conduct and reporting of the scoping review’s process and findings. Based on previous pilot searches, both quantitative and qualitative studies are likely to be included in the results of this review. A textual narrative synthesis is the most likely methodological approach that will be adopted as this is well suited for synthesising both qualitative and quantitative research evidence as it allows for a standardise reporting of each type of study [84]. This is turn makes it easier for differences and similarities between study contexts and characteristics to be identified [85].

Synthesis will be undertaken by reviewer 1 using a step-wise approach [84]. The first step is placing studies into broad sub-groups. Sub-groups are likely to be internet, social media, non-attempting suicidal ideators, and attempting ideators. Studies can and are likely to be grouped in more than one sub-group with the possibility of other sub-groups emerging or current sub-groups being refined (e.g., internet–moderated online forums, internet–unmoderated online forums, social media–Twitter) during this first step. Sub-grouping allows the reader to gain a broad overview of the research that has been identified during the review process. The review team will meet at the start of this process to discuss potential sub-groupings and after all studies have been grouped to discuss how well the sub-groupings fit the data, what new sub-groupings have emerged and data fit, and study sub-grouping overlap. Any changes to the sub-grouping process will be through consensus which may require reviewer 1 to re-visit this first step and re-group studies.

The second step is to develop a summary for each study. This is where key study aspects, findings, author conclusions, limitations, and quality assessment will be described. Quality aspects such as potential biases, confounding variables, participant characteristics, measures, research design, analytic approaches, and key findings will be summarised. Reviewer 2 will audit the same randomly selected 20% of studies for quality assessment coherence, bias, and accuracy. The final step is where reviewer 1 will compare sub-groups in terms of similarities, differences, and scope to enable a synthesis to be produced. The research team will meet at the end of this synthesis process to discuss the conclusions being drawn from the research evidence with any changes being the result of consensus. Stakeholder consultation will also occur at this point with conclusions being shared with key stakeholders for critique and comment.

Step 6 –Stakeholder consultation

Stakeholder consultation has moved from an optional aspect to a strongly recommended part of the scoping review process [57, 58]. This review will engage in stakeholder consultation with consultation focusing on two primary aims. Validation of the scoping review findings is the first with the second being to provide alternative insights, meanings, and implications of the findings. In terms of validation, up to three researchers who have conducted suicidality research will be contacted to engage in a peer review of a draft paper that will overview synthesis and review findings along with the review process. Selection will be based on researchers whose work has not included in the review and who have experience conducting scoping or systematic reviews.

Up to six stakeholders representing online and social media organisations (e.g., Facebook, Twitter), suicide prevention/intervention organisations (e.g., International Association for Suicide Prevention, Suicide Prevention Australia), and lived experience representative bodies (e.g., Roses in the Ocean) will be invited to take part in a zoom focus group where review findings will be presented for consideration from organisational, and usage perspectives. There is no guidance as to the number of stakeholders to engage, thus six has been chosen to be feasible and to allow for two representatives from each broad category to be sought. Stakeholders will be encouraged to consider alternative insights, meanings, and implications of the results. Other stakeholders may also be offered the opportunity to provide written feedback and will be selected on the basis of the review findings. For example, if results indicate that most studies have been conducted using online mental health forums, then mental health organisations that manage online forums (e.g., SANE Australia) will be invited to participate.

Reviewer 1 will complete stakeholder selection with the team meeting to ratify selection based on consensus agreement. Additional broader stakeholder consultation will occur through traditional academic outlets with the outcome of the review being submitted for open access publication and for presentation at key suicide prevention or mental health conferences.

Team composition

Levac et al [57] argue that team composition is critical to the successful conduct and completion of a scoping review. Reviewer 2 has conducted one scoping review as lead author, has published one scoping review protocol, and has experience in suicidality research. Reviewer 3 has expertise in qualitative research and has been involved as a reviewer of a previous scoping review focusing on suicide capability. In addition, reviewer 3 is a registered psychologist with over 25 years as a practitioner. Reviewer 1 has expertise in both qualitative and quantitative suicidality research, has supervised six scoping review studies which includes two published protocols and two published reviews, is currently supervising a further two scoping reviews, and has extensive experience in policy development. Reviewer 1 is also a registered psychologist with over 20 years’ experience as a practitioner.