Your first job is to distinguish evidence that exists but is hard to find from evidence that does not support the conclusion. That distinction determines whether you need a clearer reference, a new assessment or additional evidence generation. It also stops a narrow review question from becoming an unnecessary rewrite of the whole technical file.
This guide concerns technical-review findings for devices under the EU MDR. It follows the evidence across the clinical evaluation report, or CER, risk management and post-market surveillance, or PMS. It does not treat an IVDR performance evaluation as an interchangeable document.
What does the finding actually challenge?
Read the finding as a question about a specific conclusion. Before assigning work, write down the device configuration, intended purpose, claim and document revision to which the question applies. Then ask what the reviewer would need to see to assess that conclusion.
A request to substantiate a claimed patient population could mean that the reviewer cannot locate your subgroup analysis. It could mean that the analysis excludes part of the claimed population. It could also mean that you cite relevant studies without explaining whether their results apply to your device. Those are different problems, even if the response letter uses similar wording.
Use this decision table before deciding that the CER needs more prose.
| What your check establishes | What the problem means | Appropriate next work |
|---|---|---|
| The approved evidence directly supports the claim, but the submission cites the wrong section | A reference or presentation defect | Correct the reference and explain how the cited result answers the question |
| Relevant evidence exists, but nobody has assessed its applicability to this configuration or population | An assessment gap | Appraise the evidence and document the reasoning before changing the conclusion |
| The evidence excludes the population or does not measure the claimed performance | An evidence gap | Define what evidence is needed and assess the consequences for the claim |
| The finding could refer to two different claims or document versions | An unresolved scope question | Seek focused clarification while investigating both plausible interpretations |
These categories are working diagnoses, not notified-body classifications. More than one can apply to the same finding. Correcting a broken reference may expose an unsupported conclusion that the original reference had concealed.
A useful clarification request identifies the ambiguity precisely. For example, ask whether the question concerns the absence of population-specific data or the explanation of applicability in a named CER section. Simply asking what the reviewer wants leaves the technical issue unresolved.
How do you trace the effect across the file?
Start with the challenged claim and follow its dependencies in both directions. Upstream, identify the evidence and assumptions that support it. Downstream, identify the risk conclusions, user information and monitoring activities that rely on it.
The legal basis for this connected approach appears in the MDR's clinical-evaluation and PMS provisions. Article 61 in the adopted text ties clinical evidence to the device's characteristics and intended purpose, and paragraph 11 describes lifecycle updating using PMCF and PMS data. Article 83 in the adopted text describes using PMS data to update clinical evaluation, risk management, user information and technical documentation. These links reproduce the original provisions, rather than a current EU consolidation.
For your working map, record an exact document identifier, revision and section. “See risk file” is not enough to establish which conclusion you have checked. Record unchanged documents too, with the reason no revision is necessary. A dependency review should produce justified decisions, not automatic edits to every file.
What does a connected remediation case look like?
The following case is illustrative. The device, finding, document identifiers and evidence are invented to demonstrate the method, not drawn from a customer file or an accepted response.
You submit a reusable handheld device intended to measure a physiological parameter in adults and adolescents. The notified body asks how the clinical evaluation supports adolescent use. Your CER claims both populations, but its central performance study enrolled adults only. A separate bench report supports measurement performance under specified test conditions. Your instructions for use include adolescents, while the PMS summary combines all users without recording age groups.
You cannot resolve this by pointing to the bench report and repeating that the device works the same way in both groups. You first need a reasoned assessment of the relevance of the available evidence, the characteristics that could affect performance and the remaining uncertainty. The adult study might contribute useful evidence without supporting the whole claim on its own.
Here is a populated dependency map for that investigation. The revisions shown are the starting records, not completed changes.
| Finding or question | Evidence actually available | Missing work and linked decisions | Owner and internal verification |
|---|---|---|---|
| Does the CER substantiate adolescent use? | CER-01 revision 3 section 7 claims both populations. Study CL-04 revision 1 includes adults only | Clinical owner assesses applicability and identifies any additional evidence needed. RA checks the consequences for the intended population in IFU-01 revision 4 section 2 | Clinical and RA. Verify that each retained population claim has an explicit, supported conclusion |
| Does the Article 61(10) rationale explain reliance on non-clinical methods for the challenged claim? | Rationale NC-02 revision 1 cites bench report BT-07 revision 2 but does not address the population difference | Reassess whether the rationale applies to this question at all. Examine body interaction, intended performance, claims and risk results. Do not use lack of adolescent data as the justification | Clinical and RA. Verify applicability before evaluating sufficiency, with paragraph 4 considered separately |
| Does the risk-control evidence cover the claimed use conditions? | Risk record RM-01 revision 5 entry H-12 relies on a positioning instruction. Usability report UE-03 revision 1 does not evaluate the adolescent use scenario | Assess whether the evidence verifies the control for the relevant users and conditions. Decide whether further evaluation or a different control is necessary | Engineering, usability and risk owners. Trace the control to its evidence and residual-risk conclusion |
| Has PMS information informed the linked conclusions? | PMS-01 revision 2 section 4 aggregates reports without age-group information | Assess what the existing data can establish and improve future collection where justified. Review effects on the CER, risk record and PMCF planning | PMS and clinical owners. Confirm that missing subgroup information remains visible in the analysis |
The map reveals four different tasks. The clinical question needs an evidence assessment. The non-clinical rationale needs an applicability check. The risk-control question needs verification of coverage. The PMS issue needs an honest account of what the available data can show.
Do not assign all four to a writer with an instruction to make the documents consistent. Consistent wording can conceal a shared unsupported assumption just as easily as it can clarify a sound conclusion.
Should you narrow the intended population?
Changing the claim is a decision to assess, not an automatic shortcut. In the illustrative case, you might investigate whether the available evidence supports a narrower intended population. That requires coordinated clinical, regulatory and product review, including the consequences for labelling, risk management, affected configurations and any devices already supplied.
Alternatively, you might retain the proposed population and determine that additional evidence is necessary. The appropriate work depends on the device and the uncertainty. A new clinical investigation is not the inevitable answer to every clinical-evidence finding, but a stronger paragraph cannot substitute for evidence the conclusion requires.
If the assessment remains incomplete, describe the conclusion as unresolved. A planned study and an approved report establish different things.
What does Article 61(10) need to substantiate?
Article 61(10) is not a general exemption from clinical evaluation because clinical data are scarce. Its adopted wording addresses situations where demonstrating conformity with general safety and performance requirements using clinical data is not deemed appropriate, and requires an adequate justification for that exception.
The provision identifies the manufacturer's risk-management results, the specifics of device interaction with the human body, intended clinical performance and manufacturer claims. It also requires substantiation of why non-clinical testing methods alone are adequate. Its opening qualification preserves paragraph 4, so do not apply it without considering the separate requirements that paragraph addresses.
In practical terms, a rationale must explain why the chosen evidence can establish the relevant safety and performance conclusions. Naming a bench test is insufficient if you do not connect its conditions and acceptance criteria to the claim, address its limitations and explain the residual uncertainty.
Consider this illustrative diagnostic comparison, not a model justification for any device.
| Weak statement | The question your assessment must answer |
|---|---|
| There are no published adolescent studies, so Article 61(10) applies | Why is demonstration based on clinical data not appropriate, rather than merely unavailable? |
| The device passes bench testing | Which claim does each result support, and how do the test conditions represent the relevant use? |
| The device has low risk | What do the actual risk-management results establish about the adequacy of non-clinical evidence? |
| The adult and adolescent configurations are identical | Could differences in body interaction or use affect the claimed performance despite an identical design? |
For the worked case, the available facts do not establish a defensible Article 61(10) conclusion. The correct output is an assigned assessment with explicit questions, not a completed rationale manufactured from assumptions. Check the applicable EU text and device-specific requirements before relying on this route.
How should PMS information change the response?
PMS should test and update your conclusions, not sit beside them as an unrelated report. Where a finding exposes uncertainty about a population, configuration or use condition, check whether your surveillance data can distinguish that group and whether your clinical and risk assessments reflect the limitations.
In the illustrative case, the aggregate PMS summary cannot establish an adolescent-specific experience because it does not identify adolescent use. That limitation does not prove a safety problem. It also does not support a claim that no population-specific problem exists.
Review the underlying records before deciding that the information is absent. Relevant details may exist in individual reports even if the summary omitted them. If they do, reassess them with an appropriate method. If they do not, avoid reconstructing a denominator or subgroup result from assumptions.
Consider whether the PMS plan and post-market clinical follow-up, or PMCF, activities need changes to address the uncertainty. A future PMCF activity does not by itself supply the evidence needed to support a present conformity conclusion. Distinguish what the existing evidence establishes from what future monitoring is intended to learn.
If the review identifies a potential safety issue, involve the appropriate safety and vigilance owners promptly. Do not treat the notified-body response package as the only action that might be necessary.
What should the technical response contain?
Make the response traceable from question to conclusion. Explain your interpretation of the finding, the work actually completed, the evidence supporting the revised conclusion and the effect on related documents. Name the exact revisions you are submitting and distinguish them from planned work.
For a reference defect, the response can be short if the evidence genuinely answers the question. Identify the incorrect citation, give the corrected location and explain its relevance. For an evidence gap, provide the assessment and supporting results rather than hiding the unresolved issue in a longer narrative.
In the illustrative case, a useful internal handover would identify CER-01 revision 3 section 7 as the disputed conclusion, CL-04 revision 1 as adult-only evidence and IFU-01 revision 4 as the affected user-information record. It would assign the population applicability assessment before authorising replacement conclusions. That handover is useful precisely because it does not pretend the substantive work has finished.
Before submission, ask a reviewer who did not write the revisions to follow one retained claim through the CER, supporting evidence, risk assessment, instructions and PMS or PMCF decisions. Can they explain why the claim remains justified without relying on a private conversation? If not, locate the missing reasoning or evidence before adding more narrative.
An internal check does not establish notified-body acceptance. It establishes that your response says what you did, cites what supports it and leaves no known dependency unexamined.
Questions that come up during remediation
Do you need to rewrite every linked document?
No. Assess every material dependency, then revise the documents whose content or conclusions need to change. Keep a reasoned record when a linked document remains valid. A blanket rewrite creates review work without necessarily improving the evidence.
Can you challenge a finding rather than change the file?
You can explain a different interpretation using the applicable requirement and device-specific evidence. First check that you and the reviewer are discussing the same claim and revision. A disagreement about the conclusion needs technical reasoning, not simply a statement that the file already complies.
Does a technical deficiency always require a separate CAPA?
Assess that under your quality-system procedures and the nature of the issue. A wrong citation and a recurring failure to evaluate clinical evidence may need different treatment. Keep the technical response connected to any required quality-system action without confusing a completed document correction with evidence that an underlying process problem has been addressed.
Can you promise closure after one response?
No. The strength of the response depends on the evidence and the unresolved questions, while the notified body determines its assessment outcome. Focus on making every conclusion assessable rather than predicting an acceptance decision.
Put the evidence before the rewrite
The useful outcome is a coherent, supported technical position. You know which claims remain justified, which questions require more work and which document revisions follow from those decisions. That is a better starting point for a response than a list of files to rewrite.
Talk to Dovetail about your technical documentation findings.
