How to Build a Solid PMCF Plan That Will Pass Your Notified Body


Quite a few medical device manufacturers still see PMCF (Post-Market Clinical Follow-up) as a formal document they have to submit "because we were asked to." In reality, it is a central part of the MDR, and when it comes to demonstrating clinical safety and performance, there are no shortcuts.

So what really counts as good clinical evidence within PMCF? How do you map the possible types of data? And what is the difference between the different levels of clinical quality?

Let's go through the full hierarchy of 12 levels of clinical evidence, and see which levels notified bodies consider reliable and strong, and which ones do not meet the requirements.

When does PMCF become mandatory?

PMCF becomes a required element when:

  • products are moving from the MDD to the MDR and do not have enough up-to-date clinical evidence,

  • the scientific literature does not support the performance claims,

  • market data (for example, from clinical registries) points to an unusual rate of complications or malfunctions, or

  • there is a gap between the data submitted for the original approval and today's clinical reality.

In other words, any product that makes significant clinical claims needs ongoing, up-to-date support from the field.

Why is PMCF methodology split in two?

PMCF can be carried out in two main ways:

  • General PMCF: based on existing sources, such as literature reviews, user feedback, complaint data, analysis of actual use and more.

  • Specific PMCF: includes targeted clinical activities, such as Investigator Initiated Studies, endpoint-based clinical registries, patient cohort analyses and prospective studies.

This distinction matters because notified bodies look at whether the data collection method fits the level of risk and the specific product.

A hierarchy of 12 levels of clinical evidence

Notified bodies expect to see the type of evidence that supports the chosen PMCF strategy, according to its clinical strength. Here is the full breakdown:

  1. Level 1 – High-quality controlled clinical trials: The highest standard. A prospective controlled trial, usually with comparison against a control group. Considered the most reliable in terms of safety, performance and statistics.

  2. Level 2 – Good-quality but uncontrolled clinical trials: Prospective clinical trials, usually with a single arm or without randomization, but conducted according to a defined protocol.

  3. Level 3 – Clinical registries (Registries): Systematic collection of data from real patients over time, including measured clinical endpoints. Well documented, although open to certain biases.

  4. Level 4 – Quantitative clinical surveys (High-quality Surveys): Closed, quantitative questionnaires with objective, analyzable measures, carried out with a significant number of users. The data must be verifiable, and the methodology defined in advance.

  5. Level 5 – Reliable clinical equivalence data (Equivalence data): A well-founded comparison between a new device and another device with a very similar clinical profile. This level is considered reasonable only when equivalence is fully documented.

  6. Level 6 – Literature reviews and similar information (SOTA): An analysis of the existing literature (State of the Art), including a comparison of similar devices on the market. This is supporting evidence, but it does not stand on its own.

  7. Level 7 – Complaints and vigilance reports (Vigilance data): Field data from user complaints, malfunction reports and side effects. Important for monitoring, but considered a low level of proof.

  8. Level 8 – General usability surveys (General Usability Surveys): General questionnaires about the user experience, usually without measurable endpoints. The methodology is inconsistent and sometimes relies on subjective, inaccurate memory. Suitable only as supplementary data.

  9. Level 9 – Individual case reports (Case reports): Anecdotal descriptions from patients or physicians. They cannot be analyzed statistically, but they sometimes offer interesting insights.

  10. Level 10 – Compliance with non-clinical regulatory requirements: Showing compliance with non-clinical requirements such as risk assessment, technical documentation or standards. This has no direct clinical value, but it is required as part of the complete file.

  11. Level 11 – Simulated use testing: Using models or simulations to test how the device performs. Suitable for products without direct clinical use, but not considered sufficient clinical evidence.

  12. Level 12 – Laboratory and pre-clinical testing (Pre-clinical testing): Mechanical, chemical or physical testing. Required as part of development, but it does not replace actual clinical evidence.

Comparison: level 4 vs. level 8. They look similar, but they are far from it

Both types are based on surveys, but their regulatory value is very different:

Characteristic

Level 4 - High-quality clinical surveys

Level 8 - General usability surveys

Data quality

Measurable, objective data based on  endpoints

Often retrospective, memory-based, not measurable

Method

Structured questionnaire, many participants, defined protocol

General questionnaire, sometimes one-off

Validity

Statistical analysis, assessment of validity and accuracy

Usually a general impression, with no in-depth analysis

Use

Directly supports safety and performance claims

Suitable as supplementary data, not a foundation on its own

The message is clear: not every survey is clinical evidence. Using a level 8 survey as a primary source will almost always be rejected by the notified body.

What does this mean in practice?

  • The more your PMCF plan is built on levels 1-4, the higher the chance of approval from the notified body, and the less need there will be for revisions.

  • Data from levels 6-12 can be used, but only as additional support, not as the basis for demonstrating clinical performance.

  • You need transparency about data sources, a clear methodology, and the ability to perform a reliable statistical analysis.

Common red flags that get PMCF plans rejected 

Notified bodies report the same mistakes in PMCF plans again and again. Here are some of the most prominent:

  • Prospective studies that are too short (less than 24 months for implantable products)

  • No clear quantitative clinical measures (performance endpoints)

  • Surveys without closed questions or defined measures

  • Use of non-systematic real-world data, for example, no defined denominator or undocumented analysis methods

  • Relying solely on literature reviews or unfocused historical data

A special focus on Real-World Data

Real-world data can be a strong source, but only if:

  • there is a clear definition of the entire population (including those who did not complete treatment),

  • cases with a positive outcome are separated from revisions/failures, and

  • the data was analyzed consistently and is not based on anecdotes or partial numbers.

In summary

PMCF is a long-term clinical strategy. To make sure your product continues to meet safety and performance requirements after it receives the CE mark, you need to base the follow-up on clinical evidence that is high quality, measurable and of real value.

Understanding the evidence hierarchy and choosing data sources wisely are the key to a PMCF plan that holds up under a notified body's critical eye.