Meta-analysis Spurious precision? Meta-analysis of observational studies
Population
100 randomly selected articles published in 1995 and indexed in Medline with keyword 'meta-analysis'
Design
Review
Authors
Discussion
Loading...
Member takes
Implication
Indexing errors affect 40% of MeSH-tagged meta-analyses; warrants clinician caution and leaves open validity of observational data synthesis.
Key Points
- Examine the prevalence, characteristics, and validity of meta-analyses conducted on observational epidemiological studies relative to randomized controlled trials.
- Screened 566 Medline-indexed articles from 1995 assigned the MeSH term 'meta-analysis' (excluding letters).
- Randomly selected and characterized a subset of 100 articles to evaluate publication type and underlying study designs.
- Identified 60 actual meta-analyses alongside 40 methodological papers, editorials, and traditional reviews in the sample.
- Found approximately half of actual meta-analyses were based on observational studies, predominantly cohort and case-control studies assessing etiology or medical interventions.
Structured PICO
P
Population
100 randomly selected articles published in 1995 and indexed in Medline with keyword 'meta-analysis'
O
Outcome
Characteristics of the articles
Highlights the increasing prevalence of meta-analyses based on observational data and the need to understand their specific potentials and pitfalls.
Cite This Study
Egger et al. (1998) studied this question.