Toward Causally Interpretable Meta-analysis: Transporting Inferences from Multiple Randomized Trials to a New Target...

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Meta-analysis traditionally synthesizes findings from multiple studies but often lacks causal interpretability for new populations.
• Transporting causal inferences requires robust methods to address variations between trial populations and target populations.

Purpose of the Study:

• To develop and describe methods for transporting causal inferences from randomized trials to a target population.
• To establish identifiability conditions for average treatment effects in the target population.
• To provide practical tools and estimators for implementing causal inference transport in meta-analysis.

Main Methods:

• Describing methods for transporting causal inferences from individual trials and pooled trial data.
• Discussing identifiability conditions for average treatment effects in the target population.
• Proposing average treatment effect estimators using various working models and providing implementation code.

Main Results:

• Identification results for average treatment effects in the target population are provided.
• The implications of transportability assumptions on the underlying data-generating process are elucidated.
• Methods for assessing homogeneity of transported inferences and sensitivity analyses are discussed.

Conclusions:

• The proposed methods enable causally interpretable meta-analysis by transporting inferences to new populations.
• The study provides practical tools for researchers to generalize findings from randomized trials.
• Extensions address challenges like nonadherence, enhancing the applicability of causal transport methods.