A clinical evaluation report (CER) is the document in which a medical device manufacturer shows that its clinical evidence is enough to support the device's safety, performance and benefit-risk for its intended purpose. Under the EU Medical Device Regulation, Regulation (EU) 2017/745, every device needs a clinical evaluation, and the CER is how that evaluation is documented in the technical file.

It is also the document most likely to come back from a notified body with questions. This guide walks through how a CER is built, which guidance documents shape it, and where reviewers usually find gaps.

The rules that apply

Three layers of text govern a CER.

  • The MDR itself. Article 61 sets the requirement for clinical evaluation, and Annex XIV Part A sets out how it is planned and carried out. Part B of the same annex covers post-market clinical follow-up (PMCF).
  • MEDDEV 2.7/1 Revision 4. Written in 2016 for the older Medical Device Directives, MEDDEV 2.7/1 Rev 4 is still the most detailed description of the clinical evaluation process, and later MDCG guidance refers back to parts of it. Where it refers to the Directives' Essential Requirements, read the MDR's General Safety and Performance Requirements instead.
  • MDCG guidance. The two most used for CER writing are MDCG 2020-5 on equivalence, and MDCG 2020-13, the template notified bodies use to write their clinical evaluation assessment report (CEAR). MDCG documents are guidance, not law, but notified bodies work from them.

The practical consequence of that last point: write the CER so it answers the CEAR. MDCG 2020-13 is, in effect, the notified body's checklist, and every section of it maps to something your CER has to show.

The five stages

MEDDEV 2.7/1 Rev 4 describes the clinical evaluation as a sequence of stages. They are still the clearest way to organise the work.

StageWhat happensWhat it produces
0Define the scope and plan the evaluationClinical evaluation plan (CEP)
1Identify pertinent dataLiterature search protocol and results; manufacturer's own data
2Appraise each data set for validity, relevance and weightAppraisal tables
3Analyse the data and reach conclusionsAnalysis of safety, performance and benefit-risk
4Write the clinical evaluation reportThe CER

A good CER makes each stage visible, so the reviewer can follow the chain from question to evidence to conclusion.

Stage 0: the clinical evaluation plan

Under the MDR the plan is not optional. MDCG 2020-13 asks the notified body to confirm that the manufacturer's plan "meets the requirements of Annex XIV Part A Section 1a." In practice the plan should set out:

  • the device, its intended purpose, target population, indications and contraindications
  • the General Safety and Performance Requirements that need clinical data
  • the intended clinical benefits, with outcome parameters you can measure
  • the methods for examining clinical safety, including residual risks and side effects
  • the parameters you will use to judge whether the benefit-risk is acceptable, based on the state of the art
  • the clinical development plan and the PMCF plan, with milestones

The plan is where most CERs are won or lost. A benefit stated vaguely ("improves outcomes") cannot be shown in Stage 3. A benefit stated measurably ("reduces time to haemostasis compared with manual compression") can.

Stage 1: identifying the data

Clinical data comes from two places: the manufacturer's own data (clinical investigations, PMS and vigilance data, complaints, registries) and the published literature.

The literature search has to be reproducible. Write a search protocol before you search: databases (at least one major biomedical database, usually more), search strings, date limits, inclusion and exclusion criteria, and the reason for each. Then report what you found, what you excluded and why. A reviewer should be able to re-run your search and get the same result.

You will usually run two searches: one for data on your device (or the equivalent device), and one for the state of the art, meaning current treatment options, accepted benchmarks and the safety and performance of alternatives. The state of the art is what your acceptability criteria are measured against, so it cannot be an afterthought.

Stage 2: appraisal

Every data set gets appraised for methodological quality, relevance to your device and intended purpose, and the weight it should carry. MEDDEV 2.7/1 Rev 4 gives example appraisal criteria. Whatever scheme you use, apply it consistently and show the result in a table. A paper you exclude after appraisal should stay visible with the reason, not silently disappear.

Stage 3: analysis

This is where you answer the questions the plan set. For each intended benefit and each safety question: what the appraised data show, how strong the data are, and how the result compares with the state of the art. Then the overall conclusions:

  • whether the relevant General Safety and Performance Requirements are met
  • whether the benefit-risk is acceptable
  • whether the information supplied with the device (IFU, labelling, and the Summary of Safety and Clinical Performance where required) matches the evidence
  • which residual risks and open questions remain, and how PMCF will address them

Stage 4: the report

The CER pulls this together. It should be readable on its own, cite every data source, and be signed by evaluators whose qualifications are documented. MDCG 2020-13 includes a check of "clinical evaluation report authors", so attach CVs or a statement of qualifications.

Equivalence

Claiming equivalence to another device lets you use that device's clinical data. Under the MDR this route is narrower than it used to be.

MDCG 2020-5 sets out how equivalence must be shown, across technical, biological and clinical characteristics, with a comparison table for each. The notified body template asks directly: "Are the devices equivalent in accordance with Section 3 of Annex XIV including technical, biological and clinical characteristics?"

Two points catch most manufacturers:

  • Access to data. For implantable and class III devices, claiming equivalence to another manufacturer's device requires a contract that gives you ongoing access to that device's technical documentation (Article 61(5)).
  • Clinical investigations. For implantable and class III devices, clinical investigations are expected unless one of the Article 61 exemptions applies. MDCG 2020-13 asks whether such a device lacks investigations "in accordance with Article 61(4)" and, if so, expects the PMCF plan to include post-market clinical studies.

If you claim equivalence, put the full three-part comparison in the CER, not in an annex the reviewer has to hunt for.

PMCF and keeping the CER current

A CER is not finished at certification. The MDR requires the clinical evaluation to be updated throughout the device's life with data from post-market surveillance and PMCF. MEDDEV 2.7/1 Rev 4 gives a practical rule for frequency: when new PMS information could change the evaluation, and otherwise "at least annually if the device carries significant risks or is not yet well established," or every 2 to 5 years for well-established, lower-risk devices, with a justification.

Write the CER so the next update is easy: keep the search protocol, appraisal tables and benefit-risk parameters in a form you can re-run, and record what changed since the last version.

Where notified bodies usually push back

  • Benefits stated in terms that cannot be measured.
  • A state-of-the-art section that describes the disease but not the alternatives' safety and performance, so there is nothing to compare against.
  • A literature search that cannot be reproduced.
  • Equivalence claimed on technical characteristics alone.
  • IFU or SSCP claims that the clinical data do not support.
  • No PMCF, or a PMCF plan that does not address the residual risks named in the CER.
  • An "updated" CER that does not reflect the latest PMS and PMCF reports. MDCG 2020-13 asks the reviewer to "verify that this update corresponds to the most recently updated PMS/PMCF reports."

Where automation helps

Most of the time in a CER goes on the literature: screening hundreds of abstracts, extracting data into appraisal tables, and keeping the tables consistent with the text. Asthra's CER authoring drafts the CER from your device documentation, search results and PMS data, with every statement cited to its source, so the clinical evaluator spends the time on appraisal and benefit-risk judgement. The evaluator still signs the report and owns the conclusions.

FAQ

Is MEDDEV 2.7/1 Rev 4 still relevant under the MDR? Yes, as process guidance. It was written for the Directives, but its stages and appraisal approach are still widely used, and MDCG guidance refers to parts of it. The legal requirements now come from the MDR, Article 61 and Annex XIV.

What is the difference between a CER and a CEAR? The CER is written by the manufacturer. The CEAR is the notified body's assessment of it, written using the MDCG 2020-13 template.

How often does a CER need updating? Throughout the device's life, whenever PMS data could change the conclusions. MEDDEV 2.7/1 Rev 4 suggests at least annually for higher-risk or less established devices, and every 2 to 5 years for well-established, lower-risk ones.

Can I rely only on equivalence for a class III device? Rarely. Article 61 expects clinical investigations for implantable and class III devices, with limited exemptions, and claiming equivalence to another manufacturer's device needs a contract giving ongoing access to its technical documentation.

Sources