Clinical Evidence in Medical Device Marketing: What the Rules Allow and What It Costs

1. Reports

By Saroj Jha · 11 min read

What FDA and the FTC allow a device company to say about its studies, and what building that evidence costs, from federal regulations, agency guidance and peer-reviewed cost data.

A medical device startup usually learns what its evidence is worth at the worst moment: when a hospital committee asks for a study, or when a draft claim reaches a regulatory reviewer. This report sets out what the US rules allow a device company to say about its evidence, and what building that evidence costs, using only federal regulations, FDA and FTC guidance, and peer-reviewed cost studies. It is general information, not legal or regulatory advice.

What a device company says about its product is evidence of what the product is for, and any health claim needs the kind of evidence a regulator would accept. The evidence behind that claim is the largest single line in the best public cost estimate: about $30.7 million, on average, for a pivotal study of a complex device.

1. Your marketing defines the device

FDA decides what a device is intended for by looking at what the people responsible for it say and do. The regulation says intended uses “refer to the objective intent of the persons legally responsible for the labeling of an article,” and that the intent “may be shown by such persons’ expressions, the design or composition of the article, or by the circumstances surrounding the distribution of the article.” It names “labeling claims, advertising matter, or oral or written statements by such persons or their representatives” as ways to show it^[1].

For a startup this has a practical meaning. The website, the sales deck, a conference poster and a sales representative’s answer in a hospital corridor all count. A claim you would not put in a regulatory submission should not appear in marketing either.

2. Claims about the use FDA has cleared

Companies may talk about studies that support a cleared use, with limits. FDA’s June 2018 guidance on communications consistent with FDA-required labeling says that “any data, studies, or analyses relied on should be scientifically appropriate and statistically sound,” and that firms should not “fail to disclose their material limitations”^[2]. A pilot with 20 patients and no comparison group can be described honestly as that, but it cannot carry a headline outcome.

Describing the regulatory status needs the same care. For a device cleared through 510(k), the regulation says clearance “does not in any way denote official approval of the device” and that any representation creating “an impression of official approval” because of the notification “is misleading and constitutes misbranding”^[4]. The FTC’s own example is an electronic muscle-stimulation belt cleared as a Class II device, advertised as “FDA Approved” beside weight-loss claims. The FTC found the advertising deceptive because that pairing gave the impression that FDA had found the device effective for those effects^[5]. Say “FDA cleared” when that is what happened, and keep it away from outcome claims.

3. Sharing studies about uses FDA has not cleared

Sending publications about an uncleared use is the narrowest path. In January 2025 FDA issued guidance on communications from firms to health care providers about scientific information on unapproved uses, replacing its 2009 guidance on distributing medical journal articles. FDA’s page states that the guidance “is not for current implementation, pending the Office of Management and Budget’s (OMB’s) decision on the collection of information”^[3]. That wording was still on FDA’s page on October 22, 2026. Until it changes, the safe practice is to route every question about an uncleared use to a trained medical or regulatory person, and never to a salesperson with a PDF.

4. Health claims and the FTC standard

The FTC applies its own test to health benefit claims made to anyone. Its Health Products Compliance Guidance defines “competent and reliable scientific evidence” as tests, analyses, research or studies that “have been conducted and evaluated in an objective manner by experts in the relevant disease, condition, or function to which the representation relates” and “are generally accepted in the profession to yield accurate and reliable results.” It adds that “as a general matter, substantiation of health-related benefits will need to be in the form of randomized, controlled human clinical testing,” that results must be “clinically meaningful” as well as statistically significant, and that “there is no requirement for a specific number of RCTs,” although replication in an independent study adds weight^[5].

Put together, an outcome claim needs a controlled human study you could hand to a regulator, in the population your clearance covers, with a result large enough to matter to a patient. A single-site evaluation of nurse time or supply use can support a claim about nurse time or supply use. It cannot support a claim about infections.

5. What the evidence costs

The best public estimate of device evidence costs covers complex devices that need premarket approval (PMA). Researchers at Eastern Research Group built a cost model from 151 such devices among the 191 PMAs approved from January 2013 to September 2018^[6]. Their out-of-pocket figures, in 2018 dollars:

Three points are easy to miss. First, a pivotal study averaged 565 patients over about 57 months, at $54,332 a patient. Second, only 48% of devices in the feasibility sample moved on to a pivotal study, so a feasibility study is a real decision point, not a formality. Third, once failures and the cost of capital are included, the mean capitalized cost per approved device rises to $522 million, and the authors report that the nonclinical stage accounts for 85% of it. The nonclinical $20 million is an expert-opinion assumption in the model, so read the split as a reasoned estimate, not a measurement.

Real pivotal evidence is also thinner than founders often assume. A 2015 JAMA study by Rathi, Krumholz, Masoudi and Ross looked at the 28 high-risk therapeutic devices approved through PMA in 2010 and 2011. The median pivotal study enrolled 241 patients, and 13 of 30 pivotal studies (43.3%) had no comparison group^[7].

Most startups do not take the PMA route. For a 510(k), the study requirement is usually smaller, and AAJ found no primary estimate of what 510(k) clinical studies cost. What is public is FDA’s fee for the submission itself:

6. Registering a study and reporting results

If a study is large enough to matter commercially, it is probably also regulated as an “applicable device clinical trial.” Under 42 CFR Part 11 that means a prospective study of health outcomes that compares an intervention with a control, for a device subject to 510(k), PMA or humanitarian device exemption review, other than a small feasibility study^[9]. The rules that follow are short and strict:

For a device that has not yet been cleared, the registration is made public only after clearance, and a sponsor who certifies that it intends to seek clearance can delay results reporting until 30 days after that clearance. The penalty for missing these duties is set in statute at up to $10,000 and adjusted each year for inflation. In HHS’s 2025 table the figure is $15,107 for all violations in a single proceeding, and the same amount for each day a violation continues after 30 days from notice^[10].

7. Publishing, and the cheaper evidence to start with

A peer-reviewed paper is what lets a value analysis committee, and the FTC, take a result seriously. Open-access fees are modest next to the study itself: JAMA Network Open charges $4,000^[11] and PLOS ONE lists $2,477^[12].

Most startups begin with evidence that costs a fraction of a clinical study: a single-site evaluation that measures process and cost against a comparison unit, such as dressing changes or nurse time per patient. It is often what a hospital committee asks for first. It cannot support an outcome claim, and anything published from it stays inside the clearance.

What this means for a startup

1. Write the claims table before the study. List the claims you want to make, check each against your indication, and design studies to support the ones inside it.
2. Start with local evidence and plan for published evidence. A measured evaluation at a customer site wins the first committees. A registered, published study is what lets you make outcome claims later.
3. Register on time. If the study is an applicable device trial, the 21-day clock starts with the first participant.
4. Keep off-label publications out of sales hands. Treat the January 2025 guidance as not yet in force and send questions to a trained medical or regulatory person.
5. Budget for the evidence you need. The $30.7 million pivotal figure is for complex devices. Size your plan to your route and claims with your regulatory adviser.

Method and limits

AAJ read each source on October 22, 2026: the Code of Federal Regulations, FDA and FTC guidance, the published cost model and the abstract of the 2015 JAMA study, and the fee pages of two journals. FDA’s 2018 guidance was read from the PDF. The cost model’s figures are in 2018 dollars and cover complex devices approved through PMA only. The user fees are from the notice for fiscal year 2027 as posted for public inspection by the Federal Register. BMJ Open’s fee is left out because its official page could not be opened. This report is general information, not legal or regulatory advice.