LAN

Medical Device Clinical Evaluation and Performance Evaluation in Practice: CEP, CER, PEP and PER

Medical Device Clinical Evaluation Guide | EU MDR & IVDR

Jul 23, 2026

The Evidence File Problem

When MDR is governed by Article 61, we know that MDR and IVDR are governed by Article 61 and 56. The main problem in medical device clinical evaluation arises when we produce the document to the Notified Body. Like how did you map a claim? And does this document have measurable endpoints? And why does literature cover the intended purpose? Or why post-market data has not changed the conclusion? These are few gaps that arise due to the lack of evidence, not by the knowledge of the person doing it.

Clinical Evaluation and Performance Evaluation for MDR and IVDR are the same sides but different coins. We should first plan the claim and then collect the data and then sort, appraise it and then write a report which a reviewer can follow then keep a file updated with real-use data. This guide covers how CEP, CER under MDR and PEP, PER under IVDR and how PMCF and PMPF react to all these.

One Logic, Two Document Sets for Medical Device Clinical Evaluation

There is one logic under which two document sets fall: documents for medical devices and documents for IVDs requiring a performance evaluation report medical device file.

Decision	Medical devices (MDR)	IVDs (IVDR)
Legal core	Art. 61 + Annex XIV	Art. 56 + Annex XIII
Plan	Clinical Evaluation Plan (CEP)	Performance Evaluation Plan (PEP)
Report	Clinical Evaluation Report (CER)	Performance Evaluation Report (PER)
Post-market follow-up	PMCF (Annex XIV Part B)	PMPF (Annex XIII Part B)
Evidence question	Safety, performance, clinical benefit and risk	Scientific validity, analytical and clinical performance
Key guidance	MDCG 2020-5 / 2020-6 / 2020-13 / 2024-10 / 2025-6	MDCG 2022-2 / 2025-5 / 2025-6 / 2025-10

Reviewers do not grade the document titles. They grade whether the intended purpose, GSPR, methods, results, limitation and residual risk sit on a readable chain.

What the Clinical Evaluation Plan and PEP Must Have

Preparing the clinical evaluation plan (CEP) and PEP are steps which decide if the file is won or lost. Before even the study starts related to the medical device clinical evaluation plan, the plan must make sure to have the following so that it can serve well:

The intended purpose text from the plan should match text from the label, IFU and promotional claims.
The plan must contain the target population, the user persona and the specimen for use or environment.
The clinical benefits or performance claims should be validated, not just stated.
The plan should also contain specific GSPR along with the required information to be furnished.
The literature must contain databases, terms, windows, inclusion or exclusion and appraisal.
If any gaps arise, use existing data or equivalence or a new study to close the gap.
The acceptance criteria for all these in the plan should be with numbers, not just mentioned adequately with a percentage of sensitivity or something.
The plan should also contain how a PMCF or PMPF will affect or change the file after the CE marking is done.

For IVDs, the PEP should also contain characteristics as per Annex I, Section 9, which should define the following parameters: precision, trueness of the product, LOD or LOQ, measuring range, and analytical specificity, interference, stability, and clinical endpoints, which will be judged, such as diagnostic sensitivity, specificity, predictive values, and MDCG 2020-16. And this was clarified by MDCG 2025-5: the PEP is not mandatory for every performance-study application, but the authorities or the Notified Bodies can request it or even in the technical documentation if they feel so.

For medical devices, MDAI or medical device artificial intelligence, it is governed by MDCG 2025-6, or AIB 2025-1. It expects AI to be specifically validated across the themes such as accuracy, robustness, cybersecurity, transparency, human oversight, and fundamental rights to check and sit in the CEP or PEP rather than disconnect the annex.

Performance Evaluation Report Medical Device & IVD Three Pillar Evidence

The IVD three-pillar evidence must have all these three parameters as well as what must be shown in the parameter and what must be the output.

Pillar	What must be shown	Typical file output
Scientific validity	The marker is associated with the claimed state according to Article 2(38)	Scientific Validity Report from existing literature with consensus or proof of concept
Analytical performance	Correct detection or measurement under the claimed conditions and Article 2(40), from Annex I	Analytical performance studies with metrological traceability, whenever available
Clinical performance	Results should correlate with the clinical conditions under intended use. This comes under Article 2(41)	Clinical performance study which should match along with ISO 20916 or justified literature or existing data as part of the clinical performance report.

These three are separate pillars, but under a single roof. A weak scientific validity claim can collapse the other clinical performance. An analytical performance that never appears in the clinical protocol makes the performance evaluation report medical device question mark, which is self contradicting to itself. The risk management ISO 14971 and Annex I Section 3 should name the failure modes which the performance work is meant to control.

Equivalence, Literature and Gaps

Under the MDR, to find an equivalence, it is very hard, but actually, they should match the following categories, which is given under MDCG 2020-5, which says it should match under technical and biological and clinical characteristics, plus sufficient access to the equivalent data as well. Legacy devices may still require sufficient clinical evidence under MDCG 2020-6. Even having a CE mark for a long duration of years will not suffice as evidence.

Literature only helps when the search method and findings are able to be replicated whenever required and are kept ready for the review. A PRISMA-style flow is highly appreciated because it highlights what was screened out. For IVDs, Annex XIII expects a systematic scientific literature review, which feeds scientific validity and also makes a justification for clinical performance. Gaps will only generate study questions and would definitely not disappear into an optimistic narrative. This should be taken note of.

What the Clinical Evaluation Report and PER Must Address

The CER/PER is a clinical defence of a device and is a review ready clinical evaluation report which shows what is the block and what does the review checker need in it.

Block	Reviewer check
Device identity	Name of the device, the version number, and the Basic UDI-DI and where it is used and the nomenclature codes associated with the device and its accessibility
Intended purpose	Exact approved wording along with the restriction and contraindication for the device
Cross-reference	Plan and a risk file which is associated with the plan and IFU and study reports and literature protocol or report
Evidence inventory	What was identified and what was appraised and whether this was appraised as accepted or rejected
Analysis	Method, statistics, and acceptance criteria versus what is the results of the particular product
GSPR	Linked to Annex I requirements versus covered by which data
Benefit-risk	Residual risk versus demonstrated benefit by a particular product under SOTA
Limitation	What remains unknown and how PMPF or PMCF will address it
Conclusion	Clear pass or conditional or insufficient statement which is to be made by the Notified Body

MDCG 2020-13 guides for CER on how Notified Bodies will assess the medical device clinical evaluation as well as Annex XIII Section 1.3.2 guides the performance evaluation report on how to record scientific validity, analytical and clinical performance conclusion in a single place. PER related to Class C and D should be updated at least annually because it is high risk and fast iterating. MDR devices typically need annual CER refresh, which was earlier reopened on significant compliance or design change or a claim change, but now the latest laws are changed.

PMCF, PMPF, Clinical Performance Report and Update Rhythm

PMCF and PMPF are both proactive approaches. They are a part of PMS as planned under MDR Article 84 and IVDR Article 79, but it takes information from the evaluation report and under MDCG 2025-10 states the PMS as a continuous process which defines sources, methods, thresholds, and how conclusion change technical documents, risk management, or clinical performance report evaluation.

What are the things that trigger PMCF or PMPF?

New PMPF/PMCF findings or literature that changes SOTA
Vigilance signal on the trend reports or other reports
Design, software or reagent claim changes
For Class C or D, it is done annually as needed according to IVDR Article 29
For MDR Class III or IIb, the annual update rhythm is based on the PSUR according to Article 86

The transition from legacy IVD devices according to EU 2024/1860: the reclassification of legacy devices according to EU 2024/1860 does not affect the evidence work which is in progress. The revised timeline for these are for Class D devices 31st December 2027, but for Class C it is 31st December 2028. For Class B and A sterile devices it is 31st December 2029. All these device classes still even though they are A, B, C or D, despite them requiring a QMS by 26 May 2025. Still the application to Notified Body should be submitted regarding the same. And under Article 10a supply interruption duties should also be reflected in risk and surveillance thinking where many critical devices may leave the market before due to this compliance.

Rejection Reasons, Corrections and Performance Evaluation Report Fixes

What gets rejected	Why	Fix
Plan and IFU contradict on the intended purpose	Evidence will answer a different claim rather than IFU will answer a different claim and the plan would have had a different claim. This will create a clash within themselves	Freeze one purpose and keep it the same across DoC, IFU, plan and report
Acceptance criteria are qualitative	The reviewer cannot judge if it is pass or fail on the initial stages as it is up to the decision of the reviewer	Redefine numerical limits and statistical methods which we are having as a set
Negative literature omitted	When submitting literature, we need to submit everything even if it is negative or positive. If there is a bias, it gets failed. This comes under Annex XIII and Annex XIV during the appraisal	Keep the contradictory data and justify the weightages to positive data
Equivalence without data access	We provide the equivalence but it doesn’t have a link to the data of its own. It directly or automatically fails under MDCG 2020-5	Own the data, get the license or access to it and start a new investigation
Analytical and clinical protocols disconnect when three pillars are not combined together	All the three pillars support each other	Share the analysis and populations and the claim language so that three pillars have the same criteria which will support each other and the clinical performance report.
No PMCF or PMPF with measurable objectives	The lifecycle duty of each is not demonstrated separately	Objectives, sample size, logic, linked to risk, and PSUR are provided separately for each and which shows the complete lifecycle and how each acts and how it is done for PMCF and PMPF both
AI testing found outside clinical file	MDCG 2025-6 expects AI to be a part of the file, not outside the file	Make the accuracy, robustness, oversight according to the MDCG act, and then incorporate in the same file into CEP/CER or PEP/PER, so that it comes in the file
Outdated transition	This signals a weak file to a regulatory body, flags it as weak	Align all these to (EU) 2024/1860, and to current standards

FAQ

Does every In vitro Device require PEP/PER?

Is the PEP mandatory in every performance study application?

Under MDR when can you avoid new clinical investigations?

How often should CER / PER be updated?

What is the scope of MDAI?

Sources

Regulatory information verified against EUR-Lex and European Commission MDCG sources in July 2026.