Build a reproducible record by logging each database, search string, date, and hit count, then mapping every record to a screening decision and its first-failure reason. The Cochrane Handbook describes documenting searches and managing records as part of planning and running a review, and PRISMA 2020 supplies a flow diagram for reporting what was found and included. Together they give you the two halves of the record: the search log and the screening trail.
Record the searches
- For each source, log the database or platform name, the exact search string as run, the date of the search, and the number of hits returned. Keep the string verbatim so a supervisor can rerun it and compare counts.
- Note any limits applied, such as date ranges or language filters, alongside the string, because a rerun without them will return a different count.
- Save the raw export from each source before deduplication, so the hit count in your log can be reconciled against the records you actually screened.
Record the screening decisions
For every record, capture the screening stage, the decision (include, exclude, or unresolved), and the first-failure reason that ended its progress. A first-failure reason is the single criterion that caused exclusion, recorded once so a later reviewer can see why the record stopped there. Map each record to exactly one decision so the totals add up.
Build the appendix
The appendix should let a supervisor rerun each search and reconcile the numbers against the PRISMA flow counts: records identified per source, duplicates removed, records screened, records excluded with reasons, and studies included. PRISMA 2020 provides the flow diagram structure for these counts. Where a count does not reconcile, mark it unresolved and trace it to the specific export or decision entry rather than adjusting the total.
Keep the log in a versioned file with a date column, so changes to a search string or a screening decision remain visible.