Clinically Proven Claims and the Two Clinical Trials Misconception

“Clinically proven” is one of the most misunderstood claims in oral care, and it usually follows or is answered with a second misconception, the need for two clinical trials being required. Both of those questions took a considerable amount of time and research to work through, and the second one took the longest, because the answer is not where anyone expects to find it.

For those of us advising in this industry, whether as attorneys, marketing advisors or regulatory consultants, or research organization, precision on these two points matters more than anything. A six-month plaque and gingivitis trial runs well into six figures and clears half a million once you add arms and sites. Tell a client it needs a second one when the law does not require it and you have advised your client to spend that money for nothing. Tell a client a claim is out of reach when it is not and you have now created a pathway for your clients competitor who is more informed. Your client pays for the mistake either way, and the reputational cost of having given the advice lands on you.

Here, we’ll walk through those two questions in order and hopefully get you the correct answer. Starting with what the phrase means, moving to what the evidence has to look like, and only then getting to how many studies it takes to get there. The count belongs at the end because the count is a result and not a starting point.

What “Clinically Proven” Actually Means

“Clinically proven” is an establishment claim, meaning a promise that proof exists, and it is a legal category with consequences attached and not just a way of describing a strong claim. That difference follows the rest of the analysis. An ordinary performance claim tells the consumer the product works. An establishment claim tells the consumer two separate things, that the product works and that a particular kind of proof stands behind it. Once you make that second representation you must be able to establish the proof of the kind and at the level the phrase conveys to a reasonable consumer reading the package, and what your marketing team meant to convey has nothing to do with it.

Essentially, you have made two promises instead of one, and the second promise can sink you even when the first one is true. Your product can work exactly as advertised and you can still lose the case, because the question is no longer whether it does what you said. The question is whether you are holding what you said you were holding.

The governing standard on the advertising side is competent and reliable scientific evidence, which the FTC defines as tests, analyses, research or studies conducted and evaluated objectively by qualified people using procedures generally accepted in the field to produce accurate and reliable results, sufficient in quality and quantity by the standards of that field, and weighed against the whole body of relevant evidence. Nowhere in that definition is a set numerical on how many tests are required.

The doctrine starts with Pfizer, Inc., 81 F.T.C. 23 (1972), which gave us prior substantiation, meaning you need a reasonable basis at the moment the claim runs and not at the moment somebody asks you about it. Pfizer also produced the considerations everyone quotes, later restated in the FTC’s 1983 Policy Statement Regarding Advertising Substantiation and appended to Thompson Medical Co., 104 F.T.C. 648, 839 (1984), aff’d, 791 F.2d 189 (D.C. Cir. 1986), cert. denied, 479 U.S. 1086 (1987). The six factors are:

1.      The type of claim, since a health or safety claim demands more than an ordinary performance claim;

2.      The type of product, and what a consumer risks in relying on it;

3.      The consequences of a false claim, measured against the injury a deceived consumer would suffer;

4.      The benefits of a truthful claim, because the standard is not built to suppress accurate information;

5.      The cost of developing substantiation, which acknowledges that some evidence cannot reasonably be obtained; and

6.      The amount of substantiation experts in the field would consider reasonable.

What gets looked over nearly every time but it probably the most important point in this section: those six factors do not apply to establishment claims at all. The First Circuit said so directly in Removatron International Corp. v. FTC, 884 F.2d 1489, 1498 (1st Cir. 1989), affirming Removatron International Corp., 111 F.T.C. 206 (1988). Where the claim is an establishment claim the Commission does not run the Pfizer balancing, and instead asks a single question, which is what evidence would in fact establish this claim in the relevant scientific community.

That’s the biggest take away, the moment your advertising says “clinically proven” the balancing test disappears and you are left holding a scientific question that has to be put to scientists. Scientists do not answer that question by counting studies. They answer it by looking at how the study was designed, who was enrolled, how long it ran and whether the result fits what the rest of the published literature already shows. That is why no number appears in the standard, and it is why “how many trials do we need” cannot be answered on its own terms.

Specific and Non-Specific, and Why Whitening Keeps Losing

Thompson Medical splits establishment claims into two types; non-specific and specific. A non-specific establishment claim asserts that competent proof exists without describing what it is, and “clinically proven to fight plaque” is the good example of this. For that claim you need evidence sufficient to satisfy the relevant scientific community that the proposition holds.

A specific establishment claim names the study, the design or the result, as in “in a twelve-week clinical study, users saw a two-shade improvement.” That moves the bar, because now you need the specific level and type of proof your own sentence described, not evidence that satisfies the field in some general way. You need the evidence you are advertising.

Whitening advertising reaches for that specific language more than anything else in oral care, and the distance between holding good data and holding the particular data your sentence describes is where substantiation files come apart under review. A study showing improvement over twelve weeks does not support a sentence promising two shades, and once the claim has named a number no amount of additional evidence about the product’s general performance is going to close that gap. You wrote the standard yourself when you drafted the sentence.

In March 2021 the National Advertising Division (NAD) reviewed Colgate’s Optic White Renewal claims on a challenge from Procter and Gamble and found the “3% hydrogen peroxide” and “most hydrogen peroxide in a whitening toothpaste” claims supported, while recommending that Colgate discontinue “removes 10 years of yellow stains.” NAD Case No. 6914. Colgate appealed and that June the National Advertising Review Board (NARB) recommended the same thing. The supported claims were the ones Colgate could measure directly. The claim that failed was the one that promised a specific result the underlying studies had not measured in those terms.

Where the Two-Trial Misconception Comes From

When first wondering this question myself I asked around and got the same answer from everyone, the need for two well-controlled clinical trials from two independent investigators is the standard. But from my read of the actual text, I wasn’t convinced and I wasn’t able to get a confident response to my question, “where are you getting this requirement from?” The answer led me on a very interesting journey.

What I found is that it is not in the FTC Act, it is not in any Commission rule and it is not in the advertising case law. There is case law that actually leads you to believe it is a requirement, which we will discuss below. But what I eventually found is the number is real and it has a source, but the source has nothing to do with advertising. It sits in a regulation about whether an over-the-counter (OTC) ingredient is generally recognized as effective (GRASE), in a statutory plural that Congress fixed in one place and left alone in another, and in a rulemaking the FDA started and never finished. While all of these sources are important, none of them provide a law that governs the advertised language.

The Court Cases

The six decisions below all point the same direction and show you the good, the bad and the ugly of finding the ultimate answer to these two questions.

Thompson Medical Co., 104 F.T.C. 648, 839 (1984)

The FTC imposed a two clinical test substantiation standard on Thompson for its Aspercreme efficacy claims, and the D.C. Circuit affirmed, noting that the Commission had worked through the factors in its own precedents before concluding that a two clinical test standard was appropriate in that case. The court went out of its way to observe that this departed from prior FTC practice, since the agency had not previously required clinical testing to support a non-establishment claim, and it affirmed anyway. Anyone who tells you the FTC has never obtained a two-test requirement in litigation has not read this case and would be proving you the poor guidance that leads to reputational harm.

Removatron International Corp., 111 F.T.C. 206 (1988)

In Removatron the First Circuit drew the specific and non-specific distinction, and in the same sentence it wrote that where the claim is non-specific, the FTC decides what type of support is necessary, “which is usually two well-controlled scientific studies.” 884 F.2d at 1498. Set that sentence next to “clinically proven to fight plaque” and the problem is immediate, since a federal appeals court is calling two studies the usual answer for exactly the category of claim this article is about.

POM Wonderful, LLC v. FTC, 777 F.3d 478 (D.C. Cir. 2015)

In POM, the FTC wrote a flat two-RCT requirement into a litigated administrative order and POM petitioned for review. The D.C. Circuit denied most of the challenges and upheld the liability findings, then held the order valid to the extent it required at least one randomized and controlled human clinical trial for disease claims, while holding that the order failed Central Hudson scrutiny because it categorically required two RCTs for all disease-related claims and the FTC had not justified an across-the-board threshold as no more extensive than necessary to prevent deception.

That is the decision people reach for when they want to say two trials are never required, and unfortunately that is just not true; it does not say that. Three things that seem to be forgotten when citing this case. The court did not hold that the FTC could never impose a two-RCT requirement, only that it had failed to justify a categorical one on that record. The court affirmed that RCTs were required to substantiate POM’s efficacy claims and its non-specific establishment claims, which is the category “clinically proven to fight plaque” belongs to. And the court held the FTC had discretion to apply a general substantiation standard to those non-specific establishment claims. POM is a win on the categorical mandate and a loss on the broader notion that health claims can run without controlled clinical evidence.

FTC v. QT, Inc., 512 F.3d 858 (7th Cir. 2008)

In QT, Judge Easterbrook put the general proposition about as plainly as it has ever been put, writing that nothing in the Federal Trade Commission Act requires placebo-controlled, double-blind studies. The district court below had framed the requirement as competent and reliable scientific evidence consisting of at least one well-conducted, placebo-controlled, randomized, double-blind clinical study, a phrasing the Seventh Circuit left undisturbed when affirming.

FTC v. National Urological Group, Inc., 645 F. Supp. 2d 1167 (N.D. Ga. 2008)

National Urological runs the other way on design, and you should know it before you rely too heavily on QT. The district court agreed with the FTC’s expert that satisfying competent and reliable scientific evidence for weight and fat-loss claims required randomized clinical trials on the advertised products, and because the defendants had run no RCTs on the products themselves their claims were unfounded. The court also rejected a vagueness challenge to the competent and reliable scientific evidence standard along with puffery and First Amendment defenses (which I have to note was a quite interesting legal theory National Urology attempted to advance). What the court did not do is announce a count only requiring testing on the finished product.

United States v. Bayer Corp., No. 07-01, 2015 WL 5822595 (D.N.J. Sept. 24, 2015)

In Bayer, the government moved for civil contempt over structure and function claims for Phillips’ Colon Health, arguing that competent and reliable scientific evidence meant human clinical trials satisfying four design criteria, namely randomized, placebo-controlled and double-blind, run on the specific product, performed in the target population, and using validated methods with appropriate statistical analysis. Its expert further opined that two such trials were needed. Judge Linares denied the motion, refusing to read a drug-grade RCT requirement into a competent and reliable scientific evidence obligation and noting that the government had come in with a single expert who happened to prefer a stronger design. Bayer helps, but it carries three qualifications that honest advice has to include, because the product was a dietary supplement and the court expressly distinguished the regulatory treatment of drugs, the posture was civil contempt which put the government on a clear and convincing burden, and the opinion is marked not for publication.

Putting the Court Cases Together

Now that we see how the courts have interpreted the laws, we now need to find a way to piece all the decisions together to help us get to our ultimate answer. The FTC has required two studies and been affirmed. The FTC has required two studies and been reversed. A federal appeals court has called two studies the usual answer for one category of claim. Another federal appeals court has said the statute mandates no particular study design at all. A district court required testing on the finished product without ever naming a number, and another district court refused to read any count into the same statutory language.

That looks contradictory, but none of those courts required a specific number of clinical trials. Each of the courts were looking to whether the number the FTC picked fit the claim in front of it, on that record, on those specific facts.

Which brings us back to the word “usually” in Removatron, because a description of what the FTC typically does is not a statement of what the law requires. The First Circuit was telling you where the inquiry tends to come out, and the language immediately around that phrase tells you why, since the FTC decides what type of support is necessary based on what would establish the claim in the relevant scientific community. Two is often where that inquiry lands. It is not where the inquiry starts.

That is the rule, and it is not a number. The evidence has to be whatever qualified experts in the relevant field would require in order to believe the specific claim as written, which sometimes means two studies, as it did in Thompson Medical, and sometimes means one is the ceiling on what the government has managed to justify, as it was in POM. The count is an output of the formulation you walk through for your particular set of facts. It is not an input, and it is not something you can look up before you know what you are claiming.

There is no rule of general application requiring two trials, and there is no rule saying that one will always do either. Anybody who hands you either version has skipped the step that produces the answer.

The guidance documents are consistent with all of this, for whatever guidance is worth. The FTC’s December 2022 Health Products Compliance Guidance sets no number and talks instead about quality, about fit between the study and the claim and about how the study sits against the rest of the literature. The April 2023 Notice of Penalty Offenses flags calling something scientifically or clinically proven without support and speaks in terms of human clinical testing rather than a count. A Notice of Penalty Offenses creates no liability by itself, since section 5(m)(1)(B) penalties require actual knowledge that the specific practice was held unlawful in a prior litigated cease-and-desist proceeding, and the FTC’s ability to utilize this method remains contested. The proceedings the FTC lists as the basis for that substantiation notice include Removatron and Thompson Medical.

The Statute

When this question comes up somebody eventually reaches for section 505(d), and that reach is where the analysis usually goes off the rails, because section 505(d) sets the approval threshold for a new drug application. Different statute, different decision maker, different burden and different consequence. Pulling drug approval standards into an advertising question is a reaction to the word “clinical” and not an answer to the question that got asked.

For most products in this category section 505(d) never enters the analysis at all. A dentifrice conforming to the anticaries monograph answers its effectiveness question through conformance to the order and not by assembling a substantial evidence showing. A cosmetic faces no FDA premarket effectiveness standard whatsoever, which puts the claim squarely in FTC territory the moment somebody writes it. A device cleared through 510(k) has been measured against a predicate device, and that comparison tells you nothing about whether the sentence printed on the carton is accurate.

You will also hear the FDA has not finalized its 2019 or 2023 draft guidances on demonstrating substantial evidence, offered as though the standard were somehow unsettled. That observation is accurate and it establishes nothing, because guidance explains how an agency applies a rule that already exists and a rule does not go dormant while the explanation is being revised. The operative final guidance has been in place since 1998, at 63 Fed. Reg. 27093, and the statute has been in place a good deal longer than that.

Section 505(d) matters for a different reason though, but it has nothing to do with advertising. The statute originally defined substantial evidence as evidence consisting of adequate and well-controlled investigations (investigations in the plural sense), and the FDA read that plural for thirty-five years as ordinarily calling for two or more trials and said so repeatedly in its own Federal Register notices. Congress changed that in 1997, when FDAMA section 115(a) added a second sentence permitting data from one adequate and well-controlled clinical investigation plus confirmatory evidence to constitute substantial evidence where the agency determines the science supports it. The FDA published a further revised draft guidance on June 22, 2026, with comments open through September 22, and that document leans into the single-investigation pathway harder than anything the agency has published before it.

Run that as an argument from the greater to the lesser and you start to see some light at the end of the tunnel. Drug approval is the most demanding evidentiary threshold in the Food, Drug and Cosmetic Act, so if that threshold can be met on one investigation with confirmatory evidence, then the notion that an advertising standard built on what experts in the field consider reasonable silently contains a rigid two-trial floor does not survive contact with the statute. Use section 505(d) that way, and do not use it as the source of your advertising answer, because it is not one.

There is one more thing worth taking from it, and it turned out to be the piece that finally explained where the oral care convention came from. Congress amended the statute in 1997 and never touched the regulation sitting right next to it.

The Plural Nobody Amended

The provision in question is 21 C.F.R. 330.10(a)(4)(ii), which is the effectiveness provision of the OTC drug review:

Proof of effectiveness shall consist of controlled clinical investigations as defined in Sec. 314.126(b) of this chapter, unless this requirement is waived on the basis of a showing that it is not reasonably applicable to the drug or essential to the validity of the investigation and that an alternative method of investigation is adequate to substantiate effectiveness.

The noun is investigations, in the plural, and the verb is shall. Follow the cross-reference and you find that section 314.126(b) opens with “An adequate and well-controlled study has the following characteristics” and proceeds one study at a time, so the provision it points to governs quality per study. Whatever the count is, the count lives in 330.10.

Which means the NDA pathway got its single-investigation route in 1997 while the monograph pathway is still running the unamended text. A novel molecule can reach approval on one investigation with confirmatory evidence, and a fluoride dentifrice is governed by a regulation Congress never came back to. The higher-risk pathway got the flexibility and the lower-risk one did not.

So, does 330.10 impose a two-study floor on oral care? No, and being careful about why matters, because the sloppy version of this argument costs clients money in the other direction.

Look at the surrounding words around that plural to dive even further into it. Effectiveness is defined as a reasonable expectation of clinically significant relief in a significant proportion of the target population. Investigations may be corroborated by partially controlled or uncontrolled studies, by documented clinical studies from qualified experts and by reports of significant human experience during marketing. General recognition of effectiveness shall ordinarily rest on published studies, which may themselves be corroborated by unpublished studies and other data. Isolated case reports and random experience are excluded. Every clause in that provision is asking whether a scientific consensus exists, which is what generally recognized as safe and effective means, and consensus is plural by definition because one study has never been a consensus in any field about anything. The plural in 330.10 is describing a body of agreement among scientists, not counting up to a required number

There are three features of that provision which confirm that reading of the text. The requirement can be waived where controlled investigations are not reasonably applicable and an alternative method is adequate, and a provision with a built-in escape hatch is not a numeric floor. The regulation is not even internally consistent about number, since the same sentence that mandates plural investigations refers to the validity of “the investigation” and to “an alternative method of investigation” in the singular. And the drafters plainly knew how to write a count when they wanted one, because the animal drug analog at 21 C.F.R. 514.117(a) states in terms that one or more adequate and well-controlled studies are required. That language existed and was available. It is not in 330.10, and it was not in section 505(d) either, which is exactly why Congress had to add an entire sentence in 1997 instead of simply clarifying the old one.

One concession I do need to make here is The Dictionary Act at 1 U.S.C. 1, which provides that words importing the plural include the singular, applies to Acts of Congress and not to the Code of Federal Regulations, so there is no canon you can cite by number in support of reading 330.10 that way. The argument rests on the structure and purpose of the provision.

There is also a piece of this that sharpens the point instead of undercutting it. Section 505G(k)(3) directs that the regulations establishing final monographs and the procedures governing the OTC drug review under part 330 shall be withdrawn or revised to make technical changes, without notice and comment, effective on publication of a Federal Register notice, and FDA has said it intends to do exactly that once the deemed final orders are all posted. As of this writing the withdrawal has not issued and 330.10 is still sitting in the CFR, which leaves it operative and obsolete at the same time. Congress described what it was deleting as the procedures governing the OTC drug review, not the effectiveness standard and not the evidentiary threshold. That single word choice is the whole argument, since Congress classified this provision as process, which is what it always was. The plural described how the agency conducted a review. It never described how much proof a marketing claim requires.

Section 505G(a)(1)(A)(i) conditions deemed GRASE status on conformity with a final monograph issued under part 330, and the Category I route runs through the most recently applicable tentative final monograph issued under part 330, so part 330 lives on as a reference point for what was already decided. It is not a live evidentiary standard for what comes next, and it was never a standard for what you print on a carton.

The 2003 Rulemaking, and What the Industry Actually Asked For

None of this was invisible to the people who built the framework this industry has followed for twenty-three years, and the documentary record is a good deal more interesting than the version everybody repeats.

On May 29, 2003, the FDA published an advance notice of proposed rulemaking on oral health care drug products for the plaque and gingivitis indications at 68 Fed. Reg. 32232, Docket 81N-033P, proposing 21 C.F.R. Part 356. The notice ran fifty-six pages and carried the recommendations of the Dental Plaque Subcommittee of the Nonprescription Drugs Advisory Committee, and it is the single best document in this category for information. Section II.H.2.c.v, headed “Interpretation of data,” contains the sentence everyone has been paraphrasing for two decades without quoting:

Positive evidence of effectiveness should be obtained from a minimum of two studies, each conducted by an independent investigative group.

Twenty words, sitting in the Federal Register since 2003, and rarely quoted directly despite how much weight gets applied to it. Three things about that sentence matter the most and none of them travel along with it when it gets repeated.

The FDA disclaimed it on the face of the document, stating that the report was prepared independently of the agency, that the FDA had not yet fully evaluated it, and that it “does not necessarily reflect the agency’s position on any particular matter contained in it.” The unaltered conclusions were published to stimulate comment. That is the agency saying, in the same document, that the two-study language is not its position.

The same document also prescribes a single trial in a different place. At 68 Fed. Reg. 32240 the Subcommittee recommends that drug products containing Category I active ingredients formulated in dosage forms other than those it reviewed be required to demonstrate effectiveness by a single 6-month, randomized, controlled clinical trial. The document everybody treats as the two-study authority calls for one trial across an entire class of situations.

The two assumptions people mix together came from different subsections doing different work. Design, including the preference for double-blind studies with a placebo control, sits at II.H.2.c.i. Subjects sit at ii. Conduct of the study, including randomization and the recommendation that studies run at least six months, sits at iii. The two-study language sits at v. Separate provisions, separate purposes, merged into a single imagined rule somewhere along the way.

Most surprisingly, on November 25, 2003, the Joint Oral Care Task Group of the Consumer Healthcare Products Association and the Cosmetic, Toiletry and Fragrance Association filed comments on the docket. The Task Group members were Access Business Group, Church and Dwight, Colgate-Palmolive, GlaxoSmithKline, Johnson and Johnson, Pfizer and Procter and Gamble, which is essentially the entire category.

They did not argue for two trials. In section 5.3, addressing reclassification of a Category III active ingredient to Category I, the Task Group recommended that no more than two 6-month, single-site, randomized, negative-controlled clinical studies be required. That is a request for a ceiling, not an endorsement of a minimum, and it came from the seven largest companies in oral care. In section 5.2 they recommended one 6-month, single-site, randomized, negative-controlled clinical trial for a Category I active in a new dosage form, noting that this was consistent with the Subcommittee’s own recommendation at 32240. In section 5.1.2 they asked the FDA to allow a single 6-month trial as an alternative performance test for final formulations.

Their section 8 goes further and reconstructs the Subcommittee’s deliberations from the transcripts, and the reconstruction is worth reading if you ever have to argue this point. Across meetings on May 27, October 22 and December 2 of 1998, the Subcommittee returned repeatedly to one trial for new dosage forms, with Chairman Genco stating on the last of those days that if a company wants to move an ingredient from a mouth rinse into a dentifrice, “then the six month trial applies.” Singular, every time it came up.

So, the honest history is that in 2003 the advisory subcommittee recommended one trial for new dosage forms, the industry asked the FDA to cap the reclassification requirement at two, and the number that bleeds into general practice as a supposed legal minimum was neither of those things. Nobody in that record was treating two studies as a minimum requirement.

The ADA Seal

The other source of the two-trial theory comes from the American Dental Association (ADA) Seal program, and the numbers there are spelled out clearly. The 2016 acceptance requirements for chemotherapeutic products for the control of gingivitis provide that product efficacy must be demonstrated by two independent and registered clinical studies of at least three months utilizing a placebo control, and acceptance runs for five years. The pattern is not confined to gingivitis either, since the sugar-free chewing gum requirements state that for gums containing active or therapeutic agents for the reduction of cavities, two clinical studies are required.

Notice the mismatch against the rulemaking since the ANPR contemplated six-month studies and the Seal calls for three months. Same count, incompatible durations, which is exactly what you would expect from two conventions that grew up separately and got merged by people quoting each other instead of reading the sources.

If the Seal is part of your commercial strategy then budget for that testing without complaint, because you are buying something real and it has value in the marketplace. Just keep the categories separate in your head. The ADA is a private professional association and not a government body or a standards organization, and Seal acceptance is a product-specific private review. It does not function as substantiation on its own, and a company that offers the Seal in place of its underlying clinical data has offered nothing that will satisfy the FTC or the NAD panel. The Council’s review can corroborate a claim you have already supported, but it cannot be the support alone.

Why Counting Studies Is a Poor Proxy for Evidence

Two weak trials don’t fix each other. Getting the reason right matters, because the easy version of this argument doesn’t hold up. Running two trials of ten subjects each and combining them gives you twenty subjects and more statistical power than either trial had alone. Random error shrinks as sample size grows, and that part works exactly as expected.

What pooling cannot fix is an unintended bias. If both trials drew from the same narrow demographic, the second one broadened nothing, and you are left with twenty people who all look alike. That result does not support a claim addressed to everybody who walks into a supermarket. Replication buys precision, but it doesn’t fix a bad sample, since whatever’s wrong with the sample just gets copied along with everything else

Twenty subjects is not a large enough study regardless. A six-month gingivitis trial using a Loe and Silness gingival index endpoint generally needs somewhere in the neighborhood of fifty to sixty subjects per arm to detect a clinically meaningful difference with any confidence, and two ten-subject trials are never going to get there. One properly designed trial in a population that resembles your actual buyer will.

A well-run trial on the finished product, consistent with a mechanism the field already understands and sitting inside a literature that points the same direction, is more persuasive than two underpowered trials in an area where the published results contradict each other. That is not a workaround and it is not one trial plus a discount. It is more persuasive on the merits, evaluated by the same experts under the same standard. That is the reasoning behind the requirement that evidence be sufficient in quality and quantity, because quantity standing alone measures almost nothing.

The Pathway Matters Less Than People Think

Which regulatory route your product travels shapes your labeling obligations, and it does not change who polices the phrase “clinically proven,” because that answer is the FTC in every single case.

Anticaries claims occupy the most settled ground available in this category, governed by a finalized framework at 21 C.F.R. Part 355 and now carried as OTC Monograph M021 under Final Administrative Order OTC000034, deemed final by section 505G(b)(8) on CARES Act enactment. Effectiveness for that indication is answered through conformance to the order.

Antiplaque and antigingivitis claims sit in a stranger place, which from a legal aspect is quite interesting. Oral Health Care Drug Products for Over-the-Counter Human Use does have a deemed final order, M022, carrying no CFR citation because that proceeding never reached a final monograph, but M022 was deemed from the January 1988 tentative final monograph and covers the oral discomfort classes, the debriding agents and the oral antiseptics. Antiplaque and antigingivitis was a separate proceeding that never reached a tentative final monograph at all. It stopped at the 2003 advance notice discussed above, which is why that document is where this whole conversation has been living.

So what holds these products on the shelf is not a deemed order or regulatory approval. It is section 505G(a)(3), that extends lawful marketing to a drug classified in Category I under the most recently applicable proposed monograph or advance notice, so long as the product conforms to that document, to any later determination by the agency and to the general requirements for nonprescription drugs, and it holds only until FDA resolves the category by administrative order. Three formulas clear that bar, a cetylpyridinium chloride rinse, the fixed essential oils rinse and a stannous fluoride dentifrice, each within the conditions the subcommittee recommended. A Category III ingredient carrying a gingivitis or plaque claim gets nothing from that provision, and what has kept those products on shelves for twenty years is enforcement discretion rather than any marketing authority, which is a distinction worth putting in writing before a client mistakes one for the other. Either way it is a monograph question and not an advertising question, and running the two analyses together produces confusion in both.

While we are in the neighborhood, the same 2003 notice proposed that any reference to control of dental plaque be treated as a drug claim, that supragingival tartar be treated as a cosmetic claim, and that antiplaque claims not stand alone, meaning an antiplaque product would also have to show a clinically significant effect on gingivitis. None of that became binding, and the majority of the CHPA and CTFA Task Group pushed back hard on the drug status recommendation, with Colgate filing a separate minority position. All of it still shapes how the category talks about the line.

Cosmetics receive no FDA effectiveness review at all, and MoCRA did not change that, having brought facility registration, product listing, safety substantiation and adverse event reporting without touching efficacy anywhere. A cosmetic performance claim answers to the FTC from the moment somebody drafts it.

Devices reach the market through a 510(k) clearance premised on substantial equivalence to a predicate device, which establishes that your device resembles something already lawfully marketed and says nothing whatsoever about whether the performance claim on your carton is accurate.

Your Competitor Gets There Long Before the FTC Does

Most conversations about substantiation risk in this category are conversations about the FTC, and for most companies the FTC is the last party they will ever hear from, if they hear from anybody at all.

The NAD is where these fights actually happen. Panels take competitor challenges in oral care on a rolling basis and issue decisions on written submissions, with the standard track running roughly sixty to ninety business days and the Fast Track SWIFT process handling narrowly defined claims in about twenty. The questions those panels return to are the ones this article has been circling all along, and the recent oral care docket proves it. In 2025 Procter and Gamble challenged Essor Group over nano-hydroxyapatite claims for Boka toothpaste, and NAD found that while Essor had submitted studies on nano-hydroxyapatite as an ingredient, none of them tested Boka’s finished product. That same year GuruNanda challenged P&G over Crest Pro-Health Gum Detoxify and NAD found a reasonable basis for all of the challenged claims, which is what a properly built file looks like from the other side. In an earlier proceeding SmileDirectClub challenged Crest Whitening Emulsions and NAD found the comparative claims substantiated while recommending clearer disclosure of the comparator.

Participation is voluntary as a formal matter, and that formality misleads people badly, because declining to participate or refusing to comply sends the file to the FTC and the decisions issue either way. If you compete against Colgate, Procter and Gamble, Church and Dwight or Kenvue, this is the forum to prepare for.

The Lanham Act is the channel with the most teeth and section 43(a) lets a competitor take you into federal court directly, POM Wonderful LLC v. Coca-Cola Co., 573 U.S. 102 (2014), held that the FDCA does not preclude such a claim, and Lexmark International, Inc. v. Static Control Components, Inc., 572 U.S. 118 (2014), supplies the standing analysis a plaintiff has to satisfy before proceeding.

What makes an establishment claim so dangerous in that forum is the burden the plaintiff gets to carry, and this is where the doctrine turns on the exact words you chose. Ordinarily a Lanham Act plaintiff has to show the advertisement is false or misleading and not merely that it is unsubstantiated, which is a real obstacle. Sandoz Pharmaceuticals Corp. v. Richardson-Vicks, Inc., 902 F.2d 222, 227-28 (3d Cir. 1990). But where the defendant has attributed its claim to scientific testing, the plaintiff need only prove that the tests relied upon were not sufficiently reliable to permit one to conclude with reasonable certainty that they established the proposition for which they were cited. Procter and Gamble Co. v. Chesebrough-Pond’s, Inc., 747 F.2d 114, 119 (2d Cir. 1984); Castrol, Inc. v. Quaker State Corp., 977 F.2d 57, 62-63 (2d Cir. 1992); Rhone-Poulenc Rorer Pharmaceuticals, Inc. v. Marion Merrell Dow, Inc., 93 F.3d 511, 514 (8th Cir. 1996).

Set side by side, those two standards show exactly what “clinically proven” costs you. Without the establishment claim your competitor has to prove your product does not work. With it, your competitor only has to prove your studies do not say what your sentence says they say, which is a far cheaper case to build and a far shorter one to try. You handed your competitor that advantage the moment you chose the phrase.

Consumer class actions follow the same words for the same reason, since a clinically proven claim is objective and independently checkable and therefore forecloses a puffery defense before it starts. The Second Circuit’s decision in MacNaughton v. Young Living Essential Oils, LC, 67 F.4th 89 (2d Cir. 2023), shows how narrow that escape hatch has become, because the court vacated a Rule 12 dismissal of New York General Business Law claims over “therapeutic-grade” marketing, holding that whether the statements were non-actionable puffery required a fact-intensive inquiry unsuited to the pleadings stage. It did not decide the phrase was actionable. It decided the defendant does not get to end the case by calling it puffery, which for most defendants is the entire ballgame.

Working Through a Claim

The sequence matters as much as the substance, because each step narrows what the next one has to answer, and a company that starts in the middle spends money before it knows what the money is buying.

1. Write the claim out verbatim. Not the concept and not a summary of what marketing intends to convey, but the actual sentence as it will appear, in the layout it will occupy, next to the imagery that will surround it.

2. Name the claim type. Efficacy, comparative, superiority or establishment, and if establishment, specific or non-specific under Thompson Medical. Superiority stands on its own and demands head-to-head data against the named or implied comparator no matter how many trials you happen to be holding.

3. Map what the advertising implies and not merely what it states. Under the FTC’s Deception Policy Statement the message a consumer assembles from packaging, imagery and placement carries the same substantiation burden as the words you actually wrote, which means an antiplaque claim positioned next to gum health imagery may be conveying a gingivitis claim you never intended and cannot support.

4. Ask the scientific question instead of the legal one. For an establishment claim the Pfizer factors drop out and Removatron controls, so the question is what evidence would in fact establish this claim in the relevant scientific community. Put that question to people who actually work in the field, because their answer is the standard.

5. Read the whole literature, including the parts that cut against you. Substantiation gets weighed against the totality of the relevant evidence and not against the studies you would prefer to submit, and one strong trial in a field where nothing contradicts it leaves you in far better shape than the same trial surrounded by published negatives a challenger will find in an afternoon.

6. Match the study design to the claim you wrote. Finished product and not ingredient, target population and not a convenient one, a real control, a duration appropriate to the endpoint, a primary endpoint prespecified and aligned to the benefit claimed, and adequate power. A study on an ingredient does not substantiate a claim about your formula, and stacking up more ingredient literature is not going to close that gap. Ask Essor Group.

7. Only now take up the count. A specific claim needs a trial that produced the specific result your sentence describes, a comparative claim needs the comparison itself, and a Seal strategy needs the category convention regardless of what the law requires. Absent any of those, one well-designed trial supported by a consistent literature can carry a properly worded claim, and the count follows from steps four through six instead of driving them.

8. Document the reasoning now. The reasonable basis has to exist at the moment the claim is disseminated, so a file assembled after a challenge letter arrives is worth a fraction of the identical file assembled before launch, and that difference will be obvious to whoever ends up reviewing it.

The Answer

I know, forcing you to read through all of the logical reasoning from different agencies, independent bodies and court cases was almost cruel while promising you I’d get you the best answer. So without further ado here we go.

What is “clinically proven”? It is an establishment claim, which means it promises the consumer a particular kind of proof, and the company that makes that promise owns whatever the phrase conveys to a reasonable consumer reading the package. For an establishment claim the Pfizer balancing test does not apply, so the question is what evidence would establish the claim in the relevant scientific community, and if your sentence names a study, a duration or a number then that specific thing is what you have to hold.

Do you need two trials? There is no rule of general application requiring two. The FTC has obtained a two-test requirement in a litigated order and had it affirmed in Thompson Medical. It has obtained a categorical two-RCT requirement and had it vacated under Central Hudson in POM. The First Circuit has called two well-controlled studies the usual answer for non-specific establishment claims in Removatron, the Seventh Circuit has said the FTC Act mandates no particular design in QT, and a district court refused to read any count into the same statutory words in Bayer. What none of those decisions did was set a floor you can consult before you know what you are claiming, because the count is an output of the scientific question and not a rule sitting upstream of it.

The oral care version of that rule is a different animal and it is real in the way conventions are real. It traces to a plural in the OTC drug review regulation that Congress fixed in the statute next door and left alone here, to an advance notice of proposed rulemaking the FDA never finalized and expressly declined to adopt as its own position, to a private seal program running its own three-month standard, and to consent orders recording what individual companies agreed to in order to end their own cases. That convention carries genuine commercial weight, and for a particular product a second trial may well be the right call.

What matters is knowing which kind of decision you are making. A company that runs a second trial to satisfy the market, the Seal or a retailer has made a business judgment and it may be an excellent one. A company that runs a second trial because it believes the law demands one has been misadvised, and somebody handed it that advice.

The question was never how many. It was what are you claiming and what would a qualified expert in the field need to see before believing it. Answer that and the count takes care of itself. Answer the count first and you have spent six figures resolving a question nobody ever bothered to define.

---

Howard Law is a business, regulatory and M&A law firm in the greater Charlotte, North Carolina area, with additional services in M&A advisory and business brokerage. Howard Law is a law firm based in the greater Charlotte, North Carolina area focused on business law, corporate law, regulatory law, mergers & acquisitions, M&A advisor and business brokerage. Handling all business matters from incorporation to acquisition as well as a comprehensive understanding in assisting through mergers and acquisition.

The choice of a lawyer is an important decision and should not be based solely on advertisements. The information on this website is for general and informational purposes only and should not be interpreted to indicate a certain result will occur in your specific legal situation. Information on this website is not legal advice and does not create an attorney-client relationship. You should consult an attorney for advice regarding your individual situation. Contacting us does not create an attorney-client relationship. Please do not send any confidential information to us until such time as an attorney-client relationship has been established.