6C-04 - Transporting the LEADER Study Results to VA Cohorts: Assessing Liraglutide's Impact on Cardiovascular Outcomes in a Veteran Population

Background: Real-world evaluations of clinical trial results have gained substantial interest for pharmacoepidemiology applications as a bridge between drug efficacy and effectiveness.

Objectives: The LEADER trial assessed liraglutide’s impact on the composite of major cardiovascular events (MACE), including cardiovascular (CV) mortality, myocardial infarction (MI), and stroke. This study transports effect estimates from LEADER onto nested cohorts within the Veterans Affairs (VA) healthcare system.

Methods: We use semiparametric efficient estimators combined with approximate balancing weights, which yield more optimal effective sample sizes in finite sample settings, to estimate the transported target average treatment effect (TATE) of liraglutide versus placebo plus standard of care on MACE, MI, stroke, and CV mortality. Comparisons are made across VA subpopulations, from those approximating the inclusion/exclusion criteria of the LEADER trial to a broader VA diabetes cohort between 2015 and 2023. Survival outcomes are approximated using pseudo-observations of individual survival probabilities (i.e., the probability of avoiding a MACE before a given time). This indirect approach to analyzing time-to-event data enables the use of standard causal inference software tools, which often lack options for accommodating survival outcomes.

Results: Results demonstrate consistent cardiovascular benefits of liraglutide across all VA subpopulations, with of the TATE of MI and stroke similar to those observed in the LEADER trial population, and a larger TATE for CV mortality (1.50% absolute survival increase after 3 years, 95% CI: 0.62%-2.39%) and hence the composite MACE (2.01% absolute survival increase after 3 years 95% CI: 0.80%-3.22%).

Conclusions: The VA is the largest healthcare system in the US, making it a rich environment for investigating the effectiveness of new therapies in a real-world setting. We demonstrate one path for evaluating medication effectiveness by transporting treatment effects from clinical trial data to a real-world target population. The results of our analysis conclude that the CV benefits of liraglutide generalize to veterans with diabetes and that liraglutide may yield a greater improvement to MACE survival in the VA beyond what was observed in the LEADER trial.