Process & Research

A patient checking a result at home, a nurse switching between tasks on a busy ward and a carer managing someone else's medication all meet the same digital product in very different conditions. They may be unwell, worried or short of time. If a design is judged only by the team that built it, problems that matter to these people can stay hidden until launch, and in health a misread number can carry consequences beyond frustration. Process and research means getting evidence from real users while there is still time to change the design.

Why colleagues and stakeholders are not stand-ins

UK government service guidance tells teams to observe and interview actual or likely users, and to look beyond typical ones: people who struggle with existing services, and staff who help others use them. Colleagues know the product and its jargon, and rarely share the stress of the people it is for.

US guidance on medical-device usability engineering makes a sharper point. Employees can join early, exploratory evaluations, but their behaviour and opinions may mislead if they differ from intended users, already know the product or hesitate to be frank. For the final safety validation it advises against employees, apart from rare cases where every user is one. That guidance covers medical devices, software included, and is nonbinding, but the reasoning carries over.

Who counts as a user

• Patients, including functional limits their condition brings, such as the eye or nerve complications of diabetes (the FDA guidance's example).
• Carers and family members who act on someone's behalf.
• Clinicians in different roles. The FDA guidance treats healthcare professionals and lay users as separate populations, each needing its own participants.
• Administrative and support staff who use the product daily.
• Disabled people and people with limited digital skills or low literacy. GOV.UK guidance says participants should include them, and NHS guidance suggests roughly one person with access needs in every five.

Finding people takes time

Recruitment is often the slowest step. GOV.UK guidance says using a recruitment agency typically takes about ten days. Finding disabled participants, or people who use assistive technology, may take longer, and it advises allowing up to a month. For less common cognitive disabilities it advises contacting relevant organisations at least six to eight weeks before usability testing. It points to charities, professional bodies and community groups for specific groups, and calls an incentive for members of the public normal. One study of recruiting clinicians found obstacles at every stage, from getting into facilities to scheduling. Inaccurate contact information slowed it most, and about two thirds of the entries in one filtered list were ineligible.

Testing where and how the product is used

Two terms need defining. Formative evaluation means exploring a design while it is still changing, to find strengths, weaknesses and unexpected mistakes. Summative evaluation means a final check on a finished design; the FDA notes the word is defined in different ways and prefers the phrase human factors validation testing. Early on, it lists observing users in their real setting. For final testing it asks for realistic conditions such as dim lighting, distractions and multitasking, and says thinking aloud distorts natural behaviour in a validation test.

GOV.UK advises frequent small rounds over one large study: roughly one round every two weeks, usually four to eight participants for interviews and usability tests. Surveys and A/B tests need hundreds. AHRQ, a US health research agency, suggests a clinic choosing health IT let its own future users try the system, with the vendor not in the room.

Two illustrations

Imagine testing a medication-reminder flow. Colleagues at a desk will tap through it and agree it is clear. Older adults at home, with reading glasses in another room and a spouse asking a question, may misread a dose or miss the reminder. The second session is more likely to show what the design must survive. This is an illustrative scenario, not a reported study.

Co-design, where users help shape the design rather than only react to it, extends the idea. A 2022 systematic review of 25 studies with older adults found nine different methodologies, only about a third showing two-way exchange of knowledge, and few using validated usability measures. Claims of co-design deserve a look at what was done.

Consent, privacy and honest findings

GOV.UK guidance asks researchers to gain informed consent, give an information sheet in advance, collect only the personal data needed and delete it promptly. With disabled participants, it advises speaking to the person rather than an interpreter, and asking before offering help. Whether a project counts as research needing ethical review is a separate question, and the answer depends on the country. In the UK, a Health Research Authority table distinguishes research, which aims at generalisable knowledge, from service evaluation and improvement, and says which normally need NHS ethics review.

Small samples show what can go wrong, not how often. The FDA warns that early results can understate or overstate how frequent a problem is. Its minimum of 15 participants per user group applies to final validation of medical devices, not every study. A review of electronic health record vendors found user-centred design claims hard to compare because test scenarios and participant descriptions varied.

Australian context

Australia has its own framework, separate from the UK material above. A Victorian Government page says research involving people must be reviewed by a human research ethics committee (HREC) when its risk is above what the National Statement on Ethical Conduct in Human Research counts as lower risk, and that lower-risk projects may not need full HREC review. The National Statement is issued by the National Health and Medical Research Council (NHMRC). The page, last updated in October 2025, refers to the 2025 edition and says its start was postponed, so check the NHMRC site for the edition now in force.

Quality improvement and evaluation are handled differently from research. One Sydney health district describes quality improvement as measuring, monitoring and evaluating performance within a single organisation, and lists triggers for ethics review, including direct contact with patients or staff, data beyond routine collection, identifiable individuals and plans to publish. It says each health organisation decides its own review process and points to NHMRC guidance. That is one district's approach, not a national rule.

Consumer involvement is a separate expectation. Summaries of the Partnering with Consumers Standard, part of the National Safety and Quality Health Service (NSQHS) Standards from the Australian Commission on Safety and Quality in Health Care, describe consumers taking part in service planning, care design, and management and evaluation, reflecting the diversity of the community served. It governs health services, not product testing. Separately, a May 2026 announcement describes a new NHMRC and Consumers Health Forum of Australia statement on consumer and community involvement in health and medical research, expecting involvement across the whole research lifecycle.