MDCG 2025-10 Explained: Post-Market Surveillance Under the IVDR

Introduction

MDCG 2025-10 is the first guidance to describe what a post-market surveillance system has to do rather than what documents it has to produce, and the distinction is the whole point. Published in December 2025 and covering both the MDR and the IVDR, it walks Article 78 IVDR through the plan, the sources of data, the analysis and the eight processes that surveillance output has to feed — and it exposes a structural mistake that sits in a large proportion of IVD files: a plan written to describe a reporting cycle rather than a monitoring system, with indicators invented after the data arrived.

This guide follows the guidance in its own order and stops where the IVDR diverges from the MDR, because that is where files built on medical device templates fail. For the regulatory framework the obligation sits inside, start with the IVDR guide; for the performance evaluation that post-market performance follow-up feeds, the MDCG 2022-2 guide covers clinical evidence and the reports it produces.

Table of Contents

• What MDCG 2025-10 is, and the four questions it answers
• Post-market surveillance is a system, and the plan is only its instruction sheet
• The ten elements of the post-market surveillance plan
• Indicators and thresholds are set before the device is sold
• The PMPF plan lives inside the PMS plan
• What has to be collected, and from where
• Proactive is a requirement, not an adjective
• The eight destinations of post-market surveillance data
• Expected erroneous results, and the trend report that has no MDR equivalent
• Which report your class produces, and what it has to contain
• What changes if your procedures came from the MDR
• Where post-market surveillance files fail
• Frequently asked questions
• Conclusions

What MDCG 2025-10 is, and the four questions it answers

MDCG 2025-10 is titled Guidance on post-market surveillance of medical devices and in vitro diagnostic medical devices, endorsed by the Medical Device Coordination Group in December 2025. It sets itself four objectives: describe the PMS system, describe the PMS plan, describe the main activities within the system, and clarify how the system interacts with the other parts of the quality management system required by Article 10(8) IVDR.

Two things are explicitly outside its scope, and both matter for planning. It does not explain how to write a periodic safety update report — MDCG 2022-21 does that, and although that document is written for the MDR the guidance points to it as a source of presentational ideas for an IVDR post-market surveillance report as well. And it does not cover the health institution exemption under Article 5(5), while noting that health institutions are still expected to review the experience gained from in-house devices and take the necessary corrective action.

What makes the document valuable is not that it introduces anything new. It introduces nothing: every requirement in it is already in Article 78, Article 79 and Annex III. What it does is state, indent by indent, what each requirement means in practice and what a manufacturer is expected to have written down — which turns a set of one-line legal obligations into an auditable specification. The two tables it carries, one breaking down Annex III section 1(a) and one breaking down section 1(b), are the parts to work from.

Post-market surveillance is a system, and the plan is only its instruction sheet

Article 10(8)(i) IVDR requires a post-market surveillance system in accordance with Article 78, and Article 78(1) requires the manufacturer to plan, establish, document, implement, maintain and update it, proportionate to the risk class and appropriate to the type of device. The system is an integral part of the quality management system, not an annex to it, and the guidance says so at the outset: for a successful implementation the manufacturer must provide appropriate structures, procedures, processes and resources, and the effectiveness of the system should be reported to top management through management review.

The timing is worth stating precisely because it is frequently misunderstood. Planning starts during device development: the manufacturer is expected to think ahead and decide which activities will collect experience once the device is on the market. The first process cycle begins when the first device is placed on the market or put into service and closes with a report. But surveillance activities and data collection are continuous throughout the whole period between that first placing on the market and the end of the intended lifetime of the last device placed on the market — not periodic exercises that happen when the report falls due.

One PMS plan can cover a single device or a group of devices, provided the plan states its scope clearly. Devices sharing a manufacturing process, design and intended purpose, or belonging to the same device family, can share a plan. Every device must be covered by one.

The ten elements of the post-market surveillance plan

Section 1(b) of Annex III sets out what the plan must contain, and MDCG 2025-10 explains each indent with examples of what a plan should actually say. The guidance adds one structural clarification that resolves a long-running argument about how much detail belongs in the plan: the plan should define what types of methods are to be applied, while an explanation of how and by whom those methods are applied may sit in procedures referenced from the plan. A plan that reproduces the content of six standard operating procedures is not more compliant than one that names them.

The examples in the guidance are not mandatory content. MDCG 2025-10 states this twice, in footnotes to both of its tables: the examples are there to help, and it remains the manufacturer's job to select the methods, parameters, values, processes and tools appropriate to the device. A plan that copies the examples verbatim without a rationale has substituted the guidance's judgement for its own, and a reviewer will ask why those parameters and not others.

Indicators and thresholds are set before the device is sold

This is the single sentence in MDCG 2025-10 with the widest consequences, and it is easy to read past. Suitable indicators and threshold values — the limits beyond which action should be taken — are to be established in the pre-market phase. The data obtained through post-market surveillance activities is then fed back into the risk management process and used to re-evaluate those indicators and thresholds.

The order is the point. A threshold is a prediction about acceptable performance made from the risk analysis, before any field data exists, and post-market surveillance tests that prediction. A threshold derived from twelve months of observed complaint rates is not a threshold: it is a description of what happened, set at a level that by construction has not been breached, and it will never trigger anything. Reviewers detect this by comparing the revision date of the risk management file with the revision date of the plan, and by asking what evidence supported the number when it was first written.

The link runs to the risk file in both directions. The indicators and thresholds in the plan are expected to relate to hazards and their frequency of occurrence, to the estimates of the associated risks, and to the overall risk, the benefit-risk ratio and risk acceptability under section 3 of Annex I. If your ISO 14971 risk management file contains no quantity that the plan can monitor, one of the two documents is incomplete.

The PMPF plan lives inside the PMS plan

The tenth indent of Annex III section 1(b) makes the post-market performance follow-up plan an element of the post-market surveillance plan, and MDCG 2025-10 states that it is an integral part of it. Where the manufacturer determines that no specific PMPF activities are required, a justification is required in the same place. A file that produces a standalone PMPF plan referenced from nowhere, or a PMS plan that is silent on PMPF because none is planned, has missed a required element rather than made a scoping decision.

The relationship downstream is equally structured. Under the IVDR, performance-relevant information from post-market surveillance — particularly PMPF data — counts as relevant scientific data and as performance evaluation results, and Annex XIII Part B treats PMPF as a continuous process that updates the performance evaluation. The findings are analysed and documented in a PMPF evaluation report, and the conclusions of that report must be taken into account in the performance evaluation and in risk management.

The guidance also lists what PMPF data is aiming to do, and the list is worth checking a plan against: confirm safety and performance across the expected lifetime, identify previously unknown risks or limits to performance and contra-indications, identify and analyse emergent risks on factual evidence, ensure the continued acceptability of the clinical evidence and of the benefit-risk ratio, and identify possible systematic misuse. A PMPF plan whose objectives cannot be mapped onto those five has objectives that came from somewhere else.

What has to be collected, and from where

Section 1(a) of Annex III lists six categories of information the plan has to address, and the second table in MDCG 2025-10 explains what each one covers and what the resulting information is used for. The categories are serious incidents including information from PSURs and field safety corrective actions; records of non-serious incidents and data on undesirable side effects; information from trend reporting; relevant specialist or technical literature, databases and registers; feedback and complaints from users, distributors and importers; and publicly available information about similar devices.

Three of these are routinely under-populated in IVD files. Literature screening is often limited to the manufacturer's own device, when the guidance expects competent authority websites and databases including EUDAMED, device registers where available, and other published material such as clinical guidelines and health technology assessments. Feedback is often limited to complaints, when the guidance expects surveys of users, customers, distributors and importers, feedback gathered during training and workshops, and information reaching customer service and marketing through sales channels including digital ones. And publicly available information about similar devices is often absent altogether, when it is a named category with its own sources: literature, authority databases, registers, competitor instructions for use and labelling, summaries of safety and performance, information from third countries, and material presented at conferences and exhibitions.

The guidance adds a caution that cuts the other way. Data quality and integrity should be assessed before analysis, because unverifiable data can lead to over-reaction — it names social and public media as non-scientific sources. Collecting broadly is required; treating everything collected as equally probative is not.

Proactive is a requirement, not an adjective

The word proactive appears in the first indent of Annex III section 1(b), and again in Annex XIII Part B in the context of post-market performance follow-up. MDCG 2025-10 devotes a passage to it, and the definition is operational rather than rhetorical: proactive emphasises the manufacturer's obligation to actively seek out available information and not merely wait for it to arrive through channels such as complaints.

That single sentence disposes of a common file structure in which the entire surveillance system is a complaint-handling process with a report attached. Complaint handling is one input among six. The guidance lists what deliberate collection looks like: customer surveys, gathering of clinical experience, user feedback, screening of scientific literature, other clinical data sources, meta-analyses of published clinical data, evaluation of suitable registers, and post-market studies. It also says the choice of methods should reflect the level of innovation and research activity in the relevant field — which means a novel assay in an active research area cannot justify the same surveillance intensity as a decades-old clinical chemistry parameter.

For higher-risk devices, and for devices with novel features or applications, the guidance goes further: manufacturers may be required to generate data on real-world use, in most cases through the specific methods and procedures of post-market performance follow-up. Generating data is a different obligation from collecting it, and it is the one that a class C or class D plan has to confront directly.

The eight destinations of post-market surveillance data

Article 78(3) requires the information gathered by the surveillance system to be used as an input, on a continuous basis, to eight named processes. MDCG 2025-10 devotes its third table to them, and this list is the most useful auditing instrument in the whole document, because each entry is a testable claim about your quality system.

Two of the eight are habitually treated as optional and are not. Point (f), identifying options to improve the usability, performance and safety of the device, is explicitly not tied to identified risks: the guidance describes these actions as proactive and strategic, aimed at enhancing user experience, clinical outcomes or operational efficiency rather than at addressing hazards, and it requires the design and development process to define post-market surveillance as an input. Point (g), contributing to the post-market surveillance of other devices, requires that any new or increased risk found on one device be considered against every other device sharing the same or a similar intended purpose, design or processing characteristics — a horizontal review that most manufacturers have no procedure for.

Point (b) contains a requirement that is easy to miss in a device with no moving parts. Information from surveillance is used to evaluate the need to update design and manufacturing information, the instructions for use and the labelling, including where it identifies possible systematic misuse or off-label use. For an IVD, systematic misuse is rarely dramatic: it is a specimen type used outside the validated matrix, a storage condition ignored in a busy laboratory, a result read at a time point the instructions for use do not support. Finding it is a surveillance outcome; acting on it through the instructions for use is the required response.

Expected erroneous results, and the trend report that has no MDR equivalent

Article 83 IVDR requires manufacturers to report any statistically significant increase in the frequency or severity of incidents that are not serious incidents, or of expected erroneous results, where that increase could significantly affect the benefit-risk ratio and has led or may lead to unacceptable risks. MDCG 2025-10 names the category explicitly when it maps point (h) of Article 78(3): non-serious incidents, expected undesirable side-effects under the MDR, or expected erroneous results under the IVDR.

Expected erroneous results is the IVD-specific concept, and it has no counterpart in the medical device regulation. Every diagnostic test produces a known rate of false positives and false negatives; that rate is claimed in the instructions for use, supported by the clinical performance data, and accepted in the benefit-risk determination. It is expected. What Article 83 asks is whether the rate observed in the field has increased significantly against that expectation — which requires the expected rate to be stated somewhere the surveillance system can compare against, and requires a method for detecting a significant departure from it.

This is where a procedure imported from the medical device side fails silently. It has a trending mechanism for non-serious incidents and complaints, it has a statistical method, and it has an observation period. What it does not have is a category for a device performing exactly as designed at a rate that has drifted, because on the MDR side no such category exists. The plan has to define the methodology for establishing a statistically significant increase and the observation period, or reference the procedure that does — and for an IVD that methodology has to cover erroneous results as well as incidents.

Which report your class produces, and what it has to contain

The surveillance system and the plan are identical across all four IVD classes. The report is where classification bites.

Class A and class B devices produce a post-market surveillance report under Article 80. Its minimum content is a summary of the results and conclusions of the analysis of the surveillance data resulting from execution of the plan, together with a rationale and description of any corrective or preventive action taken in the period. It is updated as the plan provides, or earlier if the manufacturer judges it necessary, and it is made available to the competent authority on request.

Class C and class D devices produce a periodic safety update report under Article 81, updated at least annually, with three further elements: the conclusions of the benefit-risk determination, the main findings of the post-market performance follow-up, and the volume of sales together with an estimate of the size and other characteristics of the population using the device and, where practicable, its usage frequency. For class D the report is submitted to the notified body through EUDAMED; for class C it is made available to the notified body, and to competent authorities on request. The EUDAMED registration guide covers what is live in that system and what still runs on national arrangements.

The brevity of Article 80 is the trap. Two required elements do not mean two paragraphs of work behind them: the report summarises the execution of a plan that has ten elements, and the eight destinations of Article 78(3) apply to a class A specimen receptacle as much as to a class D blood-screening assay. What scales with class is the intensity of the activity and the formality of the report, not the existence of the obligations.

The guidance closes the loop where most plans stop. Following the outcome of the report, a review of the current plan may be necessary, and that review drives the level of surveillance activity for the next cycle — based on the class, the historical data, the similar products and the need for post-market performance follow-up. A plan that has never been revised after a report has not completed a cycle.

What changes if your procedures came from the MDR

MDCG 2025-10 covers both regulations in one document, which is convenient and also the reason IVD-specific requirements are easy to lose. The architecture is genuinely shared: Article 78 IVDR mirrors Article 83 MDR, Annex III is common to both, and the eight uses of surveillance data are the same eight. The differences are in the article numbers, in two categories of content, and in one cross-reference that produces no error message when it is wrong.

Annex XIV again. It exists in both regulations and means different things: post-market clinical follow-up under the MDR, and performance study documentation under the IVDR. A PMS plan that points its follow-up section at Annex XIV Part B is pointing at the wrong document, and the reference resolves cleanly to the wrong place. The parallel exercise on the device side is covered in the EU MDR post-market surveillance guide.

Where post-market surveillance files fail

The failures are visible in the plan, which is convenient: they can be found and corrected before a single complaint has been received, and they are expensive to correct afterwards because the surveillance data was generated under the wrong design.

The threshold failure is the most consequential and the hardest to remediate honestly, because correcting it means setting a number that the historical data may already have breached. That is uncomfortable and it is the correct outcome: a threshold derived from the risk analysis, applied retrospectively, either confirms that the device performed inside expectations or identifies a signal that was never actioned. Both are useful; only one of them is a finding.

The expected erroneous results failure is the most IVD-specific, and it is invisible to anyone auditing against MDR habits. The proactive failure is the most common: a system that receives is not a system that seeks, and the difference is legible from the list of sources alone. The similar-device failure is the easiest to fix and the most frequently absent, because it requires no internal data at all — literature, authority databases, competitor labelling and summaries of safety and performance are all public.

The two remaining failures are structural. A PMPF plan filed as a separate deliverable has misread Annex III section 1(b) in exactly the way a standalone literature search protocol misreads Annex XIII 1.1, and the correction in both cases is to move the document rather than to write a new one. And treating the brevity of Article 80 as the measure of the obligation confuses the report with the system it reports on.

Frequently asked questions

What is MDCG 2025-10?

MDCG 2025-10 is the Medical Device Coordination Group guidance on post-market surveillance of medical devices and in vitro diagnostic medical devices, published in December 2025. It describes the surveillance system, the surveillance plan, the main activities within the system and the way the system interacts with the rest of the quality management system. It covers the MDR and the IVDR together, and it is not legally binding.

When are indicators and threshold values established?

In the pre-market phase. MDCG 2025-10 is explicit that suitable indicators and threshold values — the limits beyond which action should be taken — are established before the device is on the market, and that post-market surveillance data is then fed back into the risk management process and used to re-evaluate them. A threshold derived from observed complaint rates after a year on the market is a description of the data, not a limit.

Does the PMPF plan sit inside the post-market surveillance plan?

Yes. The tenth indent of Annex III section 1(b) makes the post-market performance follow-up plan an element of the surveillance plan, and MDCG 2025-10 describes it as an integral part of it. Where no specific PMPF activity is planned, the justification belongs in the same place. A detailed PMPF protocol may sit in a separate document referenced from the plan.

What are expected erroneous results?

They are the false positive and false negative results an IVD is known to produce at a rate claimed in the instructions for use and accepted in the benefit-risk determination. Article 83 IVDR requires a manufacturer to report any statistically significant increase in their frequency or severity, alongside non-serious incidents. The MDR has no equivalent category, so a trending procedure imported from the device side will not cover them.

Which report does a class B IVD need?

A post-market surveillance report under Article 80, containing a summary of the results and conclusions of the analysis of the surveillance data and a rationale and description of any corrective or preventive action taken. It is updated according to the surveillance plan, or earlier when the manufacturer judges it necessary, and is made available to the competent authority on request. Class C and class D devices produce a PSUR under Article 81 instead, at least annually.

Can one post-market surveillance plan cover several devices?

Yes. A plan can cover a single device or a group of devices — for example devices sharing a manufacturing process, design and intended purpose, or belonging to the same device family — provided the plan clearly states which devices fall within its scope. Every device must be covered by a plan.

What does proactive actually require?

Actively seeking out available information rather than waiting for it to arrive through complaints. MDCG 2025-10 lists customer surveys, gathering of clinical experience, user feedback, screening of scientific literature, other clinical data sources, meta-analyses of published clinical data, evaluation of suitable registers and post-market studies, and expects the choice of methods to reflect the level of innovation and research activity in the field.

Conclusions

Post-market surveillance under the IVDR is a system with ten planned elements, six categories of input and eight obligatory destinations for its output. MDCG 2025-10 does not add to any of that; it makes each part specific enough to audit, which is why a plan written before the guidance existed is worth re-reading against it now.

Three points carry most of the practical value. Indicators and thresholds are established pre-market and re-evaluated with surveillance data, not derived from it. Expected erroneous results are an IVD-specific trend reporting category with no MDR equivalent, and a procedure copied from the device side will not contain them. And the post-market performance follow-up plan is an element of the surveillance plan rather than a document beside it, with the justification in the same place where no follow-up is performed.

The EU IVDR Performance Evaluation Kit includes the Post-Market Surveillance Plan covering the ten elements of Annex III section 1(b), the PMPF Plan and PMPF Evaluation Report positioned inside it, the Post-Market Surveillance Report for class A and B and the PSUR for class C and D, and the Performance Evaluation set the follow-up data reports into — all aligned with Regulation (EU) 2017/746 and MDCG 2025-10, and immediately deployable in an existing ISO 13485 quality management system.

Related articles

• IVDR Explained: What Regulation (EU) 2017/746 Requires
• MDCG 2022-2 Explained: Clinical Evidence for IVDs
• MDCG 2025-5: When an IVD Performance Study Needs Approval
• ISO 14971 Risk Management
• EU MDR Post-Market Surveillance
• EUDAMED Registration