Systematic search of the literature

Round table consultations

Six roundtable consultations were conducted: three with social media companies (n = 7), two with Australian policy makers from the Department of Health (n = 14), and one with young people (n = 7; five identified as female, mean age = 20.7). Policy makers and representatives from the social media companies were known to the research team and were recruited via email by JR. Young people were recruited via a social media advertisement posted on the #chatsafe social media accounts. Young people who had previously participated in #chatsafe activities were notified of the participation opportunity via email. All young people were paid for their time. The roundtables were conducted between June and August 2022 by JR with assistance from LH, RBC, and LLS. The discussions were structured around the following questions: 1) From your perspective what are the challenges associated with online communication about self-harm and suicide?; 2) What more (if anything) could social media platforms and policy makers be doing to keep young people safe online; 3) To what extent do you think online safety is the responsibility of government, platforms and/or individuals; and 4) What else would you like to see in the new #chatsafe guidelines or associated resources?

Each session was audio recorded and transcribed verbatim. Although only the fourth question directly asked about the guidelines, all questions prompted spontaneous responses related to online communication about self-harm and suicide, therefore, transcripts in their entirety were searched for relevant statements and were extracted by PT, LH, and RBC. A qualitative analysis of this component of the project is underway and will be reported elsewhere.

Questionnaire development

Statements were extracted from 149 peer-reviewed articles, 52 grey literature sources (including the original #chatsafe guidelines), and the six roundtables. They were then categorised thematically in a spreadsheet. A working group of three researchers (JR, PT, and LH), all of whom were experienced researchers in suicide prevention and the Delphi expert consensus method, omitted statements that contained repetitive information, and, where required, reworded statements for consistency and to contain one clear behavioural recommendation while preserving the original meaning (‘action item’). The statements from the original #chatsafe guidelines were all included in the questionnaire. The working group also created ‘action items’ based on their experience and feedback from young people on the original #chatsafe guidelines (e.g., items related to influencers). Assistance was provided, as required, by a fourth researcher (NR) who had expertise in the Delphi expert consensus method.

The Round 1 questionnaire contained 427 items and was organised into 10 sections (see Table 1). At the end of each section, panellists (see panel formation section below) were able to submit comments or suggestions to be included and rated in the Round 2 questionnaire. Two paid youth advisors (AD and EU) provided feedback on the survey to ensure that it was youth friendly.

Download:

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

Table 1. Sections included in the Delphi questionnaire and example items.

https://doi.org/10.1371/journal.pone.0289494.t001

To develop the Round 2 questionnaire, the working group reviewed the panellists’ comments, and, if they were original ideas that had not been included in the Round 1 questionnaire, they were added to the Round 2 questionnaire. The Round 2 questionnaire also included items that did not reach consensus to be included or excluded from the guidelines in the Round 1 questionnaire and required re-rating. Overall panel ratings obtained in Round 1 were reported alongside each of these items. The Round 2 questionnaire comprised 118 items to be rated for the guidelines. Again, there was an opportunity for panellists to provide final comments.

This questionnaire also included a series of complementary questions to be rated by the panels on what they thought social media companies and policymakers could be doing to improve online safety. Examples of these items included: “Social media companies should provide support to all employees and volunteers working with self-harm or suicide content”. The findings from this additional component of the survey will be analysed and reported separately.

Panel formation and recruitment

Two expert panels were recruited. One comprised suicide prevention professionals and one comprised young people.

Panels of around 20 members produce stable findings in Delphi studies, with one study reporting that 23 experts per panel produced stability in response characteristics [30, 32]. In total, our panels comprised 103 members. Of these, 29 were professionals and 74 were young people.

Delphi questionnaires

Data were collected in two questionnaire rounds via the web-based survey platform, Qualtrics. Panellists were asked to indicate if each item should be included in the new guidelines for young people on safe online communication about self-harm and suicide. For pragmatic reasons, three response options were used: “Yes”, “Unsure”, and “No”. These were defined for participants as ’Yes’ = agree, include this item in the guidelines; ’Unsure’ = unsure if this item should be in the guidelines; and ’No’ = disagree, exclude this item from the guidelines. The middle point was included because of the sensitive subject matter and because neutrality or uncertainty was considered a valid option. Different rating scales including the yes/no scale are used for Delphi consensus studies, but ultimately the process ends with a dichotomous result and consensus is generally not reached for all items due to pragmatic reasons [34–37]. The questionnaires were divided into nine sections representing common topic themes. Panellists were able to provide free text comments and suggestions at the end of each section.

Guideline development

The research team combined all included items that contained similar content and PT wrote them into prose for the final guidelines with assistance from JR and the wider research team. Given the audience, the authors used lay language to ensure the guidelines were accessible and acceptable to young people. The final draft was reviewed by three paid youth advisors (MG, AD, EU) who provided feedback on language and style. Careful consideration was given to ensuring that the final guidelines were true to the original meaning of the questionnaire items whilst still being coherent and easy to read. The draft guidelines were then provided to panellists for final feedback and endorsement. Three professional panellists requested minor changes, which were implemented (e.g., adding a few additional lines of psychoeducation, linking back to language tips, and clarifying and simplifying terms).

Systematic search results

Peer-reviewed literature.

In total, 13,705 articles were retrieved via database searching. Following initial screening, 599 full-text articles were retrieved, of which, 149 met our inclusion criteria (see Fig 1).

Download:

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

Fig 1. Prisma flow diagram.

https://doi.org/10.1371/journal.pone.0289494.g001

Questionnaire results.

The participation rate of panellists completing the two rounds of questionnaires was 65.1% (58.1% for youth and 82.8% for professionals; See Table 2). Fig 2 shows the number of items included, excluded, and rerated during the two questionnaire rounds. The panels both rated 454 items in total (427 items from Round 1 and 28 new items in Round 2 based on participant feedback in Round 1). Overall, 191 items of the 454 items (186 from the original questionnaire and five feedback items provided by panellists) were rated as “yes” for inclusion in the guidelines by at least 80.0% of both panels (S1 File shows the individual item results for the two questionnaire rounds). The correlation between the two panels was strong in both Round 1 (r = 0.90, p < .001) and Round 2 (r = 0.81, p < .001).

Download:

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

Fig 2. Flow chart of included studies and questionnaire items by Delphi round.

https://doi.org/10.1371/journal.pone.0289494.g002

Download:

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

Table 2. Round 1 and Round 2 questionnaire participation rates by panel.

https://doi.org/10.1371/journal.pone.0289494.t002

Despite the general agreement, there were some categories and individual questionnaire items where the youth and professional panels had a higher proportion of disagreement. These items did not meet consensus, and so were not included in the final guidelines. For example, items that received notably higher scores (>20.0%) by the professional panel focused on not posting content related to self-harm and scars at all. In contrast, the youth panel were more likely to endorse items that placed an increased level of responsibility on the person posting or interacting with the content: By monitoring or limiting post responses; posting trigger warnings; gaining permission from relevant parties such as family or the affected person before posting about suicide and suicide related behaviours; undertaking risk assessments or providing support to others; or actively seeking to discredit or reduce the risk of suicide hoaxes or self-harm behaviour. It is important to note that most discrepant items did not meet the threshold for inclusion in either panel, limiting their impact on the results.

The final guidelines are organised into the following eight sections: 1) General tips; 2) Creating self-harm and suicide content; 3) Consuming self-harm and suicide content; 4) Livestreams of self-harm and suicide acts; 4) Self-harm and suicide games, pacts, and hoaxes; 6) Self-harm and suicide communities; 7) Bereavement and communicating about someone who has died by suicide; and 8) Guidance for influencers. Where appropriate, self-harm and suicide guidance was combined; however, 18 items were specifically related to interactions about self-harm regardless of motive (e.g., “avoid comparing self-harm scars or injuries” and “if appropriate, urge the person to seek professional help for any inquiries”). The guidelines are publicly available via the following website: https://www.orygen.org.au/chatsafe.

How the new guidelines compare to the original ones

The most salient difference between the original and new guidelines is the inclusion of guidance on how to safely communicate online about self-harm. Self-harm is prevalent among young people and highly stigmatised (even in health services), which can deter formal help-seeking [38–41], meaning that young people may be more likely to experience and be exposed to self-harm compared to suicide and to turn to social media platforms for help and support. Several reviews have shown that online communication about self-harm is nuanced and associated with potential benefits and potential harms [29, 42–47]. Consequently, young people need guidance on how to safely create and consume online content related to self-harm, and thus #chatsafe was expanded to include self-harm. In Round 1, some items that were excluded were highly endorsed by the youth panel, but not the professional panel. Unsurprisingly, professionals did not endorse young people taking a very active role when providing support (e.g., “Post first-aid advice on how to care for [self-harm] injuries”). This reluctance to endorse intervention was also seen in items related to fear appeals such as posting about the risks of self-harm (e.g., scars and permanent injuries). Indeed, in Round 1, most discrepancies between the panels related to what young people should not do when responding to self-harm content. Six out of the 10 items had to be rerated in Round 2. Eventually, only two self-harm specific items did not reach consensus: Posting advice on how people can self-harm without injuring themselves too badly and not using humour when someone has engaged in self-harm behaviour.

Another change between the original and new guidelines is the inclusion of a specific section for influencers (also known as social media content creators). To the best of our knowledge, although some resources do exist for people posting about their lived experience, no specific and publicly available evidence-informed guidance on how to communicate safely about self-harm or suicide exists for this group. This is despite their considerable followings, and, in some cases, frequent posts about mental health including their own self-harm and suicide experiences. This can be problematic in multiple ways. For example, they may share their own self-harm or suicide experiences in unsafe ways, they may find themselves in a dialogue with vulnerable young people and not feel equipped to respond safely, or they may provide advice that is not evidence-based or even potentially harmful. Further complicating matters, often influencer content includes paid ads in disguise. This is something that social media companies are aware is a problem, and some have started to deliver training to influencers to help them talk about health related topics more responsibly [48]. Congruently, the Australian Therapeutic Goods Administration (as well as other international policymakers) have implemented strict provisions relating to influencer marketing such as rules on ad disclosures [49]. Although codes of practice currently relate to goods and services, their existence speaks to the broader issue of consumers making choices based on influencer testimonials that are not always factual or evidence based, indicating that guidance is also needed for areas such as mental health. Indeed, traditional media and journalists have codes and considerations for reporting on suicide-related content [50]. However, social media has made it possible for anyone to create, post, and share content, which is not regulated in the same manner.

The new guidelines also include sections on suicide games, pacts, hoaxes, and livestreams, all of which have caused significant concern in recent times. For example, a recent and well-known challenge that allegedly encouraged people to engage in increasingly serious acts of self-harm that ultimately led to suicide, reportedly went viral on social media, and led to significant concerns about the safety of younger users. Although some have concluded that the game itself was likely a hoax, and much of the communication about it involved users warning others about the potential harms, it received significant attention from the mainstream media, and the potential for distress and potentially harmful impacts remained [51–53]. In the current study, both panels agreed that content relating to a suicide game should not be posted or shared, and if users encounter this type of content, they should report it to the relevant platform.

Livestreams operate differently to other types of content including pre-recorded videos. These are real-time and unedited videos of people engaging in acts of self-harm, some of which result in suicide, delivered to consumers without any significant delays or time gaps. By that fact, unlike pre-recorded videos, the content creator may be more impulsive, and the consumers can engage with the content creator in real time and may not have any warning of what is about to be streamed. Again, these have received significant attention in the mainstream media [54–56]. They also present challenges for the social media platforms themselves, as on the one hand they have the potential to spread quickly and expose large numbers of people to a real-time suicide, but on the other hand if the post remains live there is the potential for the platforms (or viewers) to intervene [57]). In the current study, however, both panels agreed that, despite the potential for intervention in real time as opposed to retrospectively, acts of self-harm or suicide should not be livestreamed. This is reflected in the new guidelines, which specifically recommend that people do not engage with livestreams of self-harm or suicide acts and if people encounter them, they should avoid interacting with them, rather they should report them directly to the platform and emergency services (if appropriate).

There were also some items that were in the original guidelines that did not reach consensus for inclusion in the new guidelines. For example, directly asking someone if they are suicidal was excluded by both panels. Including trigger or content warnings ahead of an online post about suicide or self-harm was also excluded. A recent meta-analysis concluded that content or trigger warnings (albeit not related specifically to self-harm or suicide-related content) neither reduce engagement with, nor negative reactions to, sensitive content [58]. However, trigger warnings are still commonly used and as such guidance may still be useful for young people. For this reason, we decided to include some information on this, as well as explaining why it is not unsafe to ask someone if they are suicidal in the guidelines, but we presented in such a way that it was clear that they were not endorsed by the panels.

Implementing the guidelines

As before, the new guidelines will be housed on the #chatsafe website and will be disseminated via a national social media campaign that will be co-designed with young people across the country. They will also be supported by a series of adult-facing resources including for bereaved communities, family members, and educators. Funding permitting, they will be translated for an international audience as was the case with the original guidelines [21].

Until recently, there have been few (if any) studies that have actively involved young people in the co-design of a universal suicide prevention intervention [59, 60] and limited data exists regarding the effectiveness of suicide prevention public health campaigns. However, social media metrics (e.g., engagement, impressions, and reach) from the original campaign demonstrates that this approach can reach large numbers of young people quickly. Further, data from the evaluation of the co-design process showed that it was not just safe and acceptable to involve young people in developing a suicide prevention intervention, but it also had inherent benefits. For example, young people reported feeling better able to communicate safely online about suicide after participating in the co-design workshops, as well as better able to identify and respond to someone at risk of suicide [25]. Similarly, in a pilot study, social media content based on the original guidelines showed that participants’ willingness to intervene online against suicide as well as perceived self-efficacy, confidence, and safety when communicating on social media about suicide increased [26].

Strengths and limitations

This study has several strengths and limitations. A key strength is the innovative and highly translational nature of the guidelines. A further strength has been the involvement of young people, both with and without lived experience of self-harm and suicide. As noted above, despite growing evidence demonstrating the benefits of involving people with lived experience in research [61], this is still rarely done in youth suicide research [59, 60]. However, our previous studies have shown that involving young people can be both safe and beneficial for young people, [25] and, as such it was considered essential to have young people involved in this study from start to finish.

In terms of limitations, the nature of the Delphi methodology means that some items that may be important did not reach consensus, possibly because they may not be appropriate in most situations, and, as a result, are not included in the final guidelines. This includes items that were endorsed when developing the original guidelines. This means that there are areas for which guidance is lacking (e.g., the use of content warnings and how to ask someone online if they are feeling suicidal). There are some groups who were under-represented in the youth panel, such as adolescents, Aboriginal and Torres Strait Islander peoples and people from diverse cultural backgrounds. This means that the guidelines may be less applicable to these populations. Additionally, some groups were overrepresented, for example, many of our participants identified as female or gender and sexually diverse; however, LGBTIQA+ young people are overrepresented in suicide statistics. Moreover, as noted above, most youth panellists had living or lived experience of self-harm or suicide; therefore, parts of the guidelines may not resonate with the general youth population. However, again, these young people are most likely to engage in communication about self-harm and suicide and are at risk of repeat self-harm and future suicide and need assistance. A final limitation applies to Delphi studies more broadly and relates to the fact that studies such as this, that generate evidence via a process of consensus are useful when it is hard to generate the evidence using more robust study designs [30]. However, that said it remains important to then evaluate the guidelines produced to ensure that they are effective in achieving their aims. The original #chatsafe guidelines have been evaluated in a pre-test post-test study with promising results and are now being tested in a randomised controlled trial [27].