Implant registries as clinical evidence for devices
Implant registries collect standardised data on patients treated with implantable devices, and the EU Medical Device Regulation expects manufacturers and notified bodies to consider those data. How far a clinical evaluation can rest on them depends on what an implant registry holds, whether the right one can be found, whether its data are suitable for a regulatory submission, and the limits of registry data as clinical evidence.
Why registries entered clinical evaluation
Pressure to make better use of registry data for device surveillance grew as serious concerns about device safety came to light. In 2012, serious concerns about non-medical-grade silicone in Poly Implant Prothèse breast implants led the European Commission and EU countries to a Joint Plan for Immediate Actions under the existing device legislation, and one of its five actions was to support implant registries able to identify safety issues and monitor safety and performance over the long term. Later safety problems with metal-on-metal hip implants and vaginal mesh increased the pressure for closer oversight and surveillance of devices, including calls for compulsory registration of all implantable devices. Randomised controlled trials are considered the gold standard for evaluating most medical treatments, but they are not always feasible or ethical to carry out on medical devices. Registries are increasingly seen as a supplement to trial data, and in some cases they may be the only feasible approach to evaluating the long-term safety of some implantable devices.
The EU Medical Device Regulation, Regulation (EU) 2017/745, is the first regulation to require registry data to be evaluated. Its Article 108 encourages registries and registry networks built on common principles so that comparable long-term safety and performance data can be collected, and suggests that registries contribute to the traceability of implantable devices. Annex III (1.1(a)) requires manufacturers' post-market surveillance plans to consider relevant databases and registries, and Annex VII (4.11(h)) requires notified bodies to consider registry data for recertification. Manufacturers must continuously evaluate all clinical data available for their devices marketed in Europe, and with the regulation in force from May 2021 they are required to assess available implant registry data as part of clinical evaluation. This requirement will need closer collaboration between industry and registries to evaluate the safety and performance of high-risk devices.
A growing focus on the role and value of registries has led to steps to integrate registry data better into regulatory decision making. That effort needs the collaboration and input of every registry stakeholder: patients, healthcare providers, professional societies, registry custodians, researchers, reimbursement bodies, public health and regulatory bodies, and the device industry. Registries used to support regulatory requirements need to be well designed to produce valid data, and medical writers will play an important role in communicating the clinical evidence on devices that registry data generate. Medical writers should be aware of existing implant registries, understand what characteristics make a registry suitable to support regulatory requirements, and recognise both the value and the limitations of registries as a source of clinical evidence.
What an implant registry holds
The International Medical Device Regulators Forum defines a device registry as "an organized system with a primary aim to increase the knowledge on medical devices contributing to improve the quality of patient care that continuously collects relevant data, evaluates meaningful outcomes and comprehensively covers the population defined by exposure to particular device(s) at a reasonably generalizable scale (e.g., international, national, regional, and health system)". Many implant registries are led by professional medical societies or consortia. Implant registries collect detailed information about patient characteristics and clinical outcomes as well as about implants.
Registries can supply data at the level of the patient, the implant, the healthcare provider and the clinic, all of which can influence clinical outcomes. Typical data are demographics such as age, sex and comorbidities, procedure details such as the type of surgery, approach and duration, clinical data such as indication, diagnosis and previous interventions, follow-up, adverse events, and implant details such as the device identifier within the unique device identification system and the device's characteristics. More variables are not necessarily better, and the overall burden on data providers needs to be considered. Registries that collect fewer variables through a quick, easy procedure may better keep data providers motivated and data quality higher.
When an implant registry is assessed as a source of clinical evidence, harmonised implant categorisation, patient-reported outcomes and the registry's ability to link to other data sources are especially important to consider.
Registry data collection once focused on treatment procedures and outcomes in clinical files and patient records, and as implant use grew the focus shifted to device-related outcomes. To compare similar implants meaningfully, registries characterise each implant by type, size, shape, material, coating and other important attributes. Harmonised classification systems for orthopaedic implants are an important step towards comparing registries, and examples are the implant library developed by the German arthroplasty registry and adopted by the UK's National Joint Registry, and the International Society of Arthroplasty Registries' International Prosthesis Library. In arthroplasty, a standard implant classification for analysing revision rates and implant survival is a prerequisite for a registry to serve as an early warning system for implant failure.
Many registries are incorporating patient-reported outcome measures and recognise that these complement the clinical outcomes.
Data collection for registries adds to the administrative workload of health workers, and registries should be designed to collect only the minimal information needed. Registries that capture identifying patient information, where law and informed consent allow, enable linkage to external datasets and reduce the burden on data providers. For example, a patient's vital status (dead, alive, emigrated or unknown) is crucial for accurate revision rates in arthroplasty but is generally absent from registry data, and linkage to routinely collected administrative data such as mortality data can fill the gap. Linkage to other types of dataset enriches registry data and can facilitate analyses such as cost-effectiveness.
Finding a registry
Keeping a complete, current overview of every operating implant registry is hard. Well-established national registries, such as the Swedish Knee Arthroplasty Register since 1975, the UK National Joint Registry since 2002 and the Swiss National Joint Registry (SIRIS) since 2012, are generally easy to find and will have the most valuable data in quantity and quality, while smaller and newer registries may be found through professional medical societies.
Systematic reviews have mapped the implant registry landscape in Europe. In 2013 researchers identified 101 implant registries in Europe, most of them in cardiology (38) and orthopaedics and arthroplasty (29), and a review in 2017 identified 24 hip and knee replacement registries in Europe. Registries for other devices were less common: pacemakers and heart stents, breast implants, cochlear implants, insulin pumps, tubes, other stents, ophthalmological devices, brain stimulation and shunts, sacral neuromodulation, drug depots and dental implants.
A registry of patient registries is another route to the right one. The European Medicines Agency's inventory of registries, maintained by ENCePP (the European Network of Centres for Pharmacoepidemiology and Pharmacovigilance), and the cross-border PARENT (Patient Registries Initiative) list of European registries both aim to increase transparency, avoid duplication and promote collaboration among registries.
Suitability for regulatory submissions
The availability of relevant data is only one aspect of a registry's suitability to provide clinical evidence. The International Medical Device Regulators Forum's registry working group defined fifteen registry requirements, grouped into six elements, for assessing whether registry data are suitable for regulatory submissions:
- Governance: a transparent governance structure and processes.
- Quality management system: legal requirements for data collection and handling met; information on patient data protection; a policy on access to data; and the essential information available for verification, for example by a competent authority or notified body.
- Data gathering: relevant variables; unambiguous device identification, as through the unique device identification system; the ability to link with other data sources; controlled vocabularies; and a harmonised minimum data model.
- Data storage: security against hacking, alteration, deletion or theft of data.
- Methodology and data analysis: analyses across different types of analysis framework, and interpretation of the data.
- Transparency, display and distribution: publicly available reports, and their frequency and content; and a publicly accessible website with web reporting.
The weight of each element depends on the intended use of the data. For example, controlled vocabularies are recommended for post-market surveillance but highly recommended for data meant to support an initial approval or an extension of indication. Completeness of data collection, transparent quality assurance, a clear policy for data access and sharing, and the registry's sustainability also merit consideration.
Analysing registry data
Registry data with sufficiently detailed implant data collection support analyses such as those of SIRIS. SIRIS began registering hip and knee implants in September 2012 and by 2020 was the largest implant registry in Switzerland, with data collection supported by 186 healthcare institutions. Participation is compulsory for every hospital and clinic performing knee and hip arthroplasty, and by its most recent coverage estimates as of 2020 the registry included 90% to 92% of those procedures in Switzerland. Its 2019 annual report compared implant types and brands for the first time, using an implant library built from industry partners' product catalogues.
In arthroplasty the main outcome is the revision rate, a revision being the replacement of a patient's primary hip or knee implant with new components. One analysis is a funnel plot of two-year revision rates for each participating health service unit against its volume of operations, which identifies clinics whose rates lie outside the variability expected by chance. Another analysis, of specific product brands, can identify "outlier" devices or device combinations with a higher revision rate than expected compared with similar benchmark devices. The SIRIS example of this analysis compares two-year revision rates of uncemented stem-and-cup combinations used in primary total hip arthroplasty, with rates adjusted for the effects of mortality and emigration. Because registries use observational methods, many factors could contribute to an outlier status, such as patient selection, case-mix, surgical technique, surgeon experience and the characteristics of the health service. This benchmark analysis provides an initial alert that starts a more in-depth analysis of the underlying cause of a poor outcome.
Limits of registry data
Case coverage, completeness and data quality are concerns for every type of registry. Collecting high-quality device data is a particular challenge for implant registries and for disease registries that try to collect implant data. When registry data are used to answer questions about the quality of healthcare treatments or the safety of devices, careful evaluation of potential sources of bias is paramount.
Detailed knowledge of a registry's coverage is vital to calculating many outcome measures. If an implant revision is performed in a clinic that does not record the operation in the registry, for example, the revision rate is underestimated. Completeness of data is another condition for unbiased analysis, and recording every implant needs smart implant interfaces. The type of operation, such as total hip arthroplasty, defines the expected type and number of implants, which can be tracked during scanning, and warnings and error messages help make sure all expected implants for each case are captured.
Many registries do not collect data in enough detail for the sophisticated analyses of specific implants that manufacturers need to fulfil clinical evidence requirements. Some registries may collect data on general device types, such as plates, screws and external fixators, but not the details that identify a specific brand, model or reference number. Another challenge is a standard categorisation of implants, so that data can be compared across registries. International congresses and meetings, such as the International Society of Arthroplasty Registries conferences, have advanced the discussion of standardising and harmonising implant libraries, which led the UK and German registries to agree to harmonise their library definitions. Keeping these libraries current and accurate needs commitment from industry, with manufacturers classifying existing and new products to a standard system at a granularity that allows meaningful analysis.
Standardised implant libraries will reduce the administrative burden on manufacturers, who supply implant catalogues in different categories and granularity to registries in many countries. Ideally, registries may update their libraries from comprehensive implant data warehouses such as EUDAMED, the European database on medical devices, or other international databases. Local legal regulations that leave products sold in some countries and not others, and the difficulty of standardisation, hamper the development of international implant registries.