.jpg){kind=link}
- Published
- Updated
- Sources checked
A terminology decision reaches beyond downloading a dictionary. Collection, transformation, standardization and submission should preserve the same concept. A version-impact record makes late changes easier to assess and helps reviewers understand which terminology was applied to the submitted data.
Separate terminology types
FDA study-data resources support assessment of standards and terminology against the relevant catalog. Support for a standard is different from an immediate requirement to adopt every new version. Inventory clinical coding dictionaries, SDTM controlled terminology and local operational lists as distinct components.
Document the version decision
For each list, identify the version, selection rationale, affected variables and transformation owner. Show how codes and displayed labels are stored in different fields. For long studies, assess freezing or upgrading a version alongside the effect on comparability rather than treating the choice as routine maintenance.
Run the impact checklist
Find records affected by retired terms, changed labels or newly introduced codes first. Evaluate the effect using representative data. If a code’s meaning changes, assess scientific interpretation and documentation needs rather than performing a text-only replacement that could obscure a genuine semantic difference.
- Inventory terminology type and version at variable level.
- Review retired terms and changes in meaning.
- Verify upgrade decisions with representative records.
Explain historical data
Do not silently rewrite historical data with a newer dictionary. Preserve source values, applied mappings and the rationale for the version. Reviewers should understand how the delivery was produced and why unresolved differences were accepted; successful technical validation does not replace this explanation.
Frequently asked questions
- Is the newest version always the correct choice?
- No. Consider support status, study design and data consistency together.
- Can a label change be treated as immaterial?
- First assess its effect on code, meaning, transformation and reported output.