Introduction
In 2021, 15 million respiratory machines were recalled by medical technology giant Philips Respironics. Along with certain ventilators, Philips recalled their continuous positive airway pressure (CPAP) and bilevel positive airway (BiPAP) machines; devices that are used by millions of people at home, each night, who suffer from sleep apnea. The machines used a polyurethane-based sound abatement foam to reduce noise and vibrations, which was found to degrade into small foam particles that could get into the airways or be swallowed, causing serious illness or death. Although Philips reported only receiving 30 medical device reports (MDRs) of adverse events related to machine use or malfunction from 2011–2021, in April 2021, Philips informed the United States Food and Drug Administration (FDA) of its “intent to conduct a field action” due to the foam’s breakdown. In June 2021 the company announced their first recall. Over the next three years, over 116,000 MDRs were filed with FDA by patients, physicians, and hospitals, and at least 561 deaths were reported. 1 In 2024, the federal court entered a consent decree against the company that implemented a recall remediation plan and identified benchmarks to demonstrate regulatory compliance. 2 Until the decree requirements are met, Philips is restricted from participating in the respirator market; it can no longer sell new CPAP or BiPAP machines in the US. 3
Like thousands of medical devices on the market, many of the Philips recalled devices made it to the market using FDA 510(k) premarket notification, a regulatory pathway that enables novel medical devices to move quickly to market so long as there is a “predicate device” — that is, an existing device with the same intended use that is deemed substantially equivalent to the new one — already circulating in interstate commerce. The Philips respiratory machines are only one of many devices cleared through 510(k) to have been recently recalled; as will be discussed later, a series of automated external defibrillators (AED) recalls, metal-on-metal hip replacements, implanted birth control devices, and pacemakers have been other notable medical device regulation failures over the last 50 years. For years, the 510(k) premarket notification has come under increased scrutiny from patient activists who have been harmed, families who have lost loved ones due to faulty medical devices, evidenced-based medicine advocates, 4 legal and medical scholars,Reference Feldman 5 and journalists.Reference Lenzer 6 In popular media, critiques have been taken up in large scale journalistic investigations (e.g., ICIJ, 2018) and in the documentary The Bleeding Edge (2018). Adverse impact cases, which have impacted millions of people in the US alone, raise questions about the state of US medical device regulation. How it is possible for risky medical devices to routinely make it to market without sufficient clinical testing (such as randomized control trials, largely regarded as the gold standard in biomedicine) and user testing? Why are post-market surveillance and MDR systems routinely underutilized? Unlike pharmaceutical regulation, in which FDA reviews evidence on a given drug’s effects, and the drug is determined to provide more benefits than risks for the intended population 7 typically using large clinical trials, medical devices have their own regulatory pathways with wide-ranging standards for safety, efficacy, and what exactly they must demonstrate in order to be eligible to circulate on the market. In this article, we investigate the regulatory frameworks that govern medical device innovation, highlighting the values and priorities embedded in them.
Even as medical devices have proliferated since World War II (WWII), the regulatory frameworks that govern whether and how medical devices make it to market and into the hands of clinicians and patients have received less scholarly attention than drugs. While scholars have robustly studied pharmaceuticalization,Reference Abraham 8 the political economy of clinical trials and pharmaceutical testing,Reference Fisher 9 and FDA’s regulatory power, 10 the political economy and regulatory landscape of medical devices have not received the same degree of attention. Legal and medical scholars have highlighted some of the problems associated with medical device regulation,Reference Dhruva and Redberg 11 most recently with an eye toward the implications of the 510(k) pathway for the review (or lack of review) of digital devices.Reference Cortez 12 While sociologists and historians who have analyzed stakeholder participation in medical device panelsReference White 13 and how public policy shaped medical device innovation in the mid-20th centuryReference Foote 14 help create an understanding of medical device regulation as a social practice, the values and politics in contemporary regulatory frameworks, and how they came to be, have yet to be fully excavated.
In this article, we investigate the values and priorities that have shaped medical device regulation in the US historically and in the present moment, focusing on legal authority granted to FDA and considering its consequences for the public’s safety and trust. Taking a sociohistorical approach, we begin by tracing the establishment and evolution of medical device regulation over the past fifty years. We then turn to the specific risk classifications and regulatory pathways that govern how medical devices make it to market. Drawing on historical and contemporary sources, we show how a lack of legal authority throughout the mid-20th century limited FDA’s ability to regulate medical devices, positioning the agency as lagging behind the increasing development and use of medical devices in clinical care and biomedical research that marked the late 20th century.Reference Clarke and Petersen 15 We show that industry and physician concerns about stymieing the creation and use of new medical devices has always limited regulation and that the 510(k) regulatory process — initially viewed as a short-term solution to manage low-risk devices in use prior to 1976 — has now become the most used regulatory route to market for low- to high-risk devices. Although regulators try to balance ease to market alongside device safety in the creation of policies, our analysis shows that market access takes precedence in US medical device regulation. Attention to efficacy and evidence of clinical impact, that is evidence that the device will lead to better patient outcomes, is almost invisible in the content of regulatory frameworks. Although FDA has, at times, tried to strengthen the value placed on safety through post-market surveillance, by requiring health care facilities to be mandatory adverse event reporters in the 1990s, and by releasing guidance that encouraged stronger efficacy and safety testing prior to premarket submission in the late 2010s and early 2020s, such efforts have been challenged for their perceived negative effects (e.g., slowing down manufacturers’ ability to introduce new devices to the public) and for the creation of potential hurdles for biotechnology creators.
Methods
To investigate the values and contexts that animate FDA’s regulation of medical devices, we created a database that included a variety of data sources produced in the years 1906–2024. Data sources include FDA regulations related to medical devices as passed through various acts 1906–2024 as well as FDA guidance, best practice documents, and reports (n = 25), publicly available oral histories with individuals involved in the early regulation of medical devices conducted by FDA historians Suzanne Junod, John Swann, and Robert Tucker in the late 1990s and early 2000s (n = 28), congressional public hearings available through ProQuest Congressional database (n = 2), and newspaper, magazine, and media articles about medical devices and regulation from the 1960s–2024 (n = 20). We complemented the secondary sources with fieldwork at multiple medical device conferences and informational interviews with legal and historical scholars recognized as FDA experts. We also compiled databases of all FDA device clearances under the 510(k) pathway and approvals under the Premarket Approval (PMA) pathway, 1976–2024 (n = 226,289), using FDA’s publicly available records. Finally, we analyzed 510(k) clearance, De Novo clearance, and PMA summary reports for individual devices discussed throughout this paper (e.g., intrauterine devices (IUD), anesthesia machines, hip implants, and respirators, among many others).
To create the database, we used an iterative process. Utilizing oral histories and testimony at congressional hearings enabled us to trace key events and crises that served as catalysts in medical device regulation. When oral history accounts or people testifying at congressional hearings mentioned such events, for instance, the public activism around the Dalkon Shield IUD failures, we used these “insider accounts”Reference Frickel 16 to then trace references to earlier documents and conduct our own research into the events and to source historical and journalistic accounts. From there, we identified new relevant pieces of legislation, institutional and regulatory documents, policy statements, patient activists, and other elements to include in the database. Through this iterative process, we developed a sociohistorical analysis of the stakeholders, values, and contexts that coproduce medical device regulations in the US.
From the Sidelines to a Center: The Struggle to Regulate Medical Devices in the US
In the early days of food and drug regulation, medical devices were not a priority. The 1906 Pure Food and Drug Act, enforced by the USDA Bureau of Chemistry, 17 for example, contained no mention of medical devices. Instead, it focused on the interstate traffic of adulterated food and adulterated drugs, considered key problems at that time. The Federal Food, Drug, and Cosmetic Act (FD&C Act) of 1938 was the first statute to include language on medical devices. As the lack of the word “device” in the Act’s title suggests, though, devices were not yet of central concern. Instead, this regulation was created to address the use of fraudulent and adulterated medical devices circulating in the commercial market, such as height-stretching machines, nose straighteners, and heated rubber applicators promoted as a cure for prostate gland disorders, which were being sold in the US as early as 1917.Reference Rados 18 Around the same time, X-ray technology was also being used both within and outside of clinics with little guidance on radiation exposure levels for staff or clients. Although the 1938 Act listed “interstate commerce of adulterated or misbranded devices” as sites of regulatory action, the agency had no legal authority to conduct premarket reviews of medical devices.
Given that FDA had legal authority to regulate misbranded and adulterated devices, the agency primarily focused on reviewing what were labeled “quack” or fraudulent devices throughout the first half of the 1900s, as opposed to medical devices used within biomedicine. Manufacturers of “quack” devices promised a variety of unproven cures for maladies (e.g., weight loss, reduce inflammation, cure cancer) and as such engaged in misbranding, an action that FDA was empowered to regulate. Such devices were often sold outside of regular medicine even though tropes of science and scientized language might be used in their marketing materials to justify claims of healing and health. 19
The lack of attention to devices used within biomedicine began to change post-WWII with the proliferation of new, complex medical devices used by doctors to treat patients.Reference Hutt 20 In the 1950s and 60s, pacemakers, hip replacements, intraocular devices (plastic lenses surgically implanted in people’s eyes), and other implantable devices were developed by physicians, engineers, and scientists who tested them in patients, expanding the types of medical devices used in clinical practice. This period was marked by an increase in technologies being used in medical practice to manipulate the body for surgical interventions as well as technologies that would be biocompatible, meaning that they could be implanted in the body for specified (and sometimes indefinite) periods of time in order to alter organ functioning (e.g., pacemaker), fix something faulty (e.g., plastic lenses surgically implanted in eyes to replace removed natural lenses, heart valves, nails and screws for bone repair), or replace body parts that had degradation or injury (e.g., knee and hip implants). At this time, engineers began taking a particular interest in medicine, and biomedical engineering was emerging as a field.Reference Jeske 21 Engineers saw ways that developments from physics could be of use in medicine, for purposes ranging from pacemakers to imaging technologies. 22 At the time, however, physicians were largely considered to be the innovators of medical devices as they developed them for clinical use.Reference Bowrey 23
The proliferation of medical devices used in clinical practice occurred without premarket review given that FDA only had the authority to respond to reports of interstate commerce of misbranded devices. Perhaps not surprisingly, reports of adverse events started to accumulate as new devices — such as IUDs and pacemakers — were increasingly embedded in patients’ bodies. Media articles and lawsuits called attention to devices’ adverse effects. In response, President Kennedy in 1962 and President Johnson in 1964, 1966, and 1967 called on the US Congress to consider medical device regulation. The US Congress, however, did not respond to these requests with public hearings. 24 Congress considered a comparable device law when it passed the 1962 drug amendments, but this consideration failed to gain traction and Congress did not call for medical device regulation.Reference Spivak 25 Legislators focused on the recent global Thalidomide scandal, a drug that had been introduced to the market to treat morning sickness, which caused congenital malformations and infant death.Reference Pedersen 26 At the time, however, devices were indeed causing harm. In 1967 and 1968, FDA recalled suction catheters for contamination and defective seals, 27 and patients and physicians reported adverse effects from the use of pacemakers, IUDs, intraocular devices, and other technologies.
While waiting for federal legislative action, FDA continued to act within the means granted by law. FDA reviewed existing publications to identify patterns of harm from device use and pursued legal cases that targeted adverse effects from technology use in health care. 28 Drawing on how the 1938 Act positioned devices as drugs, FDA also pursued court cases that could strengthen the agency’s ability to evaluate some devices under the controls for “drugs.” The courts sided with FDA’s request to review particular devices as drugs in two key cases — the US Court of Appeals for the Second Circuit AMP Inc. v. Gardner (1968) and the Supreme Court case United States v. Bacto-UniDisk 394 U.S. 784 (1969). These cases granted FDA the ability to regulate some devices (e.g., a new method of tying off severed blood vessels and a new screening test for antibiotic effectiveness) within the category “drugs,” which enabled premarket review and for FDA to require more stringent evidence of safety and impact.Reference O’Reilly 29
Despite these wins in court cases, which set precedent for formal changes to regulatory authority in the 1970s, reviewing devices as drugs was not an easy way to ensure premarket review. With the steady increase in devices and lack of clear process to implement regulation,Reference Hutt, Daemmrich and Radin 30 FDA typically reviewed medical devices only once there was evidence of injury or illness related to device use.Reference Rogers and Lawton 31 Even as more invasive devices were introduced to clinical practice throughout the 1960s, it was generally not until device failures were reported that legal action could be taken. 32
The Cooper Report and the 1976 Medical Device Amendments (1969–1976)
In response to increasing public pressure and recommendations from the Democratic Study Group Task Force on Consumer Affairs, President Nixon established the Study Group of Medical Devices Committee in 1969, who produced a report called Medical Devices: A Legislative Plan. 33 Both the committee and the resulting report became known as the Cooper Committee and Cooper Report in honor of the chairperson, Dr. Theodore Cooper, the then director of the National Heart and Lung Institute.Reference Link 34 Reviewing testimony, reports, and published materials, the Cooper Committee concluded that there was a “medical devices hazard problem.” 35 At the time, based on an FDA tally of hundreds of published articles beginning in 1963, there were medical device reports that showed:
-
• 512 deaths and 300 injuries associated with heart valves,
-
• 89 deaths and 186 injuries associated with pacemakers,
-
• 47 deaths and several injuries associated with anesthesia machines,
-
• 28 deaths and 171 injuries associated with catheters of various kinds,
-
• over 8,000 injuries associated with IUDs,
-
• over 2,000 injuries associated with radiation equipment,
-
• and hundreds of injuries associated with a wide variety of prosthetic and orthopedic devices, dental equipment, sutures, syringes, heating pads and blankets, and contact lenses 36
The Committee considered several reasons for device harms, including clinical environment and user error, which would not be considered in the proposed legislation. They also considered device design, manufacturing, and the lack of data about a device’s potential harm to the body, areas that they deemed were appropriate for regulation. The Cooper Report would provide the foundation for the 1976 Medical Device Amendments, the first federal legislation that would centrally focus on medical devices.
The Cooper Report highlighted two different values — safety and market access — in considering potential regulation and toggled between the two; a tension that would be repeated in subsequent years. 37 Notably, efficacy and clinical impact were not identified as key priorities for regulation. Recognizing the need for better protection of the public, the Report 38 argued that safety could be produced via performance standards and cautioned against regulation that would dampen “progress in this important field.” Trying to balance safety with the ability to get new devices to market, the Cooper Report proposed a classification scheme in which medical devices would be divided into three groups: (1) devices that are exempt from standards or premarket review, (2) devices that require performance standards for safety and efficacy, and (3) devices that should have premarket review prior to use in clinical care and availability on the consumer market because they are high-risk devices or because they are novel and existing performance standards cannot be used. 39 These three groups would allow regulators to categorize and rank devices by their potential for bodily injury or adverse events. The three categories of medical devices proposed in the Cooper Report, as discussed below, became the three risk classes still used today by FDA.
Like many reports, the Cooper Report languished in inaction post-publication. After its release, there were a number of high-profile medical device failures that made local and national news. Headlines about the harms caused by pacemakers, intraocular devices, the Dalkon Shield IUD, and other devices kept up national pressure to develop regulations.Reference Cangelosi and Carstensen 40 In 1973, for example, General Electric (GE) marketed pacemakers in which the leads failed after implantation. Court documents showed that GE knew that the leads could fail, and cardiac patients died as a result. 41
Long-acting contraceptive devices, IUDs such as the Majzlin Spring and Dalkon Shield, were other high-profile cases that demonstrated the dangers of unregulated implantable devices and offered catalysts for tougher regulation on medical devices more generally. The Majzlin Spring was a device made of a long piece of stainless steel. The device was folded for insertion through the cervix. Once in the uterus it sprang open to its full size. Majzlin Springs went on the market in 1968 and over 100,000 were sold. Shortly thereafter, complaints of cramping, bleeding, and uterine perforation were reported, and the devices required surgical removal. FDA recalled the device in 1973, five years after the device went on the market.Reference Perry and Dawson 42
Between 1971 and 1974, an even larger IUD device scandal unfolded. Over 2.5 million women in the US were implanted with the Dalkon Shield, and another 1 million were reported to have been shipped abroad. Manufacturers claimed the device had a low pregnancy rate (1.1 percent) and low expulsion rate. These claims were based on studies conducted on 640 women by the device’s coinventor. Numerous study errors were later revealed, including data falsification.Reference Perry and Dawson 43 From the beginning of its widespread distribution in 1971, physicians reported that the Dalkon Shield had a higher pregnancy rate than claimed, problems with placement and removal, and a “wicking effect” whereby the strings of the IUD transported bacteria from the vagina into the uterus leading to bacterial infection. Over the next three years, spontaneous septic abortions and miscarriages with infections were reported. In cases presented to courts, Dalkon Shields were linked to a wide range of other health problems, including infertility, sterility, pelvic inflammatory disease, and perforations of the uterus. As a result of these complications, some women underwent hysterectomies and others miscarried or bore children with deformities. Advocacy organizations, such as Public Citizen, wrote to FDA asking for IUDs to be regulated as drugs, pointing to the lack of animal studies that were required of drugs that could have shown such problems with the devices. 44
The IUD recalls demonstrated the problems of retroactive regulation. As Joseph Mamana, chief of FDA’s Office of Medical Devices Compliance Section, was quoted in the National Observer, “anyone — it doesn’t even have to be a doctor — can go down to his basement, get a few hairpins, stick them together, and call it an IUD. There’s nothing we can do about it until someone is injured or dies.” Reference Perry and Dawson 45
Taken together, public demand for regulation, the increasing sophistication and proliferation of devices used in medicine, and the multiple high-profile cases of adverse events with various devices in the 1960s and early 1970s laid the foundation for the passage of the 1976 Medical Device Amendments. Although the public wanted medical devices that would not cause harm and called for regulation to prioritize safety, this priority would often take a backseat to the importance placed on introducing novel devices to clinical medicine in device regulations. In 1974, FDA established the Bureau of Medical Devices and Diagnostic Products (in 1982, the Bureau would merge with the Bureau of Radiological Health, to become the Center for Devices and Radiological Health, as it is known today). 46
After multiple attempts and a series of public hearings, the 1976 Medical Device Amendments (hereafter “1976 Amendments”) to the FD&C Act was passed. The 1976 Amendments marked the first time a statute focused singularly on medical devices. It granted FDA formal legal authority to regulate new, complex medical devices prior to their wide-scale use in clinical care. The 1976 Amendments, like the Cooper Report, primarily highlighted two values: safety and market access. Requirements for safety were typically outsourced to industry and took a back seat to practices that promoted ease to market. Efficacy and clinical impact were not prioritized as key values.
From the beginning, employees of the Bureau of Medical Devices and Diagnostic Products were concerned that regulation was too friendly toward industry. One of the regulatory models considered prior to the Cooper Report was to treat devices like drugs and require the same standards of evidence and regulation that drugs were supposed to be subjected to. Manufacturers immediately pushed back on this idea, arguing that it would be too onerous and stifle their ability to create new devices. Robert Cangelosi, one of the first hires in the newly created medical device group at FDA and an engineer by training, remarked “the device part of the industry that FDA regulated was much opposed to a drug model for devices.” He recalled that David Link, the first director of the Bureau, who had come from Hewlett-Packard’s division of medical devices, was “sympathetic to what industry was saying. […] there was a lot of pressure on Link and company to follow the drug model. So it was a fight to keep it more reasonable and sympathetic to device approvals.” 47
Expanding upon the Cooper Report, the 1976 Amendments outlined the three risk classes of devices and the regulatory requirements that are still in use today. Although media headlines and adverse events reports helped pressure Congress to act, the regulations produced have tended to value ease to market and a notion of “innovation” that highlighted newness and did not include rigorous efficacy or safety testing prior to market clearance. Unlike drugs, being a novel device was enough to justify its utility; in other words, manufacturers did not have to provide robust evidence that the device did not cause harm to patients or that it was clinically effective, particularly for 510(k) clearance, which are generally requirements for drug approval. Although regulation was instigated in part by public awareness of adverse events and concerns around medical device safety, the regulatory framework enacted prioritized market access over proven safety, made evident through its lack of codified standards for testing devices and its lack of requirement to demonstrate robust evaluation of safety and efficacy prior to market entry.
Regulatory policies are evolving practices. Medical device regulation throughout the 20th century evolved from focusing on “quack” devices used in the marketplace toward devices being developed within medicine and biosciences only after WWII, at a time when implanted and more invasive biomedical devices began to be used within clinical settings. Crucially, FDA was granted the legal authority to regulate only after the proliferation of biomedical innovations occurred, positioning regulation as lagging behind and reactive. This enabled industry and device developers to create a narrative of regulation as stymieing their ability to introduce new devices to clinical practice. In the next section, we turn to the practices of regulation, unpacking the risk classification schemes and regulatory pathways that are used to regulate medical devices in the US.
Constructing Risk Classifications and Regulatory Pathways
Today, medical devices are classified by their riskiness using the Cooper Report’s three-class classification scheme as codified in the 1976 Medical Device Amendments, passed into law on May 28, 1976. This risk classification scheme is comprised of three risk groups where risk is defined by a given group’s perceived potential to pose risk of injury or illness and/or a lack of a performance standard to ensure safety. Devices enter the market through two pathways: “approved” devices are those that have gone through the premarket approval (PMA) process and “cleared” devices have undergone the 510(k) premarket notification process. 48 Critically, the device pathways are statutory, meaning that only Congress, and not FDA, can change them.
Mapping Risk to Regulatory Pathways
The 1976 Amendments mapped the three classes onto specific regulatory pathways: Class 1 devices (e.g., bandages, handheld surgical instruments) required general controls; Class II (e.g., electric wheelchairs, CT scanners, infusion pumps for IVs) were to require performance standards of safety and efficacy; and devices classified as Class III (e.g., pacemakers, breast implants, deep brain stimulator electrodes) were to require Premarket Approval, the most stringent form of regulatory oversight in which some undetermined type of evidence of safety was to be submitted. As soon as the 1976 Amendments passed, though, it was clear that regulation was unlikely to neatly follow this scheme due to the volume of devices in all classes already on the market and the potential for regulation to negatively impact industry’s ability to market devices. For instance, regulators quickly realized that setting performance standards for Class II devices, of which there were over a thousand on the market at the time, could have adverse impacts on technology development. 49
From the get-go, then, FDA relied on manufacturers’ adherence to industry standards and international governing bodies like the International Organization for Standardization (ISO) to regulate good manufacturing practices and safety and quality of input materials (e.g., ISO 13485), and later the Clinical Laboratory Improvement Amendments of 1988 (CLIA certification) to regulate laboratory environments as proxies for more stringent device regulation. These regulatory mechanisms are important regulatory levers for medical device design and manufacturing, ensuring that the environments in which new technologies are produced are up to particular standards. However, none of these explicitly focus on the safety and efficacy of the devices themselves.
When the 1976 Medical Device Amendments passed, devices that were already circulating on the market were required to be registered with FDA. In the six remaining months of 1976 alone, 1,092 devices were submitted to FDA using the 510(k) premarket notification pathway (see Figure 1). The 510(k) pathway requires manufacturers to submit their notification to FDA 90 days prior to the introduction of their device to the market. As part of their notification, they must identify a “predicate” device, that is, a device that already exists on the market that shares the same intended use as the new device. Manufacturers must also designate the device class and indicate “actions taken” to comply with it. 50
Medical Devices cleared (510(k)) or approved (PMA) by FDA, 1976-2024.
Notably, when submitting a device through Premarket Notification 510(k) using a preamendment device, the submitting party is required to provide copies of documents that prove “the device was placed into interstate commerce (for other than research uses or as part of a plant-to-plant transfer) and was actually labeled and promoted for a specific intended use” 51 prior to May 28, 1976. If unavailable, submitters can provide an affidavit from a current or former employee of the firm that distributed the device and can attest that it was distributed prior to or provide an affidavit from a “credible person” who used the device prior to May 28, 1976 and can attest to its commercial availability. These documentation requirements make clear that what is valued most is showing that the device was commercially available and had a specific intended use for which the device was appropriately labeled and marketed prior to May 28, 1976, as opposed to showing evidence of safety and efficacy of that device. 52 These concerns are consistent with FDA’s regulatory purview prior to the 1976 Amendments, which had focused on fraudulent claims, interstate commerce, and misleading advertising rather than ensuring that devices used in medical care and procedures were safe and effective.
Originally intended to help manage the large number of devices already used in medicine in 1976, the 510(k) pathway has become the most used pathway today. 53 Figure 1 shows medical devices (which includes Class I, Class II, Class III) cleared through 510(k) premarket notification and PMA approval from 1976–2024. Although the number of devices undergoing PMA review has increased since 2006, thousands of devices enter the market using the 510(k) pathway each year (Figure 1). 54
Device classification itself — that is, what falls into Class I, II, and III — is another key site where politics play out. The 1976 Amendments specified that expert panels would be used to classify devices and explained how classification panels would be organized and operate:
(c) (1) The Secretary shall organize the panels according to the various fields of clinical medicine and fundamental sciences in which devices intended for human use are used. The Secretary shall refer a device to be classified under this section to an appropriate panel established or authorized to be used under subsection (b) for its review and for its recommendation respecting the classification of the device…. In making their reviews of devices, the panels, to the maximum extent practicable, shall provide an opportunity for interested persons to submit data and views on the classification of the devices.
What constitutes device riskiness was up to the panel, which was comprised of stakeholders with various interests and expertise. Overwhelmingly, these panels were initially comprised of physicians as opposed to engineers. Physicians utilized their professional authority to control medical device regulation, dominating engineers who were involved in medical device development in the 1950s and 1960s.Reference Bowrey 55 Physicians who worked with novel biomedical devices largely wanted to maintain their autonomy to develop technologies without federal oversight. They also drew upon their status as physicians, which relegated engineers to “technicians.” In doing this, physicians sidelined engineers’ expertise, which highlighted the need for safety engineering practices, such as design reviews and risk and failure assessments. 56
The use of panels to decide risk category has led to inconsistent categorizations. For instance, while the 1976 Amendments state that any implantable device would be considered Class III, this has not happened in practice. For example, artificial hip implants submitted in 1976 were classified as Class II, while cardiac devices such as pacemakers were classified as Class III. The classification of a device is important because it is supposed to dictate which pathway will be used — 510(k) or PMA. And critically, the review standard for these pathways differs: for 510(k), the review standard is “substantial equivalence of a new device to a legal marketed (predicate) device,” whereas the review standard for PMA is defined as “reasonable assurance of safety and effectiveness.” 57 Put simply, the 510(k) clearance does not signify that a device is safe or effective.
The 1976 Medical Device Amendments suggested that all devices classified as Class III would require PMA, with the intent that eventually all Class III devices would undergo this process. 58 However, preamendment Class III devices required PMA only after FDA publishes a regulation calling for those submissions. If FDA has not called for a PMA for a given type of device, companies must submit Class III devices through 510(k) prior to marketing the device with a “Class III Certification and Summary,” which states that the submitter has searched for information about the device type and presents a summary of the potential safety and effectiveness issues of the device it is being compared to, and a citation on which this determination is based. 59 Hence, devices that are deemed the riskiest (that is, Class III) — but are substantially equivalent to predicate devices (as determined by the submitter and agreed upon by FDA) — make it to market with widely varying levels of regulatory oversight. It is these cases in particular where the prioritization of speedy novelty and ease to market are valued over high-quality evidence and patient safety.
Although manufacturers typically make use of the 510(k) pathway to quickly get their device to market, when the PMA pathway is required, it is not as stringent as one might expect given the contemporary emphasis on large, multiple clinical trials as evidence for drugs. Given that the 1976 Amendments did not specify what counts as evidence, overtime professional norms helped define what constitutes premarket approval (PMA) as one clinical trial with small sample sizes as compared with pharmaceutical trials. A 2009 study of cardiovascular devices, for instance, found that most device trials were single studies, 20% did not report number of participants, the mean of number of enrolled participants was 308, and the number of enrolled participants ranged from 23 to 1548.Reference Dhruva and Redberg 60 This was in contrast to the size and number of clinical trials expected for drug review, which typically require three phases of clinical trials and thousands of participants. Regulators have largely seemed to agree that pharmaceutical standards are inappropriate for devices, conceding that, for instance, blinded trials of implantable devices are impractical, and that sham surgeries are unethical.Reference Faris and Shuren 61 Additionally, because most devices are categorized as iterations on existing devices, rather than “truly new,” extensive testing on iterations is thought to be unnecessary.Reference Faris and Shuren 62 Finally, because device development itself follows an iterative process — in which early events in testing lead to design changes in the device — clinical testing must happen again to “validate” changes, complicating a standard trial design. Thus, clinical trial designs, when required, for devices tend to be much more flexible in terms of data collected and acceptable outcomes data (e.g., lower performance, require fewer participants, and are shorter in duration when compared with drug trials.)
Critiques of the 510(k) Pathway
Critique of the overuse and misuse of the 510(k) pathway has been voiced within medicine and among patient advocates and citizen protection groups. Clinician-researchers, for instance, have long critiqued the lack of evidence required for medical devices to not only make it to market, but become routine tools in medicine only to later be part of medical reversals.Reference Prasad and Cifu 63 Analyses of orthopedic devices from 1992–2012 showed that 510(k)-cleared devices were 11.5 times more likely to be recalled from the market than PMA-approved devices.Reference Day 64 Other concerning analyses, such as a 2014 analysis of FDA’s high-risk List of Device Recalls from 2005–2009 showed that 78% of recalled devices were originally cleared for the market via the 510(k) process, including several automated external defibrillator (AED) devices linked to hundreds of deaths from AED malfunctions.Reference Rome, Kramer and Kesselheim 65
Public outcry was evident following class action lawsuits against manufacturers of metal-on-metal artificial hip implants in the early 2000s. Metal-on-metal ball joint hip implants, made of a cobalt-chromium alloy, promised longer lasting replacements comparted to their ceramic predecessors. There were five medical device companies that manufactured metal-on-metal hip implants, including DePuy Orthopedics, owned by Johnson & Johnson. Using past market clearances of component parts, these manufacturers were able to utilize the 510(k) pathway. Patients reported a wide range of symptoms, from muscle and tissue damage to high levels of metal in their blood, and bone erosion.Reference Dye 66 As the device was used, the rubbing of the ball joint led to metal debris, leaching toxic metal ions into the body. For some, the debris leached into their bloodstream. For others, this caused inflammation and tumor-like masses to grow around the implant. In subsequent investigations, it came to light that many of these companies, including DePuy, had not done any in-human testing — and thus did not submit any in-human clinical testing data to FDA for review — prior to their entrée onto the market.Reference Ardaugh, Graves and Redberg 67
In testing conditions, the metal-on-metal hip implants did not show any signs of rubbing. However, failure investigations shed light on the importance of user testing and the value of in-human trials. According to one safety engineer who presented on this case at a medical device conference in 2024, development testing failed to account for variation in surgeons’ placement — that is, the tolerance of how the ball and joint were placed. Testing procedures had assumed precise placement. Even placement with such slight variance that it was assumed to have been a proper implant, however, could lead to rubbing and the production of debris over time. Given the regulatory pathway, however, in-human clinical studies were not required before the device made it to market. Here, the trust that medical device companies would conduct adequate testing on their own volition proved false.
In 2011, the Institute of Medicine of the National Academy of Sciences released a report that outlined the many problems with the 510(k) clearance process and concluded that, with some exceptions, the 510(k) clearance process is “not intended to evaluate the safety and effectiveness of medical devices” and that it “cannot be transformed into a pre-market evaluation of safety and effectiveness as long as the standard for clearance is substantial equivalence to any previously cleared device.” 68 The report recommended the development of a new regulatory framework for bringing Class II devices to market instead of utilizing the 510(k) pathway, calling for the establishment of a new framework rather than reform. The committee suggested this might “boost the public’s confidence that safe and effective medical devices are being made available in a timely manner.” 69
Prominent recalls of devices and lawsuits against device manufacturers also highlight the faults of the 510(k) pathway, particularly with what we call the “predicate game” — that is, where manufacturers are able to take advantage of the 510(k) pathway by claiming substantial equivalence of parts of devices, even while the whole system has technically changed, as was the case with the metal-on-metal hip implants.Reference Ardaugh and Redberg 70 When leveraged in this way, the 510(k) becomes a loophole that enables medical device manufacturers to push their new devices through as quickly and easily as possible. This is not unique to higher risk devices, but rather any devices cleared through the 510(k). Our analysis of devices that may seem more and less risky shows the same practice consistently: identifying predicates that have come on the market within the past few years of the new device, which used a predicate of a predicate and so on. Pursuing the 510(k) pathway for any class of device, even those classified as Class III, seems to be in the manufacturers’ best interest: it is less costly, less resource intensive, and a shorter timeline to market than the PMA pathway. So long as a predicate device exists, manufacturers are incentivized to pursue the pathway of least resistance. Use of the 510(k) pathway was made easier by the 1997 Modernization Act which set up a De Novo classification. Device manufacturers are able to request De Novo classification for novel, low-risk devices. If granted, devices classified as De Novo receive clearance through an alternate 510(k) pathway, the De Novo pathway, despite not having a predicate device on the market. 71
Post-Market Surveillance
Thus far, we have focused on the three risk classes and review pathways currently used in device regulation. But regulatory systems are not static; they are evolving practices that respond to changing priorities and politics. Since the passage of the 1976 Amendments, a series of high-profile cases helped fuel the development of Acts that initially strengthened and then weakened FDA’s ability to request post-market surveillance studies. In 1983, for example, the Denver Post began reporting on deaths at a Colorado hospital, leading FDA representatives to investigate anesthesia machines developed by Puritan Bennett, finding that lethal doses of anesthesia were being administered because of device failures with a sticky spool valve. 72 The Puritan Bennett machines were used across the US. Over a five-year span, at least 200 deaths were caused by this failure. FDA representatives who investigated these cases had come from aerospace engineering prior to joining the agency. In general, they found that the medical device industry lagged far behind in terms of safety and efficacy testing. They were “aghast” that Puritan Bennett did not “put two and two together” and at how machine service work would happen repeatedly without a broader reporting or investigation of the widespread error. 73 While regulators and engineers were concerned about the lack of design reviews, risk assessments, and general safety and efficacy testing, they faced resistance in establishing stronger controls. A key source of this resistance were physicians who were developing medical devices who wanted the ability to use new devices in clinical practice without federal oversight.Reference Bowrey 74
The anesthesia machine case, as with the IUDs, highlighted both the stakes of weak regulatory standards for risky devices, as well as the failure of relying upon device developers and manufacturers to report and recall devices based on errors and adverse event reporting. At this time, there was no regulatory requirement for post-market surveillance. Thus, the 1990 Safe Medical Devices Act (SMDA) modified the 1976 Amendments, enabling FDA to require safety and efficacy testing, including clinical studies, when it deemed it necessary. 75 The SMDA aimed to improve post-market surveillance of devices by requiring user facilities such as hospitals and nursing homes to report adverse events involving medical devices in addition to manufacturers. The Act also gave FDA the authority to require manufacturers to perform post-market surveillance on permanently implanted devices if permanent harm or death could result from device failure and, critically, defined substantial equivalence.
At the same time as the SMDA increased policies related to the ensuring of safety, though, the Act changed the standard for marketing a device through the 510(k) program. This new definition meant that applicants did not need to show that their device was substantially equivalent to a pre-1976 device as had been required in the past. Instead, they could use post-amendment devices as predicate devices. 76 Demonstrating the lack of priority placed on clinical data and safety, the 1990 SMDA redefined substantial equivalence as having the “same technological characteristics as the predicate device.” If a new device had different technological characteristics compared to a predicate device, developers could provide information “including clinical data if deemed necessary that demonstrates that the device is as safe and effective as a legally marketed device.” 77
Showing how regulation toggled between an emphasis on safety and the emphasis on decreasing the regulatory burden on companies to promote ease to market, the gains related to safety were weakened by the FDA Modernization Act of 1997. In addition to the introduction of the De Novo classification discussed above, the FDA Modernization Act of 1997 changed the post-market surveillance reporting system introduced in 1990 by switching to a model in which only a representative sample of hospitals, nursing homes, and other health care facilities were required to report errors and adverse events. The 1997 Act also enabled third-party review for certain devices. These changes follow the same pattern seen in previous device regulation — maintaining access to market takes priority over stringent regulatory oversight by FDA. Importantly, the shift toward post-market surveillance had the paradoxical effect of distracting attention away from reform of the 510(k) pathway itself. 78 While post-market surveillance is crucial for monitoring the ongoing safety and effectiveness of devices in use and has led to the recall of many harmful devices,Reference Feigal, Gardner and McClellan 79 it does not replace the proactive assurance of safety and efficacy that comes from rigorous premarket review of devices, a process that has repeatedly called for by the public, evidence-based practitioners, and FDA representatives.
Conclusion
Regulatory policies are social products that emerge out of particular contexts, actors, and institutional dynamics. This article has demonstrated how particular values have shaped the establishment and evolution of medical device regulation in the US. In the first half of the 20th century, the ability to regulate medical devices was hampered by early legal definitions that initially ignored devices developed within health care contexts in favor of quack devices and limited regulatory action to interstate cases of adulteration and misbranding. As medical devices increased in number, use, and complexity post-WWII, FDA acted using available tools: reviewing medical devices where adverse events were reported (e.g., pacemakers, IUDs), reviewing the literature to determine the scope of injury or death from medical devices, and pursuing court cases to argue that some devices should undergo premarket review under the category of “drugs.” Mounting public pressure combined with a growing awareness of the potential for medical devices to cause harm finally prompted the federal government to act, first with the creation of the Cooper Committee and then with the 1976 Amendments, both of which were not easy feats. Multiple requests made by US presidents and years of congressional hearings were required before each was enacted. Subsequent regulatory changes have toggled between strengthening safety and strengthening ease to market, with ease to market often taking priority.
Promoting the introduction of new devices into medical practice is a central value that guides regulation; it does so through the use of the term innovation. However, it would be wise to question how the innovation has been consistently deployed in regulatory debates: while industry has long argued that any new device is an innovation, we could consider only those manufacturers that demonstrate evidence of safety and impact to be truly innovative. Without a more robust understanding of innovation that prioritizes safety and efficacy of medical devices, we risk promoting devices that bear resemblance to those “quack” devices that FDA originally set out to regulate.
Clinical impact is not typically part of the regulatory review process unless devices are reviewed under the PMA pathway. Even then, the rigor of clinical testing data ranges widely. As a regulatory agency, FDA requires legal definitions and federal statutes to define what they can and cannot do. This has placed FDA in a reactive position; coupled with an emphasis on ease to market, federal statutes have largely hampered their ability to proactively intervene on medical device safety and effectiveness. Our analysis advances understanding of health regulation in the US. First, the literature has importantly focused on the development, values, and implementation of regulation of pharmaceuticalsReference Abraham 80 and the political economy of clinical trials and pharmaceutical testing. 81 This article adds to the literature on the federal regulation of medical devicesReference Bowrey 82 by taking a sociohistorical approach that focuses on both the development of FDA regulations and the values that have undergirded them. Additionally, our analysis brings forward the social context, stakeholders, and values that coproduced device regulations in the US.
Although the demand for medical device safety helped spur regulatory change, the resulting regulations have not prioritized the reduction of harm; rather, they have consistently privileged market access for novel devices. This trend continues not only with focused efforts to prioritize post-market regulation rather than reform (or reimagine) the 510(k) pathway, which was called for in the 2011 IOM Report, but also through the introduction of new practices that further speed up the entrance of devices to the marketplace and clinical care: the De Novo pathway (1997) and the change in the reclassification process from rulemaking to an administrative order process (2012). This shift meant that reclassification no longer required the issuing of a regulation, but rather the publication of the proposed order in the Federal Register, consideration of comments, and a device classification panel meeting prior to issuing the decision. 83 To date, this has led to more classification downgrades than upgrades. 84
Policy is always carried out through a blend of formal written regulation, informal communication between regulators and manufacturers, and various sets of guidance that lay out suggestions for standards of evidence, manufacturing practices, and materials standards that impact technology design. Here we have focused on the statutory pathways that govern medical device development in the US. Importantly, FDA regulation of medical devices is one part of a broader regulatory landscape in which device manufacturers are expected to be operating in compliance with a set of materials standards, as governed by the ISO, Good Manufacturing Practices (GMP), and CLIA, among others. Such regulations offer assurances on material quality issues such as biocompatibility and materials manufacturing (e.g., ISO 10993 and 13485), manufacturing protocols and standards (GMP), and laboratory environments (CLIA), each of which are important for medical device design and manufacturing. However, they do not explicitly focus on the safety and efficacy of the devices manufactured. Future work should consider the broader arena of medical device regulatory practices within and across nations.
In the US, the legacy of limited legal authority to regulate medical devices prior to 1976, a focus on quack devices until the 1960s, and prioritizing the creation and use of new devices continues to haunt FDA regulatory efforts in three key ways. First, it has placed regulators as perpetually behind device development, wherein FDA is in a position of reacting to a rapidly proliferating device industry and often only able to intervene once there is evidence of harm after devices are already circulating on the market. Second, the 510(k) pathway, originally intended as a way to manage the large number of devices already in use prior to 1976, has — perhaps not surprisingly given that it is a far lower burden than the PMA pathway — been overutilized. Companies aiming to bring their device to the US market will look for the most economic path to bring their devices to market, that which is the least costly in terms of time and resource. Finally, it relies upon trust that industry (1) sufficiently self-regulates by adhering to a range of best practices in manufacturing and materials standards and (2) conducts adequate safety and efficacy testing and post-market surveillance, despite repeated evidence to the contrary. Industry has always been a key stakeholder in shaping regulation, from involvement in the Cooper Report of 1970 to the selection of the first Bureau director, and continues in the present. While there have been many moments of public pressure around high-profile device failures, such failures have been handled in reactive, singular pathways of recalls and sanctions on individual manufacturers rather than more robust changes to the regulatory infrastructure. This state of affairs is not inevitable but rather the result of particular political dynamics and values.
Acknowledgements
We thank Da Eun Jung, Joy Xu, and Heather Welty for their research assistance as well as Susan Bell for her comments on an earlier version of this article.
Disclosures
The authors have nothing to disclose.
