How do I audit screening-to-synthesis handoffs across a review's stages?

Audit the handoff by reconciling record counts at each stage and confirming that every included record has a screening decision, an extraction row, and a synthesis assignment. Build one reconciliation table that traces each record from screening through extraction to synthesis, then flag any record that is missing, duplicated, or unassigned.

Build the reconciliation table

1. List every record that passed screening as one row, carrying a stable identifier, its screening decision, its extraction row reference, and its synthesis assignment. Expected result: each included record appears exactly once, and any blank cell marks a broken handoff.
2. Compare the count of included records against the count of extraction rows. Expected result: the two counts match; a surplus extraction row points to a duplicate, and a shortfall points to a record dropped between stages.
3. Compare the count of extraction rows against the count of records assigned to a synthesis group or comparison. Expected result: every extracted record has an assignment, and any unassigned row is flagged for a decision.
4. Recheck flagged records against the original screening decision and extraction entry, then record the resolution in the table. Expected result: each flag closes with a documented correction or a stated reason for exclusion.

What the table should capture

Keep the identifier, screening decision, extraction reference, synthesis assignment, and flag status in separate columns so a reviewer can scan for gaps. PRISMA 2020 supplies a checklist and flow diagrams for reporting what studies were found and included, which gives you a reporting structure for the counts your table reconciles. The Cochrane Handbook's guidance on searching for and selecting studies covers documenting searches and managing records, which supports keeping identifiers stable across stages.

Where handoffs typically break

Duplicates usually enter when a record is re-added during extraction, and losses usually appear when a record is excluded at screening but still carried into extraction, or extracted but never assigned to a comparison. A count mismatch alone does not show which record moved; use the identifier column to locate the specific row before changing any count.

Scope

This audit checks traceability between stages. Judgments about effect measures, heterogeneity, and pooled results belong to the analysis itself, where the Cochrane Handbook's meta-analysis guidance applies.