Tirzepatide vs. Dulaglutide cardiovascular outcomes emulated in 44,671 claims records to validate real-world data methods

Real-world data (RWD) from healthcare claims offer vast potential for understanding drug effects in diverse populations, especially for conditions like Type 2 Diabetes and cardiovascular (CV) risk. However, confidently applying RWD analyses to answer clinical questions, particularly for Major Adverse Cardiac Events (MACE), remains a challenge due to inherent biases and lack of randomization. This study aims to bridge this gap by rigorously emulating a completed randomized controlled trial (RCT), SURPASS-CVOT, to establish an empirical evidence base for RWD reliability.

This non-randomized, non-interventional study, part of the RCT-DUPLICATE initiative, emulated the SURPASS-CVOT trial using healthcare insurance claims data for 44,671 individuals. The exposure group was defined by new use of tirzepatide dispensing claims, while the reference group used new dulaglutide dispensing claims. Key design features, including outcomes and inclusion/exclusion criteria, were proxied from the original trial. Randomization, which is not possible in claims data, was statistically proxied through balancing measured covariates according to standard epidemiological practice to assess the validity of RWD for trial emulation.

The study's primary objective was to build an empirical evidence base for real-world data (RWD) by conducting a large-scale emulation of the SURPASS-CVOT trial. While the abstract describes the methodological approach and goals, it does not present specific numerical findings or statistical comparisons between tirzepatide and dulaglutide from the completed emulation itself. The investigators aimed to understand the types of clinical questions for which RWD analyses can be conducted with confidence and how to implement such studies effectively. They acknowledge that many features of the original RCT, such as direct randomization, cannot be perfectly replicated in claims data. The study's completion implies the generation of data to assess how closely the RWD emulation's treatment effect estimate aligns with the RCT's reference standard. The findings, once published, will inform the utility and limitations of using claims data for predicting or replicating complex cardiovascular outcome trials.