The question must be answerable, focused, and very specific. The problem, population, intervention, comparator, and outcomes must be stated.
The inclusion and exclusion criteria must make sense and be clearly stated. Types of literature can be defined (eg, only peer-reviewed journals, books). Types of study design are precisely defined (eg, if only randomized controlled trials are included).
The sources of the search should be defined by the eligibility criteria. Hard copies of articles can be searched manually, and general, discipline-specific, and specialized databases can be searched online.
Reference lists can provide further sources of information. Other references can be suggested by experts and colleagues, librarians, authors of searched articles, and sponsors/funders of research.
The range of possible sources is wider if the search includes conference proceedings, theses, government websites, preprints, company reports/studies, and online trial registries. Languages and date span of publication must also be defined.
For online searching, it is useful to specify certain combinations of keywords (surrounded by quotation marks) and use Boolean logic (eg, AND, OR, NOT).