A narrative review's search methods section is where readers learn how the evidence base behind the argument was assembled. The task is not to imitate a systematic review's exhaustive protocol, but to make your search transparent enough that a colleague could understand your sources, your terms, your screening logic, and the boundaries of what you found. This article gives you a complete sequence for producing that section, with checkpoints, a worked example, and completion criteria you can apply before you circulate a draft.
What does a reader actually need from your search methods?
Most clinicians and researchers arrive at this task with a pile of papers and a rough sense of the literature, but no written record of how the pile was built. That is the starting state to diagnose. If you cannot reconstruct which databases you used, which terms you combined, and why some records were set aside, your methods paragraph will read as decoration rather than description.
A useful search methods section answers four reader questions in order:
- Where did the evidence come from? Name the sources you searched and, where relevant, the supplementary routes such as citation tracking or reference lists.
- How did you target the topic? Show the concepts behind your terms and how free-text words were combined with subject headings.
- How did records become included studies? Describe screening stages and how you handled multiple reports of the same study.
- How current and how bounded is the search? State date limits, language or format limits if any, and whether the search was rerun before publication.
The Cochrane Handbook's guidance on searching for and selecting studies is a sound reference point even for a narrative review: plan searches from eligibility criteria, use multiple sources, combine free-text terms with subject headings, document the process, deduplicate records, screen in stages, link multiple reports of one study, and rerun searches before publication. You can adopt these practices proportionally rather than wholesale.
What decisions must be made before you write a single sentence?
Writing the methods section is the last step, not the first. Four decisions determine what you will be able to describe.
Decision 1: What are your eligibility criteria?
Cochrane's guidance begins with eligibility criteria because they drive every search term. For a narrative review, write a short inclusion and exclusion statement in plain language: population, intervention or exposure, outcome, study designs you will consider, and any date or setting boundaries. Keep it narrow enough to be operational.
Decision 2: Which sources fit the question?
Multiple sources are recommended. For a clinical topic, a bibliographic database plus citation tracking is a common minimum. If your question spans guideline documents or grey literature, say so explicitly rather than implying your database search covered everything.
Decision 3: How will you combine terms?
Free-text terms capture new vocabulary; subject headings capture indexed concepts. Combining both is the documented approach. PubMed's field tags, phrase searching, filters, and Advanced Search give you the mechanics, and its Search Details panel shows how your terms were translated and combined — a useful artifact to save.
Decision 4: How will you record the process as you go?
Documentation is easiest when it happens during searching. Keep a running log with the date, source, exact string, filters applied, and number of records returned. This log becomes the backbone of your methods paragraph and lets you rerun the search before publication.
How do you build and document the search itself?
Treat the search as a sequence of artifacts, each of which you will later describe.
Step 1 — Concept map. Break your question into two to four concepts. For a focused example, consider a review asking how early mobilization affects delirium in hospitalized older adults. Concepts might be early mobilization, delirium, and older hospitalized adults.
Step 2 — Term harvesting. For each concept, list synonyms, spelling variants, and the relevant subject heading. Record where each term came from.
Step 3 — String construction. Combine terms within a concept, then combine concepts. Save the exact final string for each source, because strings differ between databases.
Step 4 — Execution log. Record date, source, string, filters, and yield. If you use a scholarly search interface to find candidate papers, keep the same discipline: note what you searched and when. WisPaper's scholarly search interface is one place to run topic searches for candidate papers; the log you keep around it is what makes the search describable.
Step 5 — Supplementary routes. Citation tracking and reference-list checking are legitimate sources. Google Scholar's Cited by and Related articles features, and PubMed's Similar Articles, are documented ways to move from a known paper to related work. Describe these routes separately from database searching so readers can see their contribution.
Step 6 — Deduplication. Cochrane recommends deduplicating records. Record how you did it and how many duplicates you removed.
How should you screen records and handle multiple reports?
Screening in stages is the documented approach: an initial pass on titles and abstracts, then a full-text pass. For a narrative review you can describe this proportionally, but you should still state the stages and the reason categories for exclusion.
Two practical points deserve explicit mention in your methods:
- Multiple reports of one study. Cochrane guidance calls for linking multiple reports of a single study. If three papers report the same trial, your review should treat them as one study and say so.
- Reason categories. Rather than a single exclusion count, name the categories — wrong population, wrong outcome, not a primary study — so readers understand the shape of what was set aside.
As screening proceeds, you will accumulate papers worth keeping. Saving them in an organized library makes the later write-up and any rerun far easier. WisPaper's My Library lets you save or upload papers and organize them, which supports the documentation step rather than replacing it.
Where do evidence level, applicability, indirectness, and limitations belong?
These four ideas are often blurred together in narrative reviews. Keep them distinct, and place each in the right part of the manuscript.
| Concept | What it addresses | Where to describe it |
|---|---|---|
| Evidence level | The design and strength of the underlying studies | Results or evidence summary, not the search methods |
| Applicability | Whether findings transfer to your setting and population | Discussion, tied to your eligibility criteria |
| Indirectness | Whether the included evidence matches your question's population, intervention, or outcome | Discussion, with reference to your criteria |
| Limitations | Constraints of your search and of the evidence base | A dedicated limitations passage |
Your search methods section should describe the process that produced the evidence base. Judgments about evidence level, applicability, and indirectness belong downstream, where you interpret what was found. Keeping this separation prevents the methods section from drifting into conclusions.
What does a complete search methods paragraph look like?
Here is a worked example for the early mobilization and delirium review. It is illustrative prose, not a template to copy verbatim.
We searched [database name] from inception to [month, year] for studies of early mobilization and delirium in hospitalized adults aged 65 and older. The search combined free-text terms for mobilization and delirium with the corresponding subject headings, and was limited to English-language publications. The full string is provided in the appendix. We supplemented the database search with backward and forward citation tracking from included studies. Records were deduplicated, then screened on title and abstract followed by full text; exclusions were recorded by reason category. Multiple reports of the same study were linked and counted once. The search was rerun before submission, and no additional eligible studies were identified.
Notice what the paragraph does: it names sources, shows term logic, describes screening stages, handles linked reports, and states currency. It does not claim the search was exhaustive, and it does not evaluate the strength of the evidence — that comes later.
How can a research workspace support the write-up?
A narrative review often runs alongside other work, and keeping the search record with the project prevents it from being reconstructed from memory. If you are managing the review as a defined piece of work, WisPaper's Research Projects area is where you can create and organize a project. For longer literature tasks, the Scholar Agent area is designed for extended research work including literature review use cases.
When your saved papers are assembled, you may want to interrogate them directly — for example, to check how a subset of included studies defined the outcome. Library QA supports asking questions based on papers uploaded to My Library or added from search results. Use it to interrogate your own assembled set, and record in your log that you did so, just as you would record any other step.
Before submission, verify that every reference in your bibliography is real. TrueCite checks BibTeX against real academic databases to flag hallucinated references — a concrete final check that pairs naturally with the rerun-before-publication step.
What checkpoints tell you the section is on track?
Use these checkpoints at three moments.
Before searching: You have written eligibility criteria; you have chosen at least two sources; you have a concept map with terms and headings; you have a log template ready.
During screening: Every source has a dated entry with its exact string and yield; duplicates are counted; exclusion reasons are categorized; linked reports are flagged.
Before writing: You can answer, from your log alone, which sources were used, what each string was, how many records each returned, how screening proceeded, and whether the search was rerun.
If any checkpoint fails, return to the log rather than writing around the gap.
What are the completion criteria for a finished section?
A search methods section is ready when all of the following are true:
- Every source is named, with search dates.
- The term logic is described, including the combination of free-text terms and subject headings.
- Supplementary routes such as citation tracking are described separately.
- Deduplication is stated.
- Screening stages and exclusion reason categories are described.
- Multiple reports of one study are addressed.
- Date, language, and format limits are stated or their absence is noted.
- The rerun before publication is reported.
- The full search string is available in an appendix or supplement.
- Evidence level, applicability, indirectness, and limitations are handled outside the methods section, in their proper places.
When every box is checked, a colleague can read the section and understand exactly how your evidence base was built — which is the whole purpose of describing search methods in a narrative review.




