Causally-interpretable random-effects meta-analysis

Area of Science:

• Biostatistics
• Epidemiology
• Causal Inference

Background:

• Causally-interpretable meta-analysis aims to transport treatment effects from randomized trials to specific populations.
• Heterogeneity between studies, unrelated to treatment effect modifiers, complicates synthesizing estimates.
• Existing methods may struggle with this type of between-study variation.

Purpose of the Study:

• To propose a conceptual framework and estimation procedures to account for between-study heterogeneity in causal meta-analysis.
• To develop inferential techniques for capturing excess variability in causal estimates.
• To clarify the types of treatment effects suitable for generalizability and transportability techniques.

Main Methods:

• Development of a novel conceptual framework for causal meta-analysis.
• Introduction of new estimation procedures to handle unexplained heterogeneity.
• Formulation of inferential techniques to manage excess variability in causal estimates.

Main Results:

• The proposed framework accounts for between-study heterogeneity not explained by known modifiers.
• New methods capture excess variability, improving the reliability of causal estimates.
• Clarification of the scope of treatment effects addressable by generalizability and transportability.

Conclusions:

• The developed framework enhances the interpretability and relevance of meta-analysis findings for specific populations.
• The methods provide a robust approach to synthesizing evidence despite complex heterogeneity.
• This work advances the application of causal inference in meta-analytic research.