Why this guide, and how this course works
MEDDEV 2.7/1 Rev.4 is not the law — but it is how the law gets audited.
MEDDEV 2.7/1 Revision 4 is a European Commission guidance document. It carries no legal force on its own — only the text of Directives 93/42/EEC (MDD) and 90/385/EEC (AIMDD) is legally binding. But Notified Bodies and Competent Authorities use this exact structure to assess whether your Clinical Evaluation Report demonstrates conformity. If you build your CER around anything other than this structure, you will spend your review cycle translating your document into the reviewer's checklist instead of getting to a decision.
Where this sits today
This course teaches the MEDDEV 2.7/1 Rev.4 model because it remains the clearest teaching structure for the underlying logic of clinical evaluation — a logic the EU MDR (Regulation (EU) 2017/745) carries forward and tightens in Annex XIV. If your engagement is specifically an EU MDR submission, treat this course as the reasoning layer and layer MDR Annex XIV's stricter equivalence and PMCF rules on top.
How the five stages fit together
Clinical evaluation is not one document — it is a five-stage process, and the Clinical Evaluation Report is simply the write-up of that process at Stage 4. Click any stage below to jump straight to its module.
How to use this course
Each module has: (1) what the guidance actually says, (2) what that means in a document you produce, and (3) a worked fragment from a fictional device, BPM-200, a Class IIa automated blood-pressure monitor, so you can see the guidance become sentences on a page. Hit "Explain this module" and I'll talk you through it like a tutor — not read the slide back to you — while you keep your own documents open side-by-side.
Defining scope & writing the Clinical Evaluation Plan
Everything downstream inherits the mistakes made here.
Before identifying a single paper, the manufacturer must define what is actually being evaluated: which products, models, sizes and claims are in scope, and what essential requirements the evaluation must address. This is scoping, and its output is a Clinical Evaluation Plan (CEP) — a document in its own right, separate from the eventual CER.
What must go into the scope, per Section 7
- Device description (name, models, sizes, intended purpose, target population — see Appendix A3)
- Design features needing specific clinical attention (novel materials, invasiveness, medicinal components)
- Whether equivalence to an existing device will be claimed, and to which device
- Cross-reference to the risk management file — which clinical risks remain after mitigation
- Current knowledge / state of the art for the relevant medical field
- Data sources and types to be used (own data vs. literature)
- For CE-marked devices only: any design/labelling changes since the last CER, and new PMS signals
This is the kind of paragraph a Clinical Evaluation Plan should contain — notice it answers the bullet list above in prose, not as a form:
2.1 Device under evaluation The BPM-200 is a Class IIa, non-invasive, automated oscillometric blood pressure monitor intended for home and clinical use by adult patients, operated either by the patient (lay use) or a healthcare professional. It is available in one model with two cuff sizes (22-32 cm, 32-42 cm). 2.2 Route to conformity Conformity with MDD Annex I, Section 3 (performance) and Section 6a (clinical evaluation) will be demonstrated primarily through a critical evaluation of literature relating to an equivalent device, supplemented by the manufacturer's own bench validation against EN ISO 81060-2. No new clinical investigation is planned; the justification for relying on equivalence is developed in Section 4 of this plan and detailed per Appendix A1 of MEDDEV 2.7/1 Rev.4. 2.3 Design features requiring clinical attention The oscillometric algorithm is the sole safety-relevant novel element relative to the equivalent device; cuff, bladder and electrical safety are unchanged. The risk management file (RMF-BPM200-03) identifies inaccurate reading as the principal clinical hazard, carried forward into Section 5 of this plan.
Common gap this course sees in draft CEPs
Writers frequently skip stating which Essential Requirements the evaluation addresses (MDD Annex I §1, §3, §6a for a standard device). Naming them in Stage 0 is what lets Stage 3's analysis map cleanly onto them later — do it now, not while writing the CER's conclusion section under deadline pressure.
Identifying pertinent data
Two channels, one obligation: find the unfavourable data too.
Data comes from two channels. First, everything the manufacturer already holds: pre-market investigations, PMCF studies, vigilance and complaint reports, explant analyses, field safety corrective actions, bench data. Second, published and unpublished literature — on the device itself, an equivalent device, or the state of the art. Section 8.2 is explicit that the search must be objective: it must surface unfavourable results as readily as favourable ones, and the search strategy has to be documented well enough to be reproduced.
| PICO element | BPM-200 example |
|---|---|
| Population | Adult patients requiring non-invasive BP monitoring, home or clinic setting |
| Intervention | Oscillometric automated BP measurement (device under evaluation / equivalent device) |
| Comparator | Manual auscultation (reference standard, EN ISO 81060-2) |
| Outcome | Mean measurement error and SD vs. reference; adverse events from cuff use |
Databases: PubMed/MEDLINE and EMBASE. Date limits: last 10 years plus foundational validation studies for the equivalent device regardless of age. Search terms and Boolean strings, inclusion/exclusion criteria, and the PRISMA-style flow diagram of records screened → included all belong in the accompanying Literature Search Report — not summarised from memory in the CER itself.
Documents this stage should produce
- A Literature Search Protocol (planned before execution — Appendix A5)
- A Literature Search Report (what was actually run, deviations from protocol, results)
- An inventory of manufacturer-held data: PMS/PMCF reports, complaint logs, FSCAs, bench reports
- Full-text copies retained for every document carried into Stage 2 — abstracts are not sufficient for appraisal
Appraisal — quality, relevance, weight
Write the criteria before you look at the data, or the appraisal is not defensible.
Two independent sources of uncertainty exist in every dataset: its methodological quality (is it scientifically valid?) and its relevance (does it actually speak to this device's intended purpose, population, and claims?). Section 9.2 requires that the criteria for judging both be written down in an appraisal plan before appraisal begins, and applied consistently — this is what makes the process auditable rather than a set of after-the-fact opinions.
| Field | Entry |
|---|---|
| Reference | Smith et al. 2021, J Hypertens — validation of [equivalent device] per ISO 81060-2 |
| Methods (summary) | n=85, sequential same-arm comparison against auscultation, two trained observers, blinded to device reading |
| Methodological quality | High — sample size meets ISO 81060-2 minimum (85≥85); adequate blinding; validated reference method |
| Relevance | Direct — same oscillometric algorithm family, same cuff range, adult general population matching intended purpose |
| Weighting assigned | Pivotal — contributes directly to performance requirement (MDD ER3) |
| Limitations noted | Single-centre; excluded patients with arrhythmia — carried to Stage 3 as a residual uncertainty |
Data that cannot carry a performance or safety claim
Per Appendix A6: reports missing basic methods or outcome disclosure, sample sizes too small for statistical significance, uncorrected multiple-subgroup testing, single-arm studies on subjective endpoints without controls, and studies not run to EN ISO 14155 or the Declaration of Helsinki. These can still be used to flag a previously unknown hazard — but never as proof of adequate performance or safety.
Analysis against the Essential Requirements
Four requirements. Every clinical evaluation, without exception, is judged against them.
Stage 3 pulls the appraised evidence together and reaches conclusions on each requirement above. Per Section 10.2, this means: using sound methods (qualitative where quantitative isn't possible, but justifying that choice); checking consistency across data sets; running a comprehensive analysis that names any gap; deciding whether additional clinical investigation is needed; and — critically — determining PMCF needs for whatever residual risk or uncertainty remains.
Across the appraised dataset (n=340 pooled subjects across 4 pivotal studies of the equivalent device), reported adverse events were limited to transient cuff-site discomfort (2.1%) and petechiae in patients on anticoagulant therapy (0.3%), both consistent with the known mode of action of oscillometric cuff inflation and already described as a precaution in the IFU (Section 6, "Patients on anticoagulant or antiplatelet therapy"). No serious adverse events attributable to the device were reported. Applying the sample-size reasoning of MEDDEV 2.7/1 Rev.4 Appendix A7.4, the pooled n of 340 provides >80% power to detect an undesirable side-effect with a true incidence of ≥1%; no side effect at this threshold beyond those disclosed above was observed. Conclusion: the profile of undesirable side-effects is acceptable when weighed against the intended performance, and no additional mitigation beyond the existing IFU precaution is required.
The Appendix A7.4 sample-size table, reused
Appendix A7.4 gives worked numbers: to have an 80% chance of observing at least one event of a side-effect with a true incidence of 10%, you need 15 subjects studied; for 5% incidence, 32 subjects; for 1% incidence, 161 subjects. Quoting this logic against your own dataset size, as above, is exactly the kind of statement a Notified Body reviewer is trained to look for — and its absence is one of the most common CER deficiencies.
Writing the Clinical Evaluation Report
The CER is not a new document. It is the previous four stages, made legible to a stranger.
Section 11 sets one test for the finished report: can it be read and understood by an independent party — a reviewer who was not in the room for any of Stages 0–3? Every statement has to be traceable to either a piece of evidence or an evaluator's stated opinion, and never presented as if it were the other.
Appendix A9 table of contents, condensed
- 1. Summary — the benefit/risk conclusion, for someone who reads nothing else
- 2. Scope — device description, intended purpose, regulatory route, what's new since the last version
- 3. Clinical background / state of the art — the medical condition, alternatives, unmet needs
- 4. Device under evaluation — type of evaluation, equivalence (if claimed), data summaries, appraisal, analysis vs. ER1/ER3/ER6
- 5. Conclusions — compliance statement, residual risks, what PMS/PMCF must track
- 6. Date of the next clinical evaluation, with justification
- 7. Dates and signatures of every evaluator
- 8. Qualification of the evaluators (Section 6.4 requirements)
- 9. References
This clinical evaluation concludes that the BPM-200 automated blood pressure monitor demonstrates conformity with MDD Annex I Essential Requirements 1, 3 and 6, on the basis of demonstrated equivalence to [Equivalent Device X] (Section 4.2) supported by four pivotal validation studies (pooled n=340) and the manufacturer's own bench verification to EN ISO 81060-2. No new clinical investigation is required. One residual uncertainty — measurement accuracy in patients with confirmed arrhythmia, excluded from all pivotal studies — is carried forward as a PMCF objective (Section 5) with data collection targeted for the first 18 months post-launch. The next clinical evaluation update is due no later than [date + 2 years], per Section 6.2.3 of MEDDEV 2.7/1 Rev.4, reflecting the device's well-established technology and low-risk classification.
The one line that trips up otherwise strong CERs
Section 11 requires the report to state clearly which statements are evidence and which are evaluator opinion. Writers frequently blend the two in the conclusions section ("the device is safe and effective" with no cross-reference back to Section 4). Every conclusion sentence in Section 5 should carry an implicit or explicit pointer back to the Section 4 sub-section that supports it.
Demonstrating equivalence
All three characteristics, on one device, or the claim fails.
Equivalence is the most heavily scrutinised claim in a CER because it lets a manufacturer borrow someone else's clinical data. Appendix A1 requires three characteristics to all be satisfied simultaneously, against a single device (you cannot mix-and-match clinical characteristics from one device with technical characteristics from another):
| Characteristic | Requirement | BPM-200 vs. Equivalent Device X |
|---|---|---|
| Clinical | Same condition, purpose, body site, population, comparable performance | Match — same indication, adult population, non-invasive brachial cuff site |
| Technical | Similar design, conditions of use, specs, deployment, operating principle | Match — same oscillometric principle; algorithm coefficients differ, addressed via bench validation, Section 4.2.3 |
| Biological | Same materials contacting the same tissues/fluids | Match — identical cuff/bladder material (nylon-coated PVC), same skin contact |
Every "match" row above needs a supporting sentence in the CER explaining why the difference — here, algorithm coefficients — is not expected to affect clinical performance or safety, not just a claim that it doesn't.
What kills an equivalence argument
Two failure patterns dominate Notified Body rejections: (1) claiming equivalence on clinical and biological grounds while the technical comparison is thin or missing measured specifications side-by-side; (2) insufficient access to the "equivalent" device's own technical file to actually verify the comparison — Appendix A1 explicitly tells reviewers to challenge whether the manufacturer even had that access.
Updating the CER over the device lifecycle
A CER has a due date the way a certificate has an expiry — plan for it in Stage 0, not in a panic before renewal.
The update-frequency rule
- New PMS information that could change the evaluation → update immediately, regardless of schedule
- No such trigger, but the device carries significant risk or is not yet well-established → at least annually
- No such trigger, and the device is well-established with no significant risk → every 2–5 years, with a documented justification for the chosen interval
The BPM-200 uses a well-established oscillometric technology on the market for over 15 years across multiple manufacturers, with no design changes since initial CE-marking and no open PMCF signals. The device is classified IIa (non-invasive, non-implantable), and post-market vigilance data (Section 3, PMS Report 2024-Q4) show a complaint rate of 0.04% with no unresolved trend. On this basis, a 2-year update interval is proposed, consistent with Section 6.2.3 of MEDDEV 2.7/1 Rev.4 for well-established, low-risk technology.
What the Notified Body actually checks
Write your CER as if you already have a copy of the reviewer's checklist — because you do.
The Notified Body produces its own Clinical Evaluation Assessment Report (CEAR), verifying each step of your process — not just your conclusions. Appendix A12.2.1 lists roughly twenty things it must positively conclude on. The five that most often decide whether a first review round closes clean or triggers a deficiency letter:
Five checks worth pre-answering in your own CER
- Did you identify unfavourable data with the same rigor as favourable data, and can that be seen in the search protocol?
- Is every equivalence claim supported by side-by-side technical specifications, not just a narrative assertion?
- Does the IFU's stated intended purpose match exactly what the clinical evidence actually covers — no broader, no narrower?
- Is the PMCF plan specific (what data, how much, by when) rather than a generic promise to "continue monitoring"?
- Are evaluator CVs and declarations of interest attached and current?
Worked example: BPM-200, from CEP to CER
The same fictional device, followed end to end, so you can see the document chain.
This module strings together the fragments from Modules 01–06 into the document chain a real project produces. Use it as a build order when you start your own.
| # | Document | Built from | Feeds into |
|---|---|---|---|
| 1 | Clinical Evaluation Plan (CEP) | Device file, risk management file | Literature search protocol; CER §2 |
| 2 | Literature Search Protocol | CEP's PICO question | Literature Search Report |
| 3 | Literature Search Report | Executed searches | Appraisal table; CER §4.4 |
| 4 | Appraisal table / GHTF-style forms | Full-text papers + appraisal plan | CER §4.5 |
| 5 | Equivalence comparison table | Technical files of both devices | CER §4.2 |
| 6 | Analysis narrative per ER1/ER3/ER6 | Appraisal table + state of the art | CER §4.6 |
| 7 | PMCF Plan | Residual uncertainties from step 6 | CER §5, and the RMF update |
| 8 | Clinical Evaluation Report (CER) | All of the above | Technical documentation / design dossier |
Practice task
Pick a device you already work with (or use BPM-200). Draft just row 1 and row 5 of the table above — a one-paragraph scope statement and a three-row equivalence table. That pair alone reveals whether an equivalence-based route is realistic before you invest in a full literature search.
CER release checklist, based on Appendix A10
Run this before any CER leaves your desk — client-facing or internal.
- Can an independent reviewer follow the report without needing to ask you anything?
- Are all manufacturer-held clinical data mentioned and summarised, not just cited?
- If equivalence is claimed, are all differences disclosed — not only the favourable ones?
- Has the latest PMS/PMCF data been incorporated, dated, and referenced?
- Is the state of the art current, and does the report explain why the benefit/risk profile is acceptable against it?
- Do the conclusions hold for every model, size, indication, target population and duration of use in the IFU — with any exceptions flagged?
- Is conformity to each relevant Essential Requirement (ER1/ER3/ER6 for MDD) explicitly stated, one by one?
- Do the IFU and promotional materials match what the report actually supports?
- Are all residual risks and open questions named, with a PMS/PMCF owner and timeline?
- Is the report dated, and do all evaluators' signatures, CVs and declarations of interest exist and are current?
Use this with clients
This checklist doubles as a pre-submission audit tool for consulting engagements — hand it to a client as a self-check before their document reaches you for review, and you'll spend your billable review hours on substance rather than structure.
Quick check
Five questions. Answers explain the "why," not just the "what."