PORTFOLIO · SERVICE DESIGN & UX RESEARCH · HEALTHCARE
Closing the Transition Void for Post-Discharge Psychiatric Patients
Redesigning the 48 hours after discharge, when no one formally owns the patient's care.
Helping case managers see a coordination gap that used to live only in their memory.
RoleService Designer · UX ResearcherContextExtern × Acute Psychiatric Hospital, Southeast Asia, 2026ScopeMapped the coordination breakdown between inpatient and outpatient psychiatric care, identified the 48-hour discharge window as the highest-risk gap, and designed a staff dashboard, a patient portal, and a pilot plan to close it.StatusResearch-led service-design proposal prepared for primary validation and operational pilot
Role
Service Designer · UX Researcher
Context
Extern × Acute Psychiatric Hospital, Southeast Asia · 2026
Scope
Mapped the coordination breakdown between inpatient and outpatient psychiatric care, identified the 48-hour discharge window as the highest-risk gap, and designed a staff dashboard, a patient portal, and a pilot plan to close it.
Skills
Service Design, Systems Mapping, Stakeholder Analysis, Root Cause Analysis, Prototyping, Change Management
Outputs
20 responsibilities mapped15 stakeholders2 prototypes1 pilot plan5 personas · 2 journey maps
Problem
Patients discharged from acute psychiatric care routinely lose support in the first 48 hours. The case manager tries to fill the gap, but she is working from memory, a spreadsheet, and whatever time she has left after everything else on her caseload.
Problem
What problem were we trying to solve?
Patients discharged from acute psychiatric care routinely lose support in the first 48 hours. The case manager tries to fill the gap, but she's working from memory, a spreadsheet, and whatever time she has left after everything else on her caseload.
The problem wasn't that she was bad at her job. It was that no system existed to hold the handoff.
→WHAT THE RESEARCH FOUND 20 case-manager responsibilities, 45% with no clear owner 15 stakeholders, 67% with shared or unclear ownership 7 high-stakes tasks, 5 with no backup at all A protocol for follow-up existed, with no alert if it was skipped
Research
Stakeholder mapping, responsibility mapping, fishbone analysis, 5 Whys, risk analysis, and decision mapping surfaced a consistent pattern: responsibility was concentrated on the case manager, and authority wasn't.
Research
What did the research actually show?
Stakeholder mapping, responsibility mapping, fishbone analysis, 5 Whys, risk analysis, and decision mapping surfaced a consistent pattern: responsibility was concentrated on the case manager, and authority wasn't.
This is secondary-research-led service design. It points to a proposed future state, not a validated one. It still needs testing with patients, case managers, and staff.
This was an independently scoped, secondary-research-led service-design sprint. The findings reveal structural patterns and informed a proposed future state, but they require validation with patients, case managers, and healthcare stakeholders before implementation.
Research at a glance
20
Case-manager responsibilities
Mapped across five operational domains
15
Stakeholders
Spanning clinical care, coordination, crisis response, administration, external services, and leadership
2
Prototypes
A staff dashboard and patient portal
1
Pilot plan
A People, Process, and Technology plan
Why was responsibility concentrated while authority was not?
Case-manager responsibilities · 20 mapped
11 defined
9 gap
55% defined45% ownership gap
45%
of mapped responsibilities contained a documented ownership, authority, or communication gap
Coordination carried the most ambiguity
Discharge planning, social-work collaboration, referrals, and cross-system follow-through repeatedly lacked a single designated owner.
Crisis responsibility exceeded authority
The case manager could detect deterioration and maintain safety plans but often required physician approval before activating an intervention.
External work lacked infrastructure
Community coordination and specialist follow-up relied heavily on phone, fax, manual tracking, and case-manager reconciliation.
Stakeholder ownership · 15 mapped
5 clear
3 shared
7 gap
33% clear20% shared47% gap
67%
of stakeholder relationships had either shared or unclear ownership
The case manager sat at the center of the network with maximum cross-system responsibility but limited independent authority.
40%
of stakeholders were categorized P1 critical
Six of fifteen stakeholders were essential to immediate safety, clinical authority, leadership, or patient participation.
Research Synthesis
Why did the gap stay unfixed?
7 high-stakes post-discharge tasks analyzed
71%No backup · 5 of 7
29%Partial · 2 of 7
0% of tasks had a complete secondary safety net
Post-discharge telephone contactPre-discharge psychosocial assessmentFirst-appointment actualizationMissed appointment follow-upHandover to another serviceMonitoring-level assignmentUncontactable-patient escalation
A protocol existed without an enforcement mechanism
The process specified follow-up and escalation, but the system generated no alert when those actions did not happen.
The highest-risk window had the weakest visibility
The 0–48 hour period began at discharge, but the first active contact was scheduled for the case manager’s next available working window.
Transmission did not guarantee fidelity
A handoff could route successfully while omitting risk history, caregiver fragility, or medication concerns. The system confirmed transfer, not completeness.
“The protocol was not the safety net. The case manager was.”
What conditions created the gap?
6
domains
18
contributing conditions
System fragmentation
Separate institutions, records, and governance structures prevented patient context from traveling cleanly across the handoff.
Discharge process
A patient could leave with paperwork but without an activated outpatient relationship, confirmed medication bridge, or warm handoff.
Information and communication
Medication instructions, crisis pathways, and caregiver guidance were not consistently delivered in usable, patient-facing form.
Workforce and capacity
Large caseloads and invisible coordination labor made post-discharge work the first layer to compress under pressure.
Policy and structure
The transition had no formal owner, no dedicated reimbursement mechanism, and no metric measuring whether continuity was actually achieved.
Patient and social conditions
Patients returned to housing, support, literacy, and financial conditions that could destabilize even a technically sound discharge plan.
What chain of causes kept the gap in place?
1
Two systems were designed separately, leaving the transition between them unowned.
2
Each system was funded and organized around its own episode of care rather than the patient’s continuous journey.
3
The transition produced little activity the system could bill, measure, or use to justify dedicated resources.
4
Case managers absorbed the gap through informal and largely invisible labor.
5
When that informal system failed, the harm was recorded as patient disengagement, readmission, or employee burnout rather than as a transition-design failure.
The system misdiagnosed its own failure
Outcomes were recorded, but the coordination breakdowns producing them were not. As a result, interventions repeatedly targeted patients and individual employees instead of the structural gap.
Did the pathway work the same way for every handoff?
Same system
Partial system support
The clinical system could sometimes flag a missed appointment, although follow-up still required case-manager action.
External provider
Entirely person-dependent
The case manager had to call the patient, caregiver, or clinic manually. No shared record or automatic missed-appointment notification existed.
Transferred service
Handover overlap
Routing existed, but continuity depended on the completeness, urgency, and quality of the sending case manager’s documentation.
All four critical moments relied on the case manager remembering, prioritizing, and acting without a system prompt or supervisory signal.
Research synthesis
Why did the gap stay unfixed?
01
Ownership disappeared between systems
The transition was treated as the end of one service and the future start of another, not as a phase of care with its own accountable owner.
Design requirement
Make the transition visible, assigned, and actively managed.
02
The service relied on invisible labor
Case managers compensated for missing infrastructure through memory, phone calls, manual tracking, and professional vigilance.
Design requirement
Turn informal coordination into visible workflow and measurable workload.
03
Activity was mistaken for resolution
A scheduled appointment, transmitted handoff, or logged attempt could appear complete without confirming that the patient was reached, understood the plan, or entered care.
Design requirement
Measure meaningful resolution, not only recorded activity.
04
Risk increased when context crossed boundaries
External-provider pathways lost shared records, automatic alerts, and real-time visibility, making them categorically different from same-system follow-up.
Design requirement
Design explicit checkpoints for cross-institutional continuity.
The design question became: “How might we create formal ownership, live visibility, and reliable action logic for the transition between systems?”
Current State
What did the same 72-hour service look like from each side?
Nadia, case manager, caseload of 80 — calls once, leaves a voicemail, logs “attempted,” moves to the next patient. At hour 72, her record shows Marcus as contacted and scheduled. Marcus, patient, discharged, lives alone — misses the call, doesn’t recognize the number, misses the appointment, doesn’t know who to ask. At hour 72, he’s withdrawing. Her system says resolved. His experience says otherwise. That gap is the transition void.
Role 01 · Case Manager
Nadia Kowalski
Case Manager · Caseload: 80
"I know which patients need me. What I lose track of is when."
Core tension
When the system fails, Nadia absorbs the consequence invisibly and without institutional record.
View full profile →
Role 02 · Consultant
Dr. Reuben Asante
Consultant Psychiatrist
"I don't need to see every patient. I need to see which ones are about to fall through."
Core tension
The escalation protocol exists but has no system trigger. If Nadia forgets to escalate, he never knows.
View full profile →
Role 03 · General Manager
Sandra Cheung
General Manager · CMHT
"I can't see the system failing until someone raises a complaint."
Core tension
Readmission is recorded as a patient outcome. The data that should drive structural change disappears into statistics.
View full profile →
Role 04 · Patient
Marcus Tran
Patient · High-risk · Lives alone
"I don't know if I'm doing this right. I don't know who to ask."
Core tension
Discharged with a plan that assumes capacity he doesn't yet have. The system looks compliant. He is deteriorating.
View full profile →
Role 05 · Caregiver
Linh Tran
Informal Caregiver · Marcus's sister
"Nobody told me what to watch for. I just knew something was wrong."
Core tension
Holds the most real-time clinical observation in the system. Has no formal role, no briefing, and no one to call.
View full profile →
Current State · Journey Maps
Why did the same 72 hours produce two completely different experiences?
Nadia’s map tracks a case manager working at capacity, relying on memory and protocol. Marcus’s map tracks a patient navigating recovery without a clear point of contact or a system that can detect his deterioration.
Nadia Kowalski · Case Manager
Stage
01
Discharge
Hour 0
02
Returns to caseload
Hour 0–6
03
First contact attempt
Hour 6–18
04
Voicemail logged
Hour 18–36
05
Missed appointment
Hour 36–60
06
72-hour check
Hour 60–72
Actions
Signs off discharge plan. Queues follow-up call mentally.
Returns to full caseload. Marcus is one row in a spreadsheet.
Calls Marcus. No answer. Notes it mentally and moves on.
Leaves voicemail. Logs “call attempted” in EHR. Moves to next patient.
Notices missed appointment if the system flags it. Attempts another call.
Reviews caseload. Marcus’s record shows: contacted, appointment scheduled.
Thoughts
“I’ll call him first thing tomorrow.”
“I have six urgent patients today. Marcus is stable at discharge.”
“He probably just needs a day to settle in.”
“I’ve logged the attempt. That’s the protocol.”
“I can’t tell if he’s withdrawn or just uncontactable.”
“The system says he’s scheduled. I’ll follow up next week.”
Emotion
Stretched
Pressured
Concerned
Uncertain
Worried
Falsely reassured
Feeling arc
+−
Pain points
No time built into discharge for a proper handoff briefing.
Caseload hides urgency. No signal for which patients need first contact soonest.
No automatic prompt for high-risk post-discharge contact. Memory is the only trigger.
Protocol counts an attempt the same as a completed contact. A missed patient looks resolved.
No cross-system alert for missed appointment. Flag is system-dependent.
EHR records activity, not outcomes. She cannot see Marcus’s actual state.
System record
Discharged. Follow-up scheduled.
No activity logged.
No automatic alert. No record without manual entry.
Arrives home alone. Finds paperwork confusing. Sets it aside.
Tries to follow medication instructions. Instructions unclear. Misses evening dose.
Sees missed call. Doesn’t recognise the number. Doesn’t call back.
Forgets the appointment. No reminder arrived. Stays home.
Still at home. Has not re-engaged with any service. Withdrawing.
Thoughts
“I’m fine. They said I can go home.”
“I don’t know who to call if something happens.”
“Maybe I should just wait and see how I feel.”
“I don’t want to bother anyone.”
“Maybe they forgot about me.”
“I don’t know if I’m still a patient.”
Emotion
Relief
Anxious
Uncertain
Isolated
Defeated
Invisible
Feeling arc
+−
Pain points
No verbal walkthrough. Paperwork not in plain language.
No named contact in plain sight. Crisis line feels too serious to call.
Medication instructions unclear. No prompt. No one to ask.
Doesn’t know who called or that returning the call matters for his care.
No appointment reminder. No one checks whether he is coming.
Deteriorating. No one in the system can see it.
System record
Discharged. Appointment scheduled.
No activity logged.
No activity logged.
“Call attempted.” No follow-up triggered.
Appointment missed. No automatic flag raised.
Compliant. Contacted. Appointment-scheduled.
72-Hour Post Discharge Service Design Blueprint
Service Blueprint · From discharge conversation to 72-hour system state
→At 72 hours, the EHR shows Marcus as compliant, contacted, and appointment-scheduled. In his room, he is deteriorating. The only clinical-quality observation belongs to Linh, an untrained caregiver who was never briefed on what to watch for or who to call.
The patient-facing breakdown could not be fixed without redesigning the staff workflow and supporting systems behind it.
Future State
Two proposed tools — one for staff, one for patients — designed to close the void from both sides at once. These are future-state proposals, not validated or deployed implementations.
Future State · Staff Tool
How do we make the invisible visible?
Transition Void Monitor is a staff dashboard that turns the 48-hour window into something you can see, not something you have to remember.
Three role-scoped views: the case manager sees her full queue, the consultant sees only cases needing clinical sign-off, the general manager sees team-level patterns and no patient names.
Transition Void Monitor
Role-scoped coordination dashboard · Case Manager, Consultant, General Manager views
Click to interact →
→This is a backstage solution. It improves the conditions for staff doing coordination work. But it cannot reach the patient. The void is also experienced from the other side: by patients who don't know what's happening, who to call, or whether anyone is thinking about them.
Future State · Patient Tool
How do you keep the patient from disappearing?
Bridge is a phone-based portal for a patient with reduced capacity, right after discharge. It answers three questions: what do I do, when’s my appointment, who do I call.
A daily check-in flags low mood straight to the case manager’s dashboard within two hours. Marcus doesn’t need to know the system is watching. He needs to feel like someone is.
Bridge — Patient Recovery Portal
Daily check-in, appointment info, AI-assisted call practice
Click to interact →
→When Marcus checks in, the signal appears on Nadia’s dashboard. Nadia acts, and the resolution is recorded. The loop closes between the patient’s experience and the staff workflow.
Pilot & Adoption
A People, Process, and Technology plan, because a well-designed tool that nobody trusts doesn’t close anything.
Pilot & Adoption
What does it take for this to actually get used?
A People, Process, and Technology plan, because a well-designed tool that nobody trusts doesn’t close anything.
The plan addresses who introduces the tools, how existing work transitions, what resistance is likely to emerge, and how the pilot is evaluated before broader implementation.
People · Process · Technology
People
A senior case manager leads the rollout
The rollout is led by a senior case manager, not the tech team. The champion is someone colleagues already trust when workflows break down.
Four hours of protected paid time a week supports peer demonstration, early troubleshooting, feedback collection, and translation into everyday clinical work.
The champion has authority to pause the launch if it creates an unsafe workflow.
The champion is a protected operational role with enough authority to influence implementation.
Training happens inside the work
Training would be delivered in four stages:
Stage 1
Micro-demo
A 10-minute huddle demo.
Stage 2
Peer simulation
A peer simulation showing how a Bridge patient flag surfaces in the case-manager workflow.
Stage 3
Elbow-to-elbow support
Five days of in-person support.
Stage 4
The discharge moment
A short patient walkthrough at discharge.
This approach treats training as part of service delivery rather than a separate classroom event. It allows staff to learn the tools at the moments when they will actually use them.
Process
Replace the workaround gradually, not ceremonially
The dashboard and spreadsheet run side by side for two weeks.
The dashboard becomes primary in week three.
Nothing gets deleted until the need it was meeting is understood. Estimated time saved: 15 minutes per shift, projected, not yet measured.
Nothing gets deleted until the need it was meeting is understood.
Reduce work instead of relocating it
The dashboard surfaces the 48-hour window in an urgency-sorted queue. It is designed to remove duplication, not add another interface.
Estimated time saved: 15 minutes per shift, projected, not yet measured.
Implementation estimate
Estimated time saved: 15 minutes per shift, projected, not yet measured.
Technology
The lowest-friction version is the most viable version
The dashboard opens from a link inside the existing clinical system, with no install and no login.
Patients reach Bridge through an SMS link, with no app and no account.
Check-in takes under 30 seconds. One alert per patient per day.
The lowest-friction version is the most viable version.
Accessibility meant reducing logins, downloads, alerts, and time — not merely meeting visual interface standards.
Full change management plan
The complete plan details the champion case-manager role, staged training model, spreadsheet transition, technology-access decisions, anticipated resistance, pilot metrics, and review cadence.
Opens the complete change-management plan as a PDF in a new tab.
→The implementation strategy responds directly to the research findings. Invisible coordination work becomes measurable. Person-dependent follow-up becomes a visible queue. Patient check-ins create a route back to the case manager. Existing workarounds are transitioned rather than abruptly removed. Adoption is treated as a service-design problem involving people, workflow, authority, and trust — not merely as a technology launch. The future-state service was designed not only to function, but to take root.
Reflection
What did this project change about how I think about design?
This is a research-led proposal. Real problem, rigorous research, grounded tools. Nothing here is deployed or validated with real users yet.
1
The gap is structural, not a training issue.
No case manager can close it alone, no matter how good she is.
2
Following the system, not the symptom, is what found the real failure.
Solution design came last, after stakeholder mapping and root cause analysis.
3
Role-scoped design was a safety requirement here, not a preference.
The same information shown to everyone would have created noise for the people who could act on it.
4
The adoption plan is part of the design, not an add-on.
A tool nobody trusts doesn't get used, no matter how good the interface is.
5
None of this is proven yet.
The open questions: does the dashboard get used when caseload pressure peaks, is Bridge simple enough for someone 48 hours post-discharge, do the mood alerts surface real signal or become noise.
If I kept going
I’d add caseload-weighted risk scoring to the dashboard and build a live data link between Bridge and the dashboard.
I’d also design the tool this research surfaced but didn’t build: something for caregivers, who carry real monitoring load with no formal support at all.
