Quick answer — An SmPC follows the EMA's QRD template, so the section headings and many standard statements already have agreed wording in each EU language. The translation work is the product-specific text between them, and consistency with the leaflet.
Vitra.ai Universe keeps the audit trail with the translation.
Most of the document is already decided
The Summary of Product Characteristics is not free prose. It has a fixed structure, and the EMA publishes QRD templates giving the agreed heading and standard-statement wording in every EU language. Using that agreed wording is not a stylistic preference. Diverging from it creates a finding, and findings cost review cycles late in a procedure.
So the first job is mechanical: pull the approved template text rather than translating it again. What remains is the product-specific content.
Where the actual work is
| Section | Nature |
|---|---|
| Headings and standard statements | Fixed by template |
| Therapeutic indications | Product-specific, high scrutiny |
| Posology and administration | Numbers, units, populations |
| Contraindications and warnings | Highest consequence |
| Undesirable effects | Frequency terms are standardised |
| Pharmacological properties | Dense, technical, lower risk |
Frequency terms — very common, common, uncommon, rare — map to defined percentage bands, so they are terminology rather than description and must resolve to the agreed local equivalent every time.
Consistency with the leaflet is the recurring finding
The SmPC and the package leaflet describe the same product at different reading levels, and they have to agree.
A warning phrased one way in the SmPC and differently in the leaflet is the kind of inconsistency linguistic review picks up, and it usually comes from the two documents being translated at different times by different people.
Translating them together, against one glossary, removes most of that.
Practical sequence
Lock the terminology first — product name, active substance, excipients, frequency terms, populations — and pin it so no general match can overwrite it.
Then translate the product-specific sections with the template text already in place, using document translation so the structure survives.
Then check the output rather than the process. A back-translation comparison is the right tool for contraindications and posology, because a lost qualifier or a shifted number reads perfectly and changes the meaning — quality control runs it on every segment rather than a sample.
Regulatory affairs owns the final wording. The point of the pipeline is that what reaches them is consistent, traceable and already checked, not that it replaces their sign-off.
FAQ
What is the QRD template and why does it matter for translation? It is the EMA's set of agreed templates giving standard headings and statements in each EU language. Using the published wording avoids findings, so most of an SmPC should be assembled rather than translated.
Which SmPC sections need the most translation scrutiny? Indications, posology, contraindications and warnings. These carry clinical consequence, and errors in them read fluently, which is why a back-translation comparison is worth running on every segment.
Do frequency terms need special handling? Yes. Terms such as very common, common, uncommon and rare correspond to defined percentage bands, so they are terminology rather than description and must resolve to the agreed local equivalent consistently.
Why do SmPC and leaflet inconsistencies keep appearing? Because the two documents are usually translated at different times by different people. Translating them together against one glossary removes most of the inconsistencies linguistic review would otherwise find.
