Participating before changes decisions; participating only afterwards decorates them. The distinction may look like a matter of scheduling, yet in health it determines who defines the problem, which risks are considered acceptable and what evidence will count as success. More accurate medicine does not become more humane because a survey is added at the end of a project. It does so when patients, carers and professionals can influence the work before a tool is purchased, a metric is fixed or a consultation is redesigned. That participation does not replace clinical knowledge; it places it beside the knowledge of those who live with the consequences.
Digital health promises to organise information, anticipate needs and reduce repetitive work. It can also shift invisible costs onto people: more forms, alerts that are hard to interpret, inaccessible channels or decisions presented as inevitable. When design begins by asking what technology can do, everyday experience arrives late. When it begins by asking what care must protect, technology occupies a more precise and modest place.
This article offers general criteria for governance and participation; it does not diagnose symptoms or prescribe treatment. Anyone with a health concern, adverse reaction, worsening condition or uncertainty about a clinical decision should contact a qualified healthcare professional or the appropriate emergency service. Participatory deliberation can improve systems, but it cannot replace individual assessment.
Participation is not validation of a closed decision#
A late consultation can be courteous, extensive and methodologically polished while granting no real influence. If the budget, supplier, timetable and indicators have already been fixed, responses can only adjust the language. The project may claim that people were heard even though no contribution could alter its direction. Such a process produces ornamental legitimacy: it displays participatory activity without accepting the possibility of refusal, delay or an alternative.
A late consultation can be courteous, extensive and methodologically polished while granting no real influence.
Early participation begins before a solution is formulated. The team sets out the observed problem, available evidence and genuine constraints. Patients and professionals can then discuss whether the problem has been described correctly, whether it affects everyone in the same way and whether it warrants a technological intervention. They can also propose non-digital options: longer appointments, better coordination, interpretation, training or a simpler pathway. Keeping the range open prevents a purchase from appearing to be the natural answer to every difficulty.
Participation does not mean that every opinion automatically becomes policy. It means there are public rules for collecting, comparing and answering contributions. The organisation explains what changed, what did not and why. If a proposal cannot be adopted because of safety, evidence, resources or legal duties, the reason should be documented. That traceability turns disagreement into learning and distinguishes a responsible decision from an approval ceremony.
Clinical expertise and lived knowledge need one another#
A healthcare professional contributes training, experience, evidence and the ability to recognise situations that require care. The individual contributes what no record captures completely: priorities, caring duties, tolerance of uncertainty, financial barriers, language, previous experiences and the realistic ability to follow a plan. A carer may notice daily changes that are absent from a brief visit. Administrative staff know where a process stalls. The technical team understands the tool’s limits. No single perspective is sufficient.
Responsible precision emerges when those differences are made visible, not compressed into one score. A recommendation may be statistically reasonable yet impractical for someone without transport, reliable connectivity or leave from work. An alert may identify a relevant signal but cause distress if it arrives without context or a response channel. A portal may improve access for many people while excluding those who need linguistic, cognitive or face-to-face support. Quality must be measured across the full life of the service.
Participation therefore needs diversity and material support. Inviting the same readily available people is not enough. Compensation for time, accessibility, interpretation, workable schedules and help with technical material should all be considered. Perfect representation is impossible, but foreseeable exclusion can be reduced. As far as practicable, the group should reflect those who will use the service, those who may leave it and those who will absorb extra work when it fails.
Case 1: a digital alert designed in the meeting room#
A primary care centre is considering a system that sends alerts when recorded data suggest that a person might need review. The supplier presents good laboratory performance and managers hope to reduce delays. The first proposal would send every alert automatically to the patient’s phone. Before procurement, the centre holds sessions with clinicians, patients, carers, reception staff and data-protection specialists.
The conversation changes the project. Several people explain that an unexplained alert might be mistaken for a diagnosis. Professionals warn that the centre lacks daily capacity to answer all resulting enquiries. A carer notes that the registered phone may belong to another relative. Reception staff anticipate difficult calls outside opening hours. The team decides to test an internal pathway first: a signal is reviewed alongside relevant clinical information, prioritised under an agreed protocol and only then communicated in plain language with a contact route.
Participation does not prove the system effective or authorise indefinite deployment. It does uncover conditions that technical testing missed. The pilot records false alerts, review times, workload, differences between groups and people’s experiences. It also names who can pause the system. If someone receives an alert and has worrying symptoms, professional assessment is needed; a digital message should not be used for self-diagnosis or to delay urgent care.
Case 2: an efficient portal with a doorway that is too narrow#
A hospital wants to move pre-visit questionnaires, appointments and results into a portal. The proposal looks straightforward: less paper, centralised information and fewer calls. The initial consultation includes people with different levels of digital literacy, disabled users, cultural mediators, clinicians and support staff. Testing reveals that two-step verification fails with shared phone numbers, some screen readers do not interpret the forms properly, and literal translation makes important questions ambiguous.
Instead of treating those findings as resistance, the hospital redefines the service. It retains telephone and in-person routes without penalty, provides clear authorisation for support persons, tests accessibility with real users and reviews wording with language and health professionals. The portal stops being the service and becomes one of its channels. Evaluation compares not only how many people complete a task, but who drops out, how much help they need and whether the clinical-administrative outcome arrives in time.
The case shows that early listening does not necessarily slow innovation; it prevents costly dependency. If barriers emerged after deployment, the hospital would have contracts, habits and migrated data that made correction harder. Early participation preserves choice. It also clarifies that a technical difficulty should not be solved by informally sharing credentials or sensitive information; the hospital must offer a secure route and a responsible contact.
Case 3: shared data for research into a rare condition#
A health network proposes reusing pseudonymised data to study a rare condition. The public interest is clear, but affected people raise questions about purposes, access, findings and commercial use. If consultation took place after approval of the repository, it could do little more than alter a privacy notice. By beginning earlier, the group helps define which projects are admissible, what information will be published, how applications will be reviewed and when governance must be renewed.
Patient organisations are not burdened with deciding for the whole community or assessing technical security by themselves. They participate alongside researchers, clinicians and specialists in ethics, security and data protection. The accountable body retains its legal and professional duties. Public project summaries, access controls, use logs, periodic review and understandable communication of findings are agreed. Limits are also stated honestly: pseudonymisation reduces risk, but does not make health data insensitive.
Collaboration can improve the research question and trust, but it cannot guarantee an individual benefit. Participation in a study or repository should never be understood as a promise of diagnosis, treatment or priority access. Personal decisions about tests, symptoms or treatments belong in a conversation with qualified professionals who know the particular circumstances.
Designing influence before sending invitations#
Meaningful participation needs a written mandate. It should state what is open to change, what constraints apply, who decides, how disagreements will be addressed and when a response will be published. It should also allow participants to recommend that the project does not proceed. Without that possibility, the process merely selects details. An organisation may reserve decisions that the law assigns to it, but it should name them at the outset rather than create false expectations.
Timing matters. People need time to understand material, request clarification and compare alternatives. Short plain-language documents are more useful than a sales presentation. Risks and benefits should use absolute numbers where possible, explain uncertainty and distinguish testing, prediction, decision support and a clinical decision. If the organisation cannot explain the system without jargon, it is not ready to request an informed view.
Group composition requires continuing attention. In addition to regular patients, teams should seek those who do not complete forms, miss appointments or prefer other channels, without automatically treating those patterns as lack of interest. Exclusion may arise from working hours, caring duties, housing, language, disability or past unfair treatment. Aggregate data should be reviewed across relevant groups, with care to protect privacy and avoid conclusions based on very small samples.
Knowing whether participation had consequences#
Counting attendees or completed surveys measures activity, not influence. A change log offers better evidence: initial problem, contribution received, decision made, rationale and accountable owner. It is also useful to assess whether people understood the scope, felt able to disagree and received an answer. Clinicians’ and carers’ experiences belong in the evaluation, because a system that saves minutes on one screen may move hours of work elsewhere.
Indicators should be defined before the pilot. They may include safety, time, accessibility, errors, complaints, workload, drop-out and inequalities. No isolated figure represents total quality. Short accounts help explain why a difference appears, while quantitative data help estimate its scale. Evaluation needs thresholds for correction, suspension or withdrawal, as well as a review date that stops a temporary arrangement becoming permanent by default.
There is another essential measure: the ability to challenge an outcome. A person should know whether a tool had an influence, what information it used in understandable terms, whom to contact and how to obtain human review. The answer cannot merely resend the same automated output. Where care decisions are concerned, responsibility remains with professionals and institutions that have genuine authority. A supplier may support them, but cannot absorb the duty to explain, correct and repair.
A culture able to hear bad news#
Participation fails when reporting a problem carries a personal cost. Patients who fear being labelled difficult, professionals who fear retaliation and technical staff tied to commercial targets may withhold essential information. The organisation should create safe channels, respond without defensiveness and distinguish legitimate disagreement from lack of cooperation. Hearing criticism early is a form of safety, not a reputational threat.
Results must also be returned. People who contributed experience deserve to know what happened next. A public summary can identify changes, unresolved issues, pilot findings and the next review without disclosing personal data. That feedback closes the cycle and lets the community see whether its time mattered. When nothing changed, the explanation should be specific; when the project failed, acknowledging failure supports more learning than a story of inevitable success.
Authentic participation does not promise that every proposal will be accepted; it promises that no important decision will be sheltered from evidence merely because it is already advanced.
Accurate, attentive medicine does not choose between technology and listening. It requires every tool to demonstrate value within relationships of care, rights and understandable limits. Participating early redistributes knowledge and forces alternatives to be prepared. Participating afterwards can support evaluation, but it cannot by itself compensate for the absence of initial influence.
Institutional sources for further reading#
The World Health Organization places autonomy, transparency, accountability and inclusion among the principles for ethical governance of artificial intelligence for health.
The Agency for Healthcare Research and Quality provides resources on shared decision-making and communication that bring professional evidence together with an individual’s values and preferences.
NHS England describes shared decision-making as collaboration between clinical expertise and what a person knows about their circumstances, goals and values.
The European Commission summarises the European Health Data Space and its aims for access, control, secure exchange and responsible reuse of electronic health data.
