A medical device clinical trial tests a device in people to support its safety and effectiveness, and in an FDA sample most 510(k)s included none. When you need one, you usually run it under 21 CFR Part 812: with an FDA-approved IDE for a significant-risk device, or IRB approval alone for a non-significant-risk device.
Key takeaways
- FDA describes three stages of device clinical development: exploratory (first-in-human and feasibility), pivotal, and postmarket.
- A significant-risk study needs an FDA-approved IDE and can start 30 days after FDA receives it unless FDA objects. A non-significant-risk study needs IRB approval.
- A sponsor must report its evaluation of an unanticipated adverse device effect to FDA, every IRB and every investigator within 10 working days of first notice.
- By our count, 54% of recent industry device studies were single-group and 71% open-label, against 20% and 52% of drug studies.
- FDA accepts data from studies outside the US under 21 CFR 812.28, for studies that enrolled their first subject on or after February 21, 2019.
What is a medical device clinical trial?
It is a study of a device in human subjects, run to answer a question about the device's safety or effectiveness. FDA calls these clinical investigations, and the rules for investigational devices are in 21 CFR Part 812 [1].
FDA describes device development in three stages [2]:
| Stage | What it does | Typical studies |
|---|---|---|
| Exploratory | Evaluates the device's limitations and advantages | First-in-human and feasibility studies |
| Pivotal | Develops the evidence to evaluate safety and effectiveness for the intended use | One or more pivotal studies |
| Postmarket | Improves understanding of safety and effectiveness, such as rare adverse events | Postmarket studies |
Source: FDA, Design Considerations for Pivotal Clinical Investigations for Medical Devices, final guidance issued November 7, 2013 [2].
FDA defines a pivotal study as a definitive study that gathers evidence to support the evaluation of a device's safety and effectiveness for its intended use [2]. FDA also stresses that development can be iterative: late findings can call for more studies [2]. Our 510(k) guide shows where clinical data fits in a submission.
When does a device need a clinical trial?
Less often than many manufacturers expect. In FDA's own random sample of submissions from fiscal year 2009, 14% of 510(k)s included any clinical data, against 95% of PMAs [3].
FDA's draft guidance on clinical data in 510(k)s, issued September 7, 2023, describes four situations where it may be needed [4]:
- Your indications for use differ from the predicate's.
- Your technological characteristics differ from the predicate's.
- Substantial equivalence cannot be shown by non-clinical testing.
- A new or increased risk has been identified for the predicate.
The guidance is a draft, so treat it as FDA's current thinking rather than a binding rule. Complizen's Regulatory Strategy sets out your class, pathway, predicates and the claims you can make in writing, and marks where each claim needs more evidence.
How do you plan and run a device clinical trial, step by step?
Work through these steps in order. Each is explained in the sections that follow.
- Confirm you need clinical data, using the four scenarios above.
- Agree the study with FDA in a Pre-Submission, before you commit a budget.
- Decide the study's risk level. Significant risk means an FDA IDE; non-significant risk means IRB approval.
- Design the study: endpoints, sample size, control and masking.
- Get your approvals: the IDE if needed, IRB approval at each site, and informed consent that meets 21 CFR Part 50, including the ClinicalTrials.gov statement where it applies [11].
- Register the trial on ClinicalTrials.gov if the law requires it.
- Run and monitor the study, and meet every reporting deadline.
- Close the study with final reports, and keep the records.
- Use the data in your 510(k), De Novo request or PMA.
For the budget side of each step, see our guide to medical device clinical trial costs.
How are device trials usually designed?
Usually more simply than drug trials. We ran a count on September 25, 2026 of every completed, industry-sponsored interventional study on ClinicalTrials.gov that reached primary completion from 2023 to 2025. That gave 1,747 studies listing a device intervention and 6,551 listing a drug [5].
| Design feature | Device studies | Drug studies |
|---|---|---|
| Single-group (one arm, no comparison group) | 54% | 20% |
| Open-label (no masking) | 71% | 52% |
| Double-masked or more | 16% | 45% |
Source: our analysis of ClinicalTrials.gov records [5]. Among the device studies, 35% used parallel groups and 9% a crossover design. Among phase 3 drug trials alone, 67% were double-masked or more. These include postmarket and non-regulatory studies, not only those that supported an FDA submission. By year of primary completion, 694 finished in 2023, 588 in 2024 and 465 in 2025. Sponsors can still update recent records, so treat the recent counts as provisional.
Your device decides your design. FDA's pivotal guidance covers design choices such as endpoints, control groups and blinding [2]. The typical study is small, too: the same studies enrolled a median of 56 patients, as our cost guide details.
Agree the design with FDA first. A Pre-Submission lets you get FDA's feedback on your protocol before you start. Superagent's Q-Sub Request Drafter helps you prepare the request, and our Q-Submission guide explains the process.
Does your study need an FDA IDE?
It depends on the device's risk. A significant-risk device presents a potential for serious risk to subjects, for example as an implant, a life-supporting device, or a device of substantial importance in diagnosing or treating disease [1].
Significant-risk studies need FDA's approval. A study can begin 30 days after FDA receives the IDE application, unless FDA notifies you that it may not begin, or sooner if FDA approves it by order [1].
An IDE application has a set structure. The rule lists 12 items in a set order. The main ones are [1]:
- the sponsor's name and address
- a report of prior investigations and the investigational plan, in summary or in full
- how the device is manufactured, processed, packed and stored, in enough detail to judge quality control
- a sample investigator agreement, the list of investigators, and a certification that all have signed
- each IRB's name, address and chair, and the action each has taken
- if the device will be sold, the price and why that is not commercialization
- an environmental claim or assessment, all device labeling, and all informed consent materials
Non-significant-risk studies take a lighter route. They are treated as having an approved IDE once an IRB approves them, provided you meet the abbreviated requirements, unless FDA notifies you that an application is required [1]. FDA's information sheet on significant and non-significant risk studies, issued January 2006, explains how sponsors and IRBs decide [6].
Some studies are exempt from Part 812. Examples include [1]:
- a device FDA found substantially equivalent to a device sold before May 28, 1976, studied within the indications FDA reviewed
- some non-invasive diagnostic device studies
- consumer preference testing that does not assess safety or effectiveness or put subjects at risk
Every study needs an IRB and informed consent. IRBs work under 21 CFR Part 56, and informed consent must meet 21 CFR Part 50 [11, 12].
Consent has required elements. Each subject must be told that the study is research and what it involves, its foreseeable risks and expected benefits, and the alternatives [11]. They must also learn how their records are kept confidential and that FDA may inspect them, whom to contact, and that taking part is voluntary [11]. Studies above minimal risk must also explain any compensation or treatment available if injury occurs [11].
Registered trials need a set sentence. If the trial must be registered on ClinicalTrials.gov, the consent documents must include FDA's required statement, which begins: "A description of this clinical trial will be available on http://www.ClinicalTrials.gov, as required by U.S. Law." [11]
Importing an investigational device makes someone the sponsor. The person who imports it must act as the foreign exporter's agent and as the study's sponsor, or make sure another person does [1]. Settle this before your device ships.
What does a sponsor have to do during the study?
The sponsor carries the main duties [1]. It must:
- select qualified investigators and give them what they need
- ensure proper monitoring
- ensure IRB review and approval
- submit the IDE to FDA
- promptly tell IRBs and FDA about significant new information
You cannot promote or sell an investigational device. A sponsor or investigator must not promote or test-market it before FDA authorizes it [1]. They must not charge more than needed to recover manufacturing, research, development and handling costs, or unduly prolong the study.
Many trials must be registered. An interventional device trial, other than a feasibility study, must be registered on ClinicalTrials.gov if it has a US connection [7]. That means a US site, a device made in and exported from the US, or an FDA IDE. Register within 21 calendar days of enrolling the first participant, and submit results within 1 year of the primary completion date [7].
Keep the records. Investigators and sponsors must keep study records for 2 years after the later of two dates [1]. One is the end of the study; the other is the date the records are no longer needed to support a submission.
What must you report during a device trial?
Part 812 sets fixed deadlines, counted in working days. These are the main ones [1]:
| Event | Who reports | To whom | Deadline |
|---|---|---|---|
| Unanticipated adverse device effect | Investigator | Sponsor and IRB | 10 working days after learning of it |
| Evaluation of that effect | Sponsor | FDA, all IRBs and investigators | 10 working days after first notice |
| Device used without informed consent | Investigator | Sponsor and IRB | 5 working days |
| IRB withdraws approval | Sponsor | FDA, all IRBs and investigators | 5 working days |
| Current list of investigators | Sponsor | FDA | Every 6 months |
| Progress report | Sponsor | IRBs, and FDA for significant-risk devices | At least yearly |
| Request to return or dispose of devices | Sponsor | FDA and IRBs | 30 working days |
| End of a significant-risk study | Sponsor | FDA | Notify within 30 working days; final report within 6 months |
Source: 21 CFR 812.150 [1].
An unreasonable risk means stopping. A sponsor must evaluate an unanticipated adverse device effect immediately [1]. If it presents an unreasonable risk to subjects, the sponsor must end the affected studies within 5 working days of deciding so [1]. The limit is never later than 15 working days after first learning of the effect.
Can you use a trial run outside the US?
Yes, if it meets FDA's conditions. Under 21 CFR 812.28, FDA will accept data from a clinical investigation outside the US to support an IDE, a 510(k), a De Novo request or a PMA [1]. The study must meet four conditions [1]:
- It is well designed and well conducted.
- It follows good clinical practice, including ethics committee review and informed consent.
- You submit the supporting information the rule lists.
- FDA can validate the data, if it decides to.
The rule has a start date. It applies to studies that enrolled their first subject on or after February 21, 2019. For earlier studies outside the US, FDA applies its previous standard in 21 CFR 814.15 [8].
Many device studies already run abroad: 56% of the recent industry studies we counted had no US site, and South Korea hosted sites for 61 of them [5]. For what transfers from existing approvals, see our guides on using CE-mark clinical data for a 510(k) and what transfers from an MFDS licence.
What clinical work can follow clearance?
The postmarket stage can bring more studies. Under section 522 of the FD&C Act, FDA can order postmarket surveillance for a Class II or Class III device that meets any of four criteria [9]:
- Its failure would be reasonably likely to have serious adverse health consequences.
- It is intended to be implanted for more than 1 year.
- It is intended to support or sustain life outside a user facility.
- It is expected to have significant use in children.
FDA may order prospective surveillance for up to 36 months, and longer only with your agreement [9]. FDA's final guidance on real-world evidence for medical devices, issued in December 2025, explains how FDA evaluates evidence from sources other than a new trial [10].
Once your device is on the market, adverse events fall under a different rule. Our guide to FDA medical device reporting covers those deadlines.
Frequently asked questions
What is a medical device clinical trial?
It is a study of a device in human subjects to answer a question about its safety or effectiveness. FDA calls it a clinical investigation. Investigational device studies follow 21 CFR Part 812, with IRB oversight under Part 56 and informed consent under Part 50.
Do all medical devices need clinical trials?
No. In FDA's own sample of submissions from fiscal year 2009, 14% of 510(k)s included any clinical data, against 95% of PMAs. FDA's 2023 draft guidance lists four situations where a 510(k) may need clinical data, such as different indications or technology.
What are the stages of medical device clinical trials?
FDA describes three stages: exploratory, which covers first-in-human and feasibility studies; pivotal, which develops the evidence of safety and effectiveness; and postmarket, which improves understanding once the device is sold. FDA notes that development can loop back to earlier stages.
What is the difference between a feasibility study and a pivotal study?
A feasibility study is part of the exploratory stage and evaluates a device's limitations and advantages. A pivotal study is a definitive study that gathers the evidence to support FDA's evaluation of safety and effectiveness for the intended use. Feasibility studies usually come first.
What is an IDE?
An investigational device exemption lets you use an unapproved device in a clinical study. Significant-risk studies need an FDA-approved IDE. Non-significant-risk studies are treated as having one once an IRB approves them, unless FDA says an application is required. Some studies are exempt from Part 812 altogether.
What is the difference between significant risk and non-significant risk studies?
A significant-risk device presents a potential for serious risk to subjects, for example as an implant or a life-supporting device. Its study needs an FDA-approved IDE. A non-significant-risk study needs IRB approval and must meet Part 812's abbreviated requirements. FDA's 2006 information sheet explains how to decide.
How long does FDA take to review an IDE?
A significant-risk study can begin 30 days after FDA receives the IDE application, unless FDA notifies you that it may not begin. FDA can also approve an IDE by order before then. Plan for questions, and settle the design in a Pre-Submission first.
What must a sponsor report during a device trial?
The main deadlines are 10 working days to report an evaluation of an unanticipated adverse device effect, and 5 working days after an IRB withdraws approval. Investigator lists are due every 6 months, and progress reports at least yearly. Significant-risk studies also need a final report within 6 months.
Can a clinical trial outside the US support an FDA submission?
Yes, under 21 CFR 812.28, if the study followed good clinical practice, you submit the required supporting information, and FDA can validate the data. The rule applies to studies that enrolled their first subject on or after February 21, 2019.
Can I advertise or sell an investigational device?
No. A sponsor or investigator must not promote or test-market an investigational device before FDA authorizes it. You may charge enough to recover manufacturing, research, development and handling costs, but no more, and you must not unduly prolong the study.
Do device clinical trials have to be registered?
Many do. A controlled device trial other than a feasibility study must be registered on ClinicalTrials.gov if it has a US site, a device made in and exported from the US, or an FDA IDE. Register within 21 days of the first enrollment.
How long must I keep clinical study records?
Investigators and sponsors must keep study records for 2 years after the later of two dates. One is the end of the study; the other is the date the records are no longer needed to support a submission. Plan storage for the full period.
References
- eCFR. 21 CFR Part 812, Investigational Device Exemptions. Text current as of September 1, 2026.
- FDA. Design Considerations for Pivotal Clinical Investigations for Medical Devices. Final guidance issued November 7, 2013.
- FDA. Human Subject Protection; Acceptance of Data from Clinical Studies for Medical Devices: Preliminary Regulatory Impact Analysis. Docket No. FDA-2013-N-0080.
- FDA. Recommendations for the Use of Clinical Data in Premarket Notification (510(k)) Submissions. Draft guidance issued September 7, 2023.
- ClinicalTrials.gov. Data API. Completed, industry-sponsored interventional studies listing a device intervention, primary completion 2023 to 2025. Queried September 25, 2026.
- FDA. Significant Risk and Nonsignificant Risk Medical Device Studies. Information sheet guidance, January 2006.
- eCFR. 42 CFR Part 11, Clinical Trials Registration and Results Information Submission. Accessed September 25, 2026.
- Federal Register. Human Subject Protection; Acceptance of Data From Clinical Investigations for Medical Devices. Final rule published February 21, 2018; effective February 21, 2019.
- eCFR. 21 CFR Part 822, Postmarket Surveillance. Accessed September 25, 2026.
- FDA. Use of Real-World Evidence to Support Regulatory Decision-Making for Medical Devices. Final guidance, December 2025.
- eCFR. 21 CFR Part 50, Protection of Human Subjects. Accessed September 25, 2026.
- eCFR. 21 CFR Part 56, Institutional Review Boards. Accessed September 25, 2026.