Most pharmaceutical organisations have accumulated internal clinical terminologies over years of system development: project-specific adverse event coding dictionaries, legacy clinical database value sets, local disease classifications created before standardised vocabularies were widely available. At some point, these internal terminologies must be mapped to external standards — MedDRA for regulatory pharmacovigilance reporting, SNOMED CT for clinical interoperability — and the quality of those mappings determines the quality of the downstream regulatory and analytical outputs.
Types of Mapping Relationships
Not all terminology mapping relationships are equivalent, and conflating them leads to both analytical errors and regulatory problems. An exact match exists when the internal term and the external concept have the same semantic scope — they mean precisely the same thing in precisely the same context. A broader match exists when the external concept is a parent of the internal term — the internal term is more specific than the external concept it maps to. A narrower match exists when the internal term is broader than the external concept. An inexact match covers cases where the two terms overlap but neither subsumes the other. Each mapping type has different implications for how the mapped data should be used in aggregate analyses, and each must be flagged and documented separately in a defensible mapping record.
The Mapping Methodology
A rigorous terminology mapping methodology follows several steps. First, automated candidate generation: for each internal term, a similarity algorithm (string similarity, embedding similarity, or lookup in a pre-existing cross-reference table) generates a ranked list of candidate external concepts. Second, expert review: a domain expert reviews the candidates, selects the correct mapping relationship, and documents the justification. Third, consistency validation: automated checks identify logical inconsistencies in the mapping — cases where two internal terms that are synonymous have been mapped to different external concepts, or where a term has been mapped to an external concept that is a sibling rather than a descendant of the expected concept. Fourth, version management: mappings must be maintained as both the internal terminology and the external standard evolve.
MedDRA-Specific Considerations
MedDRA mapping for pharmacovigilance purposes requires particular care because the regulatory reporting rules — which events must be reported, under which circumstances, within which timelines — depend on the MedDRA term and level at which an event is coded. Mapping an internal adverse event term to an incorrect MedDRA preferred term can result in an expedited reportable case being coded as non-reportable, or vice versa. For this reason, MedDRA mappings for regulatory reporting should be validated by a qualified MedDRA coder and reviewed against the current MedDRA release notes to identify any concept deprecations or hierarchy changes that affect the mapping.
