skip to main content
iatroX JournalClinical insight

Patient Decision Aid, Information Leaflet or AI Summary: Which Does Your Consultation Need?

Featured image for Patient Decision Aid, Information Leaflet or AI Summary: Which Does Your Consultation Need?

Use an information leaflet when the main task is explaining a condition or process. Use an appropriate patient decision aid when the task is choosing between reasonable options with different consequences. An AI summary may help organise information, but a readable summary is not automatically a validated decision aid or a substitute for the consultation.

Find the unanswered question

Consider an original fictional consultation about an elective intervention. The patient has received a detailed leaflet explaining the procedure, its preparation and recovery. They can describe what happens in hospital, yet remain unsure whether to proceed because taking time away from caring responsibilities would be difficult.

Sending the leaflet again addresses the wrong problem. The unresolved question is not "What is the procedure?" but "How should I weigh the possible benefit against disruption, uncertainty and other reasonable options?"

A useful consultation makes that distinction early. Ask what decision is being considered, whether it needs to be made now and what the patient feels is missing. Do not infer indecision from a lack of medical knowledge when the real obstacle is a practical consequence the leaflet barely mentions.

What a decision aid adds

The Ottawa Hospital Research Institute inventory, checked on 19 September 2026, provides a way to find decision aids by condition, treatment or body area. Inclusion requires a minimum set of criteria. Its current notes also describe an update to the inventory's assessment criteria, with some information still being supplied by developers.

An inventory entry is therefore a useful route to a resource, not permission to stop evaluating it. Read the aid itself. Identify its intended population, the actual decision, its evidence date, available alternatives and how it helps the user consider their preferences.

A resource that describes only one intervention attractively may be informative without supporting a balanced choice. Likewise, listing benefits and harms is not enough if the patient cannot relate them to the decision they face.

A small resource-selection exercise

Return to the fictional patient. Imagine three documents on the desk.

The first explains the procedure and recovery in plain language. The second compares proceeding now, postponing while pursuing an appropriate alternative, and choosing not to proceed, using evidence relevant to that decision. The third is an AI-generated paragraph that says the procedure is generally effective and advises discussing concerns with a clinician.

The first may be the best source for practical preparation after a decision. The second is more relevant to the current preference-sensitive choice, provided its options and evidence fit the clinical situation. The third might be a useful draft introduction, but its reassuring tone cannot compensate for missing alternatives or unsupported claims.

These are fictional document descriptions, not observations from a product test. The exercise illustrates a selection method: judge a resource against the unanswered question, not against its length, polish or technology.

Values clarification is not a hidden recommendation

Ask the patient what would make one option more acceptable. Is the concern the likelihood of benefit, recovery time, treatment burden, uncertainty, or something else? Avoid supplying a preferred answer through the wording of the question.

For example, "You would want the most effective option, wouldn't you?" narrows the discussion prematurely. "Which consequences would matter most in deciding?" leaves space for the person to explain the trade-off.

The Ottawa decision guides, checked on 19 September 2026, offer a general structure for clarifying decisions and next steps. The site explicitly distinguishes these generic guides from condition-specific patient decision aids that meet all qualifying criteria. That distinction is useful: a broad conversation aid can help without being relabelled as something it is not.

Do not assume the patient must make the decision alone. Some people want a clear clinical recommendation after their priorities have been understood. Supporting participation means responding to that preference rather than withholding professional judgement behind a worksheet.

Check the evidence and the setting

Before using a resource, compare its options with those actually available in the local service. A document built around another healthcare system may describe access routes or treatments that are not applicable. Correct that mismatch openly rather than silently editing the patient's expectations.

Examine how benefits and harms are presented. Are they measured over the same period? Are probabilities given where reliable estimates exist? Is uncertainty visible? Are important burdens omitted because they were not the primary research outcome?

Also check accessibility in the ordinary sense of use: can the patient read the text, understand its language, view its illustrations and revisit it later? An evidence-based aid that cannot be used by the person receiving it has not yet solved the communication problem.

These questions form an original consultation checklist. They do not certify a resource or replace a formal decision-aid quality assessment.

Using AI without disguising its role

An AI tool can be asked to organise a clinician's notes into questions for discussion, but the output must be checked against the evidence and intended audience. Do not label generated material "validated" because it contains references or resembles a professional leaflet.

Keep patient information within approved systems and avoid uploading identifiable details to an unapproved service. Where a summary is adapted for distribution, agree who owns clinical review, accessibility checks and subsequent updates.

This article is published by iatroX and includes its own complementary role. According to the public methodology checked on 19 September 2026, Ask-iatroX provides source-linked clinical reference. That can support a clinician's preparation, while appropriate simulation practice can rehearse explaining uncertainty and eliciting concerns. Neither function turns an unreviewed generated paragraph into a patient decision aid.

For the fictional consultation, the practical next step is to select a suitable aid, discuss the patient's priorities and agree what happens next. More information is useful only when it addresses what remains unresolved.

Frequently asked questions

Is a patient information leaflet the same as a decision aid?

No: a leaflet may explain a condition or procedure without comparing reasonable options or helping a person weigh their preferences. Its usefulness depends on the consultation's task.

Does inclusion in an inventory mean I can use an aid without checking it?

No: inspect the intended decision, population, evidence date and local applicability. An inventory helps find resources but does not remove that responsibility.

Can an AI summary be used during shared decision-making?

A checked summary may support discussion, but it should not be presented as a validated decision aid without the relevant development and evaluation. Keep its origin and limitations clear.

Prepare a source-linked clinical explanation with Ask-iatroX →

Back to Journal