US companies planning a European launch budget for testing, a notified body, and an authorized representative. The item they underestimate is the Clinical Evaluation Report. It is a document, so it looks cheap. In practice it is the most common reason a CE marking project stalls, because it is the only part of the technical documentation that asks you to prove something about patients rather than the device.
The CER answers one question: is there sufficient clinical evidence that this device, used as intended, achieves its claimed benefits with acceptable residual risk?
Where the CER sits in the technical documentation
Under MDR, clinical evaluation is required for every device in every class, including Class I. The obligation comes from Article 61 and the process from Annex XIV Part A, and the report becomes part of the technical documentation your notified body reviews. Our overview of EU MDR covers the structure.
The CER is not a standalone essay. It must be consistent with three other documents, and reviewers check the seams:
- Your risk management file. If the risk file says a hazard is acceptable because clinical benefit outweighs it, the CER must establish that benefit. Our guide to ISO 14971 risk management explains the linkage.
- Your labeling, because every claim in marketing or the IFU needs clinical support. Claims on a website but nowhere in the CER are a standard deficiency finding.
- Your post-market surveillance system, because the CER must incorporate field data as it accumulates.
How the evaluation is actually performed
The methodology notified bodies expect follows the structure of the MEDDEV guidance on clinical evaluation, adapted to MDR, in four stages.
Stage one, the plan. Before gathering anything, write a Clinical Evaluation Plan stating the intended purpose, target population, the claims to be substantiated, the acceptance criteria for the benefit-risk determination, and the literature search protocol. Writing the plan after the search is a process failure, and it is visible in the dates.
Stage two, identification of data. Three sources feed the report: published literature, your own clinical investigations, and your post-market and complaint data. The literature search must be reproducible, with named databases, search strings, date ranges, and documented inclusion criteria.
Stage three, appraisal. Each item is graded for methodological quality and relevance. A small case series on a different device is not equivalent to a randomized trial on yours, and the appraisal must say so using a stated scoring method.
Stage four, analysis. The evidence is synthesized against the plan's acceptance criteria, producing a conclusion on sufficiency and, where it falls short, a plan to close the gap.
State of the art is where most reports fail
MDR requires the evaluation to demonstrate acceptable performance and safety against the current state of the art. Your report must describe the alternatives clinicians use today, including devices, drugs, and surgical techniques, with their outcomes and complication rates, and position your device against them.
This is where American teams are caught out. A report saying the device met its bench specifications and reported no adverse events answers the wrong question. If existing therapy resolves the condition in most patients with a low complication rate, a new device must be shown at least comparable. This section ages fastest, and a report reusing a search from several years ago is rejected on that basis alone.
Equivalence, and why MDR made it hard
Under the old directives, many manufacturers avoided clinical investigation by claiming equivalence to an existing device and using its literature. MDR narrowed that route to the point where it is rarely viable for anything but a variant of your own product. Equivalence now requires three dimensions at once, with no significant difference in clinical safety and performance:
- Clinical: same condition, intended purpose, body site, similar population, same kind of user.
- Technical: similar design, conditions of use, specifications, deployment method, and principles of operation.
- Biological: same materials contacting the same tissues, for a similar duration, with similar release characteristics.
The decisive obstacle is procedural. For implantable and Class III devices, a manufacturer claiming equivalence to another company's device must hold a contract giving full ongoing access to that manufacturer's technical documentation, and no competitor grants that. In practice equivalence works only against your own predecessor product. Our article on FDA clearance versus CE marking explains why the US predicate route feels similar but behaves very differently.
When you need your own clinical investigation
If literature and equivalence leave a gap, the gap analysis produces a clinical development plan: a full investigation under ISO 14155, or something smaller such as a usability study, a registry, or a focused study on the parameter in question. Implantable and Class III devices generally require an investigation unless narrow exemptions apply. Our overview of clinical trials for medical devices covers what that involves.
PMCF turns the CER into a living document
Post-Market Clinical Follow-up is how real-world data flows back into the clinical evaluation. You write a PMCF plan, execute it through registries, surveys, follow-up studies, or structured complaint analysis, and report results in a PMCF Evaluation Report, which feeds the next CER revision.
| Device class | Typical CER update cadence |
|---|---|
| Class III and implantable | At least annually |
| Class IIa and IIb | Every two to five years, sooner if new risks emerge |
| Class I | When justified by new information |
Any significant complaint trend, field safety action, or change to the state of the art triggers an update regardless of schedule. That is why the CER is best treated as an operating process inside your quality system, connected to the duties in our article on post-market surveillance.
Cost, timeline, and who should write it
For a device with a reasonable literature base, expect a first CER to take three to six months and cost roughly $15,000 to $60,000, with updates a fraction of that. Weak literature or a novel mechanism runs well past that range, and the real cost lands in the clinical data you end up needing. See our breakdown of medical device development cost.
The evaluator must be qualified and their credentials documented. Engineers should not write their own CER. The most useful thing a development team can do is start the clinical evaluation plan early, because it exposes evidence gaps while the design and claims can still change, inside the quality system described in our guide to ISO 13485 requirements.
Projects House supports US manufacturers preparing European technical documentation, including clinical evaluation planning, gap analysis, and the design changes that follow when the evidence does not yet support the claim. Get in touch through our contact form.