Facebook Interest Targeting for Health and Chronic Disease Awareness

If your media plan includes an interest segment for a specific health condition, it is worth checking whether that segment still exists.

Meta removed sensitive detailed targeting options in January 2022, including health causes, and further restricted health-related targeting in the same period. The categories that let advertisers reach people by condition are largely gone.

More importantly, even where you can reach the right audience, your ad copy is not allowed to say so. That second rule causes more rejections than the first, and it is the one most health advertisers misunderstand.

Here is what actually applies now and how awareness campaigns work within it.

The Targeting Options No Longer Exist

Meta’s January 2022 cleanup removed thousands of detailed targeting options tied to sensitive attributes, including health causes, sexual orientation, religious practices, and political beliefs.

What survived: location, broad demographics, general interests, behaviours, Custom Audiences, and Lookalike Audiences.

What did not: condition-level interest segments. You cannot build an audience of people interested in a specific chronic illness the way you could in 2019.

Some published guidance still describes health condition targeting as available. Treat that as out of date and check Ads Manager rather than an article.

The Bigger Rule: Personal Attributes

This is the one that rejects the most ads, and it governs your creative rather than your audience settings.

Meta’s personal attributes policy prohibits ad copy that asserts or implies knowledge of a viewer’s personal characteristics. The named categories include race, ethnicity, religion, beliefs, age, sexual orientation, gender identity, disability, medical or mental health condition, financial status, trade union membership, criminal record, and a person’s name.

The trap is the second person. The moment copy implies “we know you specifically have this,” it fails.

Crucially, the policy applies regardless of whether the assertion is accurate. Even with a legitimately built audience, an ad asking “Are you diabetic?” is still asserting a health condition about the reader.

Enforcement tightened through 2026, with increased automated review and explicit coverage of indirect implication rather than only explicit statements. It applies across Facebook and Instagram, and across every format including carousel cards, Stories, Reels, and dynamic creative variants.

Why Creative Is Now Part of Targeting

Here is the mechanism that explains why the rule exists, and it changed with Meta’s newer ad system.

Meta’s AI reads ad copy, images, and video to determine who is most likely to engage. Creative content functions as a targeting signal in its own right.

So when an ad asks “Do you have diabetes?”, the system uses that language to find people it predicts will respond. Meta is then effectively selecting an audience on a health attribute, whatever you configured in the targeting panel.

Under privacy law, that can constitute a disclosure of protected health information regardless of the advertiser’s settings.

That reframes the policy from arbitrary censorship into something coherent. The creative is now part of the targeting, so the creative has to follow the same rules the targeting does.

Compliant Rewrites

The test is whether the sentence describes what you offer or claims a fact about the reader.

The pattern is consistent: reference the condition in general or third-person terms rather than assigning it to the reader.

Three reliable techniques:

• Replace “you who are X” with “people who want X.” Qualify on intent, not on attribute.
• Move qualification to the landing page. Let people self-select on the destination rather than in the creative. The form can still ask what matters.
• Lead with the outcome, not the identity. “Get help managing chronic conditions” is both stronger and safer than naming the condition as the reader’s.

That third point is worth noting: the compliant version usually reads better anyway. Copy that tells someone what they have is presumptuous even where it is permitted.

What You Can Target Instead

Since condition-level interests are gone, awareness campaigns reach audiences four ways.

Broad targeting with message-led delivery. The system finds responders based on creative signals. Counter-intuitively, this works reasonably well because relevance drives engagement.

Geography. Chronic disease prevalence varies significantly by region, and public health data is published. Geographic weighting does real work without touching protected attributes.

Custom Audiences from your own first-party data. Available in principle, and the highest-risk option legally. See the next section before doing this.

Engagement and video-view audiences. People who watched a condition-specific explainer have self-selected. This is the cleanest way to build a warm audience for awareness work.

That last one is the practical answer for most campaigns. Broad awareness content at the top, retargeting the engaged audience with deeper information.

Where the Legal Risk Sits Beyond Meta’s Policy

Meta’s policy is the floor, not the ceiling.

Health information carries regulatory protection that varies by jurisdiction and by the nature of your organisation. Sending health-related event data to advertising platforms, building audiences from patient or enquiry data, or configuring conversion tracking on pages that reveal a condition can all create exposure independent of anything in Meta’s rules.

Meta’s ad policy enforcement and its data-source classification in Events Manager are separate systems. Fixing ad copy does not resolve a data-source issue, and fixing a data-source issue does not get rejected copy approved.

This guide is general information, not legal or compliance advice. Any organisation handling health data should have this reviewed by someone qualified before launching.

A Funnel Structure for Awareness Campaigns

The shift from stage one to stage two is where self-selection happens. You are not identifying people with a condition. They are choosing to engage with content about it.

That distinction matters legally and it matters practically, because engagement-based audiences outperform inferred ones anyway.

Creative That Works Without Naming the Reader

• Third-person framing throughout. “Many people experience…” rather than “You may be experiencing…”
• Clinician or expert voice. Shifts the register from marketing to information.
• Patient stories told as the patient’s own story, not as a mirror held up to the viewer.
• Educational value first. Awareness content that teaches something gets shared, and sharing is free distribution.
• Clear, non-alarming tone. Fear-based framing performs short-term and is both riskier under review and worse for the people who see it.
• Accessible design. Captions, legible type, plain language. Health audiences skew older in many conditions.

Image and Multimodal Review in 2026

A change worth planning for: Meta’s review now evaluates image content alongside text.

An image depicting a visible condition indicator alongside copy that addresses the viewer can trigger a violation even where the text alone would pass. Thumbnail text, captions, and graphic labels all count.

Review the whole asset, not just the body copy. A compliant headline over a non-compliant image is still a rejection.

Measurement Without Condition-Level Data

You will not get condition-level reporting, and you should not try to construct it.

Track instead:

• Cost per completed video view, as a proxy for genuine attention
• Engagement rate by creative, which reveals which framing resonates
• Landing page completion rate
• Cost per action against your own economics
• Frequency, which climbs quickly with broad targeting and limited audiences

Creative-level performance becomes your main diagnostic, since audience-level segmentation is unavailable. Different framings will attract different responders, and the pattern tells you what worked.

Mistakes That Get Health Campaigns Rejected

• Planning around health interest segments removed in 2022.
• Using second-person copy naming a condition, however accurate.
• Assuming precise targeting justifies precise copy.
• Reviewing text but not images, thumbnails, or captions.
• Building audiences from health data without compliance review.
• Treating a data-source flag and a copy rejection as the same problem.
• Using fear framing because it tests well.

Where to Start

Audit existing creative first. Search your ad copy for “you” or “your” appearing within a sentence of any condition name. That single search catches most personal attributes violations before review does.

Then check whether any targeting in your plan depends on segments that no longer exist.

Only after those two things should you write new creative, because the constraints determine the approach. In health advertising, compliance shapes the campaign before strategy does.

FAQs