Medical devices used both by professional healthcare providers and the public constitute a vital part of the healthcare environment. When used according to their manufacturers’ guidance, devices are expected to be safe and effective. Given that, as discussed in Chapter 1, it is not possible to create a premarket review process that could completely ensure the safety of all devices before they enter the market, a strong surveillance system that monitors the safety of medical devices is essential. The identification of problems associated with a medical device can be an opportunity for various corrective actions, including improved device labeling, instructions for use or better user training, or, if appropriate, removal from the market.
This chapter reviews the current programs that the Food and Drug Administration (FDA) has in place, both passive and active, and some non-FDA surveillance programs that provide information on the safety and effectiveness of particular devices. It also examines how the FDA communicates information gathered through postmarketing surveillance activities to consumers, such as healthcare providers and the public.
The term postmarketing surveillance encompasses a wide array of programs, including medical device reporting by manufacturers and user facilities, third-party safety monitoring, and FDA–academic collaborations. The term postmarket surveillance refers to a specific activity defined by statute.
For medical devices, the FDA uses the term medical-device report (MDR) to encompass two types of reports—adverse-event reports are incidents resulting in a death or serious injury, and malfunction reports are incidents in which a device fails without an adverse event resulting.
Postmarketing surveillance is either “passive” or “active.” In a passive system, the regulator must depend on data from manufacturers and healthcare providers. The provision of data can be required by statute or be voluntary, but the role of the regulator is to collect and analyze the data that are provided. In an active system, in contrast, the regulator seeks information on adverse events, device malfunctions, and product effectiveness and takes advantage of opportunities to enhance data collection.
It is important to note that for most of the programs discussed below there is no reliable information about the number of devices (referred to as the denominator) on the market in clinical use. The lack of denominator information limits the ability to analyze potential safety concerns.
Mandatory and Voluntary Adverse-Event Reporting
Reporting requirements for the FDA are summarized in a 2009 Department of Health and Human Services (HHS) Office of the Inspector General (OIG) report (OIG, 2009), which states that
regulations require device manufacturers to report to the FDA (1) within 30 calendar days of acquiring information that reasonably suggests one of their devices may have caused or contributed to a death, serious injury, or malfunction and (2) within 5 working days if an event requires action other than routine maintenance or service to prevent a public health issue. Regulations also require user facilities, such as hospitals and nursing homes, to report deaths to both the manufacturer, if known, and the FDA within 10 working days. User facilities must report serious injuries to the manufacturers (or the FDA if the manufacturer is unknown) within 10 working days. User facilities must also submit annual reports to the FDA of all adverse event reports sent to manufacturers or FDA in the past year.
The vast majority of MDRs are reported by manufacturers; user facilities and others provide a small proportion of the reports received by the FDA (see Table 5-1). As a general matter, patients, caregivers, and healthcare professionals are not legally obliged to report adverse medical events. Consumers, such as healthcare providers and patients, can provide voluntary adverse-event reports to the FDA through its MedWatch program (FDA, 2009b).
|Type of Adverse-Event Report||2003||2004||2005||2006||2007|
|Percentage of total reports||90%||90%||92%||93%||94%|
|Percentage of total reports||4%||4%||4%||3%||2%|
|Percentage of total reports||7%||6%||4%||5%||4%|
NOTE: Other reports include FDA voluntary and distributor reports; percentages do not add to 100, because of rounding; 2005 and 2006 report totals do not include one report and five reports, respectively, because of missing data needed for analysis.
SOURCE: OIG, 2009.
Customary Data Sources
Two fundamental steps in all MDRs are recognizing that an adverse event or malfunction has occurred and then associating the event with one or more possible diagnostic or therapeutic interventions as causal or contributing factors. Identifying the event can be problematic for many reasons. Some problems with medical devices can become apparent immediately (for example, in a procedural situation in an operating room or an intensive-care unit), but in other cases problems can take a long time to be manifested. For example, a patient can have a procedure in a hospital, be discharged, and seek follow-up care for an emerging medical problem from a physician who does not associate the problem with the earlier procedure. In such situations, the event rarely is reported. Even when it is, the MDR often includes inadequate information about the situation or the device, which makes it difficult to link the device with the medical problem (IOM, 2011).
When a medical problem is suspected to have occurred in association with use of a medical device, a bioengineer or risk manager may or may not be available to assist in the assessment of the role and potential responsibility of the device and to take appropriate corrective action. When information is available about the role of the device in the event, healthcare facilities
are required by the Safe Medical Devices Amendments of 1990 to report the problem to the manufacturer if the problem is a device-related death or serious injury. In addition, the onset of some medical problems may be delayed, and the person using the potentially problematic device may no longer be a patient of a healthcare facility that is legally required to file an MDR.
Physicians increasingly use higher-risk devices, including implantable devices, in their offices where there is no legal requirement for them to report adverse events and device failures to the FDA. Voluntary reports made by healthcare providers have always made up a very small fraction of the reports received by the FDA. In 1993, the MedWatch program was established to improve reporting by device users, but it has had little impact: only 5.6% of reports were voluntary in 2004 (Greenfield, 2007).
The total number of events reported has increased steadily from 72,866 in 2003 to 150,210 in 2007 (OIG, 2009). Most initial reports of adverse medical events or device malfunctions lack critical information about the patient’s medical history. The Office of the Inspector General found that of the more than 140,000 MDRs filed by manufacturers in 2007, 50% of the reports were missing device identification information, such as model numbers, and 11% were missing descriptions of the adverse event or device malfunction (OIG, 1990). Follow-up inquiries are essential to fill in the gaps in information. Manufacturers are required by the FDA to perform follow-up for the majority of reports received to obtain the missing information for the medical-device reporting system. The agency also requires manufacturers to trend adverse events for reporting.
Investigations of adverse events and device malfunctions can be hampered by refusals to provide further information on the part of the patient (who may now be involved in litigation for damages), the treating physician or the physician who used or implanted the device (who may become a defendant in litigation over alleged malpractice), and any healthcare facility involved (possibly also a defendant in litigation). The device itself might not be available for examination and testing. And the manufacturer of the device may become involved in product-liability litigation. There are other reasons, not related to liability or litigation, that also affect investigations, such as the unreimbursed cost to physicians and healthcare facilities of responding to requests for information.
The next phase depends on how the manufacturer handles the information. There are regulations governing timely review and maintenance of records. Manufacturers determine whether the information meets the threshold for a reportable event. They may categorize an event as user error (OIG, 1990). There have been instances in which manufacturers have underreported serious adverse events by not reporting them to their own regulatory staff (Bren, 2003). However, whenever any agent of a company
knows of an adverse event, the company is deemed to have knowledge and is required to report. A company that does not report is in violation of the law.
Manufacturers are required to report within 30 days if a device may have caused or contributed to a death or serious injury or, if a device malfunctioned and a death or serious injury could result if the malfunction were to recur, and within 5 days if a reportable event necessitates remedial action (other than routine maintenance or servicing) to prevent an unreasonable risk of substantial harm to public health. In 2007, 31% of the more critical 5-day reports were submitted late. Similarly, healthcare facilities submitted 39% of death and injury adverse-event reports late (OIG, 2009).
The timeliness of the FDA review of MDRs is also problematic. Fewer than one-third of MDRs were reviewed for the first time within 30 days, and fewer than half were reviewed within 60 days in every year from 2003 to 2007. Documentation of the reviews is also inconsistent, and this makes it difficult to track the agency’s response to a specific event. Moreover, the FDA Office of Compliance does not link inspections to the adverse event that may have triggered them.
In addition to the previously described system for mandatory and voluntary reporting of adverse events and device malfunctions, summary reporting from industry provides an abbreviated method that relies on established codes for events that are well known and categorized. Although they allow trending, the summaries provide little in the way of new information, and may have outlived their usefulness in that they hamper the function of an already overloaded system. Similarly, the compliance of user facilities with required annual reports is difficult to assess because CDRH was able to provide fewer than half the reports to the HHS OIG (OIG, 2009). These reports can also be repetitive with respect to individually submitted events. Larry Kessler, former director of CDRH’s Office of Surveillance and Biometrics, has suggested placing more emphasis on keeping the data in the manufacturers’ quality-reporting systems (Kessler, 2010).
Each MDR is entered into an event database by an FDA contractor and then undergoes triage by a safety analyst (generally a nurse or engineer working for the FDA) to look for a signal of risk of potential substantial harm. That screening is made more difficult by the low signal-to-noise ratio, the MDR reviewers’ narrow experience with new technology, the absence of input to the reviewers by premarket staff more familiar with the device, and the sheer volume of reports, which exceeds the capacity of the current system (GAO, 2009b; IOM, 2011). Efforts to integrate the premarket and postmarket phases of review in a matrix approach have been proposed since 2005 (Schultz, 2007), but have proved difficult to sustain because staff were
in different facilities and had a heavy workload. But an integrated approach is important when reports of adverse events and device malfunctions appear shortly after introduction of a new device (Mehran et al., 2004).
As mentioned above, incomplete or poor quality MDRs makes it difficult to have an informed review of the problem. To address this issue, the FDA has safety analysts, who have relevant technological or clinical experience, attempt to contact reporters (for example, manufacturers, user facilities, or voluntary reporters) for additional information. In addition, safety analysts interact with each other for clinical or engineering expertise to more fully understand reports. All safety analysts attend internal FDA conferences or “group think” meetings where a decision on further investigation may be made (IOM, 2011). In addition, data-mining is being explored by a variety of staff as an additional tool that may help identify potential safety signals.
A rule proposed in August 2009 to require manufacturers to submit MDRs in an electronic format to allow more timely access to emerging adverse event information was met with resistance from industry, which called for a longer timeframe to implement the changes (Williams, 2010).
All signals are entered into a CDRH tracking system, but the main repository for adverse event reports is the Manufacturer and User Device Experience (MAUDE) database. This database is considered to be in need of reform because it is difficult to use and is not connected to any other CDRH database (FDA, 1999b). When William Maisel, former director of the Medical Device Safety Institute, attempted to use MAUDE for analysis of 510(k)-cleared devices, he found that the data were not well suited to analysis, because of incomplete reporting, insufficient information, and misclassification (IOM, 2011). Problems are also caused by the difficulty of aligning product-code assignments, but CDRH is trying to address this issue (FDA, 2010c). As discussed in Chapter 3, efforts to upgrade information technology (IT) facilities in CDRH in recent years have been less than successful, and, as documented in the FDA Center for Drug Evaluation and Research, the challenges are exacerbated by the lack of communication between IT management and users (Breckenridge Institute, 2006).
The Government Accountability Office (formerly the General Accounting Office) reported in 1989 and again in 2009 that the FDA was unable to manage its postmarket-surveillance responsibilities because of resource constraints, but the agency is also unable to estimate its current and future resource needs effectively because of a lack of reliable management information (GAO, 1989, 2009a).
Finding 5-1 The FDA’s current postmarketing surveillance system relies on manufacturers, importers, and healthcare facilities to collect information, to investigate, and to make mandatory reports. Voluntary reporting of adverse events and device malfunctions
depends on patients, caregivers, and healthcare providers and facilities to identify them, to associate them with medical devices, and to submit reports.
Finding 5-2 The inadequacy of the current postmarketing surveillance system and the resulting lack of data make it impossible to confidently draw broad conclusions about the safety and effectiveness of products that are on the market.
Finding 5-3 Data collected with the current postmarketing surveillance system is not systematically integrated into the premarket review process.
In addition to the general reporting requirements described above, the FDA conducts postmarketing surveillance via several other mechanisms: tracking of medical devices, the MedSun program, the MD EpiNet program, the Sentinel Initiative, and Section 522 surveillance studies.
Tracking of medical devices by manufacturers can assist in notification to users of potential problems with a device and can facilitate recalls when necessary. The FDA requires tracking of 12 implantable devices and four other devices that are used outside hospitals. They include joint prostheses, implantable pacemakers, implantable defibrillator, mechanical heart valves, ventricular-bypass-assist devices, and implantable infusion pumps (Diehl et al., 2010). Manufacturers are expected to be able to provide within 10 days information about the location of devices that have been distributed to patients and within 3 days information about devices that are in inventory.
Recognition of underreporting of adverse events led to the initiation in 2002 of a pilot program, the Medical Product Surveillance Network (MedSun), in which trained risk managers could recognize and report adverse events electronically. That direct connection with the clinical community in 350 hospitals, nursing homes, and outpatient diagnostic and treatment centers not only improved the quality of reports but also allowed the identification of “near-misses” (FDA, 2009a). In addition, focus networks were developed on specific subjects, such as LabNet, which focuses on hospital laboratories; TissueNet, on human cells, tissues, and their products;
SightNet, on ophthalmic devices; HomeNet, on device training and problems in the home environment; HeartNet, on electrophysiology laboratories; and KidNet, on neonatal and pediatric units. MedSun also conducts small sample surveys involving a small number of institutions to answer specific questions on product safety (FDA, 2009c).
A plan to implement real-time adverse-event reporting and establish pathways to interactive information exchange with healthcare providers is included in the FDA Strategic Plan for 2010 (FDA, 2010c). Additional plans include unspecified expansion of the MedSun nets, further evaluation of the regional representative pilot, and the incorporation of “large providers.” There are currently no resources allocated to implement those plans. The value of the approach is well demonstrated by the fact that 252 of the 350 MedSun facilities reported adverse events in 2007, accounting for 72% of all facilities that reported. In contrast, only 267 of the remaining thousands of other facilities reported adverse events in 2007, and that number has been declining (OIG, 2009). The regional representative pilot has established valuable linkages to healthcare providers and has led to more adverse-event reports. A 285% increase in reported medical-device concerns originating in pediatric specialties was demonstrated (Desjardins, 2011).
The MD EpiNet program, which began in February 2010, is intended to improve epidemiologic assessment of device performance by establishing an extramural network with 10 leading academic institutions that have experience with medical-device studies. The stated long-term goal is to “substantially contribute to the understanding of medical device performance and CDRH decision making, thereby improving public health, with an FY-10 focus on building a research infrastructure by linking CDRH with leading academic organizations in the country that have experience with medical device studies” (FDA, 2011b). As of May 2011, the Web site for the program lists the accomplishments as “establishment of a partnership development team to develop a research plan,” holding a “kickoff workshop with academic stakeholders” on April 30, 2010, and a workshop to discuss methodologic issues related to studying medical device performance on April 25, 2011, awarding contracts for administration of the program, and initiating pilot projects in the selected centers.
In 2007, Congress passed the FDA Amendments Act (FDAAA), Section 905 of which mandated that the FDA establish a postmarketing risk identification and analysis system to support active monitoring of postmarket-
ing drug safety. The FDA is responding to that mandate by developing the Sentinel system, which will harness clinical and administrative data held by existing health-information holders. Congress set a target for the system to include data on 100 million persons by July 2012 (FDA, 2011a). Although Section 905 of the FDAAA specifically authorized creation only of an active surveillance system for drugs, the FDA is using its general authority under Section 1003(b)(2)(c) of the Federal Food, Drug, and Cosmetic Act (FFDCA) to include medical devices in the Sentinel system. The Sentinel Initiative helps the FDA to fulfill its mission to “protect the public health by ensuring that … there is a reasonable assurance of the safety and effectiveness of devices intended for human use.”1
The FDA has already received deliverables from 10 small contracts that address various issues related to system planning and design, such as data availability, system architecture, methods, legal and privacy issues, and stakeholder engagement. In addition, it has entered a cooperative agreement with the Brookings Institution to convene meetings and workshops on active medical-product surveillance issues. Early in 2010, the FDA awarded a 5-year contract to the Harvard Pilgrim Health Care Institute to develop a “Mini-Sentinel” pilot project, a scaled-down version of the Sentinel system. The Mini-Sentinel Coordinating Center is identifying appropriate data sources, developing a scientific framework for obtaining real-time data, and developing procedures to ensure data quality and privacy protection (FDA, 2010d). As of January 2011, the electronic health records of more than 60 million people have been added to the system (Behrman et al., 2011). To protect personal information, the system relies on a distributed network architecture that keeps identifiable health data behind the existing privacy firewalls of the participating data sources; only summary results are sent to the coordinating center. The mini-Sentinel and Sentinel planned activities are public-health activities and the Common Rule2 does not require informed consent of individuals whose records are being examined (Rosati et al., 2010). These activities also fit within an existing exception to the Health Insurance Portability and Accountability Act privacy rule that allows disclosures to public-health authorities without individual authorization.3 The first year of the Mini-Sentinel effort focused on developing a data model common to data sources with an initial focus on claims data (Mini-Sentinel Coordinating Center, 2010). The Sentinel system is expected to expand and gain additional functionality over the next several years.
The Sentinel system will be able to provide data that are applicable to medical devices. For example, one Sentinel-related project identified,
1FFDCA § 1003(b)(2)(C).
2Federal policy for the protection of human subjects (Common Rule) (implemented by the Department of Health and Human Services at 45 CFR § 46.101–124).
345 CFR § 164.512(b)(1)(i).
described, and evaluated potential US orthopedic-implant registries that could participate in the creation of a national network of such registries as part of the Sentinel Initiative (Outcome Sciences, 2009). Data related to medical devices include rates of selected outcomes (for example, myocardial infarction and stroke), rates of infection, and rates of implant revision and reintervention. They also may be able to address functional status and quality-of-life outcomes. However, some 510(k)-related issues—such as software problems, manufacturing defects, out-of-box failures, misconnects and disconnects, packaging and labeling errors, and design-induced use errors—will not be captured (IOM, 2011).
As the FDA expands the scope of Sentinel to include medical devices, additional resources will be needed. Such an investment is expected to provide a rich source of clinical data from a much broader segment of the population and will prove cost-effective in the long term as adverse events associated with device use are identified and reduced (Behrman et al., 2011).
Section 522 Surveillance
Section 522 of the FFDCA is a discretionary tool that allows CDRH to require manufacturers to perform specified postmarket clinical studies of Class II and Class III products (Gross and Kessler, 1996).4 Such studies are justified when device failure is likely to cause serious health consequences, if the device would be implanted for more than a year, or if it is a life-sustaining device used outside a health facility. A Section 522 study can be used as a condition of clearance for a Class II device that is expected to have substantial use in pediatric populations.5
Section 522 studies are generally focused on only one or two aspects of performance rather than on overall risk and performance, and the duration of each study is limited to 3 years unless a longer period is agreed to by the manufacturer. That period is too short for the discovery of some late safety or effectiveness problems. Even when studies are completed, they may not require reporting of all critical information, including mortality (Lenzer and Brownlee, 2010).
Unique Device Identifiers
The FDAAA requires the FDA to develop a system of unique device identifiers (UDIs) for distribution and use. The FDA has conducted several meetings and public workshops with key stakeholders, including label manufacturers, device manufacturers, and hospitals. The agency has also
4FFDCA § 522(a)(1), 21 USC § 360l(a)(1) (2006).
5FFDCA §§ 522(a)(1)(A)–(B), 21 USC §§ 360l(a)(1)(A)-(B) (2006).
conducted pilot activities on the effects of UDI implementation on FDA and labeler-organization business processes. It has also completed a pilot test of the usability and feasibility of a prototype UDI database (FDA, 2010h).
Manufacturers want to ensure that the FDA’s standards are in alignment with those used in other regulatory systems. Advocates for the implementation of UDIs argue that UDIs could substantially improve the FDA’s ability to track medical devices once they are on the market. Implementation of a comprehensive UDI system could also reduce medical device–related errors, improve the quality of MDRs, facilitate device recalls and tracking, standardize device nomenclature among administrative databases and clinical registries, identify device use and compatibility issues, and enhance device postmarket surveillance (ERG, 2006).
Finding 5-4 Several tools, such as device tracking and Section 522 surveillance studies, are available to the FDA to improve postmarketing surveillance, but they are used only sparingly.
Finding 5-5 The FDA has postmarketing surveillance programs—such as MedSun, MD EpiNet, and the Sentinel Initiative—that are scientifically promising, but achieving their full promise will require a commitment to provide stable, adequate resources and will require resolution of various technical issues, such as unique device identifiers.
Compliance and Enforcement
When there is a problem with an FDA-regulated product, the agency’s initial effort is to work with the manufacturer to have it corrected voluntarily. If that fails, a number of legal remedies are available: the manufacturer can be asked to recall the product, federal marshals can be used to seize the product, or an imported product can be detained at the port of entry until the problem is corrected. In addition, individual company officers that deliberately violate the law can be prosecuted, with the possibility of criminal penalties. Chapter 3 contains a detailed description of the postmarket regulatory authorities that are available to the FDA.
Each center has an office of compliance that ensures compliance with regulations for premarket or postmarket studies and for manufacturing and labeling. The Office of In Vitro Diagnostic Device Evaluation and Safety (OIVD) is responsible for both premarket review and postmarket enforcement for in vitro diagnostic devices. For non–in vitro diagnostic devices, the Office of Device Evaluation (ODE) oversees premarket review, but the compliance function resides in the Office of Compliance (OC). Inspections can be made at any establishment where devices are manufactured, processed,
packed, used, or implanted or where records of results are maintained. In addition, during inspections of manufacturers, inspectors review the manufacturers’ procedures and records pertaining to reported adverse events and customer complaints. Since 2004, when there were 2,936 such foreign and domestic inspections, the number has steadily declined; there were 2,353 inspections in 2008 (Williams, 2010). Even when manufacturing problems are reported, the lack of effective action can lead to serious consequences in some cases (Lenzer, 2009).
Each compliance office works closely with the Office of Regulatory Affairs, which is responsible for the regulation of 124,000 establishments that produce, store, and transport an estimated $1 trillion dollars worth of medical products. Consumer safety officers and inspectors conduct about 22,000 inspections a year and also monitor clinical trials (Williams, 2010).
When there is a violation of the laws that the FDA enforces related to a device, process, or practice, the initial communication is likely to be a warning letter to a responsible person, manufacturer, or facility to take appropriate and prompt action. Warning letters are informal and advisory but are available to the public on the FDA’s Web site and usually receive the attention of the news media. The number of warning letters from CDRH has steadily declined from 528 in 2000 to 136 in 2009 (Williams, 2010). In addition to a warning letter, the FDA may also issue a safety alert in the form of a “Dear Doctor” letter to warn healthcare providers and consumers of the risk associated with a device.
The effect of a warning letter varies. If the letter is directing a manufacturer to stop promoting a medical device for a specific indication, it can be corrected immediately. If the letter is related to a manufacturing issue, however, the process could take several months or even years to resolve. The FDA has the authority to use warning letters to block device PMA approvals, but generally not 510(k) clearances because of a 1997 legislative enactment. The FDA may withhold clearance of a 510(k) submission for manufacturing issues only if there is a “substantial likelihood” of a serious risk to health from the specific product because of a specific good manufacturing practice violation.
Under the current FDA definition, a recall is any action to remediate a risk posed by a product already transferred from the manufacturer’s control to others (such as wholesalers, retailers, end users, or patients). Actions that may be considered recalls include a software patch, a sticker on the package,
a Dear Doctor letter, a Dear Patient letter, servicing in the field, and a revision of labeling to add warnings or clarify instructions. Recalls of medical devices are usually conducted voluntarily by manufacturers on their own initiative or after negotiation with the FDA. Manufacturers and importers are also required to report to the FDA any voluntary correction or removal of a device undertaken to reduce a health risk posed by the device.6 Occasionally, if a manufacturer or importer fails to recall a device that poses a risk to health, the FDA may issue a formal recall order to the manufacturer.7
Recalls are classified by the FDA according to the degree of hazard, which is based on injuries or deaths that have occurred, the likelihood of occurrence, the population exposed, and immediate and long-term consequences:
• Class I recalls are the most serious and are reserved for situations in which there is reasonable probability that use of or exposure to a product will cause serious adverse consequences or death.
• Class II recalls are for situations in which a product may cause temporary adverse health consequences but in which there is the possibility of severity great enough to have irreversible consequences.
• Class III recalls are for situations in which use of or exposure to a product is not very likely to cause adverse health consequences but the likelihood is great enough to justify recall.
Class II recalls make up the largest recall class and have increased progressively from 1,235 in 2004 to 2,178 in 2008 (Williams, 2010).
Off-label use of a device—its use in a manner different from that of its original approval—is not regulated by the FDA but can lead to a recall. That situation occurred in 2004 when insertion of biliary stents to treat arterial disease had adverse outcomes (FDA, 2011d); in this case, it was the instructions for use that were recalled, rather than the devices themselves.
Limitations of Recall Data
The committee explored FDA recalls of medical devices as potential indicators of their safety. A recall study can be a useful tool for identifying the point in the product life cycle at which a problem is occurring, such as problems with device design or with the manufacturing process. Recall studies have limitations, however. FDA-initiated recalls may not fully reflect the safety, or lack thereof, of devices in clinical use. Without a robust postmarketing surveillance system that can detect safety problems dependably, the absence of a recall can mean that there were no safety problems
621 CFR pt. 806.
721 CFR pt. 810.
There have been reports of device malfunctions and adverse events related to permanent and temporary (retrievable) intravascular filters (McCowan et al., 1992). Data in multiple reports in the scientific literature make a strong case for believing that the adverse-event and device-malfunction rates of permanent and temporary filters as cataloged by the FDA understate actual rates (Dorfman, 1990; FDA, 1999a, 2010g; Nicholson et al., 2010). In August 2010, the FDA issued a medical alert noting that retrievable inferior vena cava filters could move or break and perhaps cause serious problems (FDA, 2010e). The FDA cited malfunction rates that are much lower than rates cited in nearly contemporaneous reports in the scientific literature (Dorfman, 1992; FDA, 2010g; Nicholson et al., 2010). No recall has been issued.
with a device or that the device’s safety problems went undetected. The study of recall data is hampered by the lack of reliable denominator data. Recall rates often do not correlate with MDRs or with adverse-event rates reported in the scientific literature. Box 5-1 presents an example of the lack of a relationship among data in the scientific literature, MDRs, and recalls.
Two studies that evaluated recall rates of 510(k)-cleared and PMA-approved devices were presented at a July 2011 workshop hosted by the committee (IOM, 2011). William Maisel (formerly of the Medical Device Safety Institute) reported that recalls affected 510(k)-cleared devices 400–500 times a year from 2003 to 2009 (more than 48,000 devices were cleared through the 510(k) pathway in 1996–2009). He found that the annual rate of recalls of 510(k)-cleared devices was highest within the first 4 years after clearance (1.4–1.6%) and that the rate decreased after 5–6 years (to 0.9–1.1%). The reasons for recall included problems with the manufacturing process, including storage, labeling, and maintenance (28.8%); failure of a device to perform as intended despite meeting design specifications (28.4%); problems with nonconforming materials or components (16.3%); adverse changes in specifications or procedures (11.9%); employee error (7.1%); and miscellaneous (7.5%). The majority of recalls for the period reviewed (2003–2009) were of devices that went through the traditional 510(k) clearance process. Devices most likely to be recalled had a larger number of predicates (more than six), went through the special 510(k) clearance process, went through third-party review, were life-sustaining, or were Class
III devices. Three-fourths of recalled devices that were 510(k)-cleared were recalled a single time. Maisel also examined MDRs and recalls of 510(k)-cleared devices. He found that 41.8% of MDRs on 510(k)-cleared devices were associated with devices that were subject to recall.
Ralph Hall, University of Minnesota Law School, conducted the second study evaluating recall data on 510(k)-cleared devices. He presented his findings to the committee at its July 2011 workshop (IOM, 2011). Hall’s review of Class I recalls for the years 2005–2009 showed that 55% were related to postmarket issues. He suggested improving quality system regulations with better design controls, bench testing, and preclinical studies.
Medical-device recalls also were assessed in a third study, which analyzed FDA recall data from 2005 to 2009 (Zuckerman et al., 2011). For that period, 113 Class I recalls were identified. Of those, 21 were of devices approved by the PMA pathway, 80 were cleared by the 510(k) pathway, 8 were exempt from FDA regulation and had only to be registered, and 4 were counterfeit devices or categorized as “other” and did not go through approval, clearance, or registration. Of the recalled devices, 35 were cardiovascular devices (the largest category), and 23 of them had gone through the 510(k) clearance process; 13 of the 23 were Class III devices.
The committee concluded that these studies, while offering insights, have little utility in assessing the ability of the 510(k) process to assure safe products reach the market. Recall data lack strong denominator information, or even a consensus on the proper denomination group. Classification of recalls may be imprecise as well. Some important problems with 510(k)-cleared devices do not rise to the level of a Class I recall (Kessler, 2010). As noted above, there is a lack of correlation between MDR data and recalls, and moreover, data in MDRs can be incomplete and insufficient. Jeffrey Shuren, director of CDRH, commented that most recalls of 510(k)-cleared devices are Class II recalls, which have the potential to have a serious effect on public health and safety (Shuren, 2010).
Some medical-device postmarket data-collection activities are outside the FDA. They are in four major categories: postmarketing surveillance programs, administrative databases, clinical registries, and electronic health records. The FDA has partnerships with academic institutions, federal agencies, and professional medical societies. Although current collaborations are focused on devices approved through the PMA process, the partnerships provide a model of potential mechanisms for the collection of postmarketing data on devices cleared through the 510(k) process. Although non-FDA data sources have several strengths and limitations, linking the various
data sources could potentially assist the FDA in postmarket surveillance of medical devices. The development of a network of existing administrative databases, electronic health records (EHRs), and clinical registries may enhance the FDA’s ability to assess longitudinal device outcomes.
Postmarketing Surveillance Programs
The Department of Veterans Affairs (VA) Cardiovascular Assessment, Reporting, and Tracking (CART) system is a cardiovascular-disease surveillance program that is integrated into VA’s Computerized Patient Record System. The CART program is used by VA for quality, management, patient-safety device surveillance, and research (IOM, 2011). VA shares with the FDA information collected through the CART program on signals or unexpected problems with devices reported by clinicians. The strengths of this program include the level of clinical, patient, device, and outcome data; VA-wide collaboration; integration of data collection into work flow; clinical tools, such as real-time report generation; VA-wide standardization; e-mail reporting of complications; monthly site reports; national reports; and Veterans Integrated Service Network reports (IOM, 2011). Although it is successful for cardiovascular surveillance within the VA system, the utilization of this system in other organizations that have different electronic health record (EHR) systems and levels of integration may limit generalizability. Expansion of this model to other devices in the VA system may provide an additional data source for the FDA’s postmarket surveillance efforts. It should be noted that this effort is restricted to devices associated with a small number of clinical diagnoses and procedures. Thus, a small number of personnel need to be trained and integrated into the effort. Expansion of this model to other disease states and device types, even within the same institution, may bring additional challenges.
Another surveillance program is the Data Extraction and Longitudinal Trend Analysis (DELTA) system. DELTA is a software system that supports automated postmarket surveillance of cardiovascular devices (IOM, 2011). The software provides signal detection of adverse events with a generic structure, multiple analyses, and statistical applications for risk assessment (IOM, 2011). DELTA-detected signals must be followed up with interpretation and further investigation. Although successful in cardiovascular signal detection, the system has not yet been applied to other medical specialties. The FDA, which provides research support for the DELTA project, is exploring application of the DELTA system to orthopedic-device surveillance.
Several existing administrative databases could be used by the FDA for postmarket data collection on medical devices. For example, the FDA has used the Centers for Medicare and Medicaid Services (CMS) database to study the safety of implanted surgical mesh (IOM, 2011). The FDA’s MD EpiNet program has been established to leverage the use of existing administrative databases in combination with internal FDA premarket and postmarket surveillance studies (IOM, 2011). Extended use of CMS data could enhance the monitoring of new technologies in people 65 years old and older from a wide variety of healthcare facilities (Mehran et al., 2004) and has the potential for longitudinal device monitoring (Normand et al., 2010). Although CMS data are related to a large, representative sample of the older patient population, limitations in device monitoring include restriction in age range, lack of laterality, sparse clinical and implant data, and various limitations associated with administrative database coding.
Other potential databases for assessing medical-device performance include the Nationwide Inpatient Sample (NIS) (HCUP, 2010), the National Hospital Discharge Survey (NHDS) (CDC, 2010), and state hospital-discharge databases. The NIS and the NHDS provide an opportunity to assess performance of classes of devices or specific procedures, but they lack detailed device information, such as manufacturer and catalog and lot numbers, that are necessary for evaluating the performance of specific devices. In addition to the need for linking devices to procedures or diagnosis codes, these databases contain few outpatient data and availability of recent data (Torrence, 2002).
Domestic and international clinical registries provide additional opportunities for postmarket data collection. Registries are defined as “organized systems that use observational study methods to collect uniform data (clinical and other) to evaluate specified outcomes for a population defined by a particular disease, condition, or exposure, and that serve one or more predetermined scientific, clinical, or policy purposes” (AHRQ, 2010). The strengths of patient registries include known denominators and the extent of clinical and implant detail necessary for assessing device outcomes. In addition, registries can provide large samples for detecting rare adverse events and provide an alternative when randomized clinical trials are not practical, ethical, or feasible (AHRQ, 2010). Another benefit of clinical registries is the real-world assessment of medical-device performance in people who are in different care settings and have varied comorbidities.
Medical professional groups, such as the American College of Cardiology (ACC), have developed national clinical registries, including the
National Cardiovascular Data Registry for implantable cardioverter defibrillators (NCDR ICD). The NCDR ICD is an example of a successful registry that collects clinical, implant, and outcome data and allows risk stratification (IOM, 2011). NCDR ICD tracking is mandated by CMS, and this mandate results in high participation by hospitals. Other ACC registries—for example, CARE for carotid artery stenting and endarterectomy procedures and ACTION–Get with the Guidelines (GWTG) for acute myocardial infarction—have lower participation because of the lack of incentives (IOM, 2011). The FDA has established relationships with ACC and other medical professional groups to collaborate in medical-device assessment. Professional-society registries, such as the ACC NCDR, provide an opportunity to capture the device and clinical data needed for risk adjustment and device assessment. Low participation, lack of longitudinal assessment, and the cost of implementation of these registries are potential barriers in expansion to other medical devices. Another potential limitation of societal registries is the lack of standardization among specialty registries that track the same medical devices. In those cases, there is no standard for such factors as data collection, nomenclature, and performance thresholds.
In addition to professional-society registries, institutional, regional, and national device registries could be used for collecting postmarket information on medical devices. For example, in orthopedics, there are several total-joint-replacement registries, such as the University of Massachusetts Global Orthopedic Registry, the Harris Joint Registry, the HealthEast Orthopedic Joint Registry, the Hospital for Special Surgery Center for Education and Research on Therapeutics Total Joint Replacement Registry at Weill Cornell Medical College, the Kaiser Permanente National Total Joint Replacement Registry, the Mayo Clinic joint-replacement database, the MaineHealth Total Joint Replacement Registry, the Rush University Medical Center Joint Registry, and the Virginia Joint Registry (Outcome Sciences, 2009). The FDA has evaluated the quality of those registries and established collaborations to assess the safety and effectiveness of total-joint-replacement devices.
Internationally, there are numerous national orthopedic registries, including the Swedish, Norwegian, and Australian joint-replacement registries. Findings from those registries have initiated recalls and advisories in the United States. For example, a recent US DePuy ASR XL Acetabular Hip System and DePuy ASR Hip Resurfacing System recall was influenced by findings from several international total-joint-replacement registries. Problems associated with these devices were first identified in the Australian registry, which detected a higher than expected revision rate for these devices in 2007. The findings from the Australian registry and similar findings from the National Joint Registry of England and Wales prompted a recall of these devices. The ASR recall demonstrates the potential of implant registries for
postmarket data collection. Recognizing the importance of device registries, the FDA is exploring the development of the scientific infrastructure for a consortium of existing domestic and international orthopedic registries.
Registries may provide detailed clinical and device information, but they also have limitations, such as losses to follow-up, which adversely affect longitudinal tracking and potentially introduce selection bias in that they are not entirely random. Participation is also a challenge for some registries when incentives are lacking. Some clinical registries were developed for specific diseases or conditions, and this can limit their use for a wide variety of medical devices. And clinical registries are resource-intensive to develop and maintain. Despite the challenges associated with registry development, implementation, and maintenance, several regional and national efforts are under way in the United States to track orthopedic devices, such as total joints. The American Academy of Orthopaedic Surgeons American Joint Replacement Registry, a California state registry effort, and an Agency for Healthcare Research and Quality multicenter registry project may be able to assist the FDA in future postmarket surveillance efforts.
Electronic Health Records
EHRs are another possible means for monitoring medical-device performance. EHRs provide an opportunity to collect data prospectively at the point of care, integrate device information into the workflow, and assess clinical and device granularity needed for risk adjustment. Despite the potential for EHRs in medical-device postmarket surveillance, there are several challenges to implementation. First, EHRs have not yet been fully integrated into all health systems (DesRoches et al., 2008; Jha et al., 2009), although the use of electronic records by hospitals and healthcare professionals is expected to increase rapidly with resources allocated by the American Recovery and Reinvestment Act of 2009. In addition, there are more than 40 EHR systems, and they do not have standardized data-entry fields and formats. Most important, medical devices are not monitored in a consistent manner. Many EHRs lack discreet data fields for efficient data extraction and analysis. Concerns regarding privacy, security, and confidentiality are other challenges to the use of data from EHRs in postmarket surveillance. For EHRs to be used for future device monitoring and surveillance, data definitions and formats must be standardized, interoperability and data-exchange issues addressed, and privacy, regulatory compliance, and confidentiality ensured. The implementation of UDIs could substantially improve the FDA’s ability to track implants by using EHRs and standardizing device information among EHR systems.
Finding 5-6 Existing non-FDA device data sources could enhance current passive FDA postmarketing surveillance systems but are variably used by the FDA and providers.
Finding 5-7 The lack of standardization in clinical and device-specific data among existing non-FDA data sources and insufficient detail in administrative and clinical health records impede the evaluation of the performance of medical devices.
When a problem with a medical device warrants action, communicating essential information to healthcare providers and the public is challenging for the FDA. Recalls and public-health notices are posted on the FDA’s weekly enforcement report (FDA, 2011e). A report is publicized when the FDA believes that there is a potential for serious public hazard. But neither safety alerts nor recalls are distributed to appropriate recipients in a timely fashion (GAO, 1998, 2007). Recall notices may be sent to a person ordering supplies, to a loading dock, or to a billing department. Confusion about the appropriate course of action may occur when a recall notice is received. Poor communication among providers, industry, and the FDA was previously noted by the Institute of Medicine in a report that focused on device problems in pediatric patients (IOM, 2005).
Several third-party organizations offer services to ease communication between the FDA and consumers about recalls and other FDA notices, alerts, and reports. For example, the Emergency Care Research Institute is a nonprofit corporation that contracts with hospitals to provide advice on appropriate responses to FDA communications (Mantone, 2006). Commercial online tracking systems for recalls (for example, RASMAS, a recall management service for the healthcare sector [RASMAS, 2011]) also are available from other organizations.
Delays in communication of recall notices leave patients at risk and hospitals with medicolegal liability. In 2002, contaminated bronchoscopes remained in use for more than 3 months after a warning was issued, and the blame for later infections was directed at both the manufacturer and the FDA (Patterson, 2002).
The Transparency Initiative established in 2009 was intended to make public access to FDA information more widely available (FDA, 2011c). The initiative includes providing information about regulated products to the public through the CDRH Web site and making public the results of Section 522 and other postapproval studies. The site was launched in April 2010 and provides access to safety alerts, public-health notices, MedSun reports, and the MAUDE database. In the first month of its availability, however, it
received only 46 comments despite a readily accessible response form and request for feedback (FDA, 2010b). In recognition of existing limitations, a recommendation from the CDRH Task Force on the Utilization of Science in Regulatory Decision Making was to use new science and available clinical experience to communicate necessary changes in device manufacturing and labeling (FDA, 2010a). Such communications might include letters to industry, public notices, ordering of a Section 522 study, or modifying premarket requirements.
The process that has proved most effective in generating reliable MDRs also has advantages for communication of recalls (Kessler, 2010). Under the MedSun program, a trained risk manager serves as the point person for direct communication of recall information. That person can use hospital information systems to communicate risk notices to the appropriate personnel rapidly and then ensure that the appropriate corrective actions take place. The committee believes that a major improvement in the removal of affected devices from the market could result from the proposed UDI system mandated by Section 519(f) of the FDAAA, depending on how such a system is implemented. This system would permit not only more rapid analysis and collation of adverse event reports but also would focus on removal of only the affected devices when necessary. The proposed UDI rule is being drafted (FDA, 2010f), and the UDI system is scheduled to be fully implemented in 2013 (FDA, 2010c).
• Finding 5-1 The FDA’s current postmarketing surveillance system relies on manufacturers, importers, and healthcare facilities to collect information, to investigate, and to make mandatory reports. Voluntary reporting of adverse events and device malfunctions depends on patients, caregivers, and healthcare providers and facilities to identify them, to associate them with medical devices, and to submit reports.
• Finding 5-2 The inadequacy of the current postmarketing surveillance system and the resulting lack of data make it impossible to confidently draw broad conclusions about the safety and effectiveness of products that are on the market.
• Finding 5-3 Data collected with the current postmarketing surveillance system is not systematically integrated into the premarket review process.
• Finding 5-4 Several tools, such as device tracking and Section 522 surveillance studies, are available to the FDA to improve postmarketing surveillance, but they are used only sparingly.
• Finding 5-5 The FDA has postmarketing surveillance programs—such as MedSun, MD EpiNet, and the Sentinel Initiative—that are
scientifically promising, but achieving their full promise will require a commitment to provide stable, adequate resources and will require resolution of various technical issues, such as unique device identifiers.
• Finding 5-6 Existing non-FDA device data sources could enhance current passive FDA postmarketing surveillance systems but are variably used by the FDA and providers.
• Finding 5-7 The lack of standardization in clinical and device-specific data among existing non-FDA data sources and insufficient detail in administrative and clinical health records impede the evaluation of the performance of medical devices.
AHRQ (Agency for Healthcare Research and Quality). 2010. Registries for evaluating patient outcomes: A users guide. 2nd edition. Rockville, MD: AHRQ.
Behrman, R. E., J. S. Benner, J. S. Brown, M. McClellan, J. Woodcock, and R. Platt. 2011. Developing the sentinel system—a national resource for evidence development. New England Journal of Medicine 364(6):498-499.
Breckenridge Institute. 2006. Independent verification and validation of AERS II requirements process. Breckenridge, CO: Breckenridge Institute.
Bren, L. 2003. Company caught in coverup of medical device malfunctions. FDA Consumer. Nov/Dec.
CDC (Centers for Disease Control and Prevention). 2010. National hospital discharge survey. http://www.cdc.gov/nchs/nhds/about_nhds.htm (accessed 12/28/2010).
Desjardins, P. R. 2011. FDA response to information inquiry from the committee on the public health effectiveness of the 510(k) clearance process. Silver Spring, MD, 1/07/2011.
DesRoches, C., K. Donelan, P. Buerhaus, and L. Zhonghe. 2008. Registered nurses’ use of electronic health records: Findings from a national survey. Medscape Journal of Medicine 10(7):164.
Diehl, D. L., W. M. Tierney, D. G. Adler, J. D. Conway, F. A. Farraye, S. V. Kantsevoy, V. Kaul, S. R. Kethu, R. S. Kwon, P. Mamula, M. C. Pedrosa, and S. A. Rodriguez. 2010. The role of the U.S. Food and Drug Administration in device evaluation and monitoring. Gastrointestinal Endoscopy 72(1):5-10.
Dorfman, G. S. 1990. Percutaneous inferior vena caval filters. Radiology 174 (3 Pt 2):987-992.
_______. 1992. Evaluating the roles and function of vena caval filters: Will data be available before or after these devices are removed from the market? Radiology 185(1):15-17.
ERG (Eastern Research Group). 2006. Unique identification for medical devices. Prepared for FDA by Eastern Research Group, Inc.
FDA (Food and Drug Administration). 1999a. Guidance for cardiovascular intravascular filter 510(k) submissions. http://www.fda.gov/downloads/MedicalDevices/DeviceRegulationandGuidance/GuidanceDocuments/ucm073777.pdf (accessed 03/14/2011).
_______. 1999b. Managing the risks from medical product use: Creating a risk management framework. Silver Spring, MD: Food and Drug Administration.
_______. 2009a. About Medsun. http://www.fda.gov/MedicalDevices/Safety/MedSunMedicalProductSafetyNetwork/ucm112683.htm (accessed 02/3/2011).
_______. 2009b. How to report a problem (medical devices). http://www.fda.gov/MedicalDevices/Safety/ReportaProblem/default.htm (accessed 02/10/2011).
_______. 2009c. Medsun: Medical Product Safety Network. http://www.fda.gov/MedicalDevices/Safety/MedSunMedicalProductSafetyNetwork/default.htm (accessed 02/03/2011).
_______. 2010a. CDRH preliminary internal evaluations—Volume II: Task force on the utilization of science in regulatory decision making preliminary report and recommendations. Silver Spring, MD: Food and Drug Administration.
_______. 2010b. CDRH transparency website feedback summary. http://www.fda.gov/AboutFDA/CentersOffices/CDRH/CDRHTransparency/ucm214646.htm (accessed 02/03/2011).
_______. 2010c. Center for Devices and Radiological Health CDRH FY 2010 strategic priorities. Silver Spring, MD: Food and Drug Administration.
_______. 2010d. FDA awards contract to Harvard Pilgrim to develop pilot for safety monitoring system. http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm196968.htm (accessed 02/03/2011).
_______. 2010e. Inferior vena cava (IVC) filters: Initial communication: Risk of adverse events with long term use. http://www.fda.gov/Safety/MedWatch/SafetyInformation/SafetyAlertsforHumanMedicalProducts/ucm221707.htm (accessed August 30, 2010).
_______. 2010f. MDUFMA FDA performance report: Agenda and materials from January 26, 2010 FDA performance report. http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/Overview/MedicalDeviceUserFeeandModernizationActMDUFMA/ucm207964.htm (accessed 03/8/2011).
_______. 2010g. Removing retrievable inferior vena cava filters: Initial communication. http://www.fda.gov/MedicalDevices/Safety/AlertsandNotices/ucm221676.htm (accessed 01/6/2011).
_______. 2010h. Results of FDA pilot activities to explore opportunities and challenges with the implementation of a unique device identifier system. http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/UniqueDeviceIdentifiers/ucm235962.htm (accessed 02/03/2011).
_______. 2011a. FDA’s Sentinel Initiative. http://www.fda.gov/Safety/FDAsSentinelInitiative/default.htm (accessed 02/03/2011).
_______. 2011b. FDA-track CDRH Office of Surveillance and Biometrics (OSB) dashboard. http://www.fda.gov/AboutFDA/Transparency/track/ucm203271.htm (accessed 02/03/2011).
_______. 2011c. FDA transparency initiative. http://www.fda.gov/AboutFDA/Transparency/TransparencytoRegulatedIndustry/PhaseIIITransparencyReport/FDATransparencyInitiative/default.htm (accessed 02/03/2011).
_______. 2011d. List of device recalls. http://www.fda.gov/MedicalDevices/Safety/RecallsCorrectionsRemovals/ListofRecalls/default.htm (accessed 02/03/2011).
_______. 2011e. Recalls, market withdrawals, and safety alerts. http://www.fda.gov/Safety/Recalls/default.htm (accessed 02/03/2011).
GAO (Government Accountability Office). 1989. FDA resources: Comprehensive assessment of staffing, facilities, and equipment needed (HRD-89-142). Washington, DC: General Accounting Office.
_______. 1998. Medical devices: FDA can improve oversight of tracking and recall systems (HEHS-98-211). Washington, DC: General Accounting Office.
_______. 2007. Medical devices: Status of FDA’s program for inspections by accredited organizations (GAO-07-157). Washington, DC: Government Accountability Office.
_______. 2009a. FDA faces challenges meeting its growing medical product responsibilities and should develop complete estimates of its resource needs (GAO-09-581). Washington, DC: Government Accountability Office.
_______. 2009b. Medical devices: Shortcomings in FDA’s premarket review, postmarket surveillance, and inspections of device manufacturing establishments (GAO-09-370T). Washington, DC: Government Accountability Office.
Greenfield, L. J. 2007. Perspective from an academic on postmarket surveillance. In Medical device epidemiology and surveillance, edited by S. L. Brown, R. A. Bright and D. R. Tavris. Hoboken, NJ: John Wiley & Sons, Ltd. Pp. 159-169.
Gross, T. P., and L. G. Kessler. 1996. Medical device vigilance at FDA. Studies in Health Technology and Informatics 28:17-24.
HCUP (Healthcare Cost and Utilization Project). 2010. Overview of the nationwide inpatient sample (NIS). http://www.hcup-us.ahrq.gov/nisoverview.jsp (accessed 07/2010).
IOM (Institute of Medicine). 2005. Safe medical devices for children. Washington, DC: The National Academies Press.
_______. 2011. Public health effectiveness of the FDA 510(k) clearance process: Measuring postmarket performance and other select topics. Washington DC: The National Academies Press.
Jha, A. K., C. M. DesRoches, E. G. Campbell, K. Donelan, S. R. Rao, T. G. Ferris, A. Shields, S. Rosenbaum, and D. Blumenthal. 2009. Use of electronic health records in U.S. hospitals. New England Journal of Medicine 360(16):1628-1638.
Kessler, L. 2010. Letter to committee on the public health effectiveness of the FDA 510(k) clearance process regarding 7/28/2010 IOM meeting, Seattle, WA, 7/29/2010.
Lenzer, J. 2009. Watching over the medical device industry. BMJ 338:b2321.
Lenzer, J., and S. Brownlee. 2010. Why the FDA can’t protect the public. BMJ 341:c4753.
Mantone, J. 2006. Devising new tactics. Providers and medical device manufacturers struggle to find best ways to work together on recalls, advisories. Modern Healthcare 36(19):24-26.
McCowan, T. C., E. J. Ferris, D. K. Carver, and W. M. Molpus. 1992. Complications of the nitinol vena caval filter. Journal of Vascular and Interventional Radiology 3(2):401-408.
Mehran, R., M. B. Leon, D. A. Feigal, D. Jefferys, M. Simons, N. Chronos, T. J. Fogarty, R. E. Kuntz, D. S. Baim, and A. V. Kaplan. 2004. Post-market approval surveillance: A call for a more integrated and comprehensive approach. Circulation 109(25):3073-3077.
Mini-Sentinel Coordinating Center. 2010. Mini-sentinel: Overview and description of the common data model v1.1. http://www.mini-sentinel.org/work_products/Data_Activities/Mini-Sentinel_CommonDataModel_v1.1.pdf (accessed 02/03/2011).
Nicholson, W., W. J. Nicholson, P. Tolerico, B. Taylor, S. Solomon, T. Schryver, K. McCullum, H. Goldberg, J. Mills, B. Schuler, L. Shears, L. Siddoway, N. Agarwal, and C. Tuohy. 2010. Prevalence of fracture and fragment embolization of bard retrievable vena cava filters and clinical implications including cardiac perforation and tamponade. Archives of Internal Medicine 170(20):1827-1831.
Normand, S. L., D. Marinac-Dabic, A. Sedrakyan, and R. Kaczmarek. 2010. Rethinking analytical strategies for surveillance of medical devices: The case of hip arthroplasty. Medical Care 48(6 Suppl):S58-67.
OIG (Office of the Inspector General). 1990. Internal control weaknesses in the Food and Drug Administrations medical device 510(k) review process (A-15-89-00065). Washington, DC: Department of Health and Human Services.
_______. 2009. Adverse event reporting for medical devices. OEI-01-08-00110. Washington, DC: Department of Health and Human Services.
Outcome Sciences. 2009. Evaluation of potential data sources for a national network of orthopedic device implant registries: A report to the U.S. Food and Drug Administration. Cambridge, MA: Outcome Sciences.
Patterson, P. 2002. Bronchoscope recall falls through cracks. OR Manager 18(4):20.
RASMAS. 2011. About us: RASMAS. http://info.rasmas.noblis.org/?page_id=2 (accessed 06/ 15/2011).
Rosati, K., B. Evans, and D. McGraw. 2010. HIPAA and common rule compliance in the mini-sentinel pilot. Mini-Sentinel Privacy Panel.
Schultz, D. 2007. Medical device safety: FDA’s postmarket transformation initiative. Food and Drug Law Journal 62(3):593-596.
Shuren, J. 2010. Additional information regarding medical device recall studies: Follow-up to July 28 IOM workshop. Silver Spring, MD, 09/17/10.
Torrence, M. E. 2002. Data sources: Use in the epidemiologic study of medical devices. Epidemiology 13(3 Suppl):S10-14.
Williams, E. D. 2010. The medical device approval process and related legislative issues (RL32826). Congressional Research Service. 04/21/10.
Zuckerman, D. M., P. Brown, and S. E. Nissen. 2011. Medical device recalls and the FDA approval process. Archives of Internal Medicine 171(11):1006-1011.