How to write about healthcare products under FDA and Health Canada rules

A content brief comes back from legal with two thirds struck through. Efficacy line: gone. Competitor comparison: gone. Dosage tip your medical team loved: gone.

The instinct is to think the compliant version can't compete. It's the opposite. Whether you're writing under FDA rules or Health Canada regulations, the constraints legal hands you push content into the exact shape AI assistants prefer to cite: category-level, condition-first, provenance-heavy. This piece is editorial guidance, not legal advice. Every regulatory claim below links to a primary source, but run anything real past your legal team.

What rules actually bind healthcare content in the US?

US prescription drug promotion falls into three communication types. Get the type right and most of your compliance work is done before a lawyer reads a word.

A product-claim names the drug and its indication together. That triggers the fair balance requirement under 21 CFR 202.1(e)(5)(ii), meaning risk information must carry comparable prominence to benefit information. A reminder names the drug, dosage form, quantity, and price, but says nothing about what it treats (prohibited for drugs with a boxed warning). A help-seeking or disease-awareness piece discusses the condition and names no product, so it falls outside promotional rules.

The trap most teams miss: if a help-seeking piece and a reminder ad share graphics or sit adjacent on the same page, FDA can treat them as one combined product-claim ad. Keeping them legally separate means keeping them visibly separate. And 202.1(e)(4) bars promoting a drug for an unapproved use or before approval.

One thing worth clarifying, because most articles get it wrong: the Health Insurance Portability and Accountability Act (HIPAA) governs patient-data privacy, not promotional content. It has nothing to say about whether your page can call a drug "effective."

Why is compliant medical content writing suddenly urgent?

Enforcement moved from a trickle to a flood. On 9 September 2025, FDA issued roughly 100 cease-and-desist letters the same day the White House published a Presidential Memorandum on direct-to-consumer (DTC) drug advertising. In Q1 2026 alone, FDA's Office of Prescription Drug Promotion issued nine untitled letters and one warning letter.

One data point should change how you write. The share of AI medical answers that included a disclaimer fell from 26.3% in 2022 to 0.97% in 2025. Assistants have quietly stopped adding their own safety language. Whatever caution reaches the reader now has to live in your source content. A separate physician-led study found unsafe answers in 5% to 13% of 888 chatbot responses. Well-sourced material keeps you compliant and gives the assistant something careful to cite when nobody else is being careful.

What is the Health Canada bright line for prescription drug content?

Health Canada's rules are stricter than most US marketers expect. Food and Drug Regulations C.01.044 caps consumer-facing prescription drug content at four things: brand name, proper name, common name, and price/quantity. Say what the drug treats and you're over the line. Section 3(1) of the Food and Drugs Act, read with Schedule A, also bars advertising a treatment or cure for roughly thirty serious conditions (cancer, diabetes, depression, heart disease, and others).

The bottom line: if you name the drug and the condition in the same consumer-facing message, you're offside. Health Canada programs split into reminder content (names the product, says nothing about the disease) and disease-awareness content (discusses the condition, names no product). Preclearance by the Pharmaceutical Advertising Advisory Board (PAAB) and Ad Standards is voluntary. Your liability is not.

What do you write instead?

Six moves turn a struck-through brief into content an assistant will quote, covering both US and Canadian rules at once.

1. Write the category, not the brand. People ask assistants "what helps with migraine," not "tell me about [brand]." Category-level content matches the question and stays clear of product-claim territory.

2. Separate condition from product completely. Different pages, different design, so they can't be read as one combined ad.

3. Replace mechanism-of-action language with patient-decision language. Write about the questions a patient brings to an appointment, not how the drug works in the body.

4. Drop efficacy and authority words near product context. "Safe," "effective," "proven," "clinically proven" turn a page into a product claim.

5. Remove all dosage and combination references. Dosing guidance belongs to the label and the prescriber.

6. Write at a grade 8 reading level. Regulators favor it for comprehension, assistants favor it for clean extraction.

The rewrite table

Real sentences a marketer would write, the rule each one breaks, and a compliant version that keeps the value.

Provenance is the only claim you have left

When you can't make claims, provenance is your only currency. Show who stands behind the page, what it's built on, and when it was last checked. Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) rewards exactly that, and health content sits in the Your Money or Your Life (YMYL) category, held to the highest standard.

Four signals carry the most weight. Put a named clinical reviewer with credentials and a visible review date on the page. Link to primary sources (regulators, peer-reviewed literature) inline. Date your updates so freshness is legible to readers and crawlers. And make the author a real, identifiable person rather than "the team." For the deeper mechanics of building citable healthcare pages, our healthcare industry page goes further.

What happened when we ran it

One Yolando client in a heavily regulated category went from 0% to roughly 30% category relevance in AI answers within weeks. We suppressed the brand name entirely: no product names, no efficacy language, no dosages. Just category-level content with visible clinical review, built to answer the condition question rather than sell against it.

The content became citable because it matched what people actually ask assistants, and it survived their safety filtering because there was nothing promotional to filter out. This is one engagement's result, not a guaranteed outcome (Yolando client engagement, 2026, anonymised), but it shows the shape of the opportunity.

The pre-publish checklist

Run these ten checks before anything goes to legal. Fix failures first; legal review should be an editorial input, not the place your problems get discovered.

1. No named product paired with an indication on the same page.

2. No efficacy or authority words ("safe," "effective," "proven," "#1") near product context.

3. No dosage, dosing schedule, or combination guidance.

4. No mechanism-of-action claims tied to a product.

5. Help-seeking and reminder content are visibly separate.

6. For Health Canada: no drug name and condition in the same consumer-facing message.

7. No Schedule A condition advertised as treatable, preventable, or curable.

8. A named clinical reviewer with credentials and a review date is present.

9. Every regulatory or clinical claim links to a primary source.

10. The page reads at roughly a grade 8 level.

The compliant version is the citable one

An assistant can quote compliant content without tripping its own safety filters, because you already did the safety work in the source. Teams that bring legal review in early will own their categories in AI answers. Teams that treat it as a final gate will keep watching good briefs come back covered in red.

Want to see which category questions you could own today? Map your AI answer gaps with Yolando.