MDR Article 61 · Annex XIV · Articles 83 to 86
Clinical evaluation, PMCF and PMS
The CE mark is a launch date, not the end of the evidence story.
Once a device enters real use, the market starts sending signals. PMS listens across the whole product system. PMCF investigates the clinical questions that matter. Clinical evaluation decides what the combined evidence means for safety, performance and benefit-risk.
Enter the evidence control room ↓The operating principle
The market generates evidence whether the system is ready or not
Complaints, literature, service records, registries, surveys, incidents and routine clinical use do not arrive as a tidy conclusion.
The manufacturer’s job is to turn those uneven signals into a traceable decision and then return the decision to the product.
01
Clinical evaluation, PMCF and PMS form one living evidence loop #
These are not three annual documents waiting for three signatures. They are three functions in one lifecycle system.
PMS collects and analyses experience from the market. PMCF proactively develops clinical information from use of the CE-marked device within its intended purpose. Clinical evaluation appraises the relevant clinical data and determines whether the evidence still supports conformity, safety, performance and clinical benefit.
Captures quality, safety, performance and use signals across the device lifecycle
Targets unresolved or emerging clinical questions with planned methods
Integrates clinical data into a current conclusion on evidence and benefit-risk
If the CER identifies uncertainty, can it be traced into a PMS or PMCF activity? If new data arrive, can they be traced back into the CER, risk management and a controlled action?
02
Clinical evaluation is the interpretation layer #
Clinical evaluation is a continuous, methodologically sound process. It identifies, appraises and analyses clinical data, then asks whether those data provide sufficient clinical evidence for the device and its intended purpose.
The CER is the controlled record of that reasoning. It should make the path from clinical claims to evidence visible, including the state of the art, alternative treatment options, benefit-risk acceptability, residual risks, undesirable side-effects and the remaining evidence gaps.
What clinical performance or benefit does the manufacturer promise?
What amount and quality of evidence are justified by risk, novelty and intended purpose?
Are relevance, methodological quality, bias and applicability understood?
Do the combined data support safety, performance, clinical benefit and acceptable benefit-risk?
Changing the cover date does not update a clinical evaluation. New evidence must be searched, appraised, analysed and allowed to change the conclusion.
03
PMS is the wide-angle listening system #
Post-market surveillance is broader than clinical follow-up. It is the manufacturer’s proactive and systematic system for collecting and analysing information on quality, performance and safety throughout the device lifetime.
A complaint database is one channel, not a PMS system. The system must actively seek information, compare it against expected outcomes and thresholds, draw conclusions and connect those conclusions to the rest of the quality system.
Reported problems, use difficulties, service information and customer observations
Serious incidents, non-serious incidents, side-effects and statistically significant increases
Returns, repairs, nonconformities, supplier signals and production information
Registries, studies, surveys, routine use and other clinical sources
State of the art, alternatives, public safety information and competitor signals
Sales volume, population characteristics, usage frequency, misuse and off-label patterns
MDCG 2025-10 makes the interfaces explicit. PMS information can require updates to clinical evaluation, risk management, design, labelling, instructions for use, CAPA and vigilance activities.
04
PMCF is the focused clinical probe #
PMCF is a continuous process that updates the clinical evaluation. Its plan defines how the manufacturer will proactively collect and evaluate clinical data from human use of the CE-marked device within its intended purpose.
The aim is not to generate data for their own sake. PMCF confirms continuing safety and performance, identifies unknown side-effects and emerging risks, monitors known side-effects and contraindications, reassesses benefit-risk and detects possible systematic misuse or off-label use.
Outcomes observed before market entry may change across centres, users, populations and time.
Rare events and longer exposure may reveal a different clinical pattern.
Technical success does not automatically establish meaningful clinical benefit.
A previously acceptable outcome can become less persuasive when alternatives improve.
Systematic misuse or off-label use can challenge assumptions in the intended purpose.
The method should follow the question. A justified programme may combine literature, registries, surveys, structured follow-up, real-world evidence and post-market clinical investigations.
05
PMCF sits inside PMS but performs a different job #
The PMS plan must address PMCF through a PMCF plan or provide a justification why PMCF is not applicable. That relationship is structural, not cosmetic.
Monitors every relevant quality, safety, performance and use signal
Complaints · incidents · trends · service · literature · exposureProactively investigates selected clinical uncertainties and objectives
Studies · registries · RWE · surveys · structured follow-upClinical evaluation is connected to both. It supplies questions to PMS and PMCF, then receives the resulting clinical data and broader market intelligence for interpretation.
06
Plan from uncertainty backwards #
Weak plans begin with an available method. Strong plans begin with a decision that cannot yet be made confidently.
The CEP, CER and risk-management file should expose the uncertainties that matter. The PMS and PMCF plans then translate those uncertainties into objectives, data sources, endpoints, analysis methods, thresholds, responsibilities and timelines.
Which claim, population, risk or duration remains insufficiently supported?
What regulatory, clinical or risk decision will the new information support?
Which outcome, comparator, population and observation period can answer it?
What finding will maintain, challenge or change the current conclusion?
A transparent, controlled evidence gap is more defensible than a confident conclusion with no visible data behind it.
07
Match the method to the clinical question #
No method is automatically strong or weak. Its value depends on whether the design can answer the defined question with adequate relevance, quality and control of bias.
Efficient for landscape changes, but rarely enough to confirm continuing performance of the manufacturer’s own device.
Useful for specific operational questions when sampling, response bias and outcome definitions are controlled.
Powerful when device identity, data completeness, endpoints, follow-up and confounding are understood.
Can broaden population and setting, provided the data source and analysis are fit for the decision.
Appropriate when prospective design, defined endpoints or a comparator are required to reduce uncertainty.
Continues observation of an existing cohort when time is the missing evidence dimension.
The PMCF plan should justify sample size, endpoints, comparator, study design, representativeness, bias controls, statistical approach, limitations and schedule. Naming a method is not the same as designing an activity.
08
A signal becomes useful only when it reaches a decision #
Post-market information arrives with noise, missing context and unequal reliability. A controlled signal pathway prevents both overreaction and convenient dismissal.
A threshold, trend, event, publication or observation creates a signal.
Identity, denominator, duplication, data quality and clinical relevance are checked.
The signal is assessed against expected performance, risks, alternatives and prior evidence.
The conclusion may update the CER, risk file, IFU, design, training, PMCF, CAPA or vigilance.
The manufacturer confirms whether the action controlled the issue and whether monitoring should continue.
A credible system preserves negative findings, deviations, missing data and limitations, then shows why they do or do not change the benefit-risk conclusion.
09
The documents must describe the same device and the same evidence #
The MDR does not create a licence for each function to maintain its own version of reality. Intended purpose, variants, claims, residual risks, state of the art, populations, endpoints and conclusions must remain consistent across the technical documentation.
Defines scope, methods, claims, evidence needs and the clinical evaluation approach
Appraises and analyses the clinical data and documents the current conclusion
Defines data sources, methods, thresholds, responsibilities, outputs and interfaces
Converts selected clinical questions into justified activities and timelines
Analyses performed activities, deviations, positive and negative results, and impact
Summarises post-market results, conclusions, exposure and preventive or corrective actions
Receives new hazards, probabilities, residual-risk evidence and benefit-risk changes
Communicate the current intended purpose, performance, risks and clinical information
A newly identified risk cannot exist only in the PSUR. A revised claim cannot exist only in the CER. A PMCF result cannot disappear between the evaluation report and the risk file.
10
Use regulatory cadence without turning it into a waiting period #
The MDR combines periodic reporting with continuous surveillance. A scheduled report date never justifies postponing action on a material signal.
Updated using clinical data from PMCF and PMS. The review frequency should be justified, with earlier updates when evidence or product conditions change.
The plan defines a detailed reporting timetable. For class III and implantable devices, Article 61 requires an update at least annually.
Updated according to the PMS plan or earlier when the manufacturer determines that an update is needed.
Updated sooner when necessary and maintained as part of the technical documentation.
Summarises post-market analyses, benefit-risk conclusions, PMCF findings and device exposure.
Triggers that ignore the calendar
11
A reviewer follows the thread, not the folder names #
A notified body can begin with a residual risk in the CER, move to the PMCF objective intended to monitor it, examine the selected method and result, then ask where the conclusion changed the risk file and PSUR.
The system is persuasive when that journey is short, controlled and reproducible.
Do all documents cover the same devices, variants, intended purpose and populations?
Does every activity answer a defined question, with a defensible method and schedule?
Can objectives, data, conclusions and actions be followed across document boundaries?
Are unfavourable findings, deviations, limitations and missing information visible?
Did new evidence change the right documents and processes without unnecessary delay?
Were corrective or preventive actions verified rather than merely recorded?
Evidence-loop failures that attract attention
No activity is linked to a device-specific uncertainty or decision.
Rates appear comparable, but exposure populations and time windows do not match.
No own-device clinical data address continuing safety and performance where such data are needed.
Findings never reach the CER, risk management, PMS plan or next PMCF plan.
The reporting calendar is used as a substitute for continuous surveillance and timely action.
The plan generates activity but does not reduce the uncertainty that justified it.
12
Build an evidence operating system, not a document calendar #
The most efficient model gives clinical, vigilance, risk, quality, engineering and regulatory teams one controlled route for questions, signals, conclusions and actions.
- Define the device scope, intended purpose, clinical claims and evidence level
- Map CER uncertainties and residual risks to post-market questions
- Design the PMS plan as the complete listening architecture
- Design PMCF activities for the clinical questions that need active investigation
- Set data quality rules, denominators, thresholds, responsibilities and escalation paths
- Analyse positive and negative findings against the state of the art and benefit-risk
- Update every affected controlled document and external communication
- Verify actions and convert remaining uncertainty into the next evidence cycle
Official sources
- Regulation EU 2017/745 on medical devices
- MDCG 2025-10 guidance on post-market surveillance
- MDCG 2020-7 guidance on the PMCF plan template
- MDCG 2020-8 guidance on the PMCF evaluation report template
- MDCG 2020-6 guidance on sufficient clinical evidence for legacy devices
- MDCG 2020-5 guidance on clinical evaluation and equivalence
- MDCG 2022-21 guidance on periodic safety update reports
- European Commission collection of current MDCG guidance
The lifecycle conclusion
Evidence becomes valuable when it can change the product
PMS sees the broad market. PMCF examines the clinical questions. Clinical evaluation decides what the evidence means.
The system closes only when that meaning reaches risk management, design, labelling, user information, CAPA, vigilance and the next evidence plan.
A living clinical evaluation is not the document that changes most often. It is the one connected well enough to change when the evidence requires it.
Clinical evidence lifecycle
Need to connect fragmented CER, PMCF and PMS documents into one defensible operating system?
Discuss your evidence architecture ↗From the knowledge centre
