Fix an extraction mismatch by comparing each field against the source table, correcting the extraction, and flagging any derived values that change as a result.
- Line up both reviewers' tables field by field against the source table and mark the first field where the entries differ. Record the exact source location (table, row, column) that each reviewer used, so the disagreement is tied to a specific cell.
- Decide which entry matches the source table. If one reviewer misread or shifted a row, correct that extraction and note the correction. If the source table itself is ambiguous, record the ambiguity and the interpretation chosen.
- Flag every derived value that depends on the corrected field, such as totals, ratios, or pooled inputs, and recompute or mark them for review. A corrected raw value can change downstream calculations even when those calculations were done correctly.
- Document the resolution in the extraction log: the field, both original entries, the source location, the corrected value, and who resolved it. This record lets a third reviewer or an auditor follow the decision.
- Update the extraction form if the field definition was unclear. Rewrite the definition with the source location and the expected format, then re-check other records that used the same field.
Cochrane Handbook Chapter 4 covers documenting searches and managing records for systematic reviews, which supports keeping a clear extraction log. Cochrane Handbook Chapter 10 covers effect measures, missing data, and sensitivity analyses, so derived values tied to those areas deserve a second look after a correction. When the source table leaves a field genuinely unresolved, record it as unresolved and route it to a third reviewer for a decision.