PubMed Alone Is Not Enough: Multi-Database Search Reality
I reviewed a clinical evaluation last month where the manufacturer ran a perfect PubMed search. Solid Boolean logic. Clear inclusion criteria. Everything documented. The Notified Body still issued a major non-conformity. The reason? They searched only one database when the literature clearly exists across multiple sources.
In This Article
This happens more often than it should. Teams invest weeks building search strategies, refining keywords, and documenting every step. But they anchor everything on PubMed because it is free, familiar, and seems comprehensive enough.
Then the review comes back. The assessor points out that Embase contains different indexing. That conference abstracts live elsewhere. That your competitor’s clinical data sits in sources you never checked.
Suddenly your literature review has a gap. Not because you did poor work. Because you searched in only one place.
Why MDR Pushes for Multiple Databases
The MDR does not explicitly list which databases you must use. But it does require that your clinical evaluation is based on sufficient clinical evidence. And Annex XIV makes it clear that this evidence must come from a comprehensive review of the literature.
MDCG 2020-13 reinforces this. It states that the literature search should be systematic, objective, and methodologically sound. One database is rarely sufficient to meet that standard.
The reasoning is simple. Different databases index different journals. They use different controlled vocabularies. They cover different publication types and geographic regions.
PubMed focuses on biomedical literature and relies on MeSH terms. Embase has broader device and pharmaceutical coverage with Emtree indexing. Cochrane captures systematic reviews. IEEE Xplore holds engineering and technical papers. Clinical trials registries contain unpublished data.
If your device touches multiple fields—say, an implant with both surgical and material science considerations—no single database will capture everything.
A comprehensive literature search under MDR is not about checking a box. It is about demonstrating that you looked everywhere relevant clinical data could exist. One database limits your view and weakens your evidence base.
What Happens When You Search Only PubMed
Let me walk through what I see in real audits.
A manufacturer submits a clinical evaluation report. The literature search section shows a detailed PubMed query. The PRISMA flow diagram is clean. The screening process is transparent. The team feels confident.
The Notified Body assessor opens the report. First question: Which databases did you search?
Answer: PubMed.
Next question: Why only PubMed?
The manufacturer explains that PubMed is comprehensive, widely accepted, and covers the relevant medical literature.
The assessor then opens Embase. Within minutes, they find three studies that match the device scope but were not included in the clinical evaluation. These studies exist in journals indexed by Embase but not PubMed.
Now the clinical evaluation has a credibility problem. Either the evidence base is incomplete, or the manufacturer will need to justify why those studies are not relevant. Both options create delays.
In another case, the assessor finds conference abstracts in Embase that describe adverse events related to a similar device. PubMed does not index those abstracts. The manufacturer never saw them. Now the safety profile discussion is incomplete.
Relying on a single database and assuming it is comprehensive. Notified Bodies expect you to search multiple sources because different databases index different content. Missing studies means incomplete evidence.
Why Embase Adds Value Beyond PubMed
Embase is not just a larger version of PubMed. It has different strengths.
First, Embase indexes more European journals. If your device is marketed in Europe, you cannot ignore European clinical data. PubMed has a strong US focus. Embase balances that.
Second, Embase includes more conference abstracts and proceedings. Clinical data often appears in conferences before full publication. If you are reviewing recent evidence or tracking emerging safety signals, those abstracts matter.
Third, Embase uses Emtree as its controlled vocabulary. Emtree has more specific device-related terms than MeSH. For medical devices, this can make a significant difference in recall.
Fourth, Embase covers more drug and device interaction studies. If your device is used alongside pharmaceuticals, or if there are pharmacological considerations, Embase is often more complete.
I have seen cases where the same search run in PubMed and Embase returned 30% more relevant records in Embase. That is not a small gap. It is evidence you cannot afford to miss.
How to Structure a Multi-Database Search
You do not need to search ten databases. You need to search the right ones for your device and justify your choices.
Start with PubMed and Embase. These two cover the majority of biomedical and device-related literature. If your device is highly specialized, add sources that cover your specific field.
For orthopedic devices, consider adding BIOSIS or engineering databases. For software or AI-based devices, IEEE Xplore may be relevant. For systematic reviews, include Cochrane. For clinical trial data, check ClinicalTrials.gov and the EU Clinical Trials Register.
Document why you selected each database. Explain what unique content each one provides. This shows the Notified Body that your choices are deliberate, not random.
Run your search strategy in each database separately. Do not assume that one search string works across all platforms. PubMed uses MeSH. Embase uses Emtree. Adjust your controlled vocabulary accordingly.
Combine the results and remove duplicates. Use a reference manager to streamline this process. Then apply your screening criteria to the combined set.
A defensible multi-database search is not about searching everywhere. It is about selecting databases that match your device scope, adapting your search strategy to each platform, and documenting your rationale clearly.
The Cost and Access Question
PubMed is free. Embase is not. This is the practical barrier most teams face.
Embase requires a subscription. For smaller manufacturers or startups, that cost can feel prohibitive. But consider the alternative.
If your clinical evaluation is rejected because the literature search is incomplete, you will need to redo the search, update the report, and resubmit. That costs time, money, and delays market access. The cost of an Embase subscription is usually far less than the cost of a major non-conformity.
Some manufacturers try to access Embase through academic institutions or use limited trial access. This can work if you plan carefully. But if you are preparing a submission for a Class III device or a high-risk implant, invest in proper access.
Another option is to work with a clinical evaluation consultant or CRO that already has database access and expertise in running multi-database searches. This can be more cost-effective than building internal capability from scratch.
The point is not to avoid cost. The point is to weigh the cost of access against the cost of incomplete evidence.
What Notified Bodies Look For
When a Notified Body reviews your literature search, they are checking for comprehensiveness and rigor.
They will look at which databases you searched. They will check if your database choices are justified. They will verify that your search strategy was adapted to each platform.
They will compare your results to what they find when they spot-check the literature themselves. If they find relevant studies you missed, they will ask why.
They will also look at how you handled duplicates and how you documented your screening process. Multi-database searches generate more records and more duplicates. Your process needs to be clear and traceable.
If you searched only PubMed, expect questions. The assessor will want to know why you limited yourself. They may run their own search in Embase or another database to see what you missed.
If you can show that you searched multiple databases, adapted your strategy appropriately, and found no additional relevant studies in secondary sources, you are in a much stronger position.
Searching multiple databases but using identical search strings without adapting to each platform’s controlled vocabulary. This reduces the value of the multi-database approach and signals weak methodology.
Practical Steps to Avoid Common Mistakes
First, do not copy your PubMed search into Embase without modification. MeSH and Emtree are different. Your search will underperform if you do not adapt it.
Second, do not assume that more databases always mean better results. Choose databases that match your device and clinical questions. Document your rationale.
Third, plan your deduplication process before you start searching. Use a reference manager like EndNote, Mendeley, or Covidence. Manual deduplication is slow and error-prone.
Fourth, keep detailed records of your search process for each database. Include search strings, filters, dates, and the number of results. This documentation is part of your technical file.
Fifth, if you find significantly different results across databases, investigate why. It may reveal gaps in your search strategy or differences in indexing that you need to address.
When One Database Might Be Enough
There are rare cases where a single database is sufficient. But you need to justify it.
For example, if you are conducting a very narrow search on a highly specialized topic where you know all relevant journals are indexed in PubMed, you might argue for a single-database approach.
But even then, I would recommend at least checking a second database to confirm that nothing was missed. The cost of running a confirmatory search is small compared to the risk of a deficiency.
If you do choose to search only one database, document your decision clearly. Explain why additional databases would not add value. Be prepared to defend that choice during the review.
The Bottom Line
A comprehensive literature search under MDR means looking where the evidence actually exists, not just where it is easiest to look.
PubMed is a strong starting point. But it is not the finish line.
Embase, Cochrane, clinical trial registries, and specialized databases all hold evidence that can strengthen or challenge your clinical evaluation. Ignoring those sources creates risk.
The goal is not to search every database on earth. The goal is to search the right ones, adapt your strategy to each, and demonstrate that you looked where relevant data could exist.
That is what a methodologically sound literature review looks like. And that is what Notified Bodies expect to see.
If your current literature search relies only on PubMed, now is the time to expand your approach. The longer you wait, the harder it becomes to close the gaps.
Peace,
Hatem
Clinical Evaluation Expert for Medical Devices
Follow me for more insights and practical advice.
Frequently Asked Questions
What is a Clinical Evaluation Report (CER)?
A CER is a mandatory document under MDR 2017/745 that demonstrates the safety and performance of a medical device through systematic analysis of clinical data. It must be updated throughout the device lifecycle based on PMCF findings.
How often should the CER be updated?
The CER should be updated whenever significant new clinical data becomes available, after PMCF activities, when there are changes to the device or intended purpose, and at minimum during annual reviews as part of post-market surveillance.
What causes CER rejection by Notified Bodies?
Common reasons include inadequate equivalence demonstration, insufficient clinical data for claims, poorly structured SOTA analysis, missing gap analysis, and lack of clear benefit-risk determination. Structure and logical flow are as important as the data itself.
Which MDCG guidance documents are most relevant for clinical evaluation?
Key documents include MDCG 2020-5 (Equivalence), MDCG 2020-6 (Sufficient Clinical Evidence), MDCG 2020-13 (CEAR Template), MDCG 2020-7 (PMCF Plan), and MDCG 2020-8 (PMCF Evaluation Report).
Need Expert Help with Your Clinical Evaluation?
Get personalized guidance on MDR compliance, CER writing, and Notified Body preparation.
✌
Peace, Hatem
Your Clinical Evaluation Partner
Follow me for more insights and practical advice.
– Regulation (EU) 2017/745 (MDR), Annex XIV
– MDCG 2020-13: Clinical evaluation assessment report template
