A System and Its Spreadsheets
RIO is the second most used Electronic Patient Record system in the UK, and the most used EPR across Mental Health Trusts. It has dozens of modules covering everything from inpatient management to clinical assessments. One of those modules handles Mental Health Act administration: the legal side of detaining patients, tracking section expiry dates, coordinating hearings, ensuring patients are informed of their rights, and managing consent to treatment.
The MHA module was built over 15 years ago.
Entering and updating information was cumbersome, and the effort returned little practical value. The system stored the data but did little to help administrators see what was due or what needed attention.
So administrators built their own systems. Every Trust I spoke to had Excel spreadsheets tracking the same things the module should have handled: section expiry dates, upcoming hearings, consent review deadlines, rights notifications. One Trust went further and built an entire replacement system in Microsoft Access. Excel was easier to maintain and gave administrators a practical way to manage their work.
(All screenshots display fake data and not real patient information)
Navigating to MHA information meant searching for a patient, clicking into their record, scrolling through a folder tree of every clinical module in the system, finding the Mental Health Act folder, expanding it, and clicking into whichever sub-page you needed. Section details, section history, hearings, consent to treatment, rights: all separate pages, all buried.
The to-do list existed but was barely used. No filtering by ward, no sorting, no priority indicators. Every row looked the same whether it was due yesterday or overdue by five years.
I was tasked with redesigning the MHA module.
I focused the redesign on three priorities:
- Make information easier to capture and maintain.
- Make information easy to understand at a glance.
- Put captured information to work elsewhere in the system.
The Research
There was no existing research on how MHA Administrators use the module. No previous studies, no reports to build on. I had to start from scratch.
I ran 25+ remote interviews with MHA Administrators across multiple NHS Trusts. The early calls were about understanding the basics: what does an administrator actually do all day, what are the legal obligations they're tracking, how do they use the current system, where does it break down. Each call surfaced new details. Administrators showed me their screens, walked me through their processes, and pulled up their Excel spreadsheets to show me what they were really using to manage their work.
Remote interviews helped me understand the work, but familiar workarounds were easy to leave out of a conversation. I wanted to see how administrators moved between records, spreadsheets and email during an ordinary working day.
So I went on-site. Six hospital visits, spending 4-6 hours each time sitting next to administrators, watching them work. They showed me real cases, real patients with long histories, real spreadsheets with actual data. I could see the full picture: the constant switching between RIO and Excel, the manual cross-referencing, the email chains to wards about upcoming deadlines. Things that never came up in a 45-minute video call became obvious after three hours of shadowing.
After the research I understood the domain well enough to propose a first version of the key screens. I returned on-site to validate the prototype with administrators. Their feedback changed what I surfaced: 132 Rights deadlines and nearest relative information belonged in the patient list because administrators checked them constantly. The hearing calendar, shown later, also came directly from an on-site request.
The Structure I Built
The research gave me a practical basis for the redesign. Each view needed to help administrators answer a question or take action using the information already captured. The table connects the main findings to the responses in the validated prototype and the wider workflow proposal.
| Pain point | Design response |
|---|---|
| Upcoming work and urgency were hard to see. | My Week: Daily workload and deadlines grouped in one view. |
| Expiry checks required opening individual records. | Patient List: A sortable view of section expiry dates across the caseload. |
| MHA information was scattered through the clinical record. | Patient Details: A focused overview with access to the underlying records. |
| Hearing preparation was tracked in spreadsheets. | Hearing tracking: Report requests, due dates and received documents connected to progress. |
| Administrators tracked deadlines while wards performed the actions. | Cross-product proposal: Connect MHA information, ward contacts and ward tasks. |
My Week
Administrators needed to plan a week of patient-related work and legal deadlines. I focused the opening view on what they needed to do: which hearings were coming up, which sections were expiring, and which tasks needed attention. An overall compliance percentage would not answer those questions.
I designed My Week around that workload. It shows the current week by day, grouped into Hearings, Sections Expiring, Consent to Treatment Due, and Other Tasks. Each group has a count. Administrators can select a day and filter by location to focus on the wards they are responsible for.
My Week connected captured information to the work administrators needed to manage. Recording a hearing or a deadline could contribute to a useful weekly view. That was the reason to maintain information in RIO that had been missing from the old module.
A Recent Admissions table shows patients admitted in the last 14 days. Administrators told me that ward staff sometimes forgot to send documentation for new admissions. Surfacing those patients gave them a place to check arrivals and follow up on missing paperwork, rather than relying only on what had been sent to them.
Patient List
Checking section expiry dates in the old system meant opening patient records individually. I brought those dates into a sortable patient table so administrators could scan the caseload and sort by expiry date. The same view brings together the current section, consent to treatment status, 132 Rights information and location, with filters for location, section type and responsible clinician.
The expiry status bars make time remaining and overdue sections visible alongside the dates. Administrators responded positively to this during prototype validation. The table addressed a task they previously performed one patient at a time, while keeping the detailed record accessible when they needed it.
Selecting a patient opens a sidebar with section status, consent to treatment, the next hearing and report progress, 132 Rights, and nearest relative information. I designed three levels of detail: the table for scanning, the sidebar for a quick check, and the full patient page for more involved work. A routine check would not have to interrupt the administrator’s place in the list.
Patient Details
In the old system, an administrator had to navigate through a folder tree inside the full clinical record to reach any MHA-related information. Patient demographics, care plans, clinical assessments, medication - all of it was in the same view, and the MHA pages were buried several clicks deep.
I designed a dedicated Patient Details view for MHA administration. The Overview brings together Section Details, Consent to Treatment, Upcoming Hearings, 132 Rights, Nearest Relative, and Section 117 Eligibility. These were the questions administrators needed to answer about a patient’s legal status; the overview made that information available together instead of spreading it across separate pages.
The tabs provide space for detailed work within each area. The Overview supports the quick checks, while the detailed tabs keep the underlying records accessible. This carried the same list–sidebar–detail approach through to the patient record.
Documents
MHA documents were stored alongside every other clinical document in RIO. Administrators had to leave the MHA pages and filter the central document store to find them. In the redesigned frontend, I placed the MHA document view within the patient record so checking and uploading paperwork stayed connected to the task.
The Documents tab includes filters for document type, author and uploader, plus search. Draft and Final states distinguish documents still being prepared from completed versions. The design uses the existing document information in the context where administrators need it.
Patient Details - Hearings
Administrators were tracking hearing preparation in spreadsheets. On the patient page, I brought open hearings together with report progress and actions to open, reschedule or cancel a hearing. The next level of detail explains what is missing and who needs to be followed up.
Each hearing has its own Overview, Reports, Attendees, Panel and Outcome tabs. The Reports view brings together who was asked for a report, when it was requested, its due date, and whether it has been received. The prototype includes actions to add a request, record a received date and upload the document alongside that request. That captured information feeds the hearing’s report progress, making the relationship between maintaining a record and understanding readiness visible.
Hearings
The dedicated Hearings page brings upcoming hearings across patients into one list, with report progress and overdue requests visible together. Administrators can identify a hearing that needs attention, open it to see the missing reports, and follow up on the relevant request. Filters for hearing type, urgency, date, venue and notification status support different ways of reviewing the workload.
An administrator requested a calendar view during an on-site visit. I added a monthly view of the same hearing information, colour-coded by hearing type: Mental Health Tribunals, Hospital Managers Appeals, and Hospital Managers Renewals. An upcoming list gives a chronological view alongside it. This was a concrete example of validation changing the design to accommodate how administrators planned their work.
The Cross-Product Discovery
The investigation also exposed a problem at the boundary of the MHA module: administrators were accountable for deadlines, but ward staff performed many of the actions.
Administrators are responsible for making sure legal deadlines are met: sections renewed before they expire, hearing reports submitted on time, rights read to patients. But administrators can't do most of those things themselves. The wards do the actual work. And the only way administrators had to communicate with wards was email. Some Trusts had binders with ward email addresses. The administrator would look up the ward, write an email, and hope someone would act on it. If they didn't, the administrator would follow up. And follow up again.
This gap had not surfaced when I worked on InPatient. Looking at the process from the administrator’s side showed the repeated follow-up and lack of visibility into whether someone on the ward would act. Understanding both sides changed the scope of the problem.
I identified changes needed across three products. MHA needed a way to notify the people responsible for an action. RIO configuration needed ward contacts and personnel so notifications could reach the appropriate people. InPatient needed to surface upcoming section expirations, hearing deadlines and rights-related tasks in the ward dashboard. The proposal connected information captured in MHA to the people who needed to act on it.
I coordinated with the product managers responsible for these areas and could draw on my previous work redesigning ward configuration. The notifications and ward dashboard changes were proposed follow-up work, not completed features of the administrator prototype. Investigating the whole workflow made that dependency visible.
Working with AI
I used a Claude Project containing interview transcripts, RIO MHA documentation and database structure documentation to cross-reference what administrators described with how the existing system represented that information. It supported research synthesis and design exploration within the backend constraints.
I worked through the specifications and intended behaviour, then used Claude to generate an HTML mockup. I iterated on the structure and interactions before preparing an implementation prompt for our internal prototyping tool, which was similar to Lovable. The tool generated the frontend, and I modified functionality, adjusted components and refined the prototype.
This gave me an interactive prototype to take back to administrators. I could use their feedback to revise the interface and behaviour. The work shown here is that validated prototype; it does not establish a completed production release.
I also worked on the shared component infrastructure. The stack used Tailwind and shadcn/UI with custom components, but needed data tables, date/time pickers and consistent filtering patterns. I designed and built filtering components covering chips, saved views and active-filter dropdown behaviour, then wrote Markdown specifications to guide their use by our LLM tools.
The components and specifications were intended to make those interaction patterns reusable across products. This connected my MHA work to a broader design and prototyping workflow, while keeping the administrator’s tasks and the existing system’s capabilities central to the decisions.
Validation and Constraints
I validated the prototype with MHA Administrators across multiple Trusts. Their feedback supported the direction and changed details of the design, including information surfaced in the patient list and the hearing calendar. Positive responses to the expiry bars and workload views were evidence about the prototype, not measured improvements in live efficiency, compliance or patient care.
The technical boundary was a new frontend on the existing legacy backend. The redesign had to work with the existing database structure and what the current technology could support. Time was limited too, so I had to prioritise an MVP that offered practical value without treating everything in the prototype as immediately deliverable.
Cross-product notifications, ward dashboard changes and document emailing remained proposals for further work. The validated administrator prototype and those wider dependencies were different stages of the work; they should not be read as a single completed release.
My contribution covered investigation, workflow and interaction design, frontend prototyping, validation with administrators, and coordination around the dependencies between products. Working within the legacy system’s limits was part of the design problem throughout.