Short Answer
A Notified Body reviews the literature section of a clinical evaluation report (CER) against the MDCG 2020-13 assessment template and MEDDEV 2.7/1 Rev 4 Appendix A5, and asks one question: can this search be reproduced? The recurring findings are a search protocol written after the search, a single search on the device name with no state-of-the-art search, undocumented exclusion reasons, single-reviewer screening, literature on devices whose equivalence was never demonstrated, no appraisal so that every study counts equally, and an update cycle that contradicts post-market data. Each one is solved in advance by systematic-review method.
Serteser Danismanlik is run by a biomedical engineer (BME MSc) who developed a medical-AI medical device, published it in a peer-reviewed international journal, and led the methodology of four PROSPERO-registered systematic reviews. We deliver the CER literature and statistical core for the MEDDEV 2.7/1 Rev 4 literature route directly to manufacturers, and white-label, as a subcontractor, to RA consultancies and CROs; clinical evaluation sign-off stays with your named clinical evaluator. QMS, technical-file structure and product registration are not our lane.
The most common reason a CER comes back from a Notified Body is not weak clinical data on the device. It is that the report cannot explain how its literature section was built. The assessor opens the file with the MDCG 2020-13 clinical evaluation assessment report template in hand, and that template asks directly about the search: sources, terms, date range, inclusion and exclusion criteria, selection process, and appraisal. If the answers are not in the report, a finding is inevitable.
This article combines the expectations of MEDDEV 2.7/1 Rev 4 Appendix A5 (the literature search and review protocol) with the checkpoints of MDCG 2020-13, and walks through seven errors we see repeatedly, each with its fix. The goal is not a checklist but a method that passes the review the first time.
What the assessor looks at: MDCG 2020-13 and Appendix A5
MEDDEV 2.7/1 Rev 4 Appendix A5 expects the literature search to be planned in advance under a protocol, and expects the following to be documented: the sources used and why they were chosen, the search terms and strategy, the inclusion and exclusion criteria, quality control of the screening process, and full reporting of the results. MDCG 2020-13 turns each of these into a field the Notified Body assessor ticks. The standard is explicit; the problem is not that it is unknown but that its application is invisible in the report.
Systematic-review method matches this expectation one to one: the protocol is written first, the search is documented reproducibly, the selection flow is shown in a diagram, every excluded record is listed with its reason, and every included study is appraised against pre-defined criteria. PRISMA 2020 is not mandatory for a CER, but it presents everything the assessor is looking for in a single flow.
Error 1: The search protocol was written afterwards
The most frequent finding is a search that was run first, with the protocol added to the report later. It creates the impression that search terms were chosen to fit the result, and the assessor records it as a lack of reproducibility.
Fix: The search protocol sits in the clinical evaluation plan (CEP), dated and signed; databases, search strings, date range, language and filters, inclusion and exclusion criteria are fixed before the search. The dates of the protocol and the report must be consistent.
Error 2: A single search on the device name
MEDDEV 2.7/1 Rev 4 expects two separate searches: one to find data on the device and, if claimed, the equivalent device, and one to define the state of the art, meaning current treatment options, alternative devices and accepted performance and safety benchmarks. A search on the product name alone leaves the second question empty, and the benefit-risk assessment loses its comparator.
Fix: Two search strategies, two result flows. The state-of-the-art search is built on generic device type, indication and alternative-treatment terms; clinical guidelines, systematic reviews and vigilance databases join the source list.
Error 3: Exclusion reasons are not documented
If the report says "412 records found, 14 studies included" and the reason the other 398 were excluded is nowhere to be seen, the assessor cannot tell whether selection was arbitrary. This is a field marked directly in MDCG 2020-13.
Fix: Records excluded at title and abstract stage are coded by criterion; records excluded at full text are listed one by one with the reason. A PRISMA 2020 flow diagram combines the numbers in one figure. The full-text exclusion list goes into the CER annex.
Error 4: Single-reviewer screening, no quality control
Appendix A5 expects quality control of the screening process. A selection performed by one person and verified by nobody is indefensible, especially for borderline studies.
Fix: At least two independent reviewers, a pre-defined rule for resolving disagreements, and a record kept in the screening software. In small teams the second reviewer can be an external named methodologist; what matters is that the process is documented.
Error 5: Literature on a device whose equivalence was never shown
Counting publications on a similar product as evidence for the device "because it is the same technology" skips the MDR Annex XIV Part A(3) requirement to demonstrate equivalence in technical, biological and clinical characteristics. The assessor removes that literature from the device evidence and the evidence base suddenly shrinks.
Fix: If equivalence is claimed, the comparison table under MDCG 2020-5 is built before the literature section. If equivalence cannot be shown, those publications stay in the state-of-the-art context and are not used as performance evidence for the device. See equivalence in the CER and the MDCG 2020-5 evidence table.
Error 6: No appraisal, every study weighted equally
Counting every included study with the same weight puts a case series on a par with a randomized trial. Appendix A6 expects each study to be scored under a pre-defined scheme using suitability criteria (D1 to D4) and data-contribution criteria (D5 to D9).
Fix: The weighting scheme is pre-defined in the CEP; each study receives a suitability and contribution score in a table; the effect of low-scoring studies on the conclusion is discussed separately. We cover this section in literature appraisal under MEDDEV Appendix A6.
To have your CER literature section reviewed from a methods perspective before Notified Body submission, request a free 15-minute scoping call.
Error 7: An update cycle inconsistent with PMS data
MDR Article 61(11) and Annex XIV Part A(4) require the clinical evaluation to be updated with post-market data. If the literature search is three years old, the PSUR carries new vigilance signals and the CER is unaware of them, the assessor sees the inconsistency directly.
Fix: The update frequency is defined in the CEP by device class and risk profile; each update documents the new search date, added and removed records and the effect on the conclusion; PMCF and PSUR findings are cross-referenced with the literature section.
Common Mistakes
- Not copying the search string into the report: "PubMed and Embase were searched" is not reproducibility. The string itself, the date and the filters go in the report.
- Skipping grey literature and vigilance sources: For the state of the art and the safety discussion, MAUDE, EUDAMED and national vigilance sources belong on the source list.
- Applying a language restriction without justification: "English only" is possible, but it must be justified and its possible effect discussed.
- Flow diagram numbers that do not add up: If found, screened and included counts do not reconcile, the assessor doubts the whole process.
Related Articles
- How to Write the CER Statistics Section
- Literature Appraisal in Clinical Evaluation: MEDDEV 2.7/1 Appendix A6
- Equivalence in the CER: The MDCG 2020-5 Evidence Table
- Sample Size in a PMCF Study
The literature section is the cheapest part of a CER to fix and the most expensive to have returned. When the protocol is written first, the two searches are run separately, exclusions are justified and appraisal is pre-defined, the answers to the assessor's questions are already in the report. This is nothing more than systematic-review method applied to the CER; the experience behind the method is on the about page.
Scope, timeline and budget differ for every file; we settle them in a free scoping call: