Assessing Methods for Transporting Trial Treatment Effects to Target Real-World Populations Without Individual-Level Data

OBJECTIVES: Transportability methods provide a tool to extend results from randomized controlled trials or observational studies to real-world populations, typically relying on individual-level data. This study sought to validate and compare existing transportability methods that use aggregate data from target populations in the context of a different therapy area.

METHODS: Using methods previously tested for a cardiovascular prevention trial, we extended a treatment effect from the NIDA-CTN-0044 trial to a target population of U.S. substance abuse treatment seekers, represented by the 2021 Treatment Episode Data Set (TEDS): Admissions dataset. Three methods using aggregate data from the target population (weighting using simulated individual data, weighting using method of moments, and post-stratification) were compared against the gold-standard method: weighting using individual-level data. Transported odds ratios were reported for each method alongside 95% confidence intervals using robust standard errors (except for post-stratification). Seven common covariates were included as categorical or binary variables.

RESULTS: The information reported here results from secondary analyses of data from clinical trials conducted by the National Institute on Drug Abuse (NIDA). Specifically, data from NIDA-CTN-0044 Web-delivery of Evidence-Based, Psychosocial Treatment for Substance Use Disorders were included. A total of 406 participants in the NIDA-CTN-0044 study and a target population of 350,536 in the TEDS were identified. Weighting using method of moments yielded estimates closest to the gold-standard estimate transported the trial. For the NIDA-CTN-0044 trial, the expected treatment effect (odds ratio (OR)=1.32, 95% CI 0.76-2.28) obtained for the target population using the gold-standard approach was mostly closely approximated by the method of moments (OR=1.31, 95% CI 0.70-2.43), followed by the simulated data approach (OR=1.28, 95% CI 0.72-2.29) and post-stratification (OR=1.46).

CONCLUSIONS: Weighting using method of moments provided transported estimates most comparable to the gold-standard method. Future research should further validate aggregate-based transportability methods across different therapy areas and settings.