The design-by-treatment interaction model: a unifying framework for modelling loop inconsistency in network...

Area of Science:

• Statistics
• Biostatistics
• Epidemiology

Background:

• Network meta-analysis (NMA) is increasingly used to synthesize evidence from multiple treatment comparisons.
• Identifying and addressing loop inconsistencies is crucial for reliable NMA results.
• The design-by-treatment interaction model has been proposed as a framework for handling such inconsistencies.

Purpose of the Study:

• To clarify and rigorously prove the claim that the design-by-treatment interaction model includes all possible loop inconsistency models.
• To provide a strong theoretical basis for employing the design-by-treatment interaction model in network meta-analysis.

Main Methods:

• Mathematical clarification and proof of the encompassing nature of the design-by-treatment interaction model.
• Demonstration of how various loop inconsistency models can be represented within this framework.

Main Results:

• The design-by-treatment interaction model has been proven to contain all existing loop inconsistency models.
• This confirms the model's comprehensive capability in addressing loop inconsistencies in NMA.

Conclusions:

• The design-by-treatment interaction model offers a unified and robust approach to modeling loop inconsistencies in network meta-analysis.
• Researchers should consider adopting this model for enhanced accuracy and interpretability of NMA findings.