Database

In research, a database is a structured collection of records or data organized for efficient search, retrieval, and management. Academic databases such as PubMed, Scopus, Web of Science, and Embase index published research articles and provide search interfaces for locating relevant literature. Clinical databases store patient records, trial data, or registry information. The term encompasses both bibliographic databases (which index publications) and research data databases (which store datasets), each serving distinct but complementary functions in the research workflow.

Databases are the foundation of systematic literature searching, evidence synthesis, and data driven research. A systematic review that searches only one database risks missing relevant studies, since no single database provides complete coverage of the literature. PubMed covers biomedical literature but may miss social science or engineering papers indexed in Scopus or Web of Science. Regional databases such as LILACS (Latin America and Caribbean), CNKI (China), and African Journals Online extend coverage to literature that international databases may underrepresent. Understanding the scope, indexing practices, and search capabilities of different databases is essential for conducting comprehensive literature searches that meet the methodological standards of systematic reviews and evidence based guidelines.

Bibliographic databases index published articles by recording standardized metadata: titles, authors, abstracts, publication details, and controlled vocabulary terms (such as MeSH terms in PubMed or Emtree terms in Embase). Users search these databases using keyword queries, controlled vocabulary terms, Boolean operators (AND, OR, NOT), and filters for date, language, publication type, and other criteria. Advanced search features include proximity operators, truncation, field specific searching, and the ability to save and combine searches.

Each database has its own controlled vocabulary, indexing depth, and coverage scope. PubMed contains more than 40 million citations, primarily in biomedical and life sciences; MEDLINE, the part indexed with MeSH terms, is its largest component. Scopus covers over 100 million records across sciences, social sciences, and humanities. Web of Science provides citation indexing, enabling forward citation searching (finding articles that cite a known article). Embase emphasizes pharmacological and drug related literature. Cochrane Library focuses on systematic reviews and clinical trials.

Research data databases operate differently, storing actual datasets rather than bibliographic records. These include institutional repositories, domain specific archives (GenBank, the Protein Data Bank), and multidisciplinary platforms (Zenodo, Figshare). Their search interfaces allow users to find datasets by topic, creator, date, geographic coverage, and data type.

Clinical databases, such as electronic health record systems, patient registries, and claims databases, store patient level information used for observational research, quality improvement, and health services research. These databases require special governance procedures, including ethics approval and data use agreements, due to the sensitive nature of the information they contain.

Bibliographic databases should be used whenever conducting a literature search, whether for a systematic review, a grant application, a thesis, or a general exploration of a research topic. Systematic reviews require searching multiple databases to ensure comprehensive coverage. Research data databases should be used when looking for existing datasets for secondary analysis or when depositing data for sharing. Clinical databases are used when the research question can be addressed using existing patient records or administrative data rather than new data collection.

A common mistake is relying on a single database for a literature search, particularly for systematic reviews where comprehensive coverage is a methodological requirement. Another error is using only free text keywords without incorporating controlled vocabulary terms, missing articles that use different terminology for the same concept. Researchers sometimes fail to document their search strategy in enough detail for it to be reproduced, undermining the transparency of their review. Confusing database coverage (the scope of what a database indexes) with search results (what a particular search retrieves) can lead to overconfidence in the completeness of a search. Finally, neglecting to check for database specific features and limitations, such as the lag time between publication and indexing, can result in missing recent publications.

A systematic review team searching for studies on the effectiveness of mindfulness interventions for chronic pain conducts searches in PubMed, Embase, PsycINFO, CINAHL, and the Cochrane Central Register of Controlled Trials. The PubMed search uses MeSH terms ("Mindfulness," "Chronic Pain") combined with free text synonyms, yielding 1,847 records. Adding the four additional databases contributes 1,203 unique records not found in PubMed, although extra records show the value of multi database searching only if some of the studies finally included are among them. After deduplication, the team screens 3,050 unique records, of which 42 meet inclusion criteria.