October 3, 2013 | Anna Chaimani, Julian P. T. Higgins, Dimitris Mavridis, Panagiota Spyridonos, Georgia Salanti
This paper introduces a set of graphical tools and STATA routines to facilitate the interpretation and presentation of results in network meta-analysis (NMA). NMA integrates direct and indirect evidence from multiple trials comparing different interventions, potentially ranking treatments based on their effectiveness. However, NMA is often criticized for its complexity and the need for advanced statistical and computational skills. To address this, the authors provide practical examples and STATA commands to evaluate model assumptions, present the evidence base, and interpret results. The tools include network plots, contribution plots, inconsistency plots, comparison-adjusted funnel plots, predictive intervals, and ranking plots. These tools help researchers understand the structure of the evidence network, assess the impact of heterogeneity, and interpret treatment rankings. The paper also discusses the use of the `mvmeta` command in STATA for fitting NMA models and emphasizes the importance of careful interpretation of graphical and numerical results.This paper introduces a set of graphical tools and STATA routines to facilitate the interpretation and presentation of results in network meta-analysis (NMA). NMA integrates direct and indirect evidence from multiple trials comparing different interventions, potentially ranking treatments based on their effectiveness. However, NMA is often criticized for its complexity and the need for advanced statistical and computational skills. To address this, the authors provide practical examples and STATA commands to evaluate model assumptions, present the evidence base, and interpret results. The tools include network plots, contribution plots, inconsistency plots, comparison-adjusted funnel plots, predictive intervals, and ranking plots. These tools help researchers understand the structure of the evidence network, assess the impact of heterogeneity, and interpret treatment rankings. The paper also discusses the use of the `mvmeta` command in STATA for fitting NMA models and emphasizes the importance of careful interpretation of graphical and numerical results.