Background: Medical device recalls have increased in the previous two decades. Orthopedic devices are estimated to constitute 12% of all medical devices recalled. Medical devices enter the market via the Food and Drug Administration's (FDA's) premarket approval (PMA) or 510(k) pathways. This article evaluates orthopedic hip device recalls between Jan. 1, 2007, and Dec. 31, 2017. We hypothesized that the 510(k) approval process would have substantially higher recall rates for defective devices.
Methods: The FDA's device recall database was queried for all orthopedic hip devices from Jan. 1, 2007, to Dec. 31, 2017. Each recall included product description, recall number, device class, date of recall posting, date of recall termination, manufacturer, FDA-determined cause for recall, number of recalled units, distribution, product classification, and method of approval [510(k), PMA, or unspecified].
Results: In total, 774 orthopedic hip devices were recalled between Jan. 1, 2007, and Dec. 31, 2017. The 510(k) approval process constituted 85% of hip device recalls. The most common FDA-determined cause of hip device recalls was device design, which constituted 37% of 510(k)-approved device recalls but only 6% of PMA-approved device recalls. The most recalled hip devices were hip prostheses. Orthopedic hip device recalls have shown a decrease of about 10 recalls per year during the 11-year period of analysis.
Conclusion: Devices approved through the 510(k) process, compared with the PMA process, were more likely to be recalled for design defects. Although device design is the most common reason for device recall, many recalls are due to suboptimally standardized processes (e.g., packaging, process controls, device labeling). Overall, orthopedic hip device recalls decreased during the period of analysis (2007–17).
Medical device recalls have increased substantially in the previous two decades. According to the Food and Drug Administration (FDA), between November 2002 and December 2012, 20,093 devices were recalled (average of approximately 2,009 devices/year).1 Between 2013 and 2017, 13,937 devices were recalled (average of approximately 2,787 devices/year).2 In 2017, there were 3,202 device recalls—an increase of 38.5% compared with the number of devices recalled in 2002. Between 2005 and 2009, orthopedic devices constituted 12% of device recalls by the FDA and were the third most recalled medical device type behind cardiology (15%) and radiology (14%).3,,4
To enter the market, medical devices are approved through either the FDA's premarket approval (PMA) or 510(k) premarket notification process. The approval process used is determined by device class (I, II, or III) and whether a substantial equivalent of a device exists on the market. The device classes depend on intended use, as well as indications for use. Class I devices pose the least risk to the patient, whereas Class III devices pose the highest risk to the patient. Device Classes I and II require 510(k) approval for marketing unless exempt, while device Class III requires approval via the PMA process for marketing unless a substantial equivalent of a device exists on the market.5 This classification should not be confused with recall class, which classifies the recalls as I, II, or III, with Class I posing the greatest risk and Class III posing the least risk.6 Of note, only device class, and not recall class, is reported in the FDA recall database.
The PMA process requires clinical evidence for authorization and has been shown to be considerably more time consuming and costly, with an average of 8.5 years needed to recoup investments for a device.7 On the other hand, the 510(k) process can bypass this authorization by demonstrating substantial equivalence to previously approved devices. The 510(k) process requires an average of 2.4 years to recoup investments for a device.7 Based on a 2010 survey of medical technology companies, the average total concept-to-market cost of PMA devices was $94 million and that for 510(k) devices was $31 million.8
The reduced cost of 510(k) approval has allowed manufacturers to bring more devices to market. This helps promote the advancement of devices that would otherwise be hindered by the costs associated with the PMA pathway. Given the cost differential between the PMA and 510(k) approval processes, it is hypothesized that manufacturers are more likely to seek the less expensive 510(k) alternative, thereby leading to higher recall rates for defective devices.
Of note, a device also can be approved as a Humanitarian Use Device (HUD) by submitting a Humanitarian Device Exemption (HDE) to the FDA. This process is similar to other approval pathways but does not require demonstration of effectiveness. HUDs are designed to treat conditions affecting fewer than 4,000 individual per year. The HDE regulation, which was approved in 1996, allows manufacturers to have a device approved for market when research and development costs would exceed market returns for conditions affecting small patient populations.9 No information was found in the FDA recall database regarding recalls of HUDs.
Although cost-effective, the 510(k) process has been shown to be less efficacious for device longevity. Based on a study of medical device recalls between 1992 and 2012 by Day et al.,1 devices approved via the 510(k) process were 11.5 times more likely to be recalled than devices approved via the PMA process. Their analysis also showed that recalls for nonorthopedic devices approved through the PMA pathway increased from 9% to 47% during the period of study, while recalls for orthopedic devices approved through the PMA pathway increased from 6% to 12%.1 In addition, use of the less regulated 510(k) process for nonorthopedic devices decreased from 91% to 53%, while its use for orthopedic devices decreased from 94% to 88%. This trend suggests that compared with nonorthopedic device manufacturers, orthopedic device manufacturers have been favoring 510(k) approval over PMA.
The current article evaluates orthopedic hip device recalls between Jan. 1 2007, and Dec. 31, 2017. A previous study on foot and ankle devices approved by the FDA's PMA and 510(k) approval pathways concluded that devices approved via the 510(k) process were more likely to be recalled.10 However, no previous study was identified in the literature that examined orthopedic hip devices in relation to recall probability. Based on this information, we hypothesized that hip implants approved via the 510(k) process would have substantially higher recall rates.
The FDA's device recall database was queried for all orthopedic hip devices from Jan. 1, 2007, to Dec. 31, 2017 by searching for the following terms: hip, femoral head, femoral neck, bipolar, monopolar, acetabulum, acetabular, and cemented.11 Each product's name, description, and medical use was reviewed, and the products were confirmed to be hip devices that may be used by orthopedic hip surgeons. All duplicates were removed. Each recall included product description, recall number, device class, date recall was initiated, date recall was terminated, manufacturer, FDA-determined cause for recall, number of units recalled, distribution (nationwide or worldwide), product classification, and method of approval [510(k), PMA, or unspecified]. The total number of orthopedic devices approved via the PMA and 510(k) pathways was determined by searching the FDA database between Jan. 1, 2007, and Dec. 31, 2017 for devices approved by an “orthopedic panel,” which limits the search to devices approved for orthopedic use.
In total, 774 orthopedic hip devices were recalled during the specified time frame. Totals of 925 (13.7%) and 5,833 (86.3%) devices were approved via the PMA and 510(k) pathways, respectively, during the period of analysis. Of the recalled devices, 661 (85.4%) were approved via the 510(k) approval process and 33 (4.3%) via the PMA process. For 80 devices (10.3%), the FDA recall database did not specify the route of approval.
FDA-determined reasons for orthopedic hip device recalls are shown in Table 1. Similar reasons for recalling devices have been combined into common categories for analysis. In addition, there were 17 instances where the reason involved fewer than 10 recalls; these included manufacturing material removal, software design, component change control, use error, environmental control, equipment maintenance, mix-up of material/components, component design and selection, and unknown. For convenience, these low-impact recalls have been combined with recalls classified according to FDA's category of “other.”
The top three FDA-determined reasons for orthopedic hip device recalls were (1) device design (279 recalls), (2) packaging issues (184 recalls), and (3) process issues (131 recalls). Less common FDA-determined reasons for recalling devices were labeling errors (54 recalls), nonconforming materials (48 recalls), employee error (10 recalls) and “other” (55 recalls). Of note, 13 recalls were still being investigated by the FDA at the time of analysis.
A comparison was performed of FDA-determined reasons for recalls based on whether devices were approved by the PMA or 510(k) approval process (Figure 1). Of note, device design was cited as the reason for recall in 6% of PMA-approved devices and 37% of 510(k)-approved devices. In addition, a majority (73%) of PMA-approved devices were recalled due to packaging issues.
Of recalled devices related to hip surgery, the majority were hip prosthesis devices (656 recalls [85%]). Hip/femoral stems were the second most recalled device (32 recalls [4%]), followed by orthopedic instruments (19 recalls [2%]). Other less common categories of hip devices were placed into a single category named “other” (67 recalls [9%]).
From the initiation of the recall to its termination, orthopedic hip devices deemed harmful to the public took a mean of 541 days and mode of between 301 to 400 days (minimum one day, maximum 2,001 days) to complete.
In relation to device class, two recalled orthopedic hip devices were Class I, 732 were Class II, and 40 were Class III. Of note, the time to complete a recall was related to device class: Class I recalls averaged 738 days to completion of recall, Class II recalls 556 days, and Class III recalls 259 days.
Overall, the total number of orthopedic hip device recalls decreased by approximately 10 recalls per year between 2007 and 2017 (Figure 2). From 2007 to 2012, approximately 103 recalls occurred per year, while from 2013 to 2017, about 32 recalls occurred per year.
The top three manufacturers with the most recalled orthopedic hip devices were Stryker Howmedica Osteonics (207 recalls), Zimmer Holdings (157 recalls), and DePuy Orthopaedics (125 recalls).
PMA and 510(k) Approval Processes
Medical devices enter the market via one of two approval processes: the FDA's PMA or 510(k) pathway. The PMA process requires clinical evidence for authorization and has been shown to be considerably more time consuming and costly compared with the 510(k) process,7,,12 which allows devices to bypass authorization by proving substantial equivalence to preexisting devices.13
Based on the analysis performed in this article, the majority (86.3%) of orthopedic hip device recalls were approved via the 510(k) approval process compared with the more stringent PMA process. Given that 86.3% of the devices on the market during the period of analysis were 510(k)-approved devices, this suggests that 510(k)-approved devices were not more likely to be recalled when considering all FDA-determined reasons for recall. When orthopedic hip device recalls were separated by approval pathway, design issues were cited as the reason for recall for 6% and 37% of devices approved via the PMA and 510(k) approval processes, respectively.
These data suggest that the 510(k) approval process, though more cost-effective for manufacturers,7 is more likely to result in the approval of orthopedic hip devices that will end up being recalled because of design issues. We propose that the 510(k) approval process may need to be more stringent, so as to increase the safety regulation on these devices or possibly shift the number of devices approved from the 510(k) approval process to the PMA process. Further, although one might hypothesize that increasing the stringency of the 510(k) approval process would decrease design defects, it also could increase the cost of bringing devices to market and thus hinder the advancement of technology.
To further understand the effects of this hypothesis, understanding the process of device recalls is important. According to the recall process, healthcare facilities and surgeons are required to notify the manufacturer or FDA of complications that could ensue from the use of a medical device. This occurs by filing a medical device report (MDR) with the FDA, which helps to detect and resolve issues quickly and to determine the need for a recall. The agency encourages voluntary medical device reporting from all health professionals, patients, caregivers, and consumers.14 The manufacturer has the duty of notifying surgeons and healthcare facilities, while healthcare facilities and surgeons must notify patients and provide further evaluation of possible implications of a faulty device.15
Although the burden of reevaluating patients and performing revisions falls on the surgeon, the financial responsibility often falls on the manufacturer in the form of reparations. The orthopedic medical device manufacturer Stryker Howmedica Osteonics reached a lawsuit settlement in 2014 to pay more than $1 billion to help cover costs of injuries and future complications related to their recalled devices.16 In addition, the Johnson & Johnson subsidiary DePuy faced more than 10,000 lawsuits for one of its devices. The first lawsuit was settled in excess of $8 million.15
Given this information, we can better understand the effects of increasing the stringency of the 510(k) approval process. The main hypothesized disadvantage of this solution is the increased cost burden on manufacturers, which could deter investment in new technologies. However, it also should be considered that manufacturers already are financially responsible for reparation costs of recalled devices, as evidenced by the lawsuit settlements described above. This raises the question of whether increasing the stringency of the 510(k) approval process will increase manufacturers' total costs and hinder the advancement of technology or whether the net increase would be inconsequential. With more thorough inspection of devices prior to market approval, the theoretical decrease in reparation costs imposed on manufacturers could offset the increased cost to market.
This means that increasing the stringency of the 510(k) approval process could effectively decrease harm imposed on the public from recalled devices, though an increase in total cost to manufacturers could occur that might hinder advancement in health technology. Further research and investigation is needed to determine the extent to which total costs would be affected.
Time to Complete Recall
The average time from initiating to completing a recall was 541 days, with most recalls taking between 301 and 400 days. It is important to note that devices are assigned a classification (I, II, or III) to indicate the relative degree of risk. Class I devices pose the least risk, while Class III devices pose the greatest risk.5 This device classification system should be distinguished from the recall classification system, which classifies recalls as Class I, II, or III in the opposite manner (from greatest [Class I] to least [Class III] risk).6
The FDA only reports device class, and not recall class, in the recall database. In the current analysis, two Class I devices were recalled (average 738 days; minimum 421 days, maximum 1,054), 732 Class II devices were recalled (average 556 days; minimum one day, maximum 2,001 days), and 40 Class III devices were recalled (average 259 days; minimum 111 days, maximum 1,517 days). This observation suggests that devices that posed the greatest risk took the least time to recall, while devices that posed the least risk took longer to recall. The FDA database unfortunately does not provide information on recall classification, which could be useful to further understand the correlation between the severity of a recall and the time to complete a recall.
This article previously discussed the process of device recalls that involve MDRs and the importance of collaboration among manufacturers, healthcare facilities, and surgeons. The flaws present in this process must be highlighted in order to reduce the duration of public exposure to a harmful device. The first step is reducing the time between a surgeon reporting a harmful device and the initiation of the recall by the manufacturer or FDA. The second step is improving the efficiency with which the recall is processed and executed.
Regarding the first step, the FDA recognizes that MDRs have several limitations, including “submissions of incomplete, inaccurate, untimely, duplicative, unverified, and biased data” and that neither incidence nor prevalence of an event can be determined because of “under reporting of events, inaccuracies of reports, lack of verification that the device caused the reported event, and lack of information about frequency of device use.”14 Although these data are available in the FDA's public MAUDE (Manufacturer and User Facility Device Experience) database, the agency is correct in its assessment of the severe limitations of MDRs. These limitations were evidenced by the authors' inability to extract more reliable recall data from individual reports.
Surgeons do not always report problematic devices, and a multiplicity of factors can influence their decision to report. One school of thought would propose that surgeons are reluctant to report devices because of concern for litigation. However, these lawsuits are defined as product liability cases. Thus, the injured person can sue the manufacturer, distributor, and/or seller of the product, while the surgeon is not liable.17 Although surgeons may be protected in product liability cases, they may fear repercussions outside of litigation. According to Harlan M. Krumholz, MD, SM, of the Yale School of Medicine, physicians may remain silent for fear of speaking out against manufacturer devices.18 In 2008, Lawrence Dorr, MD, an orthopedic surgeon, spoke out to warn fellow surgeons of a flaw in a hip implant design by Zimmer Holdings. Zimmer responded by questioning his skills as a surgeon, which Dorr reported was detrimental to his practice for more than a year.18
Incidents such as this have provided surgeons with a reason to remain silent, which only further delays the time to initiate a recall. In addition, healthcare facilities do not always have an efficient system in place to account for recalls and distribute recall information. Some healthcare facilities have addressed this issue by assigning a recall coordinator—an individual responsible for receiving, distributing, and tracking device recall information.19 These healthcare facilities have reported substantial improvement in the efficacy of their recall process.19
Therefore, multiple factors need to be addressed to decrease the time that a harmful device remains on the market. Surgeons and manufacturers must be willing to report devices sooner and avoid incidents that may adversely affect one another. The FDA must continue to revise the quality requirements of MDRs to allow more accurate understanding of why devices were recalled and the need for recall initiation. In addition, healthcare facilities need to optimize their handling and distribution of recall information.
Reason for Recall
Device design, packaging, and process issues made up a majority of FDA-determined reasons for device recalls. Device design issues were cited as the reason for recall in a higher percentage of devices approved via the 510(k) (37%) versus PMA (6%) process. These data suggest that compared with the PMA process, the 510(k) process is more likely to result in approval of orthopedic hip devices that will later be recalled for design defects.
Packaging issues made up 73% of PMA-approved hip device recalls. Although device design issues appear to be dependent on the more stringent premarket testing of PMA compared with 510(k), more thorough packaging, process controls, and labeling of devices could lead to a substantial decrease in recalls for devices approved via 510(k) and PMA. However, the data available in the FDA database does not describe the specific defect or how the reason for recall was determined. In addition, the database specifies both an FDA-determined and manufacturer-determined cause of recall, and the causes do not always match. Because of the limitations of available data in the FDA database and the low quality of MDRs, it cannot be determined which cause of recall is more accurate. The FDA-determined cause of recall was selected for the current analysis.
Of note, a high number of metal-on-metal total hip prostheses were recalled during the time period of analysis. The following devices were identified as causing metal-on-metal complications: DePuy ASR Acetabular System, Stryker Rejuvenate Modular and ABG II Modular-Neck Hip Stems, Zimmer Durom Acetabular Component and M/L Taper with Kinectiv Technology, Wright/MicroPort Orthopedics Profemur Hip Stem and Profemur Neck Varus/Valgus CoCR, and Smith & Nephew R3 Metal Liners of the R3 Acetabular System.20 In total, 101 recalls were initiated for these devices. The DePuy ASR Acetabular System made up the largest portion of these recalls (68 recalls totaling 120,984 units).
Total Medical Device Recalls
Although the total number of medical device recalls increased by approximately 39% between 2013 and 2017,2 the current analysis showed that orthopedic hip device recalls decreased by approximately 10 devices per year between 2007 and 2017. However, as shown in Figure 2, the total number of hip device recalls did not decrease steadily but rather fluctuated from year to year. Of note, considerably fewer recalls occurred between 2013 and 2017 versus between 2007 and 2012 (average of 32 vs. 102 recalls/year, respectively). This observation suggests that hip device quality and safety have improved.
This study involved several limitations. The analysis sought to identify recalled implant products used in hip arthroplasty surgeries by searching the FDA database for specific terms. Therefore, if a recall involved a study-appropriate product recall but did not use one of the search terms, that product was not captured. We also were unable to identify the approval process for 80 products (10%) within the data set because of limitations in the FDA database. As discussed previously, limitations also exist that relate to obtaining more detailed information on the FDA- and manufacturer-determined cause of recall, including low-quality MDRs. Only information in the FDA database was available for use.
Orthopedic hip devices approved via the 510(k) approval and PMA pathways were recalled proportionately to the total number of devices entering the market. However, 510(k)-approved orthopedic hip devices were more likely to be recalled because of design defects. The data reported here suggest that the 510(k) process is more prone to result in the approval of potentially harmful orthopedic hip devices than the PMA process. Although device design is the most common FDA-determined reason for recalling an orthopedic hip device, a substantial percentage of 510(k)- and PMA-approved device recalls were due to suboptimally standardized processes (e.g., packaging, process controls, device labeling). Most devices likely remain a threat to the public for 301 to 400 days after recall initiation, while some devices may take in excess of 1,500 days to be recalled completely. The devices that posed the greatest risk tended to take the least amount of time to be recalled.
The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, or U.S. government. LCDR Cory Janney is a military service member (or employee of the U.S. government), and this work was prepared as part of his official duties. Title 17, USC, §105 provides that “copyright protection under this title is not available for any work of the U.S. Government.” Title 17, §101 defines a U.S. Government work as a work prepared by a military service member or employee of the U.S. Government as part of that person's official duties.
Wesley Peters, BS, is a medical student in the Department of Orthopaedic Surgery and Rehabilitation, University of Texas Medical Branch at Galveston, Galveston, TX. Email: firstname.lastname@example.org
Carl Pellerin, BS, is a medical student in the Department of Orthopaedic Surgery and Rehabilitation, University of Texas Medical Branch at Galveston, Galveston, TX. Email: email@example.com
LCDR Cory Janney, MD, MC, USN, is an orthopedic surgeon in the Department of Orthopedics, Naval Medical Center San Diego, San Diego, CA. firstname.lastname@example.org