Assessing Methods to Adjust for Publication Bias
Author Information
Author(s): Moreno Santiago G, Sutton Alex J, Ades AE, Stanley Tom D, Abrams Keith R, Peters Jaime L, Cooper Nicola J
Primary Institution: Dept of Health Sciences, University of Leicester
Hypothesis
Can regression-based methods effectively adjust for publication bias in meta-analyses?
Conclusion
Regression-based adjustments for publication bias and small study effects are easy to conduct and generally outperform more established methods.
Supporting Evidence
- Regression-based methods consistently outperformed the Trim & Fill estimators.
- The performance of all methods worsened as unexplained heterogeneity increased.
- Applying methods based on initial tests for funnel plot asymmetry generally provided poorer performance.
Takeaway
This study looked at different ways to fix problems in research caused by only publishing positive results. Some new methods worked better than the old ones.
Methodology
A comprehensive simulation study was conducted to assess the performance of various adjustment methods for publication bias in meta-analyses.
Potential Biases
The study acknowledges that publication bias is a persistent issue that cannot be entirely eliminated.
Limitations
The performance of methods deteriorates as heterogeneity increases and the underlying odds ratio increases.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website