A Causal Inference Guided Analysis of Lung Transplant Outcomes Using Real-world Evidence from the SRTR Database

The increasing availability of real-world data (RWD) has allowed studies to use controls external to the clinical trial involving the actively treated patients. With RWD initiatives at the FDA and the recent FDA approval of several products utilizing RWD in the US, statistical methods have become an even more visible component in analyzing RWD for regulatory purposes. We reanalyzed Erdman et al. (2022) who evaluated immunosuppressive agents' performance using a retrospective cohort study to analyze transplant-related outcomes over time in adult lung transplant recipients in the Scientific Registry of Transplant Recipients (SRTR) using Cox models. We requested the data (n = 22,085) to be reanalyzed using causal inferences' potential outcome with propensity score (PS) matching, inverse probability weighting, and doubly robust estimation (DRE) for a composite endpoint of mortality or graft failure within a 3-year period from transplant. Results generally aligned with those of Erdman et al., although the magnitude of drug effects in some cohorts diminished considerably with PS matching and DRE as expected.