Data Management and Query Resolution
Data management is a GCP activity, not an administrative afterthought. The CRF (paper or electronic) captures data that must be traceable to source, and ICH E6 defines source data as all information in original records and certified copies from which the data derive.
What a CRA must recognise
- ALCOA-C attributes: data must be attributable to a person, legible, contemporaneous, original, accurate, and complete (plus enduring and available).
- eSource and EDC: audit trails must be enabled and reviewed; users are individually credentialed; entries are never deleted outright.
- Queries: automated (edit-check) or manual (from monitors, data managers, medical reviewers). Queries are answered by the site, with correction of the CRF and, where the underlying data were wrong, correction of the source with a single-line strike-through, initial, date, and reason for change.
- Coding: adverse events and medications are coded to standard dictionaries (MedDRA; WHO Drug). Coding occurs after data cleaning and does not substitute for the verbatim term.
- Timelines: query response windows are defined in the data management plan or monitoring plan and tracked; ageing queries are a quality signal.
Traps
- Back-dating or "completing" a source note for a visit that never happened, or entering data for a visit that has not occurred.
- Deleting a wrong entry in the CRF rather than entering the correct value that preserves the audit trail.
- Answering a query with "data correct as entered" without checking source.
- Confusing a query with a protocol deviation: a query corrects data; a deviation is a departure from the protocol that must be documented and evaluated.
Worked example: the CRF says the subject took 2 tablets daily; source says 1. The site confirms 2 is correct - but the source diary legitimises 1. The correct action is to clarify with the subject, document the outcome in source, and only then update the database, noting the rationale.