Documenting and reporting your search means recording every decision and action well enough that another researcher could reproduce the search exactly. For each source you capture the database and interface, the full strategy line by line, any limits or filters, the date the search was run, and the number of records retrieved. That record is then reported against the search items of the PRISMA 2020 guideline so reviewers can audit how the evidence was found.
Reproducibility is the whole point
A search that cannot be reproduced cannot be trusted, because no one can confirm whether it was complete or biased. The difference between a systematic review and a narrative review rests largely on this: the systematic search is transparent enough to rerun. Good documentation is therefore not paperwork after the fact but part of the method itself, and it has to be built up as you search, because details like exact line counts and run dates are almost impossible to reconstruct later.
What to record for every source
The full strategy, verbatim
Report the complete search strategy for at least one database in full, copied exactly as run, including every line, operator, and field code. A paraphrased or summarised search is not reproducible. Most guidance now expects the full strategy for every database searched, usually in a supplementary file, since each was translated for that platform and differs from the others.
Dates, counts, and limits
For each source, record the date the search was run, the date range covered, any language or date limits applied with their justification, and the number of records retrieved. These counts feed directly into the next stage of reporting and let a reader reconcile the numbers at every step.
Reporting against PRISMA 2020
The PRISMA 2020 guideline sets out exactly what a search report should contain, and a dedicated extension, PRISMA-S, expands on the search specifically. Reporting against PRISMA 2020 means naming every source, giving at least one full strategy, stating the run dates and any limits, and describing supplementary methods such as grey literature searching and citation chasing. Noting that the strategy passed PRESS peer review strengthens the report further.
The PRISMA flow diagram
The record counts you captured drive the PRISMA flow diagram, which traces records from identification through de-duplication and screening to the final included studies. Every number in the diagram should reconcile with the search record, so a reader can follow the funnel from thousands of hits down to the handful of included studies without any unexplained gaps.
A worked search log entry
A search log does not need to be elaborate; it needs to be complete. A single entry for one source captures everything a later reader would need to rerun it:
- Source and interface: MEDLINE via the Ovid platform, so a reader knows which syntax and behaviour produced the count.
- Coverage and dates: the date range the database spans and the calendar date you ran the search.
- Full strategy, verbatim: every numbered line, operator, truncation symbol, and field tag, copied exactly as run.
- Limits applied: any language or date limit with its justification, never applied silently.
- Records retrieved: the count returned, which feeds the identification box of the flow diagram.
Repeat the same five fields for every database, every grey source, and each round of citation work, and the search becomes reproducible by construction rather than by reconstruction. The verbatim strategy is the part most often skimped: a paraphrased search such as “we searched for depression and therapy” is not reproducible, because it hides the headings, synonyms, and operators that actually determined the result.
Reconciling the numbers across the search
The counts in your log are not just record-keeping; they have to add up. The total identified across all sources, minus the duplicates removed at de-duplication, must equal the number that entered title and abstract screening, and each later stage must reconcile down to the final included set with no unexplained gaps. A reader who finds the numbers do not balance loses confidence in the whole review, so checking the arithmetic before submission is part of documenting the search well, not a separate task. This is also why records found through citation chasing and other methods are logged under their own source category: they enter the flow diagram by a different route and must be counted separately.
Common documentation mistakes
The failures here are almost always about timing and completeness. The first is leaving documentation until the write-up, by which point exact line counts and run dates are gone and have to be guessed. The second is reporting only a summary strategy, which falls short of reporting standards because it cannot be rerun. The third is numbers that do not reconcile between the search log, the de-duplication step, and the flow diagram, which signals to a reader that the audit trail has a hole in it. Avoiding all three is straightforward if the log is kept live, and it is the standard our literature search service delivers as a matter of course.
Building the record into your workflow
Capture the documentation at the moment you act: paste each finished strategy into a search log, note the date and count as the results load, and keep the exports in a clearly named folder. Reviewers who leave documentation until the write-up almost always lose detail. Treating the log as a live part of the systematic search workflow is the only reliable way to keep it complete and accurate.