The clinical evaluation plan (CEP) is the first document of the clinical evaluation process under MDR Article 61 and must be written before any clinical data is collected. Its required content is set out in Annex XIV Part A Section 1.1 and covers: the general safety and performance requirements that need clinical evidence, the intended purpose, the target groups and indications, the intended clinical benefits with specified outcome parameters, the methods for examining clinical safety and residual risks, the acceptability parameters for the benefit-risk ratio, the treatment of specific components, and a clinical development plan where applicable. A CER written without a pre-specified CEP is a CER that will not survive notified body review.

By Tibor Zechmeister and Felix Lenhard. Last updated 10 April 2026.

TL;DR

  • The CEP is mandatory under Annex XIV Part A Section 1.1 of Regulation (EU) 2017/745 and must exist before clinical data is collected and appraised.
  • The required contents of the CEP are listed item-by-item in Annex XIV Part A Section 1.1 and are not optional — every bullet must be addressed, scaled to the risk and novelty of the device.
  • The CEP is proportionate to risk class: a Class I device with an established measurement principle has a shorter CEP than a novel Class III implantable, but both must cover the same structural items.
  • The CEP links directly to the risk management file under EN ISO 14971:2019+A11:2021 and to the PMS system under MDR Article 83 — the three documents are written together, not in isolation.
  • The most common CEP gaps at notified body review are vague intended purpose, missing acceptability criteria for the benefit-risk ratio, reverse-engineered appraisal criteria, and an absent clinical development plan.
  • MDCG 2020-5 (April 2020) and the legacy MEDDEV 2.7/1 Rev 4 (June 2016) are useful structural references, but Annex XIV Part A Section 1.1 is the binding text.

Why the CEP exists at all

A clinical evaluation that starts with data collection and ends with a report is a clinical evaluation that got the order wrong. The regulation is explicit on this point. Annex XIV Part A Section 1.1 requires the manufacturer to plan, conduct, and document a clinical evaluation in accordance with a defined methodology — and the plan comes first.

The reason is scientific, not bureaucratic. If the appraisal criteria are written after the data has been read, the reader has already decided which studies they want to include. The criteria will — consciously or not — fit the data. A clinical evaluation built this way cannot answer the question it is supposed to answer, because the question was chosen to match the answer. The CEP protects against this by fixing the rules of the game before the game is played.

A second reason is operational. Clinical evaluation touches the intended purpose, the GSPRs in Annex I, the risk management file, the technical documentation, the PMS plan, and the PMCF plan under Annex XIV Part B. If the CEP does not specify how all of these connect, the downstream documents drift apart. Tibor has seen clinical evaluation reports whose intended purpose quietly differs from the IFU, whose benefit-risk criteria do not match the risk file, and whose PMCF plan has no visible link back to the clinical questions the CEP was supposed to raise. Every one of those gaps is a finding at notified body review. Every one of those gaps is prevented by a CEP written in the right order.

What the CEP is supposed to do in one sentence

The CEP defines — before any data is collected — what clinical questions the clinical evaluation must answer, what data sources will be used to answer them, what criteria will be applied to judge the data, and what the acceptance thresholds are for concluding that the benefit-risk ratio is acceptable for the intended purpose.

That is the whole job. Everything in Annex XIV Part A Section 1.1 is a specification of how to do that job properly.

Annex XIV Part A Section 1.1 — the required contents, section by section

Annex XIV Part A Section 1.1 lists the items the CEP shall include. The following walkthrough takes each required item and says what a compliant CEP looks like on that item.

1. The relevant general safety and performance requirements that need clinical data

The CEP identifies which GSPRs in Annex I require clinical data to be supported. Not every GSPR does. Some are supported by bench testing, some by compliance with harmonised standards, some by risk management. The CEP lists the specific GSPRs that genuinely need clinical evidence and explains why. This list drives the scope of the entire evaluation.

A CEP that skips this step produces a clinical evaluation that is either too broad (wasting work on GSPRs that do not need clinical data) or too narrow (missing GSPRs that do). Tibor's rule of thumb is: if a reviewer cannot read the CEP and map each clinical question to a specific GSPR, the CEP is not doing its job.

2. A specification of the intended purpose

The intended purpose is defined in Article 2(12) as the use for which the device is intended according to the data supplied by the manufacturer on the label, in the IFU, in promotional or sales materials, and as specified by the manufacturer in the clinical evaluation. The CEP writes this down in one or two sentences, in language that is identical to the label and the IFU.

If the intended purpose in the CEP differs from the intended purpose on the label, the clinical evaluation is evaluating a different device than the one being placed on the market. This is the most common single error in startup CEPs, and the fix is cheap at the CEP stage and very expensive later.

3. The intended target groups with clear indications and contraindications

The CEP specifies the patient populations for which the device is intended, the clinical indications, and the contraindications. "Adults with condition X" is not enough on its own — the CEP should name the age range, the disease stage, any excluded populations, and the clinical setting in which the device is used. The specification must match the IFU.

4. A detailed description of the intended clinical benefits with specified outcome parameters

This is the item startups most often hand-wave. The CEP must describe the clinical benefits the device is intended to deliver and specify the clinical outcome parameters by which those benefits will be measured. "Improved patient outcomes" is not a clinical benefit. "A reduction in wound healing time of at least X days in population Y, measured by Z assessment at day N" is a clinical benefit with a specified outcome parameter.

The outcome parameters chosen here will be the parameters the literature search, the equivalence assessment, and any clinical investigation are evaluated against. Vague parameters produce vague evidence. Specific parameters produce evidence that either passes or fails a visible threshold.

5. Methods for examining clinical safety and residual risks

The CEP specifies how clinical safety is examined — what methods will be used to identify, characterise, and evaluate the residual risks remaining after risk control measures under EN ISO 14971:2019+A11:2021. This is the link point between the clinical evaluation and the risk management file. A CEP that does not reference the risk file is a CEP that is floating free of the safety argument for the device.

6. Acceptability criteria for the benefit-risk ratio, based on the state of the art

The CEP specifies, in advance, what acceptability thresholds will be applied to the benefit-risk ratio for each indication and for the intended purpose. The thresholds are anchored to the state of the art, defined in Article 2(48) as the developed stage of current technical capability and accepted clinical practice. The CEP names the sources used to define the state of the art and the specific benchmark values against which the device will be judged.

Reviewers scrutinise this item closely. A CEP that declares the benefit-risk ratio acceptable without first specifying what "acceptable" means is a CEP that has no referee.

7. Benefit-risk issues for specific components

Where the device includes pharmaceutical substances, non-viable animal or human tissues, or other components that raise specific benefit-risk questions, the CEP addresses how those questions will be handled. For most startup devices this item is a short paragraph confirming non-applicability. For devices that do include such components, the CEP specifies the evaluation pathway.

8. A clinical development plan where applicable

For devices that require it, the CEP includes a clinical development plan that indicates the progression from exploratory investigations (first-in-man studies, feasibility and pilot studies) through confirmatory investigations (pivotal clinical investigations) to a PMCF plan as referred to in Annex XIV Part B. The clinical development plan identifies milestones and describes potential acceptance criteria.

The clinical development plan is not a marketing roadmap. It is a clinical evidence progression that the notified body can follow from first contact with patients to post-market follow-up. For a Class I device using well-established measurement methods, this item may be short. For a novel Class III implantable, it is a substantial section. Either way, it is required where applicable and cannot be reduced to a single sentence.

Proportionality to risk class — what "scaled to risk" actually means

The CEP must be proportionate to the risk and novelty of the device. Proportionality is not a permission slip to skip items — every item in Annex XIV Part A Section 1.1 must be addressed for every device. Proportionality changes the depth, not the coverage.

For a Class I device based on a well-established measurement principle with extensive published literature, a compliant CEP can be short: the GSPR list is focused, the intended purpose is narrow, the state of the art is well documented in the literature, the clinical development plan is minimal, and the benefit-risk acceptability thresholds are drawn directly from established clinical practice.

For a novel Class III implantable, every one of those items expands. The GSPR list is longer. The intended purpose description interacts with complex anatomy and population specifics. The state of the art requires active comparison with competing treatments. The clinical development plan spans exploratory, pivotal, and PMCF phases with quantitative milestones. The benefit-risk acceptability thresholds must be defended against alternative treatments.

The error startups make in both directions is the same: they write the CEP for the wrong class. A Class I startup writes a Class III CEP because the template they bought was built for that — the Berlin template trap. A Class IIb startup writes a Class I CEP because they are under-resourced and wishing the device away. The fix in both cases is the same: know your class, read Annex XIV Part A Section 1.1, and write exactly what is required for that class.

Linking the CEP to risk management and post-market surveillance

The CEP is not a standalone document. It links to two other documents that must be written in parallel.

The risk management file under EN ISO 14971:2019+A11:2021. The CEP references the risk management file for the identification of hazards, the residual risks after control measures, and the benefit-risk analysis at the risk management level. The clinical evaluation examines whether the residual risks are acceptable in light of the clinical benefits. If the risk file says one thing about residual risk and the CEP says another, one of the two is wrong. The cheapest time to catch that is when both documents are being written, not at notified body review.

The PMS system under MDR Article 83 and the PMCF plan under Annex XIV Part B. The CEP identifies the clinical questions that will remain open after pre-market clinical evaluation — the residual uncertainties, the longer-term safety and performance questions, the rare adverse events that a pre-market evidence base cannot capture. Those questions become the inputs to the PMCF plan. A CEP that does not name the open questions produces a PMCF plan that has nothing specific to monitor, and a post-market system that cannot feed meaningful data back into the next CER update.

The three documents — CEP, risk file, PMCF plan — are written together. They reference each other. They are updated together throughout the life of the device.

Common CEP gaps Tibor sees at notified body review

Across a large number of clinical evaluation files reviewed as both a consultant and a notified body lead auditor, the same CEP gaps repeat:

  • Intended purpose that drifts. The CEP says one thing, the IFU implies another, the marketing material implies a third. The clinical evaluation then covers a patchwork that matches none of them.
  • Benefit-risk acceptability criteria that are missing or circular. The CEP either fails to specify acceptance thresholds at all, or the thresholds are defined in terms of whatever the evidence happens to show.
  • Appraisal criteria retrofitted to the data. The inclusion and exclusion criteria for the literature review are suspiciously convenient. A reviewer can tell.
  • State of the art undefined. The CEP declares the benefit-risk ratio acceptable without anchoring the judgment in current accepted practice.
  • No clinical development plan where one is required. Section 1.1 requires it where applicable, and "applicable" is broader than most startups realise.
  • No link to the risk management file. The CEP and the risk file look like they were written by two different teams who never spoke.
  • No link to the PMCF plan. The open clinical questions are not identified in the CEP, so the PMCF plan is generic.
  • Outcome parameters for clinical benefits that cannot be measured. "Improved quality of life" with no instrument and no threshold.

Every one of these is fixable at the CEP drafting stage at a cost measured in hours. Left to notified body review, each one is measured in weeks or months of rework.

The Subtract to Ship angle — subtract inside the items, not from the list

Subtract to Ship applied to the CEP is strict: nothing on the Annex XIV Part A Section 1.1 list can be subtracted. Every item must be addressed. What can be subtracted is the content inside each item that does not trace to a specific requirement of the Regulation, the device, or the intended purpose.

Subtract the GSPRs that do not genuinely need clinical data from the GSPR list — but only after honestly checking which ones do. Subtract the indications the device is not actually marketed for — but only after aligning with the IFU. Subtract the outcome parameters that are not linked to a specific clinical benefit claim — but only after the claims themselves are tight. Subtract the state-of-the-art comparison with treatments that are not relevant — but only after defining the actual clinical setting.

The Evidence Pass of the Subtract to Ship framework (described in the framework pillar post) runs on the CEP first. The cheaper evidence pathway can only be chosen if the CEP is clear about what questions the evidence has to answer. A vague CEP locks a startup into the most expensive pathway by default, because there is no way to argue that cheaper sources are sufficient for a question that was never precisely asked.

Reality Check — Where do you stand on your CEP?

  1. Does your CEP exist as a written document before any clinical data collection began, or was it assembled after the CER was already drafted?
  2. Does your CEP cover every item listed in Annex XIV Part A Section 1.1, or are any of the items missing or reduced to a single sentence?
  3. Is the intended purpose in your CEP identical to the intended purpose on the label, in the IFU, and in the marketing material?
  4. Have you listed the specific GSPRs in Annex I that require clinical data for your device, and excluded the ones that do not?
  5. Are your clinical benefit claims specified with measurable outcome parameters, or are they phrased in general language that cannot be evaluated?
  6. Have you defined the state of the art for your device's clinical application, with cited sources?
  7. Are your benefit-risk acceptability thresholds pre-specified and anchored to the state of the art, or are they implicit?
  8. Does your CEP reference the risk management file under EN ISO 14971:2019+A11:2021 at the specific points where the two documents meet?
  9. Does your CEP identify the open clinical questions that will feed the PMCF plan under Annex XIV Part B?
  10. If a clinical development plan is applicable to your device, is it written with milestones and acceptance criteria, or is it a single paragraph of good intentions?

Frequently Asked Questions

What is a clinical evaluation plan (CEP) under MDR? The CEP is the planning document that governs the clinical evaluation process under MDR Article 61. Its required content is defined in Annex XIV Part A Section 1.1 and covers the clinical questions to be answered, the data sources, the appraisal methodology, the acceptability criteria for the benefit-risk ratio, and — where applicable — the clinical development plan. The CEP must be written before clinical data collection begins.

Is a CEP mandatory for every medical device? Yes. Annex XIV Part A Section 1.1 requires a documented clinical evaluation plan for every device subject to clinical evaluation under Article 61. The requirement applies regardless of risk class. The depth and length of the CEP scale with the risk and novelty of the device, but the existence of a written plan is not optional.

How long should a CEP be? There is no word count in the Regulation. A CEP for a Class I device with an established measurement principle and strong published literature can be relatively short. A CEP for a novel Class III implantable is substantial. The length is driven by the number of GSPRs requiring clinical evidence, the complexity of the intended purpose, the state of the art, and the clinical development plan.

What is the difference between the CEP and the CER? The CEP is written before data collection and specifies the methodology, the questions, the sources, and the acceptance criteria. The CER is written after the data has been collected, appraised, and analysed and documents the results and the conclusions against the pre-specified CEP. The CEP is the plan; the CER is the output. Both are required and both are updated throughout the life of the device.

Can I use MEDDEV 2.7/1 Rev 4 as a template for my CEP under MDR? MEDDEV 2.7/1 Rev 4 (June 2016) is legacy guidance written under the former Directives and is useful as a structural reference for the stages of clinical evaluation. Where it diverges from the MDR text or from MDCG 2020-5, the MDR text and the MDCG guidance take precedence. For the CEP specifically, Annex XIV Part A Section 1.1 is the binding text and must be addressed item-by-item regardless of the template used.

How often should the CEP be updated? The CEP is updated whenever the intended purpose, the clinical questions, the data sources, or the acceptability criteria change, and at the cadence required by the PMS and PMCF plans. For Class III and implantable devices, the clinical evaluation is updated at least annually per MDR Article 61(3), and the CEP is reviewed in sync with each update.

Does the CEP need to be approved by the notified body before data collection? The CEP is part of the technical documentation the notified body reviews during conformity assessment. There is no separate pre-approval step for the CEP itself, but the notified body will review the CEP alongside the CER during conformity assessment. A CEP that was written after the CER — or that does not cover Annex XIV Part A Section 1.1 item-by-item — generates findings at that review.

Sources

  1. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, Article 2(12) (intended purpose), Article 2(44) (clinical evaluation), Article 2(48) (state of the art), Article 61 (clinical evaluation), Article 83 (post-market surveillance system), Annex I (general safety and performance requirements), Annex XIV Part A Section 1.1 (clinical evaluation plan), Annex XIV Part B (post-market clinical follow-up). Official Journal L 117, 5.5.2017.
  2. MDCG 2020-5 — Clinical Evaluation — Equivalence: A guide for manufacturers and notified bodies, April 2020.
  3. MEDDEV 2.7/1 revision 4 — Clinical Evaluation: A Guide for Manufacturers and Notified Bodies under Directives 93/42/EEC and 90/385/EEC, June 2016 (legacy guidance; MDR text and MDCG 2020-5 take precedence where they diverge).
  4. EN ISO 14155:2020 + A11:2024 — Clinical investigation of medical devices for human subjects — Good clinical practice.
  5. EN ISO 14971:2019 + A11:2021 — Medical devices — Application of risk management to medical devices.

This post is part of the Clinical Evaluation & Clinical Investigations cluster in the Subtract to Ship: MDR blog. Authored by Felix Lenhard and Tibor Zechmeister. The CEP is the document where the clinical evaluation strategy is either set up to succeed or set up to fail — writing it properly, in the right order, is the highest-leverage hour in the entire Cat 3 workflow.