Understanding Safety in Online Mental Health Forums: Realist Evaluation.

Background: Online forums are used widely to facilitate mental health peer support. However, concerns exist regarding potential harm associated with their use, and little is known about forum safety from the user's perspective. Objective: This study sought to understand how users experience safety within online mental health peer support forums. Following previous research, safety was conceptualized with reference to both experiences of harm and feelings of interpersonal safety within the forum environment. Methods: Data were collected from 42 semistructured realist interviews and 504 cross-sectional survey responses from users of 3 UK-based online mental health forums. These included a forum hosted by a health service provider with subforums for anxiety, depression, and eating disorders; a corporate provider focused on young people's mental health; and a voluntary sector provider with subforums for general mental health support, eating disorders, and postpartum psychosis. A bespoke survey was used to obtain descriptive quantitative data regarding user perceptions of forum safety. Qualitative data were used to refine an initial program theories framework comprising context-mechanism-outcome configurations related to forum safety developed in previously published realist synthesis. Results: Survey responses revealed that over half of the participants felt safe to post because of online anonymity (n=202, 40.1% agreed and n=97, 19.2% strongly agreed), while a minority reported encountering distressing forum posts (n=95, 18.8% agreed and n=18, 3.6% strongly agreed) and expressed concern that talking about mental health online could make them feel worse (n=113, 22.4% agreed and n=17, 3.4% strongly agreed). Refined program theories highlight: (1) the disclosure-promoting effect of anonymity, related to the mitigation of concerns that users' mental health experiences could be linked to their offline identities; (2) the importance of proactive content moderation for addressing emerging safety issues; (3) a need for organizations to implement rule enforcement sensitively and balance between conversational openness and restricting topics likely to cause distress; (4) forum users' experiences of self-moderating their exposure to potentially distressing online content; and (5) how the perceived nonjudgement, authenticity, and similarity of other forum users generates interpersonal safety. Conclusions: This is the first realist evaluation to directly assess processes underpinning safety in online mental health forums. A key novel finding of this study is that safety emerges not only from harm reduction procedures but also from a facilitative interpersonal atmosphere defined by sensitive moderation and the sharing of lived experiences. Hosts should therefore remain attentive to both potential risks and opportunities to foster connections between community members.