Learning about treatment effects in a new target population under transportability assumptions for relative effect...

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Investigators often assume conditional relative effect measures are transportable across populations.
• Transportability of causal effects is crucial for applying trial findings to new settings.

Purpose of the Study:

• To examine the identification of causal effects in a target population under the assumption of transportable conditional relative effect measures.
• To assess the compatibility of transportability for relative versus difference effect measures.
• To develop methods for estimating treatment effectiveness in target populations.

Main Methods:

• Investigated the compatibility of transportability assumptions for relative and difference effect measures.
• Developed methods for identifying marginal causal estimands using transportable relative effect measures.
• Extended results to scenarios with partial exchangeability and limited covariate data.

Main Results:

• Transportability of relative effect measures is largely incompatible with that of difference effect measures, barring specific conditions.
• Methods were derived for identifying population-averaged causal effects under relative effect measure transportability.
• Identification is possible even when only a subset of confounding covariates is needed for transportability.

Conclusions:

• The study provides a framework for causal effect identification using transportable relative effect measures.
• Findings highlight the distinct assumptions required for relative versus difference effect measure transportability.
• Proposed estimators are implementable in standard statistical software for real-world applications.