Every strategy we deliver is built by an information specialist working with a PhD methodologist: your research question is translated into structured search logic, the electronic searches go through formal peer review, and you receive a complete, auditable search log. The methodology on this page is reviewed by Dr. Marcus Lindberg, who holds a Master of Library and Information Science and a PhD and works as an information specialist. You can commission the search strategy on its own or as the opening stage of a full systematic review service. We build searches across health sciences, nursing, psychology, education, social sciences, and beyond, and Research Gold matches you with a methodologist or information specialist experienced in your field's databases.
Every conclusion a review reaches rests on the studies it managed to find. If the search misses relevant trials, the synthesis is biased before a single result is extracted, and no amount of careful screening or statistics later can recover what was never retrieved. This is why a reproducible search strategy is treated as the foundation of a defensible systematic review rather than an administrative step. Reproducibility has a precise meaning here: a third party should be able to take your documentation, paste the exact search lines into the same databases, and return the same result set, give or take database updates over time.
That standard shapes how we work. We record the database and interface, the full search string for each source, the date each search was run, the number of records retrieved, and every limit or filter applied. A peer reviewed search strategy carries weight with journal editors and guideline panels precisely because it can be inspected line by line. When reviewers cannot reproduce a search, they cannot judge whether the evidence base is complete, and the review loses its claim to being systematic at all.
The first technical task is turning a question into searchable concepts. For clinical and health questions we most often use PICO, which breaks the question into Population, Intervention, Comparator, and Outcome. Not every concept belongs in the search; depending on your discipline, a strong strategy is usually built on the two or three concepts that are both specific and well indexed, often Population and Intervention, while sensitive outcomes are left to the screening stage so recall is not sacrificed. Every choice here comes down to balancing sensitivity and precision, so the search retrieves the relevant studies without burying the screening stage in irrelevant records.
For each concept we assemble a block of synonyms, spelling variants, brand and generic names, older and newer terminology, and relevant abbreviations. These blocks are combined with Boolean logic: terms inside a concept are joined with OR, and the concepts are joined with AND. Getting the conceptual structure right at this stage is what makes the later database search strategy development efficient rather than a guessing game. Questions that do not fit PICO cleanly, such as broad mapping or exploratory questions, are often better served by a scoping review service or a wider our evidence synthesis team, and we will tell you when that is the case rather than force a poorly fitting frame.
A search cannot simply be copied from one database to another. Each platform has its own controlled vocabulary, field tags, syntax, and quirks, so we translate the strategy for every source rather than reusing a single string. Our standard coverage includes the following.
Bibliographic databases we build for
- MEDLINE through both Ovid and PubMed, since the two interfaces handle MeSH terms, field tags, and automatic term mapping differently and a strategy tuned for one will not behave identically on the other.
- Embase with its Emtree thesaurus and Elsevier syntax, which indexes European and pharmacology literature and drug concepts that MEDLINE covers less thoroughly.
- Cochrane CENTRAL for controlled trials, which aggregates trial records from multiple sources and is central to intervention reviews.
- CINAHL for nursing and allied health literature, with its own subject headings and field structure.
- PsycINFO for psychology, behavioral science, and mental health, using the APA thesaurus.
- Scopus and Web of Science as large multidisciplinary indexes that add coverage, support powerful citation searching, and catch records the subject databases miss.
For each of these we write the search in the native syntax, map every concept to that database's indexing, and test it before it is finalized. The output is a set of parallel strategies that together give broad, controlled coverage rather than one string bent awkwardly across incompatible platforms.
Strong searches combine two retrieval mechanisms. Controlled vocabulary, such as MeSH in MEDLINE and Emtree in Embase, uses subject headings that indexers assign to records regardless of the words an author chose, which captures relevant studies even when the authors used unusual phrasing. Free-text, or title and abstract searching, catches recent records not yet indexed, terminology the thesaurus does not cover, and concepts that sit between formal headings. Relying on only one mechanism leaves predictable gaps, so every concept block is searched both ways and the results combined.
Within the free-text layer we use the full toolkit each platform offers: Boolean operators to structure the logic, proximity and adjacency operators to require that two terms appear near each other rather than anywhere in a long record, and truncation and wildcards to capture word endings and spelling variants in a single term. Applied carefully these tools raise precision without cutting recall; applied carelessly they silently drop records, which is one of the most common faults a peer review catches.
Validated search filters and hedges
For many review types there are validated search filters, sometimes called hedges, that have been developed and tested to retrieve a particular study design or population, for example randomized controlled trials, diagnostic accuracy studies, or specific age groups. Where a filter has been properly validated and suits your question, we apply it and cite its source so reviewers can see exactly what was used and how well it is known to perform. We do not attach filters uncritically; an ill-fitting or poorly validated filter can quietly exclude eligible studies, so each one is a deliberate, documented choice rather than a default. This judgment about when a filter helps and when it harms recall is exactly the kind of decision that benefits from an experienced information specialist.
Beyond the databases: grey literature, trials registers, and citation searching
Databases alone do not capture the full evidence base, and reviews that stop there risk publication bias, the tendency for studies with positive results to be published and indexed more readily than others. To counter this we extend the search in several directions.
Grey literature searching covers conference abstracts, theses, government and agency reports, and other material outside the commercial journals. Trials-register searching through ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP) surfaces completed, ongoing, and unpublished studies, which is essential for spotting outcome reporting bias. Citation searching, both backward through reference lists and forward through citing articles, plus targeted reference checking of key papers, finds studies that concept-based searches can miss because of unusual terminology. Together these methods make the search comprehensive rather than merely convenient.
Deduplication and a clean record set
Searching seven or eight sources produces heavy overlap, so results are pooled and deduplicated into a single clean set before screening begins. We do this with a documented, reproducible process and keep the raw per-database counts so the numbers reconcile all the way through your PRISMA flow. If you want to visualize that flow, our PRISMA flow diagram generator turns the record counts into a publication-ready figure. Accurate counts at this stage matter more than they appear to: they are what a reader uses to check that nothing was lost between retrieval and inclusion.
Before a search is run in full, it should be independently checked. The Peer Review of Electronic Search Strategies (PRESS) guideline (McGowan et al., 2016) provides a structured framework for a second information specialist to review the search for translation errors, missing terms, misused operators, wrong field tags, and unbalanced logic. A PRESS peer review is now expected by many journals and recommended in the Cochrane Handbook, chapter on searching (Higgins JPT, Thomas J, et al., version 6.5, 2024), and it reliably catches faults that the original searcher, close to the work, tends to miss.
Our searches go through a PRESS-based review as standard. The reviewer checks that the research question is properly translated, that Boolean and proximity operators do what they are meant to, that spelling and syntax are correct for each interface, that subject headings and free-text terms are both present, and that limits and filters are justified. You receive the completed review alongside the final strategy, so the fact that a peer reviewed search strategy was performed is documented rather than merely asserted.
Documentation for PRISMA-S and full reproducibility
The PRISMA-S reporting extension (Rethlefsen et al., 2021) sets out exactly what a search should report so that it can be reproduced and appraised. We document to that standard from the start rather than reconstructing it afterward. Your deliverable includes the complete, un-truncated search string for every database and interface, the platform and vendor for each, the dates of every search, the number of records retrieved per source, all limits and filters with their justification, the grey literature and trials-register sources searched, the citation searching performed, and the deduplication method and counts.
That package is what we mean by a complete search log: enough detail for an editor, a peer reviewer, or a future update team to re-run the search and audit it independently. It slots directly into the methods section of your manuscript and into a registered protocol. If you have not yet registered, our PROSPERO protocol registration service can lock in the search methods before searching begins, which is the sequence guideline panels expect.
Standalone or part of a full review
You can commission the systematic literature search service as a standalone deliverable, for instance when you have a capable review team but no information specialist, or when a journal or supervisor has asked for the search to be peer reviewed. Many clients then continue with us for screening, extraction, and synthesis, and where a quantitative synthesis is planned our commission a meta-analysis picks up directly from the included studies. Whether you take the search alone or the whole review, the search log is built to the same reproducible standard, because a search that cannot be reproduced undermines everything downstream of it. You can read the ask about our published systematic reviews to see the documented search strategies behind work our clients have taken to publication. The same reproducible documentation also underpins a defensible literature review writing service, where a transparent search strengthens the narrative synthesis.
If you want to sketch concept blocks and see the structure before you commission the work, our search strategy builder is a free starting point. When you are ready for an information specialist to build, peer review, and document the real thing, request a quote and we will scope it to your question and target databases.