A joint replacement implant may stay in the patient for decades, while the initial clinical investigation captures only part of its performance lifecycle. Post-market clinical follow-up has to address what remains open after market access: long-term fixation, revision, survivorship, rare complications, broader patient groups and outcomes under routine care.
CMO · Orthopedics & Spine
NHS orthopedic surgeon
in orthopedic surgery & Class III implant evidence
Chief Operating Officer · former TÜV SÜD Senior Reviewer · Senior Consultant Surgeon
in device evaluation & Notified Body review
Chief Data Officer · Biometrics & Data Systems
in clinical data management, EDC & biometrics
This page is for teams choosing between a survey, a cohort, a prospective study and a registry analysis, and needing to justify that choice.
Follow-up stopped long before the implant will.
Additional post-market data have been requested explicitly.
The residual risk sits in the long tail.
Routine practice now exceeds the studied population.
No device-specific clinical data are being generated.
Category-level data cannot substantiate your implant.
No single source answers the whole question.
Existing records may or may not be fit for purpose.
The design and endpoints need to be settled first.
The method does not match the uncertainty.
The activity has to end in a regulatory conclusion.
A shared question with three access routes.
→ Clinical evidence for arthroplasty medical devices, the parent hub, where the full evidence architecture is set out.
Under EU MDR, PMCF is a continuous process that updates the clinical evaluation. The PMCF plan defines the methods used to address the remaining clinical questions, and the results are assessed in the PMCF Evaluation Report and fed back into the wider post-market system. That sequence only works if the uncertainty is named first.
MDCG 2020-7 addresses the PMCF plan and MDCG 2020-8 the PMCF Evaluation Report template. Our PMCF services and orthopedic PMCF strategy pages cover the general framework; this page covers the arthroplasty-specific decisions.
Key message. A robust PMCF program may combine several methods, because no single source is likely to answer every short-term and long-term question.
Each answers a different class of question.
Device handling, clinical usage patterns, surgeon experience, known low-frequency issues and confirmation of routine use.
Existing treated populations, medium-term outcomes, revision and complication data, and feasibility assessment before a prospective study.
Defined clinical outcomes, PROMs, imaging, consistent follow-up, and device-specific safety and performance.
Large populations, revision and survivorship, rare outcomes, and performance across routine care.
Fixation, migration, radiolucency, loosening and device-specific radiographic uncertainty.
Potential retrospective sources include hospital electronic records, theatre records, implant logs, revision databases, radiographs, local PROM systems, discharge records, follow-up clinic records, national or regional datasets, and manufacturer-supported previous studies.
A large retrospective dataset is not automatically stronger than a smaller prospective study if device identity, revision completeness or outcome definitions are unreliable.
The distinction determines what the evidence can actually support.
Public registry evidence informs benchmarks and study design. PMCF value: context and state of the art.
A separate study using compatible variables and outcome definitions. PMCF value: future comparability.
The analysis is performed within the registry dataset. PMCF value: large-scale real-world outcomes.
Study data are connected to registry or claims data. PMCF value: enhanced follow-up and revision capture.
Prospective study processes are incorporated into registry infrastructure. PMCF value: efficient long-term evidence where feasible.
Key message. Published registry results may support context and benchmarking, but they do not automatically demonstrate the safety and performance of the manufacturer’s own device. Access to a national registry, or the existence of published registry evidence, does not by itself provide device-specific PMCF evidence. The terminology matters too: « registry study » should not be used to describe a literature review of a registry annual report, and an analysis cannot become device-specific simply by naming a brand after the data have already been grouped at category level.
See our general pages on device registries under EU MDR and PMCF and real-world evidence.
A DACH plan is not three unrelated country activities, and it is not one activity copied three times. The question can be shared; the data-access route cannot.
Registry access, data-use permissions and device representation are confirmed before any plan is allowed to depend on them.
The German Arthroplasty Registry collects hip and knee information and systematically links procedural and implant data with routine healthcare data. It describes survival analysis as central to evaluating arthroplasty outcomes and currently reports follow-up extending to approximately ten years for eligible data. Participation is voluntary, and the methodology acknowledges differences in participation by hospital volume and limits in the proportion of data suitable for linked survival analysis.
The Swiss national implant-registry system. Current public materials cover hip, knee, shoulder and spine programs, with procedure, follow-up and PROM documentation for the relevant modules, including primary and revision forms and patient-reported measures such as OHS, OKS and OSS.
Austria should not be treated as a registry copy of Germany or Switzerland. Options may include prospective multicenter PMCF studies, retrospective hospital cohorts, institution-level implant databases, local quality datasets, PROM programs, imaging follow-up, routine-care data where accessible, and registry sources only where their current status, coverage and access are confirmed.
Important limitation. EPRD- and SIRIS-based analyses may be considered where the device, sample size, access route and data permissions allow. Neither is a guaranteed source of manufacturer-specific PMCF evidence, and neither is automatically available to every manufacturer.
Germany, Austria and Switzerland do not offer the same post-market data. The route is chosen from the residual uncertainty, not from whichever dataset is easiest to obtain.
A survivorship analysis depends on definitions that are easy to leave implicit: the index arthroplasty, the start of follow-up, the revision endpoint, censoring, death, loss to follow-up, bilateral procedures, component-level versus construct-level analysis, all-cause versus cause-specific revision, the follow-up distribution and the completeness of revision capture.
Key message. A reported survival percentage is difficult to interpret without its follow-up duration, confidence interval, endpoint definition and data-completeness assessment. Revision, reoperation and complication rates should never be used interchangeably.
A practical asymmetry worth planning around. A registry may identify the implant construct more reliably than routine records, while clinical notes may give more detail about the reason for revision. The optimal analysis therefore often needs more than one source.
Device identification and traceability. Registry evidence depends on being able to identify the manufacturer, product family, component, catalog number, size, lot or serial number where relevant, fixation method, bearing combination, modular construct, laterality, primary or revision use, and the UDI where available. The recurring problems: legacy product names, multiple catalog-number formats, product-family pooling, mixed-manufacturer constructs, unrecorded component exchange, incomplete revision records, changes in branding and merged device generations.
The instrument logic is covered on arthroplasty endpoints and PROMs; acquisition, core-lab reading and RSA on imaging, core lab and RSA for arthroplasty studies.
Key message. Registry and imaging data should be connected only where patient, implant, timepoint and outcome linkage are reliable.
Where a survey is used, quality depends on a clear respondent population, a defined denominator, device-specific experience, a stated recall period, avoidance of leading questions, duplicate-response controls, defined handling of incomplete responses, a prespecified analysis and transparent limitations. See PMCF surveys for medical devices.
Key message. A survey should not be selected simply because it is faster than a clinical study.
Revision and survivorship at scale, across routine care.
Pain, function and patient-perceived benefit.
Fixation, migration and loosening: the mechanism behind a signal.
Handling, usage patterns and surgeon experience.
Safety signals arising outside the study.
State of the art and comparator context.
Key message. Each stream carries a defined role, and together they feed one integrated PMCF conclusion. The same outcome should not be counted twice merely because it appears in more than one dataset.
Key message. Large numbers do not remove confounding.
Any dataset proposed for PMCF should be assessed for relevance, representativeness, completeness, accuracy, consistency, traceability, timeliness, device granularity, endpoint validity, follow-up completeness, revision completeness, missingness, data provenance and quality-control processes.
Missing data are not all the same. Distinguish a patient lost to clinical follow-up, no PROM response, no imaging, a revision performed at another center, death, emigration, hospital non-participation, a missing component identifier, an incomplete revision cause and a record-linkage failure. Then answer: is missingness related to poor outcome? are revision events captured independently? can routine data reduce loss to follow-up? does death compete with revision? can missing device identity be resolved? which sensitivity analyses are required?
Two key messages. Kaplan-Meier and competing-risk estimates answer related but different questions where death prevents later revision. And a threshold should trigger review, not an automatic conclusion that the device has failed. Analysis is produced by our biostatistics team against a prespecified SAP.
EPRD feasibility, hospital and surgeon selection, registry or routine-data permissions, prospective PMCF sites where registry data are insufficient, German-language documentation, and data-protection and contracting requirements.
Site-led prospective or retrospective evidence, confirmation of the available institutional or national data sources, ethics and data-use requirements, German-language materials, and multicenter feasibility.
SIRIS feasibility, Swiss data-access and consent requirements, PROM availability, site and registry coordination, local regulatory and data-protection considerations, and multilingual documentation where required.
Delivered with site feasibility and selection, study start-up, on-site and remote monitoring, EDC and data management and our operational infrastructure.
Registry work, prospective follow-up, imaging and vigilance only become evidence once a single report explains which uncertainty each of them addressed.
The report should explain which residual uncertainty the activity addressed, why the chosen method was proportionate, which data were collected, the data quality and limitations, the population and device applicability, the safety and performance findings, the revision and survivorship outcomes, the PROM and imaging outcomes, the comparison with expected performance, any new or changed risks, the remaining uncertainty and the required next actions.
PMCF results rarely stop at the CER. They also affect the PMS report, the PSUR, the risk-management file, the SSCP, the IFU, device claims, training materials, CAPA, trend reporting, vigilance activities and the next PMCF plan. The 2025 MDCG post-market surveillance guidance reinforces that these have to operate as a connected system rather than as isolated documents.
Key message. The CER should not merely state that PMCF was completed. It should explain how the findings changed, confirmed or limited the clinical conclusions, a recurring theme among MDR clinical evaluation non-conformities.
Unclear residual evidence gap. Output: prioritized PMCF roadmap.
Considering EPRD, SIRIS or another registry. Output: access, variable and sample-size assessment.
Existing site or hospital datasets. Output: fitness-for-purpose report.
A new observational study. Output: synopsis, protocol and endpoint framework.
The registry lacks clinical depth. Output: linked PROM or imaging work package.
A focused usage or handling question. Output: survey, sampling and analysis plan.
Completed activities. Output: integrated PMCF conclusions.
A multicountry evidence need. Output: end-to-end delivery and reporting.
Post-market clinical follow-up is a loop, not a deliverable. What the report cannot answer becomes the residual uncertainty the next activity has to address.
PMCF questions are reviewed by a team combining orthopedic clinical expertise and former Notified Body review experience, led by Dr Nikhil Khadabadi.
Registry outputs are interpreted alongside implant design, surgical practice, PROMs, imaging and residual risk.
Strategy, protocol, sites, data management, statistics, PMCF reporting and CER integration are planned together.
Commission a registry feasibility review, a single PMCF activity or a full DACH program.
We work with sponsor systems, registry platforms, site databases or an agreed EDC and study workflow.
Timelines follow the definition of the residual uncertainty, registry access, data-use approvals, device representation in the dataset, site contracts, ethics and data-protection review, retrospective data remediation, prospective follow-up duration, PROM and imaging components, record linkage, statistical complexity, and the timing of the PMCF Evaluation Report and CER update. Costs follow the number of countries and sites, registry-access fees, the number of records, manual data abstraction, device-identification remediation, PROM collection, imaging and core-lab review, the length of prospective follow-up, the number of endpoints, comparative analysis, and medical writing and CER integration. We quote in defined work packages, on the same basis as our European clinical trial cost benchmarking.
Related Eclevar pages across the orthopedic program.
Not automatically. A registry may answer revision and survivorship questions at scale, but PMCF must address the specific residual uncertainty for your device. Where the device is not identifiable at component level, or the registry lacks the relevant outcome, another method is needed.
When the outcome must be defined in advance, when PROMs or imaging are required, when the denominator and implant traceability must be certain, or when the question concerns a population or endpoint that routine datasets do not capture.
Yes, where the data pass a fitness-for-purpose review covering device identification, revision completeness, follow-up consistency, outcome definitions and data-access permissions. Size alone does not make a dataset suitable.
A revision involves the removal, exchange or addition of implant components. A reoperation is a further procedure on the operated joint that may leave the components unchanged. They should be defined and counted separately.
Through time-to-event methods with a prespecified endpoint, defined censoring, stated follow-up duration and confidence intervals, and competing-risk methods where death prevents later revision.
EPRD-informed or EPRD-based analyses may be considered where the device, sample size, access route and data permissions allow. Access is not automatic, and feasibility should be confirmed before the plan depends on it.
SIRIS may support Swiss arthroplasty PMCF where the device is represented and the required permissions, variables and follow-up are available. As with any registry, feasibility comes before commitment.
By selecting the model from actual data availability and permissions, typically through site-led prospective or retrospective evidence, rather than by assuming an EPRD or SIRIS-equivalent pathway exists.
For handling, usage patterns, surgeon experience, training needs and confirmation of intended-use conditions. It is not appropriate as the main evidence for survivorship, revision incidence, fixation or comparative performance.
Often, yes. Revision-free survival does not describe pain, function or satisfaction, so PROMs add the patient-perceived benefit that registries usually cannot supply at the required depth.
When the residual uncertainty concerns fixation, migration, radiolucency, osteolysis or loosening, or when a registry signal needs a mechanism. Registries may identify failure, while imaging may explain it.
Through the PMCF Evaluation Report, which carries the findings into the CER update, the benefit-risk conclusion and the risk-management and labeling review, with the PSUR and PMS report reflecting the same conclusions.
Yes. Published registry evidence is well suited to benchmarks, comparator context and device-category outcomes, but it does not by itself demonstrate the performance of your own device.
Death should normally be treated as a competing event rather than simple censoring, and loss to follow-up should be quantified, explained and tested through sensitivity analyses.
Share your device, existing clinical evidence, current PMCF plan and target DACH markets. We review the residual uncertainties and outline a proportionate strategy using prospective studies, retrospective data, registries, PROMs, imaging or focused surveys.
Your documents are reviewed confidentially. An NDA can be put in place before we receive any technical or clinical information. You can also reach the team through the contact page.
