Methods for detecting, quantifying, and adjusting for dissemination bias in meta-analysis are described

Area of Science:

• Biostatistics
• Medical Research Methodology
• Evidence Synthesis

Background:

• Dissemination bias, or the nonpublication of studies, can significantly impact meta-analysis results.
• Identifying and accounting for this bias is crucial for accurate evidence synthesis.

Purpose of the Study:

• To systematically review methodological articles on detecting, quantifying, and adjusting for dissemination bias in meta-analyses.
• To evaluate the empirical application and validation of these methods.

Main Methods:

• Comprehensive literature search of Medline, Cochrane Library, and Web of Science.
• Inclusion of 150 full-text articles describing methods for dissemination bias.
• Categorization of methods into graphical, statistical, and recent approaches.

Main Results:

• A wide array of methods for addressing dissemination bias were identified.
• Graphical methods (e.g., funnel plots) and various statistical approaches were common.
• Limited empirical validation of these methods, particularly using unpublished data from regulatory agencies.

Conclusions:

• Existing methods for dissemination bias have limitations and require further validation.
• Thorough literature searches remain essential in systematic reviews.
• Increased availability of all research results is necessary to mitigate bias.