Health sites in India sit under two sets of rules at once. The search and AI systems treat medical content as high stakes and look hard at who wrote it and whether it can be trusted. The professional regulations treat doctors as practitioners who may inform but may not solicit, and the advertising law restricts what can be claimed about treatments at all. Structured data touches both, because it is the most literal statement a site makes about itself, and it is read by machines that do not understand nuance.
This is a guide to doing it properly. The broader trust picture for healthcare sites is in healthcare SEO in India; this piece is only about the markup.
First, what medical schema does not do
It does not produce a medical rich result. Google does not show special search features for MedicalCondition, MedicalProcedure or Physician markup the way it does for recipes or events. Anyone promising that marking up a condition page will get it a special panel is describing something that does not exist.
It does not override the trust assessment. A page with perfect markup and no identifiable author, no credentials and no review date is still a page with no identifiable author.
What it does is remove ambiguity. It tells a crawler or an assistant that this page is about a clinic rather than a company that happens to use the word clinic, that this named person is a physician with this specialty at this organisation, and that this article was reviewed by a named clinician on a stated date. Entity clarity is the whole value, and for AI assistants assembling an answer about doctors or treatments in a city, it is a meaningful one. The general evidence on citation is in does schema markup help you get cited.
The types worth using
Keep to a small set and use each where it fits.
MedicalOrganization, Hospital, MedicalClinic. One of these for the organisation, on the home page and the contact page, with address, phone, opening hours, medical specialties offered, and the departments or branches as separate entities where they have their own pages. A hospital with six branches is six MedicalClinic or Hospital entities with a shared parent organisation, not one entity with six addresses.
Physician. One per doctor, on the doctor’s own page. Name, medical specialty, the organisation they work at, qualifications as text, and the registration body membership where the doctor is comfortable stating it. Link it to the organisation entity.
MedicalWebPage. On condition, procedure and health information pages. The useful properties are the subject of the page, the audience, who reviewed it and when it was last reviewed. This is the markup that does the most work for trust, because it names an accountable clinician.
MedicalCondition and MedicalProcedure. As the subject of a MedicalWebPage rather than as the page type itself. Keep them to the name and an identifier where one exists; the detailed sub-properties for symptoms, causes and risk factors are rarely worth the maintenance and are where regulated claims creep in.
FAQ markup, breadcrumbs and article markup apply as on any site, with the author pointed at the Physician entity rather than a generic byline.
A doctor page, marked up
{
"@context": "https://schema.org",
"@type": "Physician",
"@id": "https://example-hospital.in/doctors/a-sharma#physician",
"name": "Dr A Sharma",
"medicalSpecialty": "Cardiology",
"description": "Consultant cardiologist. MBBS, MD (General Medicine), DM (Cardiology).",
"worksFor": {
"@type": "Hospital",
"@id": "https://example-hospital.in/#organization"
},
"url": "https://example-hospital.in/doctors/a-sharma",
"image": "https://example-hospital.in/images/doctors/a-sharma.webp"
}
Notice what is absent. No rating. No “best cardiologist in the city”. No count of procedures performed. No success rate. The description states qualifications as a fact, which is permitted, and stops.
A condition page, marked up
{
"@context": "https://schema.org",
"@type": "MedicalWebPage",
"@id": "https://example-hospital.in/conditions/atrial-fibrillation",
"name": "Atrial fibrillation: symptoms, diagnosis and treatment options",
"about": {
"@type": "MedicalCondition",
"name": "Atrial fibrillation"
},
"audience": {
"@type": "MedicalAudience",
"audienceType": "Patient"
},
"reviewedBy": {
"@id": "https://example-hospital.in/doctors/a-sharma#physician"
},
"lastReviewed": "2026-09-14",
"dateModified": "2026-09-14",
"publisher": {
"@id": "https://example-hospital.in/#organization"
}
}
The reviewedBy reference points at the same Physician entity as the doctor page, by identifier, so the two pages agree about who the person is. The lastReviewed date should be real and should change when a clinician actually re-reads the page. A review date that updates automatically on every template change is worse than none, because it claims a review that did not happen.
Where the advertising rules meet the markup
The professional conduct regulations for doctors in India restrict solicitation and self-promotion, and the law on advertising of drugs and remedies restricts claims about treatment for a list of conditions. The detail of what is permitted is a matter for the hospital’s legal and compliance people, not for a markup guide, and it changes. What a markup guide can say is this: anything that would be a problem in the visible page copy is a bigger problem in structured data, because structured data is designed to be extracted and repeated without its surrounding context.
The practical consequences:
- Do not put comparative or superlative claims in any schema field. Name, specialty, qualifications, organisation, address, hours.
- Do not put treatment outcome claims, cure language or success rates anywhere in markup, including the description of a MedicalProcedure.
- Do not mark up prices for procedures unless the hospital is comfortable with those prices being quoted in isolation, which is how an assistant will quote them.
- Do not mark up testimonials or patient stories as reviews. Patient testimonials are themselves restricted in many contexts, and review markup on a doctor is self-serving under the search guidelines in any case.
- Do not attach aggregate ratings to Physician entities. If the organisation displays third-party reviews, the rating belongs to the organisation entity, must match what is visible, and must come from a source the site does not control.
A useful test: read each field as if it were a sentence in a newspaper advertisement bearing the doctor’s name. If it would not be allowed there, it does not go in the markup.
Authorship for health content
The piece of markup that most affects how health content is treated is the author and reviewer. A health article attributed to the marketing team, or to nobody, is treated accordingly. An article with a Physician author entity, a reviewer entity, qualifications in text and a real review date is treated as what it is.
Where content is written by a medical writer and reviewed by a clinician, say so: author is the writer as a Person, reviewedBy is the Physician. That is honest and it is what the trust assessment is looking for. Where a clinician wrote it, the author is the Physician. Never attribute authorship to a doctor who did not write or review the piece, which happens more often than hospitals admit, usually because a template assigns a default author.
Keeping it consistent across the site
The fault we see most on hospital sites is not wrong markup but inconsistent markup. The doctor page says one specialty and the department page says another. The organisation entity on the home page has a different name from the one on the contact page. Three doctor pages share one identifier because somebody copied the block.
Use stable identifiers, one per entity, and reference them rather than restating the entity. Generate the markup from the same database fields that render the visible page, so the two cannot drift. If the site runs on a CMS, generate it from templates and validate it on every build, which is the approach in automating structured data from your CMS. Hand-written JSON-LD blocks on a site with two hundred doctor pages will be wrong within a year.
Departments, branches and specialties
A department page is a MedicalClinic or a MedicalOrganization entity of its own with a parent organisation, a specialty and the physicians who work in it, each referenced by identifier. A branch is the same with an address. A specialty landing page that lists the doctors for that specialty marks up the list of physician references and nothing more.
The useful outcome is a graph: organisation, branches, departments, doctors, condition pages reviewed by those doctors. An assistant asked who treats a condition at a hospital in a city can follow it. A hospital site with that graph answers questions that a site with a hundred loose pages cannot.
Testing
Validate every template with the standard validator, then do the step most teams skip: read the extracted entities back as plain text and check them against the regulations with somebody who knows them. The validator tells you the markup is well formed. It does not tell you the description field contains a claim that should not be made.
Then check the cross-references. Every reviewedBy and worksFor identifier should resolve to an entity that exists on the site. A reference to a doctor who left two years ago is both a trust problem and, potentially, a regulatory one.
What is a waste of effort
Marking up every symptom, cause and risk factor of every condition with the full set of medical sub-properties. Nobody consumes it, it is hard to keep accurate, and it multiplies the places where a claim can go wrong.
Adding review and rating markup to doctor pages to get stars in results. It is against the guidelines for self-serving reviews and it is the first thing a regulator would screenshot.
Using a plugin’s generic LocalBusiness output for a hospital. The medical types exist and they are the ones that say what the organisation is.
Chasing medical rich results that do not exist.
A sensible order of work
- Decide the entity model: organisation, branches, departments, physicians, pages. Assign one stable identifier to each.
- Mark up the organisation and branches with address, hours and specialties.
- Build the Physician entity for each doctor from the same fields that render the doctor page, with no claims beyond qualifications and role.
- Add MedicalWebPage markup to condition and procedure pages with a real reviewer and a real review date.
- Strip anything comparative, outcome-related or price-related from every field, and have compliance read the extracted text.
- Automate generation and validation so the markup cannot drift from the pages.
- Re-check cross-references whenever a doctor joins or leaves.
Dental practices have a narrower version of the same problem, covered in SEO and GEO for dental clinics. If your hospital site’s markup has never been read by anyone who knows the regulations, get in touch.