Clinical Evaluation Report (CER) for Medical Devices: The Complete Guide

What Is a Clinical Evaluation Report?

A Clinical Evaluation Report (CER) is a living document that systematically assesses the clinical data relating to a medical device in order to verify the device's clinical safety, clinical performance, and acceptability of its benefit-risk profile when used as intended by the manufacturer.

It is not a summary. It is not a literature review. It is a structured, methodical analysis — conducted according to a defined plan — that synthesizes all available clinical evidence and draws conclusions about whether that evidence is sufficient to demonstrate conformity with the relevant General Safety and Performance Requirements (GSPRs) of the EU Medical Device Regulation (EU) 2017/745.

Every medical device placed on the European market must have a clinical evaluation. Every class. No exceptions. The CER is the document that records that evaluation and its conclusions, and it sits within the technical documentation for the device.

If you have been in this industry for any length of time, you have encountered a CER. But the gap between a CER that passes Notified Body (NB) review and one that comes back with a multi-page deficiency letter is enormous. This guide covers the full scope: regulatory basis, methodology, document structure, equivalence arguments, literature search strategy, PMCF integration, Notified Body expectations, and the practical realities of writing a CER that actually withstands scrutiny.

Regulatory Basis for Clinical Evaluation

EU MDR Article 61

Article 61 of the MDR is the primary legal basis for clinical evaluation. Its key requirements:

• Article 61(1): The clinical evaluation and its documentation must be updated throughout the lifetime of the device with clinical data obtained from post-market surveillance (particularly PMCF) and post-market clinical follow-up.
• Article 61(2): For devices where demonstration of conformity with GSPRs based on clinical data is not deemed appropriate, adequate justification must be provided. This applies to certain low-risk Class I devices where clinical data may be replaced by performance testing and bench data — but the manufacturer must still justify this approach within the clinical evaluation.
• Article 61(3): Confirmation of conformity with GSPRs under normal conditions of use, the assessment of undesirable side-effects, and the acceptability of the benefit-risk ratio must be based on clinical data providing sufficient clinical evidence.
• Article 61(4): For implantable devices and Class III devices, clinical investigations must be performed unless reliance on existing clinical data from equivalent devices is adequately justified and the NB confirms this approach during conformity assessment.
• Article 61(5): A clinical investigation may also be necessary for Class IIa and IIb devices where existing data is insufficient.

Annex XIV — Clinical Evaluation and Post-Market Clinical Follow-Up

Annex XIV provides the procedural framework for clinical evaluation. It is split into two parts:

Part A — Clinical Evaluation requires manufacturers to:

1. Define the scope of the clinical evaluation through a Clinical Evaluation Plan (CEP)
2. Identify pertinent data through a systematic scientific literature review and any other sources of clinical data
3. Appraise all pertinent data by evaluating its suitability for establishing the safety and performance of the device
4. Analyse all pertinent data and draw conclusions about conformity with GSPRs

Part B — Post-Market Clinical Follow-Up establishes the requirements for PMCF plans and PMCF evaluation reports, which feed directly back into the CER.

MEDDEV 2.7/1 Rev.4

While technically a guidance document from the MDD era, MEDDEV 2.7/1 Rev.4 ("Clinical Evaluation: A Guide for Manufacturers and Notified Bodies under Directives 93/42/EEC and 90/385/EEC") remains the most detailed and widely referenced methodology document for conducting clinical evaluations. The EU MDR does not prescribe a specific methodology in the same level of detail, so MEDDEV 2.7/1 Rev.4 continues to serve as the de facto standard for CER methodology.

Notified Bodies expect CERs to follow this guidance. If your CER does not align with the structure and methodology described in MEDDEV 2.7/1 Rev.4, prepare for questions.

Key elements from MEDDEV 2.7/1 Rev.4 include:

• Staged evaluation process (identification, appraisal, analysis of clinical data)
• Detailed criteria for establishing device equivalence
• Literature search protocol requirements
• Data appraisal methodology (relevance, quality, contribution to the evaluation)
• Specific requirements for the qualification and experience of evaluators

MEDDEV 2.7/1 Rev.4 Appendix-by-Appendix Reference

The appendices of MEDDEV 2.7/1 Rev.4 contain some of the most practically useful content in the entire document. Many manufacturers focus on the main body text and overlook these appendices — which is a mistake, because Notified Bodies use them as benchmarks. Here is what each key appendix covers and how to apply it:

Appendix	Title	What It Contains	How to Use It
A1	Demonstration of Equivalence	Detailed criteria for clinical, technical, and biological equivalence. Specifies that equivalence can only be demonstrated against a single CE-marked device, and that all three characteristics must be fulfilled simultaneously.	Build your equivalence comparison table directly against these criteria. If your table does not address every parameter listed in A1, the NB will identify the gap.
A2	Appraisal of Clinical Data — Pivotal Data	Guidance on identifying which data items constitute pivotal evidence versus supportive evidence in the evaluation.	Explicitly label pivotal vs. supportive data in your appraisal tables. The NB expects this distinction.
A3	Identification of Devices Where Clinical Investigation Is Needed	Decision framework for determining when a clinical investigation is required versus when existing data is sufficient.	Reference this appendix when justifying that a clinical investigation is not required — particularly for Class IIb devices where the answer is not always clear.
A4	Data Generated in the Pre-Market Phase	Covers clinical data generated before the device enters the market, including first-in-human studies, feasibility studies, and pivotal trials.	Use this to ensure your CER correctly categorizes and presents pre-market clinical investigation data.
A5	Literature Search and Literature Review Protocol — Key Elements	The most detailed guidance available on how to construct and document the literature search protocol. Specifies required elements including data sources, search terms and parameters, selection/exclusion criteria with justification, deduplication methodology, data integrity procedures (e.g., independent second reviewer), and appraisal criteria for each data source.	This appendix is essentially the specification your literature search protocol must meet. Map your protocol against it point by point. See the Literature Search Protocol Template section below for a practical implementation.
A6	Appraisal of Clinical Data — Red Flags	Provides specific examples of studies that lack scientific validity for demonstrating clinical performance or safety. Red flags include: studies lacking basic information (methods, patient numbers, product identity), studies with no statistical analysis or inappropriate statistics, studies with unexplained high dropout rates, studies where conclusions are not supported by the data presented, and studies with undisclosed conflicts of interest.	Use this appendix as a screening checklist during data appraisal. If any of your included studies trigger these red flags, you must either exclude them (with justification) or explicitly discuss the limitations and reduce the weight assigned to that evidence.
A7	Types and Purposes of Clinical Investigations	Framework for different clinical investigation designs and when each is appropriate.	Reference when justifying the design of pre-market or post-market clinical investigations in the CER.
A8	CEP — Proposed Content	Detailed proposed content for the Clinical Evaluation Plan.	Cross-check your CEP against this appendix before finalizing.
A9	CER — Proposed Table of Contents	Nearly six pages of guidance on CER structure, including proposed section headings and examples of information to include within each section. This is the closest thing to an "official" CER template.	Structure your CER to match this proposed table of contents. If your CER deviates structurally from A9, be prepared to justify why. Many NBs use A9 as a de facto checklist when reviewing CERs.
A10	Proposed Checklist for Release of the CER	A quality checklist for verifying the CER before release. Key questions include: Is the report understandable to a third party? Does it provide sufficient detail? Is all data generated, mentioned, and summarized? If claiming equivalence, are differences adequately disclosed? Are unfavorable data discussed? Are conclusions justified by the data?	Run your CER through this checklist before submission. Have an independent reviewer complete the checklist as a final quality gate. If any answer is "no," address the gap before the NB sees it.

MDCG Guidance Documents

Several Medical Device Coordination Group (MDCG) guidance documents supplement the regulatory framework:

MDCG Document	Topic	Relevance to CER
MDCG 2020-1	Guidance on clinical evaluation (MDR) / Performance evaluation (IVDR) of medical device software	Essential for software CERs — addresses clinical evidence expectations for SaMD
MDCG 2020-5	Clinical Evaluation — Equivalence	Detailed criteria for demonstrating equivalence; supplements MEDDEV 2.7/1 Rev.4
MDCG 2020-6	Sufficient clinical evidence for legacy devices	Critical for MDD-to-MDR transitions
MDCG 2020-13	Clinical evaluation assessment report (CEAR) template	Shows what the NB assessment looks like — reverse-engineer your CER to answer these questions
MDCG 2024-2	PMCF Plan and PMCF Evaluation Report templates	Directly links PMCF outputs to CER updates

Clinical Evaluation vs. Clinical Investigation

These terms are frequently conflated. They are not the same thing.

Aspect	Clinical Evaluation	Clinical Investigation
What it is	A systematic assessment of all available clinical data for the device	A prospective study designed to generate new clinical data about a device
Regulatory basis	Article 61, Annex XIV Part A	Articles 62–82, Annex XV
When performed	Always — every device, every class	When existing clinical data is insufficient to support the clinical evaluation
Data sources	Literature, clinical investigations, post-market data, equivalent device data, registry data	The clinical investigation itself — protocol-defined endpoints, controlled data collection
Output document	Clinical Evaluation Report (CER)	Clinical Investigation Report
Relationship	The CER synthesizes all data including clinical investigation results	Clinical investigation data is one input to the CER
Ongoing requirement	Updated throughout device lifetime via PMCF	Conducted when needed; not inherently ongoing

A clinical investigation generates clinical data. A clinical evaluation assesses all available clinical data — including any data generated by clinical investigations, but also literature data, post-market surveillance data, and data from equivalent devices.

For many devices — particularly those with a well-documented clinical history, those with demonstrated equivalence to an existing device, and lower-risk Class I and IIa devices — a clinical evaluation based on literature data and post-market experience is sufficient without conducting a new clinical investigation.

For implantable devices and Class III devices, Article 61(4) creates a strong presumption that a clinical investigation is necessary unless the manufacturer can justify reliance on existing data from an equivalent device. This is a higher bar than under the MDD, and Notified Bodies are enforcing it rigorously.

The Clinical Evaluation Process

The clinical evaluation is a staged process. MEDDEV 2.7/1 Rev.4 and MDR Annex XIV Part A together define a workflow that proceeds through distinct phases.

Stage 0: Define the Scope — The Clinical Evaluation Plan (CEP)

Before touching a single data source, you need a plan. The Clinical Evaluation Plan (CEP) defines the scope, objectives, and methodology of the clinical evaluation. Under MDR Annex XIV Part A, the CEP must be part of the technical documentation.

The CEP should include:

• Device identification: Device description, intended purpose, intended patient population, indications, contraindications, target groups and intended users
• Clinical background and state of the art: Current knowledge about the medical condition, existing treatments, alternative devices, and accepted clinical outcomes
• Scope of the evaluation: Which GSPRs will be addressed through clinical data, and which through non-clinical data (with justification)
• Equivalence claim (if applicable): Identification of equivalent devices with preliminary justification
• Literature search protocol: Databases, search strings, inclusion/exclusion criteria, time period
• Clinical data types to be considered: Literature, clinical investigation data, post-market data, registry data
• Appraisal criteria: How data relevance, quality, and contribution will be assessed
• Evaluator qualifications: Identification of the individuals conducting the evaluation and their qualifications

The CEP is not a static document. It should be reviewed and updated when the scope of the evaluation changes — for example, when new indications are added, when the state of the art evolves, or when PMCF findings necessitate a change in evaluation methodology.

Practical tip: Write the CEP before you begin the literature search. If you write it retrospectively to justify searches you have already conducted, your methodology will have gaps and your NB reviewer will notice.

Stage 1: Identification of Clinical Data

This stage involves finding all clinical data relevant to the device. The data comes from three source categories defined in MEDDEV 2.7/1 Rev.4:

Source 1 — Clinical data generated from the device under evaluation:

• Clinical investigations (pre-market and post-market)
• PMCF studies and registries
• Complaint data and vigilance reports
• Post-market surveillance data

Source 2 — Clinical data generated from equivalent or similar devices:

• Published literature on equivalent devices
• Clinical investigation reports from equivalent devices (if accessible)
• Post-market data from equivalent devices

Source 3 — Clinical data generated from non-clinical sources that have clinical relevance:

• Bench testing results with clinical implications (e.g., fatigue testing predicting implant lifetime)
• Preclinical studies (animal studies, cadaver studies)
• Biocompatibility data
• Computational modeling and simulation data with clinical relevance

Literature Search Strategy

The literature search is the backbone of most CERs. A poorly executed search undermines the entire document. Here is what a rigorous search looks like.

Databases to search:

Database	Coverage	Why It Matters
MEDLINE (via PubMed)	Biomedical literature, ~36 million citations	The primary database for medical device literature; mandatory for any CER
Embase (via Embase.com or Ovid)	Biomedical and pharmacological literature, ~40 million records	Broader European journal coverage than PubMed; captures device-specific literature that PubMed may miss
Cochrane Library	Systematic reviews and controlled trials	Gold standard for evidence synthesis; essential for identifying existing meta-analyses and systematic reviews relevant to your device
ClinicalTrials.gov / EU Clinical Trials Register	Registered clinical trials (completed, ongoing, terminated)	Identifies unpublished studies; required to avoid publication bias
FDA MAUDE / Eudamed (when available)	Adverse event databases	Post-market safety data for equivalent and similar devices
Manufacturer's own data	Internal clinical investigation reports, PMCF data, complaint data	Source 1 data — you cannot rely solely on published literature

Building search strings:

Your search strategy must be systematic and reproducible. Document every step.

1. Define the PICO framework:

   • P (Population): The intended patient population
   • I (Intervention): The device under evaluation (or equivalent device)
   • C (Comparator): Standard of care, alternative treatments, predicate devices
   • O (Outcomes): Clinical outcomes relevant to the device's intended purpose — safety outcomes (adverse events, complications) and performance outcomes (clinical efficacy, functionality)

2. Develop search terms: Use a combination of Medical Subject Headings (MeSH terms) for MEDLINE, Emtree terms for Embase, and free-text keywords. Include synonyms, brand names, generic device names, and alternative spellings.

3. Apply Boolean operators: Combine search terms with AND, OR, NOT to create precise queries.

4. Define inclusion and exclusion criteria:

   • Include: peer-reviewed publications, clinical investigation reports, systematic reviews, meta-analyses
   • Exclude: editorials, letters to the editor (unless they contain relevant clinical data), non-English publications (with justification), animal-only studies (unless directly relevant to a biological endpoint)
   • Define time period — justify if you limit the search to a specific date range

5. Document the search: Record the exact search strings, date of search, database version, number of hits at each stage, and the screening process (title/abstract screening followed by full-text review).

Literature Search Protocol Template

MEDDEV 2.7/1 Rev.4 Appendix A5 specifies the required elements for a literature search protocol. The following template covers every element that Notified Bodies expect to see. Treat this as a minimum — not a maximum — for your protocol documentation.

Section 1 — Objectives and Scope

Protocol Element	What to Document
Search objective(s)	At minimum two distinct objectives: (1) identify clinical data on the safety and performance of the device under evaluation and equivalent/similar devices, and (2) identify data defining the current state of the art for the relevant medical condition and available treatments
Scope alignment	Explicit statement that the search objectives are consistent with the scope defined in the CEP
PICO/PICOS framework	Population, Intervention, Comparator, Outcome — and optionally Study design. Each element must be defined with inclusion and exclusion boundaries. For devices where time-to-event outcomes are relevant, consider PICOT (adding Time).

Section 2 — Data Sources

A minimum of two databases is required. Searching only PubMed/MEDLINE is insufficient.

Database	Rationale for Inclusion
MEDLINE (via PubMed)	Primary biomedical literature database; mandatory
Embase	Broader European journal coverage; captures device-specific literature that PubMed misses. MEDDEV 2.7/1 Rev.4 specifically emphasizes the use of European databases such as Embase.
Cochrane Library (CDSR + CENTRAL)	Systematic reviews and registered trials
ClinicalTrials.gov / EU Clinical Trials Register / WHO ICTRP	Registered and ongoing trials — required to address publication bias
Device-specific registries	Where applicable (e.g., national joint registries, cardiovascular registries)
Regulatory databases (FDA MAUDE, MHRA, BfArM)	Post-market safety signals for equivalent/similar devices
Grey literature sources	Conference proceedings, manufacturer white papers, HTA reports — document what was searched and why

For each database, document: the database platform used (e.g., PubMed.gov vs. Ovid MEDLINE), the database version or date coverage, and the date the search was executed.

Section 3 — Search Terms and Strategy

Element	Requirements
Controlled vocabulary	MeSH terms (MEDLINE), Emtree terms (Embase), and their mapping
Free-text keywords	Device brand names, generic device names, alternative spellings, abbreviations, manufacturer names, competitor product names
Boolean logic	Document how AND, OR, NOT operators are combined. Use parentheses to define grouping. Use quotation marks for exact phrase matching where needed.
Search string construction	Provide the complete, copy-pasteable search string for each database. Search strings will differ between databases due to different controlled vocabularies.
Filters applied	Language filters (justify any language restriction), date filters (justify any date restriction), study type filters (justify if applied)

Section 4 — Selection (Inclusion/Exclusion) Criteria

Criterion Type	Examples
Inclusion criteria	Peer-reviewed publications; clinical data on the subject device, equivalent device, or similar devices; study populations matching the intended patient population; outcomes relevant to safety or performance
Exclusion criteria	Editorials and letters without original data (justify); non-human studies (unless biologically relevant); languages not covered by the evaluation team (justify and discuss potential impact); duplicate publications; publications with insufficient methodological reporting

Critical requirement per MEDDEV 2.7/1 Rev.4: the reason for exclusion of each screened publication should be documented to demonstrate that both favorable and unfavorable data have been considered.

Section 5 — Screening and Selection Process

Document a two-stage screening process:

1. Title/abstract screening: Initial review for relevance against inclusion/exclusion criteria
2. Full-text review: Detailed assessment of shortlisted publications against all criteria

Specify: who performs each screening stage, whether independent dual review is used (recommended), how disagreements are resolved, and how the process is documented at each stage.

Section 6 — Data Integrity and Quality Control

QC Element	Implementation
Deduplication	How duplicate records across databases are identified and managed
Independent verification	Whether a second reviewer independently screens a sample (or all) records
Reproducibility	The protocol must be sufficiently detailed that a third party could reproduce the search and obtain substantially the same results
Version control	How protocol amendments are documented and justified

Critical point: Notified Bodies will scrutinize your literature search for two things: completeness and bias. If your search is narrowly constructed to find only favorable data, or if you excluded unfavorable studies without transparent justification, you will receive a deficiency. The search must be designed to find all relevant data — favorable and unfavorable.

PRISMA flow diagram:

Use a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram to document the screening process. This is not technically required by the regulation, but it has become an expected standard. It shows the NB exactly how you went from thousands of initial hits to the final set of included studies, and where each exclusion occurred.

Stage 2: Appraisal of Clinical Data

Once you have identified the relevant data, each data item must be appraised for its suitability. MEDDEV 2.7/1 Rev.4 defines three appraisal dimensions:

Relevance:

• Is the data directly applicable to the device under evaluation?
• Does it address the intended purpose, patient population, and clinical conditions defined in the CEP?
• If data comes from an equivalent device, has equivalence been demonstrated?

Quality:

• What is the study design? (Randomized controlled trial > prospective cohort > retrospective cohort > case series > case report)
• Was the study adequately powered?
• Are the statistical methods appropriate?
• Is there risk of bias?
• Was the study conducted in accordance with good clinical practice (GCP)?
• Is the publication peer-reviewed?

Contribution:

• What does this data item contribute to the overall clinical evaluation?
• Does it provide evidence of safety, performance, or both?
• What is the weight of this evidence relative to other data items?

Each data item should be appraised individually and the appraisal documented in a structured format — typically a data appraisal table or form. This creates the audit trail that Notified Bodies expect.

Weighting the evidence:

Not all clinical data is created equal. A well-designed randomized controlled trial with 500 patients provides stronger evidence than a case series with 12 patients. Your CER must reflect this hierarchy. The following table illustrates the evidence hierarchy as applied to medical device clinical data:

Level	Study Design	Weight	Typical Use in CER
1	Systematic reviews and meta-analyses of RCTs	Highest	Strongest basis for safety and performance conclusions; rare for device-specific data
2	Randomized controlled trials (RCTs)	High	Primary evidence for clinical safety and performance when available
3	Non-randomized controlled studies (prospective cohorts with comparator)	Moderate-High	Common for device evaluations; valuable when RCTs are not feasible
4	Prospective cohort studies (single-arm)	Moderate	Frequently available; useful when combined with other data sources
5	Retrospective studies	Moderate-Low	Can provide longer follow-up data; subject to bias
6	Case series	Low	Supportive evidence; insufficient alone for higher-risk devices
7	Case reports	Very Low	Anecdotal; may be useful for identifying rare adverse events
8	Expert opinion, bench-to-clinical extrapolation	Lowest	Never sufficient alone; may support other evidence

Formal Quality Assessment Tools for Data Appraisal

The evidence hierarchy table above provides a general framework, but Notified Bodies increasingly expect CERs to use recognized, validated quality assessment tools — not ad hoc quality judgments. The choice of tool depends on the study design being appraised. The following tools are widely accepted in CER data appraisal:

For Randomized Controlled Trials:

Tool	What It Assesses	Scoring
Jadad Scale (Oxford Quality Scoring System)	Methodological quality across three domains: randomization (description and appropriateness), blinding (description and appropriateness), and withdrawals/dropouts	0–5 points. Studies scoring 0–2 are considered low quality; 3–5 are considered high quality. Simple to apply but limited in scope — does not assess allocation concealment, intention-to-treat analysis, or selective reporting.
Cochrane Risk of Bias Tool (RoB 2)	Seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, other bias	Each domain rated as Low Risk / Some Concerns / High Risk. More comprehensive than Jadad but more time-intensive. Preferred for CERs that include pivotal RCT data.

For Non-Randomized Studies (Cohort, Case-Control):

Tool	What It Assesses	Scoring
Newcastle-Ottawa Scale (NOS)	Three quality parameters: Selection (4 items), Comparability (1 item, max 2 points), and Outcome/Exposure (3 items)	Maximum 9 stars. Studies with 7–9 stars are considered high quality; 4–6 moderate; 0–3 low. Slightly different versions exist for cohort and case-control studies.
ROBINS-I (Risk of Bias in Non-Randomized Studies of Interventions)	Seven domains of bias: confounding, selection, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes, selection of reported result	Each domain rated Low / Moderate / Serious / Critical / No Information. More rigorous than NOS but significantly more time-consuming.

For Case Series and Case Reports:

No universally accepted validated tool exists. Use a structured assessment covering: clarity of case selection criteria, adequacy of case description, completeness of follow-up reporting, and transparency about limitations. The Joanna Briggs Institute (JBI) critical appraisal checklists for case series and case reports provide a reasonable framework.

Applying Weights in Practice:

Weighting in a CER is not a single numerical score. It derives from two independent dimensions that must both be assessed for each data item:

1. Methodological quality: Assessed using the tools above. Higher-quality studies receive greater weight in the overall evidence synthesis.
2. Relevance to the subject device: A high-quality RCT on a device with significant technical differences from yours is less relevant than a moderate-quality prospective study on the exact device under evaluation.

The overall weight of a data item is the combination of these two dimensions. A practical approach is to use a matrix:

	High Relevance (subject device or demonstrated equivalent)	Moderate Relevance (similar device, partial overlap)	Low Relevance (analogous device, different indication)
High Quality (RCT, well-designed prospective study)	Pivotal evidence — highest weight	Important supporting evidence	Contextual evidence — moderate weight
Moderate Quality (retrospective cohort, single-arm prospective)	Key supporting evidence	Supporting evidence	Background evidence — low weight
Low Quality (case series, case reports)	Supplementary evidence	Limited contribution	Negligible weight — include for completeness only

Document which quality assessment tool was used for each study type, present the individual scores or ratings, and explain how the quality and relevance assessments together determined the weight assigned to each data item. This level of methodological transparency is what distinguishes a CER that passes NB review from one that receives an appraisal methodology deficiency.

Stage 3: Analysis and Conclusions

This is where the CER earns its value — or fails. The analysis stage requires you to synthesize all the appraised data and draw conclusions about:

1. Clinical safety: Have the risks associated with the device been identified and quantified? Are the residual risks acceptable? How does the safety profile compare to the state of the art?

2. Clinical performance: Does the device achieve its intended clinical performance? Is this performance consistent with the claims made in the IFU and promotional materials?

3. Benefit-risk profile: When all benefits and risks are weighed, is the overall benefit-risk profile acceptable? This must be considered in the context of the state of the art and available alternatives.

4. Acceptability of undesirable side effects: Are the known side effects acceptable in relation to the intended benefits and when compared to alternatives?

5. Identified gaps: Is the clinical evidence sufficient, or are there gaps that need to be addressed — through PMCF, additional clinical investigations, or other means?

The analysis must be objective. It must address unfavorable data. If three studies show excellent outcomes and one study reports a higher complication rate, the CER must discuss that study and explain why the overall conclusion is still favorable — or update the risk analysis accordingly.

Practical tip: The analysis section is where most CERs fail NB review. Manufacturers write extensive literature reviews but then produce conclusions that are generic statements like "the device is safe and performs as intended." This is insufficient. The conclusions must specifically address each GSPR claimed to be met through clinical data, reference specific evidence, and quantify outcomes wherever possible.

Equivalence Demonstration

Claiming equivalence to another device allows you to use clinical data generated from that device to support your own CER. This is one of the most powerful — and most scrutinized — elements of a CER.

The Three Pillars of Equivalence

Under MDR Annex XIV Part A and MDCG 2020-5, equivalence must be demonstrated across three dimensions simultaneously:

Dimension	Criteria	What You Must Show
Clinical	Same clinical condition, same intended purpose, same intended site in the body, same target population (including age, anatomy, physiology), same intended user, similar severity and stage of disease	The device is used for the same clinical purpose in the same type of patient
Technical	Same design, same specifications and properties (including physicochemical properties, intensity of energy, deployment method, principles of operation), similar use conditions, similar specifications	The device has similar technical characteristics such that there would be no clinically significant difference in safety and performance
Biological	Same materials or substances in contact with the same human tissues or body fluids, for the same contact duration and intensity	The materials in contact with the patient are the same or have equivalent biocompatibility

All three dimensions must be met. You cannot claim equivalence on clinical and biological grounds if the technical characteristics are substantially different, or vice versa.

The MDR Tightened Equivalence Claims

Under the MDD, equivalence arguments were relatively common and accepted with moderate documentation. The MDR raised the bar significantly:

• Article 61(5): Manufacturers claiming equivalence to another manufacturer's device must have a contract in place with that manufacturer giving them sufficient access to the technical documentation to justify the equivalence claim. This requirement effectively prevents equivalence claims to competitor devices unless the competitor agrees to share documentation.
• For Class III and implantable devices: Reliance on equivalence alone (without your own clinical investigation) requires explicit confirmation from the Notified Body during conformity assessment, and the justification must be thorough.

The practical effect: for many devices — especially higher-risk devices — manufacturers can no longer rely solely on equivalence to a competitor's device. They need either their own clinical data, a contract with the equivalent device manufacturer, or a clinical investigation.

Documenting Equivalence

A robust equivalence demonstration should include:

1. Device-to-device comparison table: A detailed, feature-by-feature comparison across all three equivalence dimensions. This is not a high-level summary — it is a granular comparison of specifications.

2. Discussion of differences: Any differences between the subject device and the equivalent device must be identified and assessed for clinical significance. The CER must explain why each difference does not affect the validity of using the equivalent device's clinical data.

3. Supporting evidence: Test reports, design specifications, material certificates, and other documentation that substantiates the claimed similarities.

4. Access to equivalent device data: If the equivalent device is made by another manufacturer, documentation of the contractual arrangement providing access to technical documentation per Article 61(5).

State of the Art Analysis

The state of the art analysis is a frequently underestimated section of the CER. It establishes the baseline against which the device's safety and performance are evaluated.

The state of the art analysis should cover:

• The medical condition or indication: Epidemiology, pathophysiology, clinical presentation, natural history
• Current treatment options: Medical, surgical, and device-based alternatives. Include their outcomes, risks, and limitations.
• Relevant clinical guidelines and standards of care: What do professional societies and clinical guidelines recommend?
• Available devices on the market: Competitor devices, their design approaches, known safety issues, market history
• Accepted clinical outcomes and benchmarks: What complication rates, performance thresholds, and clinical endpoints are accepted as benchmarks in the relevant clinical literature?

The state of the art analysis serves two purposes: (1) it demonstrates that the evaluator understands the clinical context in which the device operates, and (2) it provides the benchmark against which the device's benefit-risk profile is assessed. If the accepted complication rate for a procedure is 3% and your device data shows a 5% complication rate, the CER must address this discrepancy.

Practical tip: The state of the art analysis is not a textbook chapter on the disease. It is a focused, evidence-based summary of what is currently accepted clinical practice, specifically tailored to the context of the device being evaluated. Keep it relevant, keep it current, and reference primary sources.

CER Document Structure and Template

While no single template is mandated by the MDR, the structure outlined in MEDDEV 2.7/1 Rev.4 and the expectations set by Notified Bodies have converged around a standard framework. The following template is widely used and accepted:

Recommended CER Structure

Section	Content	Approximate Length (for a Class IIb device)
1. Executive Summary	High-level summary of the device, clinical evaluation methodology, key findings, and conclusions	2–3 pages
2. Scope and Definitions	Scope of the CER, abbreviations, definitions, reference documents	1–2 pages
3. Device Description	Detailed technical description, intended purpose, indications, contraindications, target population, accessories, IFU summary	3–5 pages
4. Clinical Background and State of the Art	Medical condition overview, current treatments, clinical guidelines, alternative devices, accepted clinical outcomes	5–10 pages
5. Clinical Evaluation Plan Summary	Summary of the CEP — objectives, methodology, literature search protocol, appraisal criteria, equivalence approach	3–5 pages
6. Equivalence Demonstration	Device-to-device comparison (clinical, technical, biological), discussion of differences, justification	5–10 pages (if applicable)
7. Literature Search and Review	Search protocol execution, PRISMA diagram, list of included/excluded studies, search strings	5–8 pages
8. Data Appraisal	Individual appraisal of each data item (relevance, quality, contribution) — often in tabular form	10–30 pages (depends on volume of data)
9. Data Analysis — Safety	Analysis of all safety data — adverse events, complications, side effects, contraindications	10–20 pages
10. Data Analysis — Performance	Analysis of clinical performance data — efficacy, functionality, performance claims	10–20 pages
11. Benefit-Risk Analysis	Structured benefit-risk determination considering alternatives and state of the art	5–10 pages
12. GSPR Mapping	Mapping of clinical evidence to each relevant GSPR with gap identification	3–5 pages
13. PMCF Integration	Summary of PMCF plan, PMCF results (if available), how PMCF data has been incorporated	3–5 pages
14. Conclusions	Overall conclusions on clinical safety, performance, and benefit-risk acceptability; identified gaps and mitigation	2–3 pages
15. Evaluator Qualifications	CV summaries and declarations of the clinical evaluation team	1–2 pages
16. References	Complete bibliography	Variable
Appendices	Literature search outputs, data appraisal tables, PRISMA diagram, equivalence comparison table	Variable

The total length of a CER varies enormously depending on device risk class and complexity. A Class I non-sterile, non-measuring device CER might be 30–50 pages. A Class III implantable device CER might be 200–400 pages, with extensive appendices.

Section-by-Section Guidance

Executive Summary: Write this last. It should stand alone as a document that a Notified Body reviewer can read in 10 minutes to understand the device, the evaluation approach, the key evidence, and the conclusions. Include quantitative outcomes — do not limit yourself to qualitative statements.

Device Description: This must match the device description in the rest of the technical documentation. Inconsistencies between the CER device description and the IFU, the risk management file, or the design dossier will trigger questions. Include model numbers, variants, sizes, configurations, and accessories.

State of the Art: Reference clinical guidelines from recognized professional societies (e.g., ESC, AHA, NICE, WHO). Use the most current editions. Outdated guidelines signal a stale evaluation.

Equivalence Demonstration: If claiming equivalence, this section must be airtight. Use a three-column or multi-column comparison table. Every claimed equivalence parameter must be supported with data. Any difference — however minor — must be discussed and justified.

Literature Search: The search must be current. If your CER was last updated 18 months ago but you are submitting it for NB review today, the search must be updated. Many Notified Bodies expect the literature search to be no older than 6–12 months at the time of review.

Data Appraisal: This is the most labor-intensive section. Each included study should have a structured appraisal entry. Consider using a standardized appraisal form or table with the following fields:

Field	Description
Study ID / Citation	Full bibliographic reference
Study design	RCT, cohort, case series, etc.
Patient population	Number of patients, demographics, indications
Device evaluated	Subject device, equivalent device, or similar device
Relevance to subject device	Direct / Partial / Indirect
Quality assessment	Strengths and limitations (bias, confounders, follow-up)
Key safety findings	Adverse events, complication rates
Key performance findings	Efficacy outcomes, performance metrics
Contribution to CER	What conclusion does this study support?

Benefit-Risk Analysis: This is not a restatement of the risk management file. It is a clinical assessment of whether the clinical benefits demonstrated by the evidence outweigh the clinical risks identified in the evidence, in the context of the state of the art and available alternatives. Reference the risk management file for the full risk analysis, but the CER's benefit-risk section should be independently reasoned based on clinical data.

GSPR Mapping: Create a table that maps each clinically relevant GSPR (particularly Annex I, Chapter I, GSPRs 1–9, and applicable GSPRs from Chapter II and III) to the clinical evidence that supports conformity. This makes it immediately clear to the NB reviewer which requirements are covered by clinical data and whether any gaps exist.

PMCF Integration

Post-Market Clinical Follow-up is not separate from the CER — it is a continuation of it. The MDR makes this explicit: the clinical evaluation and its documentation must be updated throughout the lifetime of the device with clinical data obtained from PMCF.

The PMCF-CER Feedback Loop

The relationship works as follows:

1. CER identifies gaps in clinical evidence (e.g., insufficient long-term follow-up data, small sample sizes for rare complications, limited data in specific subpopulations)
2. PMCF plan is developed to address these gaps through defined activities (PMCF studies, registries, surveys, literature updates)
3. PMCF activities generate new clinical data
4. PMCF Evaluation Report summarizes the findings
5. CER is updated to incorporate the new PMCF data and reassess conclusions
6. Updated CER identifies new gaps (or confirms that previous gaps have been addressed)
7. Cycle repeats

What PMCF Activities Feed Into the CER

PMCF Activity	Data Generated	CER Impact
PMCF clinical studies	Prospective clinical data on safety and performance	Strengthens clinical evidence base; may close evidence gaps identified in the CER
Patient registries	Long-term outcome data, large population data	Provides real-world evidence; addresses long-term safety questions
Published literature monitoring	New publications on the device or equivalent devices	Updates the state of the art; may identify new risks or confirm existing safety profile
Complaint and vigilance data analysis	Trend data on adverse events, malfunctions	Identifies emerging safety signals; validates or challenges benefit-risk conclusions
Clinician surveys	User experience data, off-label use patterns, usability findings	May reveal unanticipated risks or uses not covered in the original evaluation

Update Frequency

The MDR does not prescribe a specific update frequency for the CER, but Notified Bodies have established clear expectations:

• Class III and implantable devices: CER updated at least annually
• Class IIb devices: CER updated at least every 2 years (some NBs expect annually)
• Class IIa devices: CER updated at least every 2–5 years (depending on NB and device risk profile)
• Class I devices: CER updated when significant new data becomes available or when the device changes

These are minimum expectations. The CER should also be updated whenever material new information becomes available — a serious safety signal, a relevant new publication, a field safety corrective action, or a change to the state of the art.

PMCF Study Design: Choosing the Right Approach

One of the most common weaknesses in CERs is a PMCF section that lists generic PMCF activities without specifying study designs, endpoints, sample sizes, or follow-up durations. Notified Bodies expect specificity. The PMCF plan — and its summary in the CER — must describe concrete, methodologically sound activities with defined parameters.

The Three Categories of PMCF Studies

PMCF studies fall into three distinct regulatory categories under the MDR, each with different requirements. The distinction is important because it determines which regulatory provisions apply, what approvals are needed, and how the study must be conducted.

Category	Description	Regulatory Framework	When to Use
Type 1: Device use outside the intended purpose	The study involves using the device beyond its CE-marked intended purpose — e.g., a new indication, an expanded patient population, or a use condition not covered by the conformity assessment.	Full clinical investigation requirements apply: Articles 62–81, Annex XV. Notification to competent authority and ethics committee required. EUDAMED registration required.	When the CER identifies a gap that can only be addressed by generating data outside the current intended purpose — e.g., evaluating the device in a paediatric population when the current intended purpose covers only adults.
Type 2: Intended purpose with additional burdensome or invasive procedures	The device is used within its approved intended purpose, but the study protocol requires additional procedures beyond normal clinical use that are invasive (e.g., additional blood draws, biopsies) or burdensome (e.g., additional imaging, questionnaires causing discomfort).	Article 74(1) applies. Must meet Article 62 requirements (paragraphs 4b–k, m), Articles 75–77, Article 80(5), and Annex XV. Notification to competent authority required (typically 30 days before initiation). Ethics committee approval required.	When the CER identifies evidence gaps that require collecting data through procedures not part of routine clinical care — e.g., imaging at specific follow-up intervals to assess device migration, or additional blood tests to monitor a biomarker.
Type 3: Intended purpose without additional burden	The device is used within its approved intended purpose and data is collected from routine clinical care without imposing any additional procedures on the patient.	The MDR clinical investigation provisions do not apply. National regulations govern the study conduct. Ethics committee notification is still required in most EU member states.	Most PMCF studies fall into this category. Appropriate when the evidence gap can be addressed by systematically collecting and analysing data that is already generated during routine use — e.g., complication rates, device performance metrics, patient-reported outcomes collected as part of standard follow-up.

Study Design Selection by Evidence Gap

The study design should be driven by the specific evidence gap identified in the CER, not by convenience or cost. The following framework maps common evidence gaps to appropriate study designs:

Evidence Gap Identified in CER	Recommended PMCF Study Design	Key Design Parameters
Insufficient long-term safety data	Prospective single-arm cohort study or patient registry	Follow-up: minimum 2–5 years for implantable devices (longer for permanent implants). Sample size: powered to detect complications with expected incidence rates (see below). Endpoints: device-related adverse events, device survival/failure rates.
Small sample size in pre-market studies	Prospective cohort study (multicentre)	Follow-up: 12–24 months minimum. Sample size: calculated to provide adequate statistical power for the primary endpoint. Endpoints: primary safety and performance endpoints from the pre-market study, replicated with larger N.
Limited data on specific subpopulations	Prospective cohort with stratified enrolment or subgroup registry	Follow-up: matched to the clinical outcome timeline. Sample size: powered for the subgroup analysis. Endpoints: same as overall study, with pre-specified subgroup comparisons.
Rare complications not captured in pre-market data	Patient registry (large N, long-term) or retrospective database study	Follow-up: ongoing (registries). Sample size: large enough to detect events with incidence of less than 1% (typically requires N > 300 for events at 1% incidence with 95% CI). Endpoints: specific rare adverse events of interest.
User experience and usability in real-world settings	Clinician survey or structured user feedback program	Sample size: sufficient to represent the user population diversity. Endpoints: device usability, off-label use patterns, training adequacy, user-reported issues.
Comparison with current standard of care	Prospective comparative study (randomized or matched cohort)	This is a Type 1 or Type 2 study in most cases. Requires full clinical investigation application. Sample size: powered for non-inferiority or superiority analysis. Endpoints: matched to the comparator's established clinical endpoints.

Sample Size Considerations for PMCF Studies

Notified Bodies increasingly expect PMCF study sample sizes to be justified with a statistical rationale — not arbitrarily chosen. The approach depends on the study objective:

For confirming a known complication rate:

Use a precision-based sample size calculation. If the expected complication rate from pre-market data is p, and you want to estimate the true rate with a 95% confidence interval of width +/- d, the required sample size is approximately:

n = (Z^2 x p x (1-p)) / d^2

Where Z = 1.96 for 95% confidence. For example, if the expected complication rate is 5% and you want a 95% CI of +/- 3%, you need approximately 203 patients.

For detecting a safety signal (rare event monitoring):

Use the Rule of Three: to be 95% confident of observing at least one event with a true incidence of 1/N, you need approximately 3N patients with zero events. To detect a complication occurring at 0.5% incidence, you need approximately 600 patients.

For registry-based studies:

Power calculations are less straightforward because registries are typically ongoing. Define interim analysis points and the minimum dataset size at which meaningful conclusions can be drawn. Reference the MDCG 2024-2 PMCF plan template for expected documentation.

Endpoint Selection

PMCF study endpoints must trace directly to the evidence gaps identified in the CER. Generic endpoints like "safety and performance" are insufficient. Define:

• Primary endpoint(s): The specific clinical outcome that addresses the primary evidence gap. Must be measurable, clinically meaningful, and have a defined success criterion (e.g., "device-related serious adverse event rate below 2% at 12 months").
• Secondary endpoint(s): Additional outcomes that provide supporting evidence. May include patient-reported outcomes (PROMs), device longevity metrics, or subgroup analyses.
• Safety endpoints: All device-related adverse events, serious adverse events, device deficiencies, and device-related re-interventions. These are expected regardless of the study's primary objective.

Follow-Up Duration by Device Type

Device Type	Minimum Expected Follow-Up	Rationale
Permanent implant (e.g., joint prosthesis, cardiac valve)	5–10+ years; ongoing registry recommended	Long-term failure modes, wear, migration, and late complications may not manifest for years
Resorbable implant	Through complete resorption plus 12 months	Need to capture the full resorption timeline and any delayed reactions
Short-term implant (e.g., temporary fixation)	12–24 months post-removal	Need to confirm no long-term sequelae after device removal
Non-implantable Class III (e.g., drug-coated balloon)	12–24 months	Must capture clinical outcomes over the therapeutically relevant period
Non-implantable Class IIb	6–12 months (longer if clinically indicated)	Sufficient for most short-term safety and performance outcomes
Active device / software	12–24 months	Must capture any drift in performance, software-related incidents, and user adaptation effects

Notified Body Expectations

Understanding what Notified Bodies look for in a CER is as important as understanding the regulatory requirements. Here is what NB reviewers focus on:

Evaluator Qualifications

MEDDEV 2.7/1 Rev.4 requires that the clinical evaluation be performed by individuals with specific qualifications:

• A degree in medicine, pharmacy, nursing, biomedical sciences, or a related discipline
• Documented knowledge of the device technology and its clinical application
• Documented experience in clinical research methodology (including literature searching, data appraisal, and biostatistics)
• A minimum of 5 years of professional experience (though this is guidance, not a hard legal requirement under the MDR)

Notified Bodies routinely check evaluator CVs. If the CER for a cardiovascular stent was written by someone with no cardiovascular or implantable device experience, the NB will question whether the evaluation was conducted with adequate expertise. The evaluator does not need to be a cardiologist — but they need demonstrable relevant experience.

Objectivity and Independence

The evaluator should be able to demonstrate objectivity. While the MDR does not require the evaluator to be external to the manufacturer, many Notified Bodies prefer — or require — that the CER author has a degree of independence. This can be demonstrated through:

• A declaration of conflicts of interest
• Peer review of the CER by an independent clinical expert
• Use of external clinical evaluation service providers

What Triggers Deficiency Letters

Based on publicly available NB observations and industry experience, the most common deficiency findings in CERs include:

Deficiency Category	Common Finding	How to Avoid
Literature search	Search too narrow, outdated, or biased; search strategy not documented; key databases omitted	Follow a pre-defined protocol; search multiple databases; include unfavorable data; document everything
Equivalence	Equivalence claimed without sufficient technical/clinical/biological comparison; differences not discussed; no access contract for third-party devices	Complete three-pillar comparison; discuss every difference; secure Article 61(5) contract if needed
State of the art	Outdated or absent; does not cover current clinical guidelines; does not identify alternative treatments	Update to current guidelines; cover alternatives; reference primary sources
Data appraisal	Studies not individually appraised; quality assessment missing; no structured methodology	Appraise each study systematically using defined criteria; document strengths and limitations
Benefit-risk	Generic conclusions; not evidence-based; does not compare to alternatives; does not quantify outcomes	Quantify benefits and risks with data; compare to state of the art; address unfavorable data
GSPR mapping	Missing or incomplete; GSPRs claimed to be met without evidence mapping	Map each GSPR to specific evidence; identify and address gaps
PMCF	CER does not reference PMCF plan; PMCF data not integrated; no identified evidence gaps	Include a PMCF section; describe how PMCF data updates the CER; identify gaps explicitly
Device description	Inconsistent with IFU or risk management file; incomplete variant coverage	Cross-check device description across all technical documentation
Evaluator qualifications	Evaluator has no relevant clinical/device experience; qualifications not documented	Appoint qualified evaluators; include CVs and declarations
Currency	CER has not been updated since last conformity assessment cycle	Update literature search and conclusions before submission

Quantitative Data on NB Deficiency Patterns

Notified Bodies have repeatedly indicated that clinical evaluations and clinical investigations are the single most frequent source of MDR non-conformities across all device classes — outpacing Quality Management System (QMS), risk management, and labeling issues combined. While individual NBs do not publish granular deficiency-rate breakdowns, aggregated industry data provides a quantitative picture of the challenge.

Industry survey data (68 MedTech professionals across manufacturers of medium- and high-risk devices):

Finding	High-Risk Devices	Medium-Risk Devices
Manufacturers who experienced inconsistencies in clinical data acceptance across different Notified Bodies	46%	40%
Manufacturers who invested significantly in clinical evaluation training under MDR	42%	10%
Manufacturers planning to remove devices from the EU market due to increased clinical evidence requirements	54%	0%
Manufacturers shifting to US-first or non-EU-first launch strategies	41.7%	20%
Manufacturers who do NOT outsource CER generation	76% (overall)	—
SMEs that outsource CER preparation vs. large enterprises	50% (SME)	18% (large)
Primary reason for outsourcing	"Lack of internal expertise" (62%)	—

What these numbers mean for your CER: The 46% inconsistency rate across NBs means that a CER accepted by one Notified Body may not pass review at another. This is not an argument for lower quality — it is an argument for exceeding the minimum bar. A CER built to the highest standard of any NB is portable across all NBs. The fact that 54% of high-risk device manufacturers are removing devices from the EU market underscores how significantly the MDR has raised the clinical evidence threshold. Those who remain in the market must meet that threshold.

Most common documentation-level deficiencies (aggregated from publicly available NB observations, BSI, TUV SUD, and NSAI feedback):

Most non-conformities raised by Notified Bodies do not stem from the absence of data but from poor documentation structure, insufficient justification, or lack of traceability between documents. The pattern is consistent:

1. Vague or non-measurable clinical objectives — The CER defines clinical safety and performance objectives that cannot be quantitatively assessed against the available evidence. Example: "The device is safe" rather than "The device demonstrates a major complication rate below 3%, consistent with the state-of-the-art benchmark of 4.2% reported in [guideline reference]."

2. Incomplete state-of-the-art benchmarking — The CER does not establish accepted performance thresholds from current clinical guidelines, making it impossible for the NB reviewer to assess whether the device's clinical outcomes are acceptable. This is flagged in the CEAR template (MDCG 2020-13, Section D).

3. Unsubstantiated equivalence claims — Equivalence is stated at a high level without the granular, parameter-by-parameter comparison across clinical, technical, and biological dimensions that MDCG 2020-5 and MEDDEV 2.7/1 Rev.4 Appendix A1 require.

4. Non-systematic literature searches — Searches that do not follow a pre-defined protocol, use only a single database, apply unjustified date or language restrictions, or do not document the screening and exclusion process. These deficiencies are immediately visible because the NB reviewer can compare the documented search methodology against the MEDDEV 2.7/1 Rev.4 Appendix A5 requirements.

5. Missing or superficial data appraisal — Studies are listed or summarized but not individually appraised for methodological quality, relevance, and contribution. No recognized quality assessment tool (Jadad, NOS, RoB 2) is used. No structured appraisal table is provided.

6. Weak benefit-risk analysis — Conclusions not grounded in quantified evidence; no comparison to alternatives or state of the art; unfavorable data not addressed.

7. PMCF disconnection — The CER identifies no evidence gaps (unrealistic for any device), or identifies gaps but the PMCF plan does not address them, or PMCF data exists but has not been integrated into the CER.

8. Inconsistency across technical documentation — The intended purpose in the CER does not match the IFU; the risk profile in the CER does not align with the risk management file; the PMCF plan addresses different gaps than those identified in the CER.

CER for Legacy Devices vs. New Devices

The transition from MDD to MDR has created a distinct challenge for legacy devices — devices that were certified under the MDD and now need to transition to MDR certification.

Legacy Devices

Legacy devices typically have significant post-market experience but may lack the structured clinical evaluation documentation that the MDR demands. Common issues include:

• MDD-era CERs were often thin. Under the MDD, many manufacturers — particularly for Class I and Class IIa devices — produced CERs that were little more than literature summaries with a brief conclusion. These do not meet MDR standards.
• Equivalence arguments may no longer hold. Under the MDR, equivalence to a competitor's device requires an Article 61(5) contract. If the legacy CER relied on equivalence to a competitor without such a contract, the equivalence argument must be reconstructed — potentially with own clinical data.
• Post-market data is an asset. Legacy devices often have years of complaint data, vigilance data, and post-market surveillance records. This data is highly valuable for the MDR CER if it is properly collected, analysed, and presented.
• MDCG 2020-6 provides specific guidance on sufficient clinical evidence for legacy devices, including a framework for assessing whether existing clinical data — combined with available post-market data — meets MDR requirements.

New Devices

New devices face different challenges:

• Limited clinical data. Without market history, the clinical evidence base is inherently smaller. The CER must be transparent about data limitations and define a PMCF plan to address them.
• First-in-human considerations. For truly novel devices, the CER may rely heavily on preclinical data, bench testing with clinical extrapolation, and data from analogous (not equivalent) devices. The argument must be carefully constructed.
• Clinical investigation may be required. For Class III and implantable devices, the presumption under Article 61(4) is that a clinical investigation is necessary. The CER must incorporate the clinical investigation report.
• PMCF is critical. For new devices, the PMCF plan is not a formality — it is the mechanism by which the CER's preliminary conclusions will be confirmed or revised with real-world data.

Comparison: Legacy vs. New Device CER Challenges

Aspect	Legacy Device	New Device
Clinical data availability	Extensive post-market data; may lack structured clinical investigation data	Limited; may rely on preclinical data and analogous device data
Equivalence	May have historical equivalence claim that needs updating	May establish equivalence to a marketed device (with contract)
Literature	Likely abundant published literature	May have limited device-specific literature
PMCF	Can leverage existing PMS data; PMCF may confirm existing profile	PMCF is essential to build the evidence base post-market
Main challenge	Upgrading documentation quality to MDR standards	Building sufficient evidence for a device with no market history
Regulatory expectation	NB expects thorough transition documentation; MDCG 2020-6 applies	NB expects rigorous pre-market evidence; clinical investigation often required

SSCP Requirements

The Summary of Safety and Clinical Performance (SSCP) is a public-facing document required under MDR Article 32 for:

• Implantable devices (except sutures, staples, dental fillings, dental braces, tooth crowns, screws, wedges, plates, wires, pins, clips, and connectors)
• Class III devices

The SSCP is derived from the CER and the risk management file. It is written for a dual audience: healthcare professionals and, where relevant, patients. It must be validated by the Notified Body and uploaded to EUDAMED.

Relationship Between CER and SSCP

CER	SSCP
Comprehensive, detailed technical document	Summary document written in plain language (where applicable)
Part of the technical documentation (confidential)	Publicly available via EUDAMED
Covers the full clinical evaluation methodology and findings	Summarizes key safety and performance information
Updated throughout device lifetime	Updated when the CER is updated
Audience: manufacturer, Notified Body, Competent Authority	Audience: healthcare professionals and patients

The SSCP is not a marketing document. It must be objective and balanced. It must include known risks, contraindications, and residual risks — not just favorable performance data. Notified Bodies review the SSCP to ensure consistency with the CER; if the CER identifies a safety concern that is absent from the SSCP, this will be flagged.

SSCP Content Requirements

The SSCP must include at minimum:

• Device identification (UDI-DI, trade name, manufacturer)
• Intended purpose and indications
• Device description
• Risks and warnings relevant to the healthcare professional and patient
• Summary of clinical evaluation (key findings on safety and performance)
• Possible diagnostic or therapeutic alternatives
• Suggested profile and training for users
• Reference to harmonized standards and common specifications applied
• Revision history

MDCG 2019-9 provides a detailed template for the SSCP content. Following this template closely will streamline NB review.

CER Differences: Class III vs. Class IIb Devices

The MDR does not provide a single set of clinical evaluation requirements that apply uniformly to all devices. The expectations escalate with risk class, and the differences between Class III and Class IIb devices are substantial. Manufacturers who treat these two classes identically in their clinical evaluation approach are making a mistake that Notified Bodies will identify.

Clinical Investigation Requirements

Requirement	Class III / Implantable Devices	Class IIb (Non-Implantable)	Class IIb Implantable
Clinical investigation presumption	Mandatory under Article 61(4), unless reliance on equivalent device data is justified and confirmed by the NB	Not mandatory — clinical investigation required only when existing data is insufficient (Article 61(5))	Same requirements as Class III implantable devices under Article 61(4)
Equivalence as alternative to clinical investigation	Permitted but requires explicit NB confirmation during conformity assessment. For third-party devices, Article 61(5) contract is mandatory.	Equivalence may be used more readily, but the same three-pillar demonstration (clinical, technical, biological) and Article 61(5) contract requirements apply	Same heightened scrutiny as Class III
Justification burden	Manufacturer must affirmatively justify why a clinical investigation is not needed. The default expectation is that one is required.	Manufacturer must justify that existing data is sufficient. The default expectation is more permissive.	Same as Class III

The Clinical Evaluation Consultation Procedure (CECP) — Article 54

The CECP is an additional layer of scrutiny that applies specifically to:

• Class III implantable devices
• Class IIb active devices intended to administer and/or remove a medicinal product (ARMP devices)

Under Article 54, the Notified Body must submit the Clinical Evaluation Assessment Report (CEAR) and supporting documentation to an expert panel for independent scientific review. The expert panel then determines whether to issue a scientific opinion.

Quantitative data from the CECP to date:

Metric	Value
Total devices submitted by NBs as potentially within CECP scope	353 (across 13 Notified Bodies)
Devices determined to fall within CECP scope (non-exempted)	36 (10.2%)
Devices exempted from CECP	317 (89.8%)
Most common exemption reason	Modification of a device already marketed by the same manufacturer for the same intended purpose, with modifications not adversely affecting the benefit-risk ratio (314 of 317 exemptions, 99.1%)
Expert panel opinions issued	4 of 35 non-exempted applications received (11.4%)
Non-exempted devices by class — Class III implantable	34 (94.4%)
Non-exempted devices by class — Class IIb ARMP	2 (5.6%)

The majority of scientific opinions rendered by the expert panels indicated limitations in the currently available and planned clinical evidence, with recommendations to collect additional data. No Notified Bodies submitted justification for a decision to diverge from expert panel recommendations — meaning that when the expert panel identifies a gap, the manufacturer must address it.

What this means for your CER: If your device is a Class III implantable or a Class IIb ARMP device, your CER will likely be reviewed by an expert panel unless it qualifies for the modification exemption. The expert panel focuses overwhelmingly on the clinical evidence domain — 10 out of 10 published scientific opinions addressed the quantity and strength of clinical evidence, while only 3 out of 10 addressed technical or biological aspects. Your CER's clinical data analysis must be exceptionally strong for these device categories.

For standard Class IIb non-implantable devices, the CECP does not apply. This is a meaningful difference in the review pathway.

CER Update Frequency and Ongoing Obligations

Obligation	Class III	Class IIb
CER update frequency	At least annually	At least every 2 years (some NBs expect annually for higher-risk Class IIb devices)
PSUR (Periodic Safety Update Report)	Required; updated at least annually	Required; updated at least annually
PMCF plan	Mandatory; must include specific PMCF studies for most devices	Mandatory; general PMCF activities (literature monitoring, complaint analysis) may be sufficient for lower-risk Class IIb devices, but specific studies may be required when evidence gaps exist
SSCP	Required for all Class III devices	Required only for Class IIb implantable devices

Practical Differences in CER Content Expectations

CER Section	Class III Expectations	Class IIb Expectations
Clinical data volume	Extensive; typically requires clinical investigation data (own or equivalent device). A CER based solely on literature is rarely sufficient.	Literature-based CERs are common and accepted when the clinical history is well-documented and the device is not novel.
Equivalence	Equivalence arguments are scrutinized intensively. The NB must explicitly confirm that reliance on equivalent device data is justified.	Equivalence arguments are still scrutinized but with a somewhat lower threshold — particularly for well-established device types.
Benefit-risk analysis	Must be comprehensive, quantitative, and compare against all available alternatives. The higher the risk, the higher the clinical benefit must be to justify it.	Quantitative analysis expected, but the acceptable evidence base may be broader (e.g., supportive evidence from similar devices).
State of the art	Must be current, comprehensive, and reference the most recent clinical guidelines. Given the expert panel review, outdated guidelines will be flagged.	Must be current, but the depth of the state of the art analysis can be proportionate to the clinical complexity of the device.
Total CER length (typical)	150–400+ pages with extensive appendices	60–150 pages

Tips for Writing a CER That Passes NB Review

After covering the technical requirements, here is practical guidance — drawn from the patterns that distinguish CERs that pass review from those that do not.

1. Write the CEP First, and Follow It

The single most common structural weakness in CERs is a mismatch between the Clinical Evaluation Plan and the Clinical Evaluation Report. The CEP defines the methodology. The CER executes it. If the CEP says you will search MEDLINE, Embase, and Cochrane, but the CER only documents a PubMed search, the NB will flag this. If the CEP defines specific appraisal criteria, but the CER uses different criteria (or none at all), the NB will flag this.

Write the CEP before you begin the evaluation. Follow it. If you need to deviate, document the deviation and justify it.

2. Do Not Ignore Unfavorable Data

This is arguably the most important piece of advice. Notified Body reviewers are experienced. They know the literature. They will conduct spot-checks. If a well-known study reports a complication that your CER omits, the NB will question the objectivity of the entire evaluation.

Include unfavorable data. Discuss it. Explain how the overall body of evidence — including the unfavorable data — still supports an acceptable benefit-risk profile. If it does not, address the gap in the PMCF plan.

3. Be Specific in Your Conclusions

"The clinical evaluation demonstrates that the device is safe and performs as intended" is not a conclusion. It is a meaningless statement that every CER ever written has included.

A proper conclusion looks like this: "Based on 14 clinical studies comprising 3,847 patients, the [device name] demonstrated a primary patency rate of 87.2% at 12 months (range across studies: 82.1%–91.4%), which is consistent with the accepted benchmark of greater than 80% reported in the ESC/ESVS 2024 clinical practice guidelines. The overall major adverse event rate was 4.3% (95% CI: 3.1%–5.8%), which compares favorably to the 6.2% rate reported for the current standard of care (open surgical repair). Three studies reported device migration in a total of 7/1,204 patients (0.58%), which is addressed in the risk management file under hazard H-014 and monitored through the PMCF plan."

That is a conclusion that tells the NB reviewer exactly what the evidence shows.

4. Keep the Literature Search Current

A CER with a literature search that is more than 12 months old at the time of NB review is likely to receive a deficiency. Many NBs expect the search to be within 6 months. Plan your submission timeline accordingly and build in time for a literature search update before submission.

5. Cross-Reference Technical Documentation

The CER does not exist in isolation. It must be consistent with:

• The Instructions for Use (intended purpose, indications, contraindications, warnings)
• The risk management file (hazards, risk mitigations, residual risks)
• The PMCF plan (evidence gaps, planned activities)
• The PMS plan and PMS reports (complaint trends, vigilance data)
• The SSCP (for applicable devices)

Inconsistencies between these documents are among the most common NB findings. Before submitting, systematically cross-check the CER against each related document.

6. Invest in Qualified Evaluators

The evaluator's qualifications are checked. Do not assign the CER to someone who lacks relevant experience simply because they are available. If your internal team lacks the necessary clinical or device-specific expertise, engage external clinical evaluation consultants. The cost of a qualified external evaluator is a fraction of the cost of a deficiency cycle.

7. Use the MDCG 2020-13 CEAR Template as a Mirror

MDCG 2020-13 provides the template that Notified Bodies use when assessing a CER — the Clinical Evaluation Assessment Report (CEAR). Read this document. Understand what questions the NB assessor will ask. Structure your CER to pre-emptively answer them.

If the CEAR template asks "Has the manufacturer identified the state of the art for the relevant medical condition?" and your CER does not include a state of the art section, you will receive a deficiency. The CEAR template is essentially a checklist of NB expectations. Use it.

8. Treat the CER as a Living Document

The CER is not a one-time deliverable. Under the MDR, it is updated throughout the device lifecycle. Build your processes — and your document management system — to support regular CER updates. This means:

• Periodic literature search updates
• Integration of PMCF data as it becomes available
• Revision when the state of the art changes
• Update when the device is modified (new variants, expanded indications, design changes)
• Update following any significant safety signal or field safety corrective action

9. Structure for Readability

NB reviewers assess dozens of CERs. A well-structured, clearly written CER with consistent formatting, logical flow, and clear cross-references is far easier to review — and far more likely to pass — than a dense, poorly organized document. Use:

• Consistent heading hierarchy
• Numbered sections with a table of contents
• Cross-references between sections (e.g., "see Section 6.3 for the equivalence comparison table")
• Summary tables and figures to present complex data
• Appendices for voluminous data (appraisal tables, search outputs) that would interrupt the narrative flow

10. Address Device Variants and Accessories

If your device family includes multiple variants, sizes, configurations, or accessories, the CER must address how the clinical evaluation applies to each. Do not assume the NB will accept that clinical data for one variant applies to all variants without justification. Discuss the clinical relevance of variant differences and justify extrapolation where appropriate.

Common Mistakes to Avoid

Beyond the deficiencies listed above, here are errors that manufacturers make repeatedly:

Treating the CER as a literature review. The CER is an evaluation, not a review. A literature review summarizes publications. A clinical evaluation appraises data, analyses it in the context of the device's intended purpose, and draws conclusions about safety, performance, and benefit-risk.

Copy-pasting from previous CER versions without updating. Legacy CERs that were adequate under the MDD are often recycled with minimal updates for MDR submissions. Notified Bodies can tell. The level of rigor expected under the MDR is qualitatively different from the MDD. A genuine re-evaluation is required.

Omitting Source 1 data. Some manufacturers focus entirely on published literature (Source 2 and Source 3) while neglecting their own post-market data (Source 1). Your complaint data, vigilance reports, and PMCF findings are clinical data. They belong in the CER.

Claiming equivalence without substance. Stating that "Device X is equivalent to Device Y" without a detailed, parameter-by-parameter comparison across all three dimensions is insufficient. Every claimed equivalence point must be supported.

Failing to identify evidence gaps. No clinical evaluation is perfect. There are always limitations — limited follow-up duration, small sample sizes for certain subpopulations, lack of comparative data. Acknowledging these gaps and addressing them through the PMCF plan demonstrates methodological maturity. Failing to acknowledge them suggests either ignorance or concealment — neither of which helps with the NB.

Not involving clinical expertise. A CER written entirely by regulatory affairs professionals without input from clinicians who understand the device's clinical context will lack clinical depth. The analysis of clinical outcomes, the state of the art assessment, and the benefit-risk evaluation all benefit from clinical insight.

Putting It All Together

The clinical evaluation report is the central clinical document in your EU MDR technical documentation. It connects the device description to the clinical evidence, the clinical evidence to the GSPRs, the GSPRs to the benefit-risk profile, and the benefit-risk profile to the PMCF plan that will sustain the evidence base over the device's lifetime.

Writing a CER that passes Notified Body review is not about finding a template and filling in the blanks. It requires:

• A clear plan (CEP) executed with methodological rigor
• A thorough, unbiased literature search across multiple databases
• Systematic appraisal of every data item
• An honest, evidence-based analysis that addresses both favorable and unfavorable data
• Specific, quantified conclusions tied to GSPRs
• Integration with PMCF and the broader post-market surveillance system
• Qualified evaluators with relevant expertise
• Consistency with all other technical documentation

The investment in getting the CER right pays dividends. A robust CER reduces conformity assessment timelines, minimizes deficiency cycles, and — most importantly — provides genuine assurance that the device is safe and performs as intended for the patients who depend on it.