Extending treatment effects from a randomized trial using observational data: an application to coronary thrombus aspiration

Background Observational data can be used to complement inferences about treatment effects from randomized trials. To increase confidence in the use of observational data for this purpose, one can first benchmark, that is, demonstrate the observational analysis can replicate an index trials findings, before using the observational data to estimate what the index trial could not estimate. Methods We use observational Swedish registry data to emulate a target trial similar to the Thrombus Aspiration in ST-Elevation myocardial infarction in Scandinavia (TASTE) randomized trial, which found no difference in the risk of death or myocardial infarction by 1 year when comparing percutaneous coronary intervention with and without thrombus aspiration among individuals with ST-elevation myocardial infarction. We benchmark the emulation estimates against the trial estimates at 1 year, then extend the emulations follow-up to 3 years and estimate effects in subpopulations that were underrepresented in the trial. Results Like TASTE, the observational analysis found no difference in the risk of death or myocardial infarction by 1 year in the groups with or without thrombus aspiration (risk difference 0.7 (-0.7, 2.0) and -0.2 (-1.3, 1.0) for death and myocardial infarction respectively), so benchmarking was considered successful. We additionally show no difference in the risk of death or myocardial infarction by 3 years, or within subpopulations by 1 year. Conclusions Benchmarking before using observational data to extend treatment effects from a randomized trial allows us to understand if the observational data, and assumptions made when analyzing these data, can be trusted to deliver valid estimates of treatment effects.