Description of network meta-analysis geometry: A metrics design study

Area of Science:

• Biostatistics
• Network Meta-Analysis
• Graph Theory

Background:

• Network meta-analysis (NMA) reporting, including network plots, requires transparency.
• Existing NMA visualizations may lack standardized descriptive metrics.

Purpose of the Study:

• To propose quantifiable metrics from graph theory and social network analysis to describe NMA geometry.
• To enhance the transparency and interpretability of NMA network plots.

Main Methods:

• A systematic review of NMAs was conducted, and network plots were reproduced.
• Eleven geometric metrics were evaluated using statistical correlation and concordance tests.
• Reproducibility of network plots was assessed based on provided information.

Main Results:

• Only 167 out of 477 NMAs had reproducible network plots.
• Key metrics describing NMA geometry include density, median thickness, common comparators, and strong edges.
• Certain metrics like mean thickness and average path length may yield misleading conclusions.

Conclusions:

• Seven simple metrics are recommended for reporting NMA geometry.
• Editors and peer reviewers should enforce strict NMA reporting guidelines.