Background: The U.S. Food and Drug Administration (FDA) evaluates medical devices by two main pathways. The more stringent Premarket Approval (PMA) review requires clinical trials, and the Premarket Notification 510(k) process generally exempts devices from clinical trials if they prove to be substantially equivalent to existing devices. We hypothesized that orthopaedic devices are more likely to be cleared through the 510(k) process and thus are more susceptible to being recalled.
Methods: Using the FDA’s public database, we searched for the following: PMA and 510(k) clearances for orthopaedics and non-orthopaedic specialties, including General & Plastic Surgery, Gastroenterology/Urology, Obstetrics/Gynecology, and Ear Nose & Throat, from 1992 to 2012. Additionally, we searched for all device recall events from 2002 to 2012. For the top-twenty recall companies, we calculated the odds ratio that compares the likelihood of recall for 510(k)-approved devices with that for PMA-approved devices.
Results: From 1992 to 2012, the proportion of non-orthopaedic devices cleared via the 510(k) process decreased from 91% to 53%. However, that of orthopaedic devices decreased only from 94% to 88%. Furthermore, we found that from 2002 to 2012, the percentage of recalled devices was 17.8% for 510(k)-cleared devices and 1.6% for PMA-approved devices. When stratified on the basis of recall class, the odds ratios were 3.5 for class-I devices, 13.2 for class-II devices, and 8.5 for class-III devices.
Conclusions: Given that 510(k)-cleared devices were 11.5 times more likely to be recalled than PMA-approved devices, it is concerning that most orthopaedic devices are cleared through the 510(k) process with limited clinical trials data.
Clinical Relevance: When orthopaedic surgeons are considering using a new device clinically in their patients, it is important for them to consider how the new device was approved by the FDA. If the device was approved by the 510(k) pathway, then it may have been approved without additional clinical studies confirming efficacy or safety.
1Department of Orthopaedic Surgery, Beth Israel Deaconess Medical Center, Boston, Massachusetts
2Harvard Medical School, Boston, Massachusetts
3Tufts University School of Medicine, Boston, Massachusetts
E-mail address for C.S. Day: email@example.com
In 1976, the U.S. Food and Drug Administration (FDA) enacted the Medical Device Amendments in light of the proliferation of faulty medical devices, particularly, the Dalkon Shield intrauterine device, which resulted in more than 700 deaths and 10,000 patient injuries1. Since then, medical devices have been classified into three categories based on their risks. Class-I devices are associated with low to moderate risk and include surgical apparel, arthroscopes, and goniometers. Class-II devices include wheelchairs and infusion pumps, which pose moderate to high risk and therefore require additional surveillance. Class-III devices, such as pacemakers and implantable joint prostheses, require the most surveillance2 as they “may present a potential unreasonable risk of illness or injury.”
Currently, the accepted pathway for regulatory clearance of class-II devices is the 510(k) Premarket Notification process3. In contrast to the more stringent Premarket Approval (PMA) application process, which requires clinical trials, this 510(k) process generally exempts devices from clinical trials if they prove to be substantially equivalent to existing devices4. Although the utilization of the PMA process would ensure safety and efficacy of high-risk medical devices, a 2009 Government Accountability Office (GAO) report of all FDA reviews from 2003 to 2007 found that more high-risk devices were cleared through the 510(k) process than were approved through the original PMA process5.
Under the 510(k) process, manufacturers can update devices via incremental innovations rather than waiting to release a whole new product that would require the more substantial review by the PMA process6. Furthermore, the 510(k) process even allows recalled devices to serve as predicates7. Consequently, because these 510(k)-cleared devices seldom have their safety established clinically before they are marketed, there is a greater propensity to malfunction8. When a product is found to malfunction or presents a serious threat to health, the device is designated, on the basis of its potential for causing adverse health consequences, as either class I (reasonable probability), class II (remote probability), or class III (not likely) and is recalled9.
In 2011, in response to an increase in device recalls, the FDA asked the Institute of Medicine to evaluate its 510(k) process. The Institute of Medicine concluded that the 510(k) process is flawed and lacks the legal basis to be a reliable premarket screen of the safety and effectiveness of moderate-risk devices10. The orthopaedic industry, in particular, has been recently publicized for safety issues of medical devices. For example, Medtronic’s spine fusion device11 and Stryker’s hip implant device12 have been recalled and have been investigated for their potentially serious complications. Furthermore, the DePuy Orthopaedics division of Johnson & Johnson recalled its Articular Surface Replacement hip implant device, which was implanted in about 93,000 patients and “ranks as one of the most-flawed medical implants sold in recent decades”13.
Given the number of highly publicized recalls in the orthopaedic industry, we aimed to evaluate the process by which orthopaedic devices are introduced into the market. We hypothesized that orthopaedic devices compared with non-orthopaedic devices are more likely to be cleared through the 510(k) process rather than through the more regulated PMA process. To do this, we examined the number of orthopaedic devices approved through the PMA process compared with those cleared through the 510(k) process from 1992 to 2012. Subsequently, we compared the 510(k) clearance rate for orthopaedic devices with that for devices of other surgical specialties. Additionally, we examined the top-twenty companies with the highest number of devices recalled over the past ten years and stratified them by their respective fields of medicine. We also hypothesized that the PMA review’s rigorous product evaluation is more likely to detect deficiencies related to a product than the 510(k) process would. To evaluate this hypothesis, we calculated the likelihood of a device being recalled on the basis of its approval process.
Materials and Methods
FDA Database Extraction
Using the FDA’s public database, we searched for device approvals under the PMA regulatory process and clearances under the 510(k) regulatory process from 1992 to 201214,15. We extracted the following from the databases: PMA approvals for all non-orthopaedic specialties and for orthopaedics and 510(k) clearances for all non-orthopaedic specialties combined, orthopaedics alone, and the individual surgical specialties, including General & Plastic Surgery, Gastroenterology/Urology, Obstetrics/Gynecology, and Ear Nose & Throat, as designated by the FDA.
The FDA’s PMA database allows users to search for device approvals by decision date, product code, advisory committee (medical specialty), and supplement type14,16. In general, all PMA applications are handled similarly by the FDA; however, for a first-of-a-kind device, the FDA may refer the PMA to an outside advisory panel based on the device type and medical specialty17. We searched for PMA device approvals for all non-orthopaedic specialties from 1992 to 2012. As the database displays a maximum of 500 device approvals per search, we obtained device approvals by month to capture all of the results. We performed a similar search to extract PMA approvals for orthopaedic devices from 1992 to 2012.
The 510(k) database allows users to search for device clearance by decision date, product code, and panel (medical specialty)15,16. We searched for 510(k) clearances for all non-orthopaedic specialties and orthopaedics from 1992 to 2012 by month. Finally, we searched for medical devices cleared through the 510(k) process for each surgical specialty: General & Plastic Surgery, Gastroenterology/Urology, Obstetrics/Gynecology, and Ear Nose & Throat.
Analysis of Device Approvals and Clearances
Using the data extracted from the FDA’s PMA and 510(k) databases, we analyzed the respective trends in device approval and clearance over time. We first compared the slopes in PMA approvals for all non-orthopaedic specialties with PMA approvals for only orthopaedics from 1992 to 2012. Similarly, we compared 510(k) device clearance for all non-orthopaedic specialties with that for orthopaedics.
Then we determined the proportion of devices approved through the PMA process compared with those cleared through the 510(k) process in orthopaedic and non-orthopaedic specialties from 1992 to 2012. Lastly, we analyzed the trend in 510(k) clearance for orthopaedics compared with that for other surgical specialties. For each of the aforementioned comparisons, we used an approval process × time interaction to evaluate whether there was a difference in slopes between the two arms being analyzed. A Wald test with binomial distribution based on a generalized linear models strategy was used to determine the significance of the differences in slopes. Significance was set at p < 0.05.
Analysis of Device Recalls
The device recalls were obtained using the FDA Medical Device Recalls database. We extracted recall events from 2002 to 2012, as the database provides information beginning in November 2002. The database allows searching by recall class, date posted, product name, and recalling firm18. The FDA stratifies its recalls by company or firm name. Therefore, we aggregated all recall events by the FDA from 2002 to 2012 and selected the top-twenty companies with the highest number of recall events. These companies were then categorized by medical specialty based on the area of medicine that they serve. We were then able to extrapolate and to group the recall events to assess the proportion of recalls for orthopaedics compared with that for non-orthopaedic specialties within the top-twenty companies.
Calculation of the Odds Ratio
The FDA allows for searching the PMA and 510(k) databases by company name. We determined the number of devices introduced through the PMA and 510(k) processes from 1992 to 2012 by the top-twenty companies with the highest number of recalls. A given malfunctioning device may lead to multiple recall events. Therefore, we grouped recall events into individual devices and determined the number of devices recalled by each of the top-twenty companies. We then determined the likelihood of a device being recalled by calculating an odds ratio based on the regulatory process of PMA compared with 510(k) for that device and we then stratified the odds ratio by recall classes.
From January 1992 to December 2012, the FDA approved a total of 19,139 medical devices under the PMA process. Of this total, only 646 (3%) were orthopaedic devices. In the same time period, 77,164 medical devices were cleared through the 510(k) process, of which 8345 (11%) were orthopaedic devices, and of those orthopaedic devices, 7826 (94%) were categorized as class-II devices.
PMA Approval and 510(k) Clearance Trends
In this twenty-one-year period, PMA approvals for all non-orthopaedic specialties increased by a mean of eighty-six devices per year; in contrast, PMA approvals for orthopaedic devices increased by a mean of 2.5 devices per year (Fig. 1-A). The difference in slopes for PMA approvals for all non-orthopaedic specialties compared with orthopaedics alone was determined to be significant (Wald test: 10.8; p < 0.001). During the same period, 510(k) clearances for all non-orthopaedic specialties decreased by 108 devices per year, but 510(k) clearances for orthopaedics increased by twelve devices per year (Wald test: 583.69; p < 0.001) (Fig. 1-B).
Orthopaedic and Non-Orthopaedic Device Approval Trends
The proportion of PMA-approved orthopaedic devices increased from 6% to 12%, whereas those cleared via the 510(k) process decreased from 94% to 88%, from 1992 to 2012 (Fig. 2-A). However, the proportion of PMA-approved non-orthopaedic devices increased from 9% to 47%, whereas those cleared via the 510(k) process decreased from 91% to 53% (Fig. 2-B). The difference in device approval proportions for orthopaedic devices compared with non-orthopaedic devices was significant (p < 0.001).
510(k) Clearance Trends: Orthopaedics Compared with Other Surgical Specialties
We then compared device clearance in orthopaedics with that in other surgical specialties. Although 510(k) device clearance for orthopaedics increased by twelve devices every year, the other specialties had a mean decrease in 501(k) device clearance of 8.3 devices for General & Plastic Surgery, 14.0 devices for Gastroenterology/Urology, 4.8 devices for Obstetrics/Gynecology, and 9.1 devices for Ear Nose & Throat (Fig. 3). We tested the slope of the trend in device clearance for orthopaedics against those for General & Plastic Surgery (Wald test: 95.84), Gastroenterology/Urology (Wald test: 189.15), Obstetrics/Gynecology (Wald test: 48.52), and Ear Nose & Throat (Wald test: 137.61), and found them all to be significant (p < 0.001 for all).
From November 2002 to December 2012, there were a total of 20,093 recalls initiated by 1641 manufacturers. The top-twenty companies with the highest number of recalls in this ten-year period were identified, and, among them, six companies, including the top two, were orthopaedic device manufacturers (Stryker Howmedica Osteonics, DePuy Orthopaedics, Medtronic, Zimmer, Biomet, and Smith & Nephew) (Fig. 4-A). These six companies accounted for 19% of total recalls during this period. The top-twenty companies accounted for 46% of total recalls (9226 recall events) during the ten-year period; of these, orthopaedic devices accounted for the largest proportion (41%), followed by devices categorized under general hospital (25%), diagnostics (21%), cardiovascular (9%), anesthesia (2%), and radiation oncology (2%) (Fig. 4-B).
For the top-twenty companies with the highest number of recalls, from 1992 to 2012, there were 8202 devices cleared through the 510(k) process and 7789 devices approved through the PMA process. Additionally, from 2002 to 2012, these twenty companies initiated 1461 recalls of 510(k)-cleared devices and 121 recalls of PMA-approved devices. The percentage of recalled devices among all devices was 17.8% for 510(k)-cleared devices and 1.6% for PMA-approved devices. The calculated odds ratio, which compares the likelihood of recall for 510(k)-cleared devices with that for PMA-approved devices, was 11.5. When stratified on the basis of recall class, the odds ratios were 3.5 for class-I devices, 13.2 for class-II devices, and 8.5 for class-III devices.
The PMA process has long been recognized as the standard for the screening of high-risk devices. In 1992 to 2012, PMA approvals for the entire non-orthopaedic device industry increased by eighty-six devices per year. However, in orthopaedics, PMA approvals increased only by 2.5 devices per year. Conversely, for the less-regulated 510(k) process, the orthopaedic industry had an increase of twelve devices per year, but the overall non-orthopaedic device industry had a decrease of 108 devices per year. These trends demonstrate that the orthopaedic industry has been increasingly clearing its products via the 510(k) process.
The PMA process is required for class-III devices that carry the greatest potential risk to health. However, Zuckerman et al. showed that many high-risk implantable orthopaedic devices are miscategorized under class II and thus do not undergo PMA review19. From 1992 to 2012, 94% of orthopaedic devices were categorized as class II. Furthermore, during that same time period, orthopaedic devices were approximately thirteen times more likely to be cleared through the 510(k) process (n = 8345) rather than the PMA process (n = 646).
Devices cleared through the 510(k) process may carry greater risk to patients because of lack of clinical data. For example, Mahomed et al.20 reported that only 15% of 510(k)-cleared total hip replacement devices had published data on clinical effectiveness. Additionally, Gartland21 reported that ten of ten articles on the long-term follow-up of patients after total hip arthroplasty with FDA-approved devices were found to be “deficient in terms of design, to be flawed by confusing data, and to contain results of doubtful validity,” calling into question the reliability of post-market clinical outcomes provided by the orthopaedic industry. Furthermore, Castro et al.22 estimated that <5% of device-related complications were reported to the FDA. Taken together, the lack of clinical data, the questionable nature of post-market clinical results, and the inadequate reporting of device malfunctions may have led to increased recall of orthopaedic devices cleared through the 510(k) process.
From 2002 to 2012, the top-twenty companies with the highest number of recall events, ranging from 183 (Hospira) to 1139 (Stryker), accounted for approximately 46% of total device recall events. Six of these top-twenty companies specialize in orthopaedic devices and accounted for the greatest proportion of these recalls (41%). Contrary to our findings, Zuckerman et al.19 demonstrated that cardiovascular devices accounted for the highest number of recalls from 2005 to 2009. However, their article focused solely on high-risk recalled devices, which account for only a small fraction of total devices recalled. For instance, recalls of high-risk devices accounted for only 7% of total devices recalled from 2002 to 2012; therefore, we evaluated all recalled devices. Although a device may be recalled for reasons other than harmful effects on patients, it is still concerning that these devices are being cleared for daily practice. The consequences of device recall range from lost time and cost to patient injury.
Recent media coverage of recall events has raised concerns over the effectiveness of device regulatory processes. To determine whether the orthopaedic industry’s increasing use of the 510(k) process may be responsible for the increased number of recall events, we compared the efficacy of the 510(k) clearance process with that of the PMA approval process. Specifically, we compared the likelihood of a device being recalled given its regulatory process. We found that a 510(k)-cleared device was 11.5 times more likely to be recalled than a PMA-approved device. The likelihood of recall remained higher in 510(k)-cleared devices even after stratification by class. For example, class-I recalled devices that pose the greatest threat to patient health were 3.5 times more likely to have been cleared through the 510(k) process.
Our analysis of the FDA’s device approval process and recall data had several potential limitations. First, we did not consider non-surgical specialties that also utilize high-risk implantable devices. Additionally, although it would have been ideal to perform a twenty-one-year recall analysis, we were limited by the fact that the FDA did not start publishing recall information until November 2002. Finally, although we have grouped device manufacturers on the basis of the area of medicine that they serve, no manufacturer strictly serves one specialty. For example, Stryker, labeled as orthopaedics, manufactures some non-orthopaedic devices. Likewise, Boston Scientific, labeled as non-orthopaedics, manufactures some orthopaedic devices. Therefore, we categorized companies according to the specialty where the majority of their devices are of use.
Our study raises concerns about the FDA’s device regulatory process, although we acknowledge that device manufacturers and surgeons also have culpability in medical device recalls. There may be two reasons for the increased use of the 510(k) process: (1) the orthopaedic device industry’s technology may be more mature than the technology of other specialties, with the consequence of more orthopaedic devices qualifying as predicates; and (2) the high burden of PMA user fees on manufacturer cost may have led to the preferential use of the 510(k) process, possibly slowing down innovation. Because 510(k)-cleared devices are at a higher risk of being recalled than PMA-approved devices, increased use of the 510(k) process may predispose the orthopaedic industry to higher recalls, as our data suggest. However, it is important to note that fewer PMA-approved device recalls may not necessarily indicate longitudinally increased clinical effectiveness, as PMA clinical studies are performed over a minimum two-year period, which may be insufficient to demonstrate clinical longevity. Furthermore, there has been concern over the conflict of interest that may arise with the close relationships of manufacturers and members of the Orthopaedic and Rehabilitation Devices Panel, as demonstrated in the approval process of Menaflex by ReGen Biologics23,24. However, it is worth noting that the FDA goes to great lengths in monitoring this sometimes inevitable conflict of interest.
As health-care providers, we are concerned that many orthopaedic devices are introduced with limited safety and efficacy data. The government has taken some action and passed the FDA Safety and Innovation Act (MDUFA III), which allows companies that list their devices with the agency to “increase the efficiency of regulatory processes with a goal of reducing the time it takes to bring safe and effective medical devices to the U.S. market”25. However, this law focuses on reducing the time of the review process and does not encourage more clinical trials. Today, 510(k) devices are approved if a clinically relevant predicate exists. This is often the pathway pursued as it makes more sense economically for the manufacturers, and shareholders may see little to no benefit in pursuing the PMA route, as the return on investment may not be adequate.
In taking steps toward change, there are instances in which the FDA has modified the 510(k) process, for example, with the approval of the Birmingham Hip Resurfacing (BHR) joint replacement system. This “Super 510(k)” required peer-reviewed clinical studies that demonstrated successful use of the device outside of the United States as well as studies evaluating the learning curve, training program, and longer-term safety and efficacy based on the experience of a single physician, and also included a surveillance system of quarterly investigator teleconferences for the first two years26. Although the generalizability of such a system can be questioned, it does demonstrate the FDA’s ability to modify the 510(k) process, if necessary. In the meantime, if substantial changes cannot be immediately implemented in the regulatory processes, the onus may be placed on industry-supported and FDA-monitored education of health-care institutions and providers, who ultimately decide which devices are implanted in a patient. Both educational and financial support for surgeons for the conduct of clinical trials may be instituted as a contingency in the 510(k) process to promote provider participation in clinical trials to improve post-market surveillance of medical devices. This novel model can maintain synergy among the FDA, manufacturers, and physicians to ensure higher safety of medical devices for our patients.
Investigation performed at the Department of Orthopaedic Surgery, Beth Israel Deaconess Medical Center, Boston, Massachusetts
Disclosure: No external funding was received for this study. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article.
1. Dhruva SS, Redberg RF. FDA regulation of cardiovascular devices and opportunities for improvement. J Interv Card Electrophysiol. 2013 ;36(2):99–105. Epub 2012 Dec 21.
5. U.S. Government Accountability Office. FDA should take steps to ensure that high-risk device types are approved through the most stringent premarket review process. 2009 . http://gao.gov/new.items/d09190.pdf
. Accessed 2015 Nov 12.
6. 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):385–91.
7. Ardaugh BM, Graves SE, Redberg RF. The 510(k) ancestry of a metal-on-metal hip implant. N Engl J Med. 2013 ;368(2):97–100.
8. Feigal DW, Gardner SN, McClellan M. Ensuring safe and effective medical devices. N Engl J Med. 2003 ;348(3):191–2.
19. Zuckerman DM, Brown P, Nissen SE. Medical device recalls and the FDA approval process. Arch Intern Med. 2011 ;171(11):1006–11. Epub 2011 Feb 14.
20. Mahomed NN, Syed K, Sledge CB, Brennan TA, Liang MH. Improving the postmarket surveillance of total joint arthroplasty devices. Open Rheumatol J. 2008;2:7–12. Epub 2008 Feb 25.
21. Gartland JJ. Orthopaedic clinical research. Deficiencies in experimental design and determinations of outcome. J Bone Joint Surg Am. 1988 ;70(9):1357–64.
Copyright 2016 by The Journal of Bone and Joint Surgery, Incorporated
22. Castro FP Jr, Chimento G, Munn BG, Levy RS, Timon S, Barrack RL. An analysis of Food and Drug Administration medical device reports relating to total joint components. J Arthroplasty. 1997 ;12(7):765–71.