|The third edition of the Registries for Evaluating Patient Outcomes: A User’s Guide includes a new chapter dedicated to medical device registries.|
Patient registries can help bridge the knowledge gap between how medical devices perform in the optimal situations studied in clinical trials and how they perform in routine practice over time. The third edition of the Registries for Evaluating Patient Outcomes: A User’s Guide includes a new, stand-alone chapter dedicated to medical device registries. Published in 2014 by the Agency for Patient Research and Quality, the guide is a widely used resource for the design and analysis of patient registries, or repositories of information collected outside of a clinical trial setting for a defined scientific purpose.
Whether the goal is to fulfill a postmarket safety commitment for a regulatory body, perform a comparative effectiveness assessment, or learn how medical devices work in certain subgroups of product users, medical device professionals can and should leverage registries to provide clinically meaningful insights into how medical devices perform in a real-world setting. Since registries often include information on large numbers of patients treated in diverse clinical settings, they enable users to understand patients’ experience with medical devices over time—an invaluable insight for device developers, clinicians, patients, payers, and regulators.
While it might be tempting to approach the construct and use of a medical device registry as one would a drug registry, there are a number of special considerations that merit consideration. This article reviews these considerations, including challenges facing the development of medical device registries and potential solutions. It also discusses future opportunities for such registries. When approached properly, registries can be powerful tools that offer a wealth of product knowledge, enabling manufacturers to enhance medical device development.
Unique Challenges Facing Medical Device Registries
Registries need to be designed with special attention to the products of interest and to the generalizations that users would like to make. For example, do you want to draw conclusions about the effectiveness of a medical device, perhaps based on whether it was implanted by an amateur or an expert? Are you interested in near-term performance, prompting you to ask, “Can I hear better with this device or another device?” Or do you want to demonstrate how well a device performs over time, causing you to ask, “Will I still be able to rely on this knee replacement in 20 years?” Do you want to know about all patients, or are you particularly interested in a subgroup, such as patients that received a hip revision?
Once the professional has decided which target population to study and in which situations—in primary care or urgent care facilities, for example—the next step is to focus on the medical device itself. At this point, it will quickly become apparent that the approach to understanding the benefits and risks of a given medical device differs substantially from researching a typical prescription drug. While drugs are rarely revised during their patent life once FDA has approved them, medical device technologies are modified frequently to incorporate new materials and design modifications.
Therefore, it is not sufficient to simply record that a patient received a device with a particular name because without knowledge of the version, the manufacturing site, the lot, and other information, it can be difficult to draw reliable conclusions about the device’s performance. As if that weren’t challenging enough, it is important to understand that device performance is also affected by the skill and experience of the person who implanted the device. Such ancillary factors as the setting in which the device was implanted may also play a role in its safety and effectiveness.
To help overcome these challenges, registries include unique medical device identifiers, data on device modifications and components, and information on the operator and institution at which the device was implanted, since a small rural hospital will produce different results from a tertiary care facility in a large city. In addition, registries provide information on ancillary technology and therapies such as drug exposures, information on performance issues, follow-up information, and information on the healthcare provider’s experience and training.
Unique Device Identifiers. Although unique device identifiers (UDIs) are available for some medical devices in the form of global trade item numbers or Health Industry Bar Code identifiers, they are not routinely captured in observational data sources such as health insurance claims data or most medical records. However, such sources are often the first choice for registries. If a specific device of interest cannot be identified in a registry, how is it possible to rely on evaluations that are derived from combining data on many different devices? When devices lack an identifier, their risks or benefits can be attributed incorrectly, leading to the widespread use of faulty products or false suspicions toward safe products.
Recognizing the lack of UDIs as inhibiting registry research, FDA has issued a draft rule requiring that manufacturers establish UDIs for their products. While this requirement will be phased in over several years, several approaches can be used in the meantime to capture device identity in the absence of a UDI.
One approach is to use such identifiers as catalogue, model, serial, and lot numbers that are unique to a particular manufacturer’s device. While these identifiers are not standardized and several components from different manufacturers may have similar catalogue or model numbers, these numbers—taken in aggregate—can be used to identify and track medical devices. Resources such as the Society of Thoracic Surgeons Adult Cardiac Database are taking this ability one step further by providing their membership with an exhaustive list of heart valve devices, enabling registry contributors to provide information that could be relevant to understanding device performance and helping to guide practice.
Capturing Device Performance. It can be difficult to capture all potential problems, failure modes, and adverse events in a single device registry because these performance issues may be related to software, hardware, biomaterials, sterility, or a variety of other issues. At the same time, these issues can also have a variety of root causes, some of which may not be immediately evident, such as a breast implant rupture. It is of great importance in registry design to carefully consider anticipated categories of device performance issues.
Automated surveillance reporting from a registry can be carried out periodically to identify clinically evident performance issues. Surveillance, however, is a complex endeavor, and standardized data elements and consistent and timely reporting practices are required. Larger datasets, such as those from multiple institutions or registries, are more informative than analyses based on smaller groups of patients and physicians. A good example of a successful registry implementation for performing surveillance is the Data Extraction and Longitudinal Trend Analysis (DELTA) network study for cardiovascular devices.
Differing Device Systems and Components. In many cases, a device of interest for a registry is either part of a larger system of devices or one that contains multiple components that are themselves considered devices. Issues associated with the lack of unique identifiers are compounded by the challenge of determining which component is responsible for a performance issue. Because different components in the same system have varying lifecycles or expiration dates, they may have to be replaced at different rates. And with the increasing availability of device/drug combination devices over the past decade, it has become challenging to determine whether the technology’s benefits and risks are more likely attributable to the drug or the device.
Registries can address component differences by collecting data on the device as a whole and then attempting to obtain information about each component, as warranted. Eventually, UDIs will capture component expiration dates, eliminating this challenge. For combination devices, special considerations in registry design will include recording drug dosages separately and flagging possible interactions with other drugs.
It is also important to bear in mind that some implantable devices can only be inserted in the body with the assistance of such procedural or ancillary devices as imaging equipment. In such cases, it is also desirable to obtain information about these devices. For example, in hernia repair procedures, it may be necessary to collect information on the method of mesh adhesion—whether staples, glue, or sutures—since these adhesives could interact with the mesh and affect device performance. Researchers should consider how such interactions can be captured in the data-collection process.
Obtaining Sufficient Follow-Up Information. While many implantable medical devices are supposed to work for decades, they are studied during relatively short periods of use prior to marketing. This discrepancy presents a unique challenge to device researchers because once a device is implanted, the patient may not return to the doctor unless an issue arises. In contrast, patients taking medications must keep in touch with their physicians, if only to obtain periodic refills. Thus, how do medical device manufacturers obtain data about their products?
A long-term follow-up registry maintained by a clinical practice can answer questions concerning safety and effectiveness issues over the medical device’s full product lifecycle. Long-term data can be obtained directly from patients or healthcare providers. Alternatively, it can be obtained by linking registries with electronic healthcare records in integrated delivery networks or other data sources that independently collect clinically relevant data on patients for other reasons. Administrative billing data can also be useful, although this information is imprecise and may be more reflective of billing practices than the diagnostic or curative procedures themselves.
Provider Experience and Training. The impact that the provider has on device performance is commonly underestimated. Surgeons generally prefer some medical devices and surgical techniques to others, and although this is sometimes based on clinical appropriateness, it can also be based on marketing pressures or familiarity with a particular brand. Group purchasing organizations, costs, provider contracts, reimbursements, and other market forces may also influence selection. Moreover, surgeons may not have a choice of devices in some cases. A provider’s experience and surgical and procedural skill levels can also influence a medical device’s safety and effectiveness.
Because of the impact that the healthcare provider can have on device performance, it may be important for a registry to collect individual provider identifiers. Device-specific training is also important. Experience-oriented factors can also figure into analyses and training evaluations. These factors can include the amount of time that has elapsed since the surgeon received training and how the training was provided; practitioner annual volume; practitioner lifetime volume; facility volume; and such facility characteristics as academic teaching status.
Although device registries are operationally similar to other registries, it is critical to consider the specific challenges they face during the registry design and analysis phases. Careful review of the unique features of medical device registries can result in useful, high-quality studies of device performance.
Automated Data: The Future of Device Registries?
Technology advances continue to reshape the healthcare industry’s approach to the development of biopharmaceuticals and medical devices. The registries for these technologies will be similarly transformed. New technologies in development can enable medical devices to transmit data directly to electronic medical records and other patient management systems. Ultimately, these types of data may be sent directly to patient registries, ensuring that they remain up to date and reducing the burden of data entry.
An exciting example of this potential is the use of automatic measurement capability and programming adjustments to provide optimal settings for pacemakers and other implantable devices. The feasibility of this innovative technology has been demonstrated by the Automaticity registry, which aims to evaluate physicians' acceptance of automatic algorithms to capture ventricular activity, perform automatic sensing, and automatically optimize sensor settings. The Automaticity team concluded that automated programming can allow doctors to avoid reprogramming steps and increase the effective use of hospital time and resources. This technology would give registries a significant boost because it enables all automated changes to be collected at one time rather as they occur.
Diagnostics for implantable devices are another area of technical improvement. Such implantable devices as pacemakers and cardiac resynchronization therapy devices can track heart rate, heart rate variability, respiration rate, atrial tachyarrhythmia, ventricular tachyarrhythmia recurrence and duration, intrathoracic impedance, symptom markers, and patient activity. This diagnostic information can be provided directly to electronic medical records and fed into registries—in many cases continuously. Although the clinical application of this capability is still being examined, its ability to provide efficient, timely data capture is clear.
The use of registries—device and otherwise—is experiencing a well-deserved resurgence as sponsors realize that these tools save them both time and money in getting effective products to appropriate patients quickly. Medical device registries can provide useful information about the safety as well as the short- and long-term effectiveness of products under real-world conditions. They can also provide information about the effects of such factors as surgical technique, training, and hospital and patient characteristics. Like all studies, medical device registries have some limitations, but careful consideration of the unique nature of medical devices in designing, analyzing, and interpreting registries will help manufacturers extract maximum value from their devices.
Nancy Dreyer is chief of scientific affairs and senior vice president at Durham, NC–based Quintiles Outcome. With more than 25 years of experience in epidemiological research, she leads a team of researchers who design, conduct, and interpret observational research on comparative effectiveness, safety, and quality-improvement programs. As CEO of Epidemiology Resources Inc., she launched and published the peer-reviewed journal Epidemiology and led the New England Epidemiology Institute, a summer program that trained more than 5000 scientists through collaboration with international medical schools and schools of public health. Dreyer is a Fellow and board member of the International Society of Pharmacoepidemiology. She received a Master’s degree and a PhD in epidemiology at the University of North Carolina in Chapel Hill. Reach her at firstname.lastname@example.org.