Reviewer(s):
Juilet Brown
Alan Lovell
MS Copilot
Full Reference:
Hartling, L., Featherstone, R., Nuspl, M., Shave, K., Dryden, D.M., & Vandermeer, B. (2016). The contribution of databases to the results of systematic reviews: a cross-sectional study. BMC Medical Research Methodology, 16:127. DOI: 10.1186/s12874-016-0232-1.
Short description:
This cross-sectional study examined the impact of selective database searching on the results of meta-analyses within systematic reviews. The authors analyzed 129 Cochrane reviews from three clinical areas—Acute Respiratory Infections (ARI), Infectious Diseases (ID), and Developmental Psychosocial and Learning Problems (DPLP)—to assess how restricting searches to Medline and one additional database affected the identification of relevant studies and the statistical outcomes of meta-analyses.
Findings showed that most relevant studies were indexed in a limited number of databases, with Medline and Embase or BIOSIS yielding the highest coverage for ARI and ID, and Medline plus PsycINFO for DPLP. Restricting searches to these combinations rarely changed the statistical significance of meta-analyses, and changes in effect estimates were generally small and non-systematic. The study suggests that selective searching may be sufficient for many systematic reviews, especially in mainstream healthcare topics, and could inform more efficient approaches to rapid reviews.
Limitations stated by the author(s):
Results may not generalize beyond the three clinical areas studied or to non-conventional interventions.
The study used completed reviews and did not assess the quality of the original search strategies.
Analyses were based on whether studies were present in databases, not whether they could be retrieved by search strategies.
Some studies may have been indexed after the original searches, potentially inflating database coverage.
CENTRAL and citation databases (e.g., Scopus, Web of Science) were excluded to avoid bias.
Only Medline via Ovid and Embase via Ovid were assessed, excluding other interfaces like PubMed or Embase.com.
Limitations stated by the reviewer(s):
The retrospective design limits insights into real-time search performance and retrieval effectiveness.
The study does not evaluate the impact of grey literature or non-indexed sources.
The focus on primary outcomes may overlook variability in secondary outcomes.
The exclusion of citation databases omits a potentially valuable source of relevant studies.
The study assumes that presence in a database equates to discoverability, which may not hold in practice.
Study Type:
Cross-Sectional Quantitative Analysis (Retrospective Evaluation of Database Contributions to Systematic Reviews)
Related Chapters:
Tags:
Databases