Real-world cardiovascular effects of liraglutide: transportability analysis of the LEADER trial

Abstract Appropriate use of recently approved type 2 diabetes treatments depends on external validity of landmark clinical trials (RCTs) in real-world populations that may differ from trial participants. This study transported effect estimates from the Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) trial, a placebo-controlled RCT of liraglutide on cardiovascular outcomes, onto real-world cohorts within the Veterans Affairs (VA) healthcare system. Risk differences (RD) in survival outcomes, approximated using pseudo-observations of individual survival probabilities, were estimated with augmented inverse probability weighting after balancing baseline characteristics between RCT and target samples using approximate balancing weights. Transported effects of liraglutide compared to placebo on major adverse cardiovascular events (MACE) and all-cause mortality in veterans (“VA-weighted LEADER”) were larger than, though statistically consistent with, the treatment effects observed in LEADER: MACE RD at 3 years of 4.6% [95% CI 2.2, 7.0] in VA-weighted LEADER versus 1.6% [0.3, 2.9] in LEADER; all-cause mortality RD at 3 years of 2.9% [0.8, 5.1] in VA-weighted LEADER versus 0.9% [-0.09, 1.9] in LEADER. These estimates of the effects of liraglutide in veterans with diabetes provide real-world evidence that can guide diabetes treatment decisions and formulary policies for a high-risk population underrepresented in RCTs.