Why Translation Quality in FDA and EMA Submissions Is Not a Language Problem
30–37% of FDA submissions receive a Complete Response Letter. Translation errors in regulatory documents don't just delay approval — they trigger full review cycles worth months and tens of millions. Here's what regulatory translation actually requires.
Table of Contents
TL;DR — Key Takeaways
- 1.30–37% of FDA new drug applications receive a Complete Response Letter — regulatory agencies cite documentation quality, including translation, among the most common deficiencies.
- 2.Regulatory terminology follows ICH (International Council for Harmonisation) guidelines with precise definitions. Translating a term like 'adverse event' with a non-standard equivalent isn't just wrong — it creates a classification discrepancy that requires formal response.
- 3.EMA submissions require documentation in EU official languages for patient-facing materials, meaning translation quality affects both regulatory approval and market access simultaneously.
- 4.The cost of a single CRL cycle is not just the direct response cost — it's months of delayed market entry, which in a patent-protected drug means tens of millions in lost revenue.
The Real Cost of a Complete Response Letter
A Complete Response Letter from FDA doesn't reject an application — it identifies deficiencies that must be addressed before approval can be granted. The process of receiving, analyzing, responding to, and resubmitting is a minimum 6–12 month cycle. For a drug with a $1B revenue potential, each month of delay represents tens of millions in lost revenue from a time-limited patent window.
Translation quality contributes to CRL triggers in ways that are often indirect but consequential. Inconsistent terminology in clinical study reports creates interpretation ambiguity — a term translated differently across sections of an NDA can lead reviewers to question whether different events are being described. Ambiguity triggers requests for clarification, which triggers a CRL.
This isn't hypothetical risk management. Regulatory agencies publish guidance explicitly requiring consistent use of MedDRA (Medical Dictionary for Regulatory Activities) terminology throughout submission documents. Submissions that deviate — including through translation — are deficient by definition.
Why Regulatory Terminology Is a Controlled Vocabulary
ICH guidelines establish standardized terminology for clinical trial documentation across FDA, EMA, and other major regulatory bodies. Terms like 'adverse event,' 'serious adverse event,' 'suspected unexpected serious adverse reaction' have precise definitions and hierarchical relationships in MedDRA. Using non-standard terms in translation — even synonyms that are technically accurate — creates classification mismatches that require formal resolution.
The problem is compounded by language pairs. Korean medical terminology has evolved with its own conventions, and the correct translation of Korean medical terms into English regulatory language isn't always the most literal translation. 'Death' as an outcome term in Korean clinical documentation may be expressed with formality conventions that machine translation renders as a clinical synonym that isn't the ICH-standard term.
EMA adds another layer: its submission requirements vary by document type. Some documents require translation into all 24 EU official languages; others require only English and the language of the member state where the application is filed. The rules are specific and penalties for non-compliance include submission invalidation.
The Consistency Problem Across Large Submission Packages
An NDA or CTD submission package contains hundreds of documents produced over years by multiple teams. Clinical study reports, preclinical data packages, quality documentation, and labeling drafts often come from different internal teams and external CROs. Each may have translated terminology independently, producing a submission where the same concept is expressed differently across documents.
Regulatory reviewers read across multiple documents to assess the overall submission. Inconsistent terminology between a clinical study report and the labeling draft — even when both are scientifically accurate — signals that the submission may not have been coherently assembled. This is a credibility risk that translates directly into review questions.
The technical solution is a master regulatory glossary maintained across the entire submission — one source of truth for how every regulated term should appear in English across every document. This isn't standard practice in most Korean pharma companies' translation workflows because the volume makes it operationally difficult without dedicated tools.
How to Structure Regulatory Translation to Avoid CRL Risk
Build a master regulatory glossary before translation of any submission documents begins. The glossary should include MedDRA terms, ICH-standardized terminology, and your compound's specific terms (drug name, mechanism description, indication language). This glossary must be enforced across every document in the package — not as a reference, but as a constraint.
Separate the translation workflow from the medical writing workflow. Medical writers who produce the source Korean documents should not be the same people checking translation accuracy — they're too close to the content. Regulatory translation QA requires someone who knows both the target language conventions and the regulatory context to evaluate whether the translation is correct for regulatory purposes, not just linguistically accurate.
Tools like leapCAT can apply a regulatory termbase consistently across large document sets and flag deviations for review. The combination of glossary enforcement, MQM-based quality evaluation, and human regulatory review is the practical approach for submission packages at the volume and complexity of an NDA.
Frequently Asked Questions
Get expert-level translation without the expert cost
43 AI agents run the full professional translation workflow — analysis, terminology, translation, review, QA — starting at $0.01/word.
Try it free