The clinical evaluation process under MDR is a defined sequence that begins with scoping the evaluation, runs through a pre-specified clinical evaluation plan (CEP), identifies clinical data from literature, equivalence, and clinical investigations, appraises and analyses that data against the relevant general safety and performance requirements, documents the conclusions in a clinical evaluation report (CER), links the output to the risk file and the post-market surveillance system, sets the post-market clinical follow-up (PMCF) cadence under Annex XIV Part B, and plans the updates that keep the evaluation alive for the life of the device. The steps are prescribed by MDR Article 61 and Annex XIV Part A, and they run in order.

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

TL;DR

  • The clinical evaluation process under MDR Article 61 and Annex XIV Part A is a nine-step sequence: scope, CEP, data sources, search and appraisal, analysis, CER, link to risk and PMS, PMCF cadence, and update plan.
  • Steps run in order. A step skipped early is a nonconformity found late, almost always at Notified Body clinical review.
  • The process is governed by MDR Article 61 and Annex XIV Part A for clinical evaluation and by Annex XIV Part B for PMCF, with MDCG 2020-5 (April 2020) for equivalence and MDCG 2023-7 (December 2023) for Article 61(4)-(6) exemptions on implantable and Class III devices.
  • The clinical evaluation process never finishes. Article 61(3) requires it to be updated throughout the life of the device with PMCF and PMS data, and for Class III and implantable devices the CER is updated at least annually.
  • EN ISO 14155:2020+A11:2024 governs any clinical investigation that feeds the process, and EN ISO 14971:2019+A11:2021 is the risk management standard the clinical evaluation must reference at every stage.

Why the sequence matters

A clinical evaluation that is assembled out of order will not survive Notified Body review. Tibor has seen this pattern repeat across the files he has audited: a literature search runs first, a CER is drafted around whatever the search returned, the CEP is written last to make the data look like it answered a pre-specified plan, and the PMCF plan is pasted in at the end with boilerplate activities nobody funded. Every one of those shortcuts is visible to a trained reviewer, and every one of them generates findings.

The MDR is explicit on sequence. Annex XIV Part A Section 1.1 requires the clinical evaluation to be planned before data is collected. Article 61(2) requires a defined, methodologically sound procedure. Article 61(3) requires the evaluation to be updated with post-market data. The steps below are not a process Felix invented — they are the workflow the Regulation already prescribes, written out in the order a resource-constrained startup can actually execute.

Running the steps in order is also cheaper. Each step feeds the next. A tight scope produces a tight CEP. A tight CEP produces a focused search. A focused search produces a shorter CER. A shorter CER is easier to defend and easier to update. Out-of-order work accumulates rework, and rework on a clinical file is measured in weeks of runway, not hours.

Step 1 — Define the scope of the evaluation

The first step is to write down exactly what the clinical evaluation is evaluating. The inputs are the intended purpose under MDR Article 2(12), the classification under Annex VIII, the device description from the technical file, and the list of clinical claims the manufacturer plans to make on the label, in the IFU, and in promotional material.

The output of Step 1 is a one-page scope statement that names the intended purpose in the exact wording that will appear on the label, the target population with indications and contraindications, the clinical claims and clinical benefits to be substantiated, the relevant general safety and performance requirements from Annex I that require clinical evidence to be supported, and an explicit exclusion of the GSPRs that do not need clinical data because they are covered by bench testing, harmonised standards, or risk management.

The mistake at Step 1 is vagueness. "Adults with chronic condition" is not a scope. "Adults aged 40-75 with stable type 2 diabetes managed in primary care, using the device for continuous glucose monitoring over a 14-day wear period" is a scope. The narrower and more specific the scope, the cheaper every later step becomes.

Step 2 — Write the clinical evaluation plan (CEP)

The CEP is the written plan that governs the clinical evaluation. Its content is prescribed by Annex XIV Part A Section 1.1 and covers the relevant GSPRs requiring clinical data, the intended purpose, the target population with indications and contraindications, the clinical benefits with specified outcome parameters, the methods for examining clinical safety and residual risks, the acceptability criteria for the benefit-risk ratio anchored to the state of the art, treatment of specific components, and — where applicable — a clinical development plan from exploratory investigations through confirmatory investigations to PMCF.

The CEP is written before data is collected. Appraisal criteria, inclusion and exclusion criteria, and acceptance thresholds are fixed in the CEP and not adjusted once the evidence starts coming in. A CEP written after the CER is not a plan; it is a post-hoc justification, and reviewers can tell.

For the section-by-section walkthrough of the CEP, the dedicated post covering Annex XIV Part A Section 1.1 is the reference. Step 2 of the process is complete when the CEP exists, has been reviewed internally, and is under version control alongside the rest of the technical documentation.

Step 3 — Identify the data sources

The CEP specifies which of the three legitimate clinical data sources will be used and what weight each one is expected to carry. The three sources are prescribed by Annex XIV Part A.

Scientific literature. Published peer-reviewed literature on the device, on devices of the same generic group, on the underlying technology, on the clinical condition, and on established measurement methods the device relies on. Literature is often the cheapest source and is under-used by startups who assume clinical evidence means new trials.

Equivalence to an existing device. Clinical data from a device demonstrated to be equivalent under the technical, biological, and clinical dimensions set out in MDCG 2020-5 (April 2020). For implantable and Class III devices, MDCG 2023-7 (December 2023) clarifies the "sufficient levels of access" requirement — typically an ongoing contractual relationship with the manufacturer of the equivalent device, not a one-off data purchase.

Clinical investigation of the device itself. A study conducted on human subjects under EN ISO 14155:2020+A11:2024 and MDR Articles 62-82 and Annex XV. This is the most expensive source and is appropriate where literature and equivalence cannot close the evidence gap.

Step 3 is complete when the CEP names, for each relevant GSPR, which of the three sources will answer the clinical question and why. Running Step 3 honestly is where the Evidence Pass of Subtract to Ship produces its largest savings — cheaper sources are chosen first, and new clinical investigations are scoped only for the gaps.

Step 4 — Search and appraise the data

Step 4 executes the methods specified in the CEP. The literature search runs on the databases, date ranges, and search strings the plan identified, with a PRISMA-style audit trail of records retrieved, excluded, and included. Equivalence analysis compares the device to the claimed equivalent across the technical, biological, and clinical dimensions under MDCG 2020-5 and documents sufficient levels of access where implantable or Class III devices are involved. Clinical investigation data is gathered from studies conducted or reported under EN ISO 14155:2020+A11:2024.

Appraisal is the part Step 4 that startups most often skip. Every included data set must be appraised against the pre-specified criteria in the CEP — methodological quality, quality of results, relevance to the device, relevance to the intended purpose, and the weight assigned to the set. A narrative paragraph that describes the study is not an appraisal. An appraisal is a documented decision against a documented grid, reproducible by a second reader working from the same CEP.

Step 4 is complete when every included data set has a documented appraisal and an audit trail exists from search to final inclusion decision.

Step 5 — Analyse the data against the GSPRs

Step 5 is the intellectual centre of the process. The appraised body of evidence is analysed against the specific Annex I general safety and performance requirements that Step 1 identified as requiring clinical evidence. Each relevant clinical GSPR is traced to the appraised data that supports it, gaps are named explicitly, residual risks are mapped to the risk management file under EN ISO 14971:2019+A11:2021, and the benefit-risk ratio is determined against the pre-specified acceptance criteria from the CEP.

The output of Step 5 is a GSPR traceability table that lists each relevant clinical GSPR in one column and the supporting appraised data in the next. This table is the single most valuable artefact in the entire clinical file for a Notified Body reviewer — it shows, at a glance, which data supports which requirement. A clinical evaluation without this table is an argument the reviewer has to reconstruct themselves, and reviewers do not reconstruct arguments in the manufacturer's favour.

Step 6 — Write the clinical evaluation report (CER)

The CER documents the output of Steps 1-5 in the structure required by Annex XIV Part A. Administrative information, scope, device description and state of the art, explicit reference to the pre-specified CEP, the data identified from literature, equivalence, and clinical investigations, the appraisal of each data set, the analysis against the GSPRs with the traceability table from Step 5, the benefit-risk determination and conclusions, the PMCF plan reference, and the qualifications of the evaluators.

The CER is the document the Notified Body reviews most carefully in the clinical file. It must be consistent with the CEP, with the risk management file, with the PMS plan under Annex III, with the PMCF plan under Annex XIV Part B, and with the label and IFU. A CER that reads well in isolation but contradicts any of these other documents is a finding waiting to be written up.

Step 7 is where the clinical evaluation stops being a standalone document and becomes part of the integrated technical file. Three cross-references are mandatory.

Risk management file under EN ISO 14971:2019+A11:2021. Residual risks identified in the risk file must be consistent with the clinical conclusions in the CER. Clinical data that changes the estimated probability or severity of harm must feed back into the risk file. The two documents reference each other at the specific points where they meet.

Post-market surveillance system under MDR Article 83 and Annex III. The PMS plan describes how post-market data will be collected and assessed; the CER names the clinical uncertainties that the PMS system must address and feeds into the periodic safety update report (PSUR) under Article 86 for Class IIa, IIb, and III devices.

PMCF plan under Annex XIV Part B. The CER identifies the open clinical questions that PMCF will answer, and the PMCF plan defines the activities that will answer them.

Step 7 is complete when the four documents — CER, risk file, PMS plan, PMCF plan — cross-reference each other explicitly and consistently.

Step 8 — Set the PMCF cadence

Annex XIV Part B governs PMCF. Step 8 sets the cadence at which PMCF data is collected, analysed, and fed back into the clinical evaluation. The cadence depends on the risk class, the novelty of the device, and the nature of the clinical claims.

For Class III and implantable devices, Article 61(3) requires the clinical evaluation to be updated at least annually with new clinical data, and the PMCF cadence is aligned with that annual update. For Class IIb devices with significant residual uncertainty, an annual or bi-annual cadence is typical. For Class IIa devices with established technology, a less frequent cadence may be proportionate, documented in the PMS plan. For Class I devices, PMCF is still required under Annex XIV Part B but the activities can be proportionate to the risk — ongoing literature surveillance, complaint analysis, and user feedback are often the core activities.

The test of a PMCF cadence is whether the activities are actually feasible within the resources the company has committed. A PMCF plan that promises post-market studies the company cannot fund is a plan that will fail at the first PMS review cycle.

Step 9 — Plan the updates

Step 9 is the step most startups forget. The clinical evaluation process never finishes — Article 61(3) requires updates throughout the life of the device with PMCF and PMS data. The update plan specifies when the CEP, the CER, the risk file, the PMS plan, and the PMCF plan will be reviewed, who is responsible for each review, what triggers an out-of-cycle update, and how the updates are linked to the PSUR and Eudamed reporting cycles.

Triggers for out-of-cycle updates include significant new clinical data from PMCF, a vigilance event reportable under MDR Articles 87-92, a change to the intended purpose or the IFU, a change to the state of the art, a new version of a harmonised standard relevant to the clinical evidence, and Notified Body findings at surveillance audits.

Step 9 is complete when the update plan exists, names responsible owners, and is integrated into the QMS procedures for design changes, PMS review, and vigilance handling.

The Subtract to Ship angle — run the steps honestly, once

Subtract to Ship applied to this process is simple: run every step, in order, once, at the depth proportionate to the risk class. Do not skip steps. Do not repeat steps because the first pass was rushed. Do not inflate steps with content that does not trace to a specific Annex XIV Part A requirement.

The Evidence Pass of Subtract to Ship lives inside Step 3 and Step 4. Literature first, equivalence under MDCG 2020-5 second, clinical investigations under EN ISO 14155:2020+A11:2024 only for the gaps the cheaper sources cannot close. For a Graz-based company we worked with, running Step 3 honestly — checking harmonised standards and published literature before defaulting to new clinical investigations — saved EUR 400,000-500,000 and 1-1.5 years. The saving came from running Step 3 properly, not from skipping it.

The discipline of the framework is to cut the content that does not belong in each step while keeping every step intact. A shorter CER built from a disciplined process is always stronger than a longer CER built from a process that skipped scoping or appraisal. Reviewers reward methodological rigour, not volume.

Reality Check — Where do you stand on the process?

  1. Can you name the nine steps of the clinical evaluation process in order, and point to the specific MDR article or annex paragraph each one traces to?
  2. Did your clinical evaluation actually start with a written scope statement and a CEP, or did it start with a literature search and work backwards?
  3. Are your appraisal criteria in the CEP word-for-word identical to the ones applied in the CER, or did they drift during the project?
  4. Do you have a GSPR traceability table that maps each relevant Annex I clinical GSPR to the appraised data supporting it?
  5. For equivalence claims, have you run the MDCG 2020-5 three-dimensional analysis and documented sufficient levels of access where the device is implantable or Class III?
  6. For any clinical investigation feeding the evaluation, was the study conducted or reported under EN ISO 14155:2020+A11:2024?
  7. Is your CER cross-referenced to the risk management file, the PMS plan under Annex III, and the PMCF plan under Annex XIV Part B, and do the four documents agree with each other?
  8. Is your PMCF cadence feasible within the resources your company has committed, and does it match the risk class of the device?
  9. Does your update plan name owners, triggers, and review cycles, or is "update annually" the entire plan?
  10. If a Notified Body reviewer walked into the room today, could you defend every step by pointing to a specific MDR article, annex paragraph, or MDCG provision?

Frequently Asked Questions

What is the clinical evaluation process under MDR? It is the defined sequence prescribed by MDR Article 61 and Annex XIV Part A for generating, collecting, appraising, and analysing clinical data to demonstrate conformity with the relevant general safety and performance requirements. The process runs from scoping through a pre-specified CEP, data source selection, search and appraisal, analysis against the GSPRs, documentation in the CER, linkage to risk management and PMS, PMCF cadence setting under Annex XIV Part B, and lifecycle updates.

How many steps are there in the clinical evaluation process? There is no fixed count in the Regulation, but the process breaks cleanly into nine operational steps: define scope, write the CEP, identify data sources, search and appraise, analyse against the GSPRs, write the CER, link to risk and PMS, set the PMCF cadence, and plan updates. Each step maps to specific text in Article 61 or Annex XIV Part A or Part B.

Does the clinical evaluation process ever finish? No. Article 61(3) requires the clinical evaluation to be updated throughout the life of the device with clinical data from the PMCF plan and the post-market surveillance plan. For Class III and implantable devices, the CER is updated at least annually. The process is a lifecycle loop, not a one-off project.

Can the steps be run in parallel to save time? Some steps overlap in practice — the CEP, the risk file, and the PMCF plan are often drafted by the same small team in the same review cycle — but the logical dependencies run in the order above. Writing the CER before the CEP inverts the process and makes the appraisal criteria indefensible. Running Step 4 before Step 2 produces literature searches with no pre-specified inclusion criteria. The steps can be compressed in time but not reordered.

What is the difference between the clinical evaluation process and a clinical investigation? The clinical evaluation process is the overarching workflow defined in Article 61 and Annex XIV Part A. A clinical investigation is one possible data source inside that workflow — a study on human subjects conducted under EN ISO 14155:2020+A11:2024 and MDR Articles 62-82 and Annex XV. Not every clinical evaluation includes a new clinical investigation; every device on the EU market requires a clinical evaluation.

Where does MDCG 2023-7 fit into the process? MDCG 2023-7 (December 2023) clarifies the exemptions under MDR Article 61(4)-(6) from the general requirement to perform pre-market clinical investigations for implantable and Class III devices. It matters at Step 3, when the data sources are selected, for implantable and Class III devices that may qualify for one of the four exemption cases. For devices that are neither implantable nor Class III, Article 61(4)-(6) and MDCG 2023-7 do not drive the decision.

Who in a startup should own the clinical evaluation process? Annex XIV Part A requires the evaluation to be conducted by suitably qualified individuals. In a startup, this usually means a clinical evaluation lead with relevant training and experience, working with the PRRC under Article 15, the risk management owner, and the QMS owner. Ownership cannot be handed wholesale to an external consultant — the manufacturer retains the obligation, and the qualifications of the internal or external team must be documented in the CER.

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 61 (clinical evaluation), Article 83 (post-market surveillance system), Annex I (general safety and performance requirements), Annex III (technical documentation on post-market surveillance), Annex XIV Part A (clinical evaluation), 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. MDCG 2023-7 — Guidance on exemptions from the requirement to perform clinical investigations pursuant to Article 61(4)-(6) MDR and on 'sufficient levels of access' to data needed to justify claims of equivalence, December 2023.
  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 clinical evaluation process is prescribed by the Regulation in a specific order, and running the nine steps honestly — once, proportionate to the risk class — is the difference between a clinical file that survives Notified Body review and one that comes back with findings that cost months of runway.