Transportability in healthcare research: A review of practical applications and statistical methods

Transportability assesses if study results (from Randomized Controlled Trials (RCTs) or Real-World Data (RWD)) can be applied to different target populations, including other real-world settings, geographic regions, or specific subgroups. Although Health Technology Assessment agencies advocate for transportability to optimize resources and improve decision-making, there is little guidance on specific methods to achieve it. This review aims to identify statistical methods to achieve transportability applied from RCTs or RWD to real-world populations.

A literature review was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. We identified articles by searching MEDLINE and Embase until 18th October 2023. The review focused on studies describing and applying statistical methods for transporting evidence from RCTs to real-world populations and from RWD to other RWD contexts. To be included, a manuscript should use statistical methods to address transportability and present practical, health-related examples of transportability. If the publications focused on methodological aspects without practical examples or aimed to test model validity, they were excluded.

The search identified 441 records and 13 were included for analysis. Of these, ten used Inverse Odds of Sampling Weights (IOSW), one used Inverse Probability Weighting (IPW), one used G-computation and another used standardization. Most studies applied transportability methods to transport RCT findings to real-world populations, two to transport between RWD sources for applications across disease areas or geographic locations. Common limitations reported included unmeasured confounding, differences in patient management and motivation, and challenges related to sample size.

IOSW was the primary method used for transportability, predominantly within the same country when individual-level data from both the study population and target population could be combined. Our findings reveal that practical applications of transportability remain largely limited to national contexts, with few cross-country implementations observed.