Premarket Approval

Premarket Approval of High-Risk Medical Devices

Premarket approval, usually shortened to PMA, is the US Food and Drug Administration (FDA) decision that allows a Class III medical device, such as a pacemaker or an implantable cardioverter-defibrillator, to be sold once the agency finds reasonable assurance that it is safe and effective for its intended use. The FDA calls it "the most stringent type of device marketing application", yet it answers a narrower question than a payer or health technology assessment body, which asks whether the device is worth funding. For health economists the PMA file, and above all its public Summary of Safety and Effectiveness Data (SSED), is often the main clinical evidence available at launch. Its comparators, follow-up and study populations therefore shape the uncertainty in early device cost-effectiveness models.

Where premarket approval sits among US device pathways

US law sorts devices into three classes by the controls needed to give reasonable assurance of safety and effectiveness. Under 21 CFR 860.3 a device is Class III when general and special controls are insufficient and it is life-supporting or life-sustaining, of substantial importance in preventing impairment of human health, or presents a potential unreasonable risk of illness or injury. Part 814 covers Class III devices that were not on the market before 28 May 1976 and are not substantially equivalent to an earlier device or one in a lower class.

Most devices reach the US market by other routes. A 510(k) premarket notification ends in an order that the device is substantially equivalent to a legally marketed predicate, with the same intended use and, if its technology differs, no different questions of safety and effectiveness (21 CFR 807.100). The De Novo route places a new type of device with no predicate in a lower class than Class III (21 CFR 860.200).

From 1987 to 2020, Darrow and colleagues found a median of 32 novel devices granted premarket approval each year against a median of 3,404 510(k) clearances.

What a PMA must show and how it is reviewed

The evidential test sits in 21 CFR 860.7. The FDA relies only on valid scientific evidence, which runs from well-controlled investigations through partially controlled studies and trials without matched controls to well-documented case histories. Effectiveness evidence must consist principally of well-controlled investigations unless the agency accepts other valid evidence, and safety is shown when probable benefits outweigh probable risks. The statute requires the FDA to consider the least burdensome appropriate means of evaluating effectiveness and whether premarket data can be reduced through reliance on postmarket controls, a trade between evidence before launch and evidence after it.

The application itself must summarise the indications for use, alternative practices or procedures for the condition, and the clinical investigations, including adverse reactions, device failures and replacements, and the results of statistical analyses (21 CFR 814.20). Clinical studies of an unapproved device can be run under an investigational device exemption, and the PMA must identify any investigation conducted that way.

The FDA decides within 45 days whether to file the application, the 180-day review period runs from filing, and an advisory panel may review the application at a public meeting. Approval must be refused if there is no showing of reasonable assurance of safety or effectiveness under the proposed labelling. On approval the FDA makes public a detailed summary of the safety and effectiveness information behind the order, the SSED.

Supplements, postapproval studies and recalls

An approved PMA is a living file. A PMA supplement must be approved before a change that affects safety or effectiveness, such as a new indication or a design change (21 CFR 814.39), and the FDA may impose postapproval requirements, including continuing evaluation and periodic reporting on safety, effectiveness and reliability (21 CFR 814.82).

Much of the evidence arrives after launch, if at all. For cardiac implantable electronic devices, Rome and colleagues counted 77 original PMAs and 5,829 supplements approved from 1979 to 2012, a median of 50 supplements per original approval, and only 23% (15 of 64) of the 180-day supplements approved from 2010 to 2012 included new clinical data. Rathi and colleagues found that about 13% of initiated postmarket studies of high-risk therapeutic devices approved in 2010 and 2011 were completed 3 to 5 years after approval. Of 193 original high-risk therapeutic devices approved through PMA from 2014 to 2023, Mooghali and colleagues found that 68 (35.2%) had at least one serious recall by 31 July 2025.

Evidence at launch and the gap with HTA needs

The evidence that clears the regulatory bar is often thin by HTA standards. Dhruva and colleagues reviewed 78 PMAs for high-risk cardiovascular devices and found that 33 of 123 supporting studies (27%) were randomised, 51 of 78 approvals (65%) rested on a single study and 187 of 213 primary end points (88%) were surrogate measures. Among the studies that formed the basis of approval for high-risk therapeutic devices in 2010 and 2011, Rathi and colleagues found that 13 of 30 had no comparator and that median follow-up of the primary effectiveness end point was 3.0 months.

Each of these features maps onto a modelling problem. A single-arm study leaves the comparator effect to be borrowed from elsewhere, a surrogate endpoint must be linked to survival or quality of life, and a few months of follow-up must be extended over a lifetime time horizon by extrapolation, a source of structural uncertainty. The MedtecHTA project added device-specific challenges, such as device and user interaction and incremental innovation, and recommended high quality registries and iterative evaluation over time, so real-world evidence is expected to fill some of the gaps that the PMA leaves open.

Medicare coverage after FDA approval

In the United States the coverage decision is taken separately, mainly by the Centers for Medicare and Medicaid Services (CMS). In an August 2026 notice CMS explained that the FDA applies a statutory standard of safety and effectiveness while CMS asks whether an item is reasonable and necessary for people with Medicare, so FDA authorisation alone does not entitle a device to coverage. CMS added that older beneficiaries with multiple comorbidities may be underrepresented in the studies the FDA reviews, and that it looks for evidence on health outcomes in the Medicare population and on the durability of those outcomes.

Where evidence is limited, a national coverage decision can be made under coverage with evidence development, which pays for a technology while studies continue. The Transitional Coverage for Emerging Technologies pathway, finalised in August 2024, expedites national coverage for certain Breakthrough Devices, with no more than five candidates a year expected. The same August 2026 notice proposed the RAPID pathway for Breakthrough-designated Class II and Class III devices, with a proposed national coverage determination released on the day of FDA authorisation and a final one about 90 days later for Class III devices. CMS noted that higher-risk devices are more likely to reach authorisation with evidence gaps remaining.

Worked example of durability and the ICER at launch

All numbers in this example are illustrative. A Class III implant costs USD 30,000 more per patient than medical therapy, including the implant procedure. Its main premarket study followed patients for 12 months and showed a gain of 0.05 QALYs in that year. Costs and effects are left undiscounted, and no downstream cost offsets are counted, to keep the arithmetic visible.

If the benefit is counted only for the 12 months observed, the ICER is

$$\text{ICER} = \frac{\Delta C}{\Delta E} = \frac{30000}{0.05} = 600000$$

where $\Delta C$ is the incremental cost in USD and $\Delta E$ is the incremental QALYs per patient. If the benefit of 0.05 QALYs a year lasts for 5 years, $\Delta E = 5 \times 0.05 = 0.25$ and

$$\text{ICER} = \frac{30000}{0.25} = 120000$$

where the incremental cost is unchanged because the device is paid for once.

If instead the benefit wanes by 0.01 QALYs each year to zero after 5 years, $\Delta E = 0.05 + 0.04 + 0.03 + 0.02 + 0.01 = 0.15$ and

$$\text{ICER} = \frac{30000}{0.15} = 200000$$

Against an illustrative threshold of USD 150,000 per QALY, only the constant-benefit case looks good value. The 12-month study cannot distinguish the second case from the third, so the decision turns on a durability assumption that the approval evidence does not test. Value of information analysis can then show whether longer follow-up, through postapproval studies or coverage with evidence development, is worth paying for.

Premarket approval and EU conformity assessment

The EU has no central approval decision for devices. Under the Medical Device Regulation (EU) 2017/745, a Class III device undergoes conformity assessment by a notified body, a conformity assessment body designated under the Regulation, through Annex IX or through Annex X coupled with Annex XI. Clinical investigations must be performed for implantable and Class III devices except in defined cases, mainly modified versions of a marketed device whose equivalence the notified body endorses, and for Class III implantable devices the notified body must also consult an expert panel on the clinical evaluation. The outcome is CE marking rather than an agency approval order, so conformity assessment is the closest EU counterpart to PMA, as marketing authorisation is for medicines.

The EU then links this step to HTA. Under Regulation (EU) 2021/2282, applied from 12 January 2025, Class IIb and Class III devices that received an expert panel opinion under that consultation may be selected for joint clinical assessment, with selection by implementing act at least every two years.

Common misreadings of premarket approval

Approval is not evidence of value. A PMA says nothing about cost, the comparators payers care about or affordability, which early health technology assessment tries to anticipate during development.

PMA does not imply a randomised controlled trial. The regulations accept partially controlled studies, and fewer than a third of the cardiovascular studies reviewed above were randomised.

An approved model is not necessarily the studied model. For cardiac implantable electronic devices many models in use were approved through supplements, mostly without new clinical data, so an economic model built on the original study may describe a device that is no longer sold.

Clearance and approval are different findings. A 510(k) clearance rests on substantial equivalence to a predicate device, so describing a cleared device as "FDA approved" overstates the evidence behind it.

Sources

• Centers for Medicare and Medicaid Services. Medicare program; Transitional Coverage for Emerging Technologies. Final notice. Federal Register. 2024;89(155):65724-65754. https://www.federalregister.gov/documents/2024/08/12/2024-17603/medicare-program-transitional-coverage-for-emerging-technologies
• Centers for Medicare and Medicaid Services. Medicare program; Regulatory Alignment for Predictable and Immediate Device (RAPID) coverage pathway. Notice with comment period. Federal Register. 2026;91(153):51710-51718. https://www.federalregister.gov/documents/2026/08/11/2026-16368
• Darrow JJ, Avorn J, Kesselheim AS. FDA regulation and approval of medical devices: 1976-2020. JAMA. 2021;326(5):420-432. https://doi.org/10.1001/jama.2021.11171
• Dhruva SS, Bero LA, Redberg RF. Strength of study evidence examined by the FDA in premarket approval of cardiovascular devices. JAMA. 2009;302(24):2679-2685. https://doi.org/10.1001/jama.2009.1899
• European Parliament and Council of the European Union. Regulation (EU) 2017/745 of 5 April 2017 on medical devices. Official Journal of the European Union. 2017;L117. Articles 2, 52, 54 and 61. https://eur-lex.europa.eu/eli/reg/2017/745/oj
• European Parliament and Council of the European Union. Regulation (EU) 2021/2282 of 15 December 2021 on health technology assessment. Official Journal of the European Union. 2021;L458. Articles 7 and 36. https://eur-lex.europa.eu/eli/reg/2021/2282/oj
• Federal Food, Drug, and Cosmetic Act, sections 513 and 515, 21 USC 360c and 360e. Legal Information Institute text, accessed 14 October 2026. https://www.law.cornell.edu/uscode/text/21/360c
• Food and Drug Administration. Premarket approval (PMA). Content current as of 16 May 2019. https://www.fda.gov/medical-devices/premarket-submissions-selecting-and-preparing-correct-submission/premarket-approval-pma
• Food and Drug Administration. Medical device user fee rates for fiscal year 2027. Federal Register. 2026;91(145):48134-48142. https://www.federalregister.gov/documents/2026/07/30/2026-15335
• Food and Drug Administration. 21 CFR Part 814, Premarket approval of medical devices; 21 CFR Part 860, Medical device classification procedures; 21 CFR 807.100. eCFR, current to 9 October 2026. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-H/part-814
• Mooghali M, Arshad F, Rathi VK, et al. Premarket clinical evidence strength and recalls of high-risk therapeutic medical devices. JAMA Health Forum. 2026;7(8):e262626. https://doi.org/10.1001/jamahealthforum.2026.2626
• Rathi VK, Krumholz HM, Masoudi FA, Ross JS. Characteristics of clinical studies conducted over the total product life cycle of high-risk therapeutic medical devices receiving FDA premarket approval in 2010 and 2011. JAMA. 2015;314(6):604-612. https://doi.org/10.1001/jama.2015.8761
• Rome BN, Kramer DB, Kesselheim AS. FDA approval of cardiac implantable electronic devices via original and supplement premarket approval pathways, 1979-2012. JAMA. 2014;311(4):385-391. https://doi.org/10.1001/jama.2013.284986
• Tarricone R, Torbica A, Drummond M; MedtecHTA Project Group. Key recommendations from the MedtecHTA project. Health Economics. 2017;26(Suppl 1):145-152. https://doi.org/10.1002/hec.3468