MedDRA Coding Guide for Pharmacovigilance Teams
MedDRA coding converts reported medical language into standardized terminology used in pharmacovigilance, clinical safety, aggregate reporting, and regulatory exchange. Good coding is not a simple search-and-replace task. It requires attention to the reported verbatim, the level of medical specificity, the current MedDRA version, and the difference between what was stated and what might be clinically inferred.
Start with source data quality
The quality of MedDRA coding depends on the quality of the source narrative. A vague report such as "felt bad" cannot be coded with the same confidence as a report that states "severe headache and vomiting six hours after dose." When the verbatim is incomplete, the coding process should flag the limitation instead of silently adding assumptions.
Pharmacovigilance teams should preserve the original wording, capture context, and separate reported events from medical history, indication, product quality issues, medication errors, and outcomes. This separation prevents overcoding and keeps the case review defensible.
Select terms for reported information
A robust MedDRA workflow should code the medical concepts that are actually reported. It should avoid adding a diagnosis when only signs and symptoms were reported unless the source explicitly supports that diagnosis.
This is where automated MedDRA coding software must be controlled carefully. Suggested terms are useful, but the reviewer must understand the source evidence, the alternative terms, and the clinical impact of choosing a broader or narrower code.
Review LLT, PT, and hierarchy impact
MedDRA term selection normally starts at the Lowest Level Term. The selected LLT maps to a Preferred Term and higher-level groupings used for retrieval and analysis. A small difference in selected LLT can affect signal review, case grouping, and downstream analytics.
PV teams should review whether a suggested term represents the most accurate current term for the reported concept and whether the resulting Preferred Term is operationally acceptable.
How the VigiHelp AI engine supports coding
The VigiHelp AI engine can extract candidate adverse events from verbatim text, suggest structured case fields, and support MedDRA coding review. Reviewers can use these suggestions as a starting point while keeping final coding decisions under human control.
This workflow is useful when teams process high volumes of unstructured intake. It reduces repetitive reading and transcription, but it should be paired with documented coding conventions, quality checks, and medical review rules.
Official references
These official sources informed the regulatory and terminology context of this guide. Teams should always confirm current requirements against their own procedures and target authorities.
Evaluate VigiHelp for your pharmacovigilance workflow
Use representative cases to test adverse event extraction, MedDRA coding review, causality support, existing case quality control, and preparation for E2B(R3), CIOMS, ESG, or EudraVigilance-related workflows.
Request a demoBack to resources