Start from the baseline, because it is not the one a medicines checklist assumes. A drugs pack asks whether your claim about a product is substantiated. The NMC Code of Ethics starts further back: a registered practitioner may not solicit patients, directly or indirectly. Your brochure, your landing page and your scan-package post all sit on that.
A hospital ad is not a product ad with a different subject
Start from the baseline, because it is not the one a medicines checklist assumes. A drugs pack asks whether your claim about a product is substantiated. The NMC Code of Ethics starts further back: a registered practitioner may not solicit patients, directly or indirectly. Your brochure, your landing page and your scan-package post all sit on that.
So your exposure is rarely a missing footnote. It is the sentence written to persuade a patient to choose your facility over another one. That takes two shapes, and both are stop-work rather than fix-later: a claim of superiority over a named or implied practitioner or facility, and a guarantee of a clinical outcome, cure or recovery.
This is also why the wrong tool gives false comfort. Screen a diagnostics or fertility brand against a medicines pack and it will pass copy a clinical-services pack stops on sight, including three lines that are not a regulatory matter at all.
Which lines are criminal, not just regulatory?
Three prohibitions here carry their own statutes, and none of them has an equivalent in a medicines pack.
These are not items for a risk-weighing conversation with the brand team. There is no version of them that becomes publishable once a qualifier is attached, and no campaign date that changes it. Take them out, and take out the paraphrases with them, because drafts rarely arrive written in the plain words a keyword list expects.
Prenatal sex determination, or any service implying it. Advertising sex selection is an offence under the PCPNDT Act 1994. The risk is rarely a blunt sentence. It is a friendly one: a gender-reveal line in imaging copy, or a scan page that hints the report answers the question.
The supply of, or demand for, human organs. Advertising organ trade is an offence under the Transplantation of Human Organs and Tissues Act, 1994. Transplant pages that mention donor availability are the usual way in.
Commercial surrogacy and the recruitment of surrogates, prohibited under the Surrogacy (Regulation) Act 2021. Fertility pages that price a surrogacy package, or invite women to enrol, are the copy to go looking for.
What can you publish instead?
Plenty, and it holds up better than the copy it replaces. Four kinds of statement do almost all of the work.
The pattern under all four is the same. Each swaps a persuasive adjective for a fact that carries its own provenance, and each needs that provenance on file before it goes live. Outcome, success-rate, accreditation and capability claims all need a source attached, not a colleague's recollection of where the number came from.
Verifiable capability. The equipment installed and when it went in, the units you run, the hours a department keeps, the procedures your registered team performs. Capability can be checked by anyone who walks in, which is exactly why it survives review.
Named and qualified practitioners. Give the qualification and the registration alongside the name, and do not imply a specialisation that is not registered. A consultant described by their actual credential reads stronger than one described as a leading name.
Outcomes with their denominator. Say whether the figure is per cycle or per patient, the age band it covers, the period it was measured over, and where it came from. A bare percentage is not a claim anyone can check, and an uncheckable number is worth less to you than no number.
Prices with what they exclude. Name the exclusions in the same breath as the figure: implants, consumables, room category, complications and follow-up. A starting-at price with a silent exclusion list is next quarter's complaint, written in advance.
Superiority is the claim that keeps slipping through
Most review processes catch the loud version. Best hospital in the city, number one for cardiac care, top surgeon. Those get flagged and rewritten. The quiet version is what reaches the site: comparative without naming anyone. The only centre in the region that does this. A safer choice than what you were offered elsewhere. An award badge doing the job of the sentence you deleted.
Guaranteed outcomes travel the same road. Assured recovery, painless procedure, complete relief. Softening the verb does not change the promise a patient reads, and the promise is the thing the code stops. The swap is a capability statement: if you commissioned a new scanner, say which one and when it went in, and let the reader draw their own conclusion.
Two neighbouring habits belong in the same sweep. A testimonial or a before-and-after image must not be used to solicit patients or to imply a typical clinical outcome, which rules out the recovery-story format an agency will propose first. And no screening or procedure decision should be pushed with urgency or fear. Limited slots on a cancer screening page is a sentence you do not want read back to you.
Accreditation, telemedicine and diagnostics each need their own line
Accreditation is granted per facility and it expires. So a reference to NABH, NABL or JCI has to name the accrediting body, the entity that actually holds it, and the current validity. A group-level badge sitting on a branch page is a claim about the wrong entity, and it is precisely the thing a patient or a rival checks.
Telemedicine and digital-health content needs an explicit line saying the service is not for emergencies, with urgent cases directed to in-person care, per the Telemedicine Practice Guidelines 2020. It belongs on the page where someone books, not buried in terms nobody opens.
Diagnostic and screening copy needs to say what a result does and does not confirm. A test that flags something for further investigation is not a diagnosis, and copy that lets a reader believe otherwise is the failure people come back angry about. If the service collects reports, images or records, state how that data is handled, consistently with your DPDP obligations.
What happens when an assistant rewrites your page?
It compresses it. An assistant answering where to get an MRI nearby, or what IVF costs, does not reproduce your page. It reads that page alongside a rival's comparison, an aggregator listing and a forum thread, then writes one short reply. The first things compression drops are the qualifiers: the denominator on the outcome, the age band, the exclusion list under the price, the not-for-emergencies line.
That is a compliance problem wearing a marketing costume. A page that passed your review can become an answer that would not have, and you cannot edit the model to correct it. What you can do is treat the answer as an output of its sources and work on the sources.
In practice that means four things. Keep an approved, source-backed set of facts about the facility, since every outcome, accreditation and capability claim needs a source attachment anyway. Sample what assistants actually say about you, per engine, across the questions patients ask. Flag the answers where a qualifier went missing or a superiority framing appeared that you never published. Then correct the strongest source the engine is reading and measure the same questions again, so you can show the wrong version is gone. That last step is the difference between hoping and evidence, and it is the one compliance will ask about.