Designing with patients, not for them

When I started my master’s thesis, on a conversational interface to support IBS self-management, I thought the hard part would be the technology. It was the easy part. The hard part was recruiting patients.

I invited 99 people with IBS to take part in the study. Eight made it through to sessions. A 12-to-1 ratio. And even those eight asked for reschedules, remote sessions, and a level of care in the contact that I hadn’t expected.

This post is about what I learned. It applies to any design work with patients, not just IBS.

Why it’s different from “regular” research

When we recruit for a UX study on a consumer product, participants are there because they want product feedback, like testing things, or want the incentive. When we recruit patients with a chronic condition, they’re there despite:

Not feeling well some days.
Having shared their story a thousand times already.
Fear of judgment about food, mental health, social choices.
Saturation with tracking, forms, “how do you feel today”.
Stigma around conditions whose symptoms aren’t “photogenic”.

This asks for extra care in three areas: recruitment, consent, and session logistics.

Recruitment: clear criteria

For the thesis I defined inclusion and exclusion criteria before sending any invite:

Selected:

All genders, 18-50 years old.
Diagnosed with IBS.
Followed by a doctor for at least 12 months.
Current health state not sensitive.

Excluded:

Patients with other inflammatory bowel diseases (not IBS).
Patients with low digital literacy.

Criteria aren’t there to close doors. They’re there to protect participants (someone in an acute phase shouldn’t be in a research session) and to keep the data coherent (IBS and inflammatory bowel disease have different paths; mixing them obscures patterns).

The rule: if you can’t justify why a criterion is there, drop it.

The informed consent form should say:

Nature and purpose of the study.
What data will be collected (audio, video, transcripts, screenshots).
How it will be anonymised.
Where it will be stored, for how long.
That the participant can withdraw any time, no questions, no penalty.
Who the data controller is.

In Portugal and the EU this isn’t a nice-to-have, it’s GDPR. But even outside that context, it’s the ethical baseline for working with people in a vulnerable state.

If your project is academic, there’s a good chance you don’t need ethics-board approval if you’re not engaging public or private organisations. If you are, prepare for a months-long process. Ask before starting.

Anonymisation: it’s not enough to remove the name

When I interviewed IBS patients, the audios contained:

First names (mentioned by me or by them).
Names of doctors, hospitals, workplaces.
Specific details (age, profession, city) that, combined, identify someone.

Anonymisation means: transcribe without first names (P1, P2, P3…), replace places with categories (“a public hospital in Lisbon” → “a public hospital”), keep raw audios in a separate protected location, hold a mapping key only you have access to.

Small technical note: don’t publish full transcripts in thesis appendices without doing this work. I’ve seen it happen and it leaves identifiable people who had trusted you.

Remote sessions, by default

I decided (rightly) that all sessions for the thesis would be remote. Reasons:

Bad days are frequent for people with chronic conditions. Travelling to a physical location can be impossible.
Bathroom proximity matters. For IBS patients, being near a known toilet lowers anxiety.
Energy is limited. 60-minute remote sessions are bearable. In-person plus travel doubles the cost.
Geographic inclusion. I can talk to someone in the south, in the north, or in another country, with no logistical friction.

Tools I used: Zoom for sessions (with recording consent), Otter.ai for automatic transcription (then manually reviewed), Miro for co-design (with participants screen-sharing).

Logistics: small frictions matter

Some practices that stack up:

Book wide windows. Not “Tuesday at 2pm”, but “Tuesday between 2 and 4”. Lets the participant start when they feel up to it.
Confirm 24h before with a light message, not a formal reminder.
Allow easy rescheduling. Cancelling once is normal. Twice, three times, still re-book.
Short session. 45-60 minutes max. Breaks if needed.
Don’t ask everything. Focus on 3-5 core questions. The rest will appear in the stories that come up spontaneously.

What changes in the study design

Working with patients also changes the study structure:

Empathise ≠ 30-min interview. With a chronic condition, story matters more than answers to questions. Longer, open, narrative sessions.

Define includes per-participant empathy maps. Not only consolidated personas. Each person brings unique patterns that get lost if aggregated too early.

Ideate benefits from co-design. Instead of generating ideas for the patient, generate with them. It was the most valuable phase of my thesis.

Test is with real people, not friends. Patients have intuitions your PM doesn’t.

Covered in Design Thinking in healthcare: what changes.

To get started

Define inclusion and exclusion criteria before inviting anyone.
Write a consent form you’d sign yourself.
Schedule sessions in flexible windows and remote by default.
Prepare for a 10:1 ratio or worse between invites and participations.
Treat the data with care. Anonymisation isn’t optional.

More on the background in the Design for Health guide. On how to map what you learn from these participants, see Patient care journeys. On the ideate phase (co-design workshop) and the rest of the process, see Design Thinking in healthcare. On research in general, see Product designers and research.