Audit a decision brief by pulling the exact passage from each claim's original source, comparing the claim's wording against that passage, and marking the claim supported, overstated, or unsupported with the revision it needs.
Run the audit in this order
- Build a traceability table with one row per claim and columns for the claim as written, the cited source, the exact quoted passage, the source type, and the verdict. Expected result: every claim has a row, and any row with no locatable passage is flagged immediately.
- Retrieve the original source and copy the passage that carries the claim. ICMJE advises authors to verify references against original sources and to cite direct sources where possible, so a review summary or secondary report does not close the row on its own. Expected result: each row holds a verbatim passage from the primary document.
- Compare the claim's wording to the passage and assign a verdict. Mark it supported when the passage states the claim; overstated when the passage supports a narrower or weaker version; unsupported when the passage does not carry the claim. Expected result: every row has one verdict and, where needed, the specific revision.
- Check whether the cited source has been retracted or corrected before you finalize. ICMJE guidance includes identifying retractions. Expected result: retracted or corrected sources are replaced or the claim is withdrawn.
- Rewrite overstated and unsupported claims to match the passage, or drop them, then re-check the revised wording against the same passage. Expected result: the final table shows only supported claims or claims revised to what the source states.
Where secondary summaries enter
A systematic review or meta-analysis can support a claim about the review's own conclusion, and PRISMA 2020 sets out reporting items and flow diagrams for how such reviews are reported. When a brief uses a review to support a claim about an underlying study, trace that claim to the study itself. Cochrane Handbook Chapter 15 covers how review results should be interpreted, including confidence intervals, exact P values, absolute and relative effects, applicability, indirectness, and certainty, which helps you judge whether a summary's wording matches what the analysis shows.
What the finished table should show
Each claim carries a verdict, the exact passage, and the source type, so a reviewer can see at a glance which claims rest on primary documents and which need revision before the brief is handed off.