Substantiating clinical outcomes and physician credentials in marketing

The billboard that needs a filing cabinet

A hospital wants "94% cancer survival rate" on a highway billboard. Before that number goes up, someone has to build a file: the source dataset, the cancer type and stage, the time window, the peer comparison, and wording that says what "survival" counts as. Miss a piece and you have bought mediabought mediaAny media you pay for: display ads, search ads, social ads, and sponsorships. You buy access to someone else's audience on a per-click, per-impression, or flat-fee basis.View full definition → for a regulatory liability.

This lesson is about the file. What proof supports a survival, volume, safety or "top-ranked" claim, and how credentials and rankings have to be qualified before a patient reads them. That objective claims need a reasonable basis held before launch is the ground the advertising-law lesson lays; the question here is what that basis is made of.

What counts as evidence, and in what order

Substantiation is not one tier. When a state attorney general or a plaintiff's expert asks what you held, these sources answer with very different force.

• A peer-reviewed publication of your own cohort, methods section included, is the strongest thing a hospital can hold. Someone else has already checked the arithmetic.
• A submission to an audited external registry comes next: the National Cancer Database, the STS National Database for cardiac surgery, ACS NSQIP for general surgery. The value is standard definitions and risk adjustment you did not write yourself.
• National benchmark data (SEER, run by the National Cancer Institute) substantiates context, not your performance. It tells you what average looks like.
• An internal dashboard or an EHR extract is the weakest. It can still work, but only with the query, the run date, the inclusion and exclusion rules, and a named analyst behind it.

One overlay belongs here: if the ad names a device or a drug ("we use the XYZ robotic system for prostate cancer"), FDA promotional rules attach and the manufacturer's cleared indication caps what you may say. A robot cleared for a procedure has not been cleared to produce better cancer outcomes, and hospital ads that blur those two things attract letters.

The FTC's own guidance is the clearest free primer on evidence quality: FTC Health Products Compliance Guidance.

Building the evidence file for a survival claim

Take "94% cancer survival rate." Here is what the file must contain before launch.

1. The exact metric definition

"Survival rate" is ambiguous. Specify:

• 5-year relative survival, the standard oncology measure: patients alive five years after diagnosis, adjusted for deaths unrelated to the cancer.
• The cancer type and stage. A 94% figure is plausible for localized breast or prostate cancer and impossible for late-stage pancreatic cancer. A blended "cancer" number across all sites is meaningless and reads as deceptive.

2. The source and the denominator

Your own registry, or a national benchmark used for comparison, or both. Then the size question. A claim built on 28 patients over two years carries a confidence interval wide enough to swallow the national average whole, which means the point estimate is advertising, not evidence. Under roughly 100 cases, quote a range or say nothing. Volume claims have the mirror problem: "over 5,000 procedures" counted across eleven system sites, by billing code rather than distinct patients, is a number that will not survive being asked to show its query.

3. The worked calculation

Say the file shows 500 patients diagnosed with localized prostate cancer, 470 alive at five years, adjusted for unrelated deaths.

Relative 5-year survival = survivors / diagnosed cohort
                         = 470 / 500
                         = 0.94  ->  94%

Now the honesty test: is 94% better than average? SEER data has long put localized prostate cancer 5-year relative survival near 99% (NCI estimate; check the current SEER release for its as-of date). If the national figure beats yours, advertising 94% as excellence is misleading by implication, and implied claims are treated exactly like express ones.

This is why Martini-Klinik in Hamburg, a prostate-only clinic performing radical prostatectomies on the order of 2,000 a year, does not market survival. In localized disease almost everyone survives five years, so the metric cannot separate anyone. It reports patient-reported continence and erectile function instead, the outcomes that actually vary between surgeons. Picking a metric where you can honestly differ is upstream of substantiating it.

4. The comparison claim trap

"Better survival than the national average" needs an apples-to-apples comparison: same cancer, same stage, same measure, same window, cited source. Comparing a localized-only cohort to a national all-stages average is cherry-picking the denominator.

The subtler version is case mix. A hospital that refers its complex and comorbid patients to an academic centre will show better raw outcomes than the centre that accepts them, and an ad built on unadjusted rates rewards that referral pattern. If your registry offers risk-adjusted figures, use those and say so; if you advertise the unadjusted number because it looks better, the file records the choice.

5. The disclaimer

Disclaimers must meet the clear-and-conspicuous standard the advertising-law lesson sets out, and format decides whether they do. A line nobody can read at 65 mph does not cure a billboard headline. A six-second pre-roll has no room for a qualifier at all, so the claim itself has to be self-limiting. Instagram truncates captions, so anything after "more" is not a disclosure. And a disclaimer cannot rescue a claim that is deceptive on its face; it clarifies a truthful claim that needs context ("5-year relative survival, localized prostate cancer, [source], [year]").

🎬 [VIDEO: "Truth In Advertising: The Basics" - youtube.com - FTC-style overview of substantiation and the clear-and-conspicuous disclaimer standard]

Substantiating physician credentials

The second common claim: "our surgeons are board-certified experts." Credentials have their own file.

Board certification is verifiable, so verify it

"Board-certified" is a defined status, not marketing language. Verify through the ABMS certification database before publishing, and keep a dated record showing each named physician is currently certified in the specialty you are advertising.

Failures that turn into deceptive claims:

• Advertising a physician as "board-certified in oncology" when the certification is internal medicine and only the practice is oncology.
• Running the ad after certification lapses or the physician resigns. Print and out-of-home have no kill switch, and a bus shelter naming a surgeon who left in March is still recruiting patients in September.
• Using "expert" or "leading" as if they were credentials. These are puffery up to a point, but attach a specific ("leading, with a 94% success rate") and they become objective claims needing proof.

"Top doctor" badges and US News rankings

If an asset carries a US News & World Report badge, the file holds three things: the licence permitting use of the mark, the exact scope of the recognition, and the edition year. Scope is where hospitals slip. US News ranks a limited number of hospitals nationally in each specialty and separately rates many more as "high performing" in specific procedures and conditions. A high performing rating in heart failure does not support "top-ranked hospital in America," and it says nothing about your maternity service. Rankings are published once a year, so the qualifier is part of the claim: name the specialty and the edition.

State medical boards separately govern how physicians describe their qualifications. Misstating specialty status can trigger board discipline whatever the FTC does.

Testimonials and patient stories

A patient saying "this surgeon saved my life" is a testimonial, and under FTC rules a testimonial implies typical results. If the outcome is unusual you need a clear statement that results are not typical, or better, data showing what typical is. The 2023 endorsement guides tightened this area, including incentivized and fabricated reviews. Do not run a story you cannot back with the same file you would build for the number.

Files go stale, and claims outlive them

Most enforcement exposure in hospital marketing is not a false claim. It is a true claim still running after its evidence expired.

1. Safety grades have short lives. The Leapfrog Hospital Safety Grade is issued twice a year, so an "A-rated for safety" asset has roughly six months before it needs re-checking. CMS star ratings run on their own annual refresh.
2. Rankings expire annually. Tag every badge asset with its edition year and an automatic review date.
3. Credential records need a re-verification window before each campaign, plus an expiry on the record itself.
4. Survival figures age out of relevance when treatment changes. A cohort diagnosed before a therapy shift describes a hospital that no longer exists.
5. Version drift kills otherwise clean files. The approved asset and the trafficked asset diverge in production, and the substantiation on record covers a headline that never ran.
6. Equity screening runs in parallel under the rules the fair-treatment lesson sets, and sign-off sequence and ownership belong to the pre-launch review lesson. The evidence file is an input to that gate, not a substitute for it.

Why the file matters more than the claim

The legal test is not "was the claim true." It is whether you had a reasonable basis at the time you made it. A survival rate that turns out accurate but had no supporting file when it ran is still unsubstantiated.

There is a second-order consequence worth pricing. A specific outcome claim can be read as a promise, and plaintiff's counsel will put the billboard next to the consent form and ask a jury which one the patient believed. Geisinger's ProvenCare bundle for coronary artery bypass, and later its ProvenExperience refund offer, are the disciplined version of that exposure: the marketing claim and the financial commitment were designed together. If your organisation is not prepared to stand behind a number that way, it should not be on a billboard.

Key takeaways

• Substantiate before you publish. Evidence gathered after a complaint is not a reasonable basis.
• Rank your sources. Peer-reviewed cohort, then audited registry, then benchmark data, then internal extract with its query attached.
• Pick a metric that can honestly differ. Near-universal survival cannot separate providers; function and complication outcomes can.
• Comparisons fail on denominators and case mix. Localized-only against national all-stages is the obvious version; unadjusted rates flattered by referral patterns are the quiet one.
• Credentials and rankings are dated facts. Name the specialty, name the edition year, hold the licence, and re-verify on a fixed clock.

*This lesson is educational and not legal or medical advice. Confirm current SEER, ABMS, FTC, Leapfrog and US News rules against their official sources for your launch date.*