Concepts and Estimation in Network Meta-Analysis

Network meta-analysis (NMA) extends conventional meta-analysis by letting researchers compare three or more interventions simultaneously and combine evidence from multiple studies. Unlike pairwise meta-analysis, which compares two interventions at a time, NMA integrates direct evidence from head-to-head comparisons and indirect evidence through a common comparator. This approach is particularly useful when clinical trials or other experimental studies have not directly compared all interventions. Students will learn how to interpret the structure of a comparison network, in which interventions are represented as nodes and direct comparisons as connecting links. They will also examine the assumption of transitivity, which requires that differences in study characteristics and potential effect modifiers do not systematically undermine the validity of indirect comparisons.

The second component focuses on NMA's statistical foundations, including consistency and inconsistency between direct and indirect evidence. Students will explore how consistency supports integrating evidence from different sources, while inconsistency may indicate disagreements that require further investigation. The session introduces the basic principles of frequentist and Bayesian approaches to NMA, emphasizing how these frameworks estimate relative effects across multiple interventions. Students will also learn how intervention rankings can be generated to summarize the relative performance of competing treatments or strategies. However, rankings are inherently uncertain and should be interpreted alongside effect estimates, confidence or credible intervals, the quality of the evidence, and the assumptions underlying the model.

The final component addresses the limitations and practical interpretation of NMA results. Sparse or imbalanced networks, limited direct comparisons, differences in study populations, and violations of transitivity or consistency can reduce the reliability of estimated relative effects. Students will learn why an intervention ranked first is not necessarily the most effective or appropriate option in every context, particularly when differences between interventions are small or highly uncertain. Interpretation should therefore consider the magnitude and precision of effect estimates, the structure and completeness of the evidence network, and the research question being addressed. By the end of this session, students should be able to explain the purpose of NMA, distinguish direct from indirect evidence, describe its key assumptions and statistical approaches, interpret relative effects and intervention rankings, and recognize the limitations of using rankings alone to guide scientific or clinical decisions.