Clinical Data Definition MDR: What Counts as Clinical Evidence

Written by HATEM RABEH, MD, MSc Ing

Your Clinical Evaluation Expert And Partner

in
S

The MDR shifted the definition of clinical data in ways that many manufacturers still misunderstand. What counted as supporting evidence under the MDD often fails under MDR scrutiny. The difference is not subtle. It affects your equivalence claims, your SOTA analysis, and your entire clinical evaluation structure.

Understanding what qualifies as clinical data under MDR is not an academic exercise. It determines whether your technical file survives review or gets suspended mid-certification.

What MDR Article 2(48) Actually Says

MDR Article 2(48) defines clinical data as information concerning safety or performance obtained from the use of a device. This sounds broad. It is not.

The regulation specifies that clinical data includes information derived from clinical investigations, scientific literature, and registries or post-market surveillance. But each of these sources has qualification criteria that most submissions overlook.

Clinical data must relate to the clinical use of the device. Not bench testing. Not simulations. Not biocompatibility panels run in isolation. The data must demonstrate how the device performs or behaves in contact with human tissue or in clinical application.

This distinction matters because many manufacturers still pad their clinical evaluation reports with benchtop studies and animal trials, then wonder why reviewers reject the dataset as insufficient.

The Three Recognized Sources of Clinical Data

MDR and MDCG 2020-5 identify three acceptable sources for clinical data: clinical investigations conducted on the device, relevant scientific literature, and data from equivalent devices when equivalence is properly demonstrated.

Each source has its own qualification process. You cannot substitute one for another without meeting the specific criteria.

Clinical Investigations

This is data you generate yourself. Prospective studies. Clinical trials. Observational studies conducted under defined protocols with ethical approval and proper consent.

Clinical investigation data is the strongest form of clinical evidence under MDR. It directly addresses your device, your intended use, and your target population. No inference. No extrapolation. No equivalence assumptions.

The problem is cost and timeline. Running a clinical trial for a Class IIa device is often impractical. That is why manufacturers rely on literature and equivalence. But that reliance must be justified.

Scientific Literature

Published studies count as clinical data if they meet specific quality and relevance criteria. The literature must describe devices similar to yours used in comparable clinical settings. The study design must be robust. The endpoints must align with your intended use.

MDCG 2020-13 outlines the appraisal process for scientific literature. It requires systematic search strategies, critical analysis of methodology, and transparent documentation of inclusion and exclusion decisions.

What does not count: in vitro studies published in peer-reviewed journals. Animal studies. Bench testing. Computational models. These may support your technical rationale, but they are not clinical data.

Data from Equivalent Devices

If you demonstrate equivalence to another device under MDR Annex XIV, you can leverage the clinical data of that equivalent device. But equivalence is not similarity. It is not close enough. It is not within the same product family.

Equivalence requires identical technical characteristics, identical biological characteristics, and identical clinical performance. If you cannot demonstrate all three, you cannot rely on the equivalent device’s clinical data.

Most equivalence claims fail on biological or clinical characteristics. The devices may share the same materials and similar design, but differ in contact duration, tissue type, or clinical indication. That difference invalidates the equivalence claim.

When equivalence fails, the clinical data from the other device becomes irrelevant to your submission. You are left without sufficient clinical evidence.

What Reviewers Actually Look For

Notified Bodies and competent authorities do not just count publications. They assess the dataset for relevance, quality, and sufficiency.

Relevance means the data addresses your device’s specific risks, intended use, and target population. A study on a similar device used in a different clinical setting may be high quality but irrelevant to your submission.

Quality refers to study design, sample size, methodology, and bias control. Observational studies with undefined protocols and uncontrolled variables get flagged. Case reports and opinion pieces do not qualify as clinical evidence.

Sufficiency means the dataset covers all identified risks and claims. You cannot ignore a significant risk because no literature exists for it. You must either generate new data or justify the gap with technical rationale and risk management.

The Gray Zone: Post-Market Surveillance and Registries

MDR Article 2(48) includes post-market clinical follow-up and registry data as potential sources of clinical data. This creates confusion because PMCF is often retrospective and uncontrolled.

PMCF data qualifies as clinical data when it is systematically collected, analyzed according to a defined plan, and addresses specific clinical questions. Ad hoc complaint logs and vague customer feedback do not meet this standard.

Registry data can be valuable if the registry tracks relevant clinical outcomes for devices similar to yours. But the registry must be credible, the data must be accessible, and the population must align with your intended use.

In practice, most manufacturers cannot rely on registries at the time of initial submission. The data is either unavailable or not specific enough. PMCF becomes more useful during recertification and periodic safety updates.

Why This Matters for Your Clinical Evaluation

The clinical data you include determines whether your clinical evaluation report demonstrates conformity with MDR Annex I. If your dataset is weak, your entire evaluation collapses.

Reviewers will challenge every non-clinical citation. They will question every equivalence assumption. They will flag every gap between your claims and your evidence.

I have seen submissions with comprehensive technical files, strong risk management, and robust quality systems still get rejected because the clinical data definition was misunderstood. The manufacturer thought biocompatibility data counted as clinical evidence. It does not.

Another common mistake is citing preclinical studies to support clinical claims. You can reference those studies in your technical rationale, but they cannot appear in your clinical data section. Mixing the two signals to the reviewer that you do not understand the regulatory framework.

How to Build a Defensible Clinical Dataset

Start by identifying what clinical questions need answers. What are the significant residual risks? What are the clinical claims? What does the state of the art require you to demonstrate?

Then map each question to a data source. Can it be answered with literature? Do you need clinical investigation? Can equivalence apply if properly justified?

Conduct your literature search according to MDCG 2020-13. Document your search terms, databases, inclusion criteria, and appraisal methodology. Do not cherry-pick favorable studies. Include contradictory findings and explain how they affect your conclusions.

If you rely on equivalence, document it rigorously. Provide side-by-side comparisons of technical specifications, material properties, and clinical use. If any characteristic differs, explain why it does not affect clinical performance. If you cannot explain it, equivalence is not valid.

For gaps in the dataset, acknowledge them openly. Explain why the gap exists. Describe how risk management addresses the uncertainty. Outline your PMCF plan to collect the missing data post-market.

Transparency builds trust. Trying to hide a weak dataset with volume or vague language does the opposite.

What Happens When You Get It Wrong

If your clinical dataset is insufficient, the Notified Body will issue a non-conformity. You will need to provide additional data or justify the gap. If you cannot, the certification process stalls.

In some cases, you can address the deficiency with a better literature search or refined equivalence justification. In other cases, you will need to conduct a clinical investigation. That adds months or years to your timeline.

I have seen manufacturers lose market access because they could not generate sufficient clinical data within a reasonable timeframe. The product was technically sound. The risk management was solid. But the clinical evidence was missing, and there was no fast path to generate it.

The earlier you understand what counts as clinical data, the earlier you can plan how to obtain it. Waiting until the Notified Body review to discover the problem is too late.

Final Thoughts

The definition of clinical data under MDR is narrower and more demanding than most manufacturers expect. It excludes preclinical work. It requires systematic appraisal of literature. It sets high bars for equivalence.

Understanding this definition is not optional. It is the foundation of your clinical evaluation. If you misinterpret it, every downstream decision in your submission is compromised.

Take the time to map your clinical data sources correctly. Challenge your assumptions about what qualifies. And when in doubt, focus on data derived from human use in clinical settings. That is what the regulation requires. That is what reviewers will demand.

Peace,
Hatem
Clinical Evaluation Expert for Medical Devices
Follow me for more insights and practical advice.