JMIR Publications

ABSTRACT

Background:

Emergency departments (EDs) are critical access points for people experiencing mental health crises yet transitions to community-based care often break down. Mental Health Clinicians (MHCs) support discharge planning and referral coordination, but this work occurs within fragmented information environments. Digital and artificial intelligence (AI)-supported tools may help, but little is known about what MHCs would trust and use in high-stakes transition workflows.

Objective:

This study aimed to characterize current ED-to-community mental health transition workflows from the perspective of MHCs and identify design requirements for future digital or AI-supported transition support tools.

Methods:

We conducted a qualitative formative study using rapid qualitative methods. MHCs with direct experience supporting ED or urgent mental health discharge planning across three large academic hospital settings in Toronto, Ontario, participated in individual semi-structured interviews over Zoom. The interview guide was informed by service blueprinting concepts and explored transition workflows, decision points, information needs, tools used, referral pain points, and perceptions of digital or AI-supported tools. Data were analyzed using rapid qualitative analysis. Structured summaries were organized into matrices, including a current-state service blueprint and technology-supported design opportunities.

Results:

phases: consult trigger, chart review, patient assessment, consideration of collaboration needs, resource search, referral or discharge, and post-discharge. Across interviews, transition planning involved substantial manual effort, reliance on informal clinician knowledge, diverse collaboration with other care providers, and limited visibility into post-discharge outcomes. Four recurring patterns were identified: relationship-driven and locally embedded transition work; fragmented, outdated, and under-trusted information systems; largely invisible post-discharge outcomes; and conditional, mechanism-dependent trust in digital tools. Six technology-supported opportunities were identified: a trusted resource directory, wait-time and availability visibility, referral tracking, shared psychosocial and social work notes, a human provider directory, and AI-assisted resource matching. Participants suggested that AI may be most useful for filtering verified information, surfacing source-linked options, or identifying missing transition information, rather than generating autonomous or unverifiable referral recommendations.

Conclusions:

MHCs play a unique role in helping patients move from ED mental health care to community-based support. Digital and AI-supported tools may help address transition gaps when they are designed to strengthen, rather than replace, the trusted coordination work performed by clinicians. Future tools should prioritize source transparency, human verification, shared transition documentation, direct provider contact, and referral tracking. Future research should build on these findings through co-design and evaluation with MHCs, patients, peer workers, community agencies, and other care partners. Clinical Trial: N/A