conduct-a-literature-search

# conduct-a-literature-search Design the search from the question, then retain enough history for another person to understand and rerun it. ## When to use - Use for literature reviews, technical surveys, evidence briefs, and research scoping. - Use a formal review protocol when completeness, bias control, or regulatory use demands one. ## Procedure 1. Define the question, concepts, populations, interventions or exposures, outcomes, settings, and date range. 2. Write inclusion and exclusion criteria before seeing results. 3. Select databases, repositories, registries, citation indexes, and gray-literature sources that match the field. 4. Build concept blocks using controlled vocabulary, synonyms, spelling variants, abbreviations, and Boolean logic. 5. Pilot the query against known relevant and known irrelevant records; revise without optimizing for a preferred conclusion. 6. Run each search and record platform, database, full query, filters, date, result count, and export file. 7. Deduplicate by identifiers and metadata while preserving source provenance. 8. Screen titles and abstracts against the criteria, recording decisions and uncertain cases. 9. Use backward citations, forward citations, author searches, and related-record tools to check coverage. 10. Report the search boundary, selection flow, unresolved access gaps, and last-search date. ## Failure plan - Do not claim a search is exhaustive unless the protocol and coverage justify it. - If a database changes syntax or coverage, preserve the original query and document the adaptation. - Separate unavailable full text from excluded evidence. - For current or high-stakes questions, refresh searches near delivery. ## Done - The search strategy, platforms, dates, exports, criteria, and decisions are recorded and reproducible - Coverage is checked with relevant vocabulary and citation chaining - Limitations, access gaps, duplicates, and the update cutoff are explicit