Linear inference for mixed treatment comparison meta-analysis: A two-stage approach

Area of Science:

• Biostatistics
• Medical Research Methodology
• Evidence Synthesis

Background:

• Mixed treatment comparisons (MTC) meta-analysis integrates comparative evidence from multiple treatments in randomized controlled trials (RCTs).
• Understanding consistency between direct and indirect evidence is crucial for robust MTC.
• Existing methods require examination of MTC properties and consistency concepts.

Purpose of the Study:

• To examine the properties of mixed treatment comparisons (MTC) estimates.
• To elucidate the concept of consistency between direct and indirect evidence in MTC networks.
• To present a two-stage framework for consistent MTC synthesis.

Main Methods:

• Decomposition of MTC synthesis into two stages: direct meta-analysis and weighted least squares regression.
• Utilizing a specific design matrix to represent consistency conditions for optimal estimates.
• Employing leverage statistics and regression residuals for influence diagnosis and inconsistency assessment.

Main Results:

• Consistent MTC estimates are derived as linear combinations of direct estimates.
• A likelihood-ratio statistic allows for testing overall evidence consistency under normality assumptions.
• The two-stage framework facilitates detailed diagnostic analysis of MTC data.

Conclusions:

• The proposed two-stage method provides a consistent framework for MTC meta-analysis.
• This approach enhances the ability to assess evidence consistency and identify data influences.
• The methodology is applicable and illustrated with examples from medical research.