The authors reweighted individual-level results from the LEADER placebo-controlled trial of liraglutide onto Veterans Affairs cohorts with type 2 diabetes, balancing baseline characteristics between the trial and the target population. Transported three-year risk differences were larger than the trial's own, 4.6% (95% CI 2.2-7.0) for major adverse cardiovascular events and 2.9% (0.8-5.1) for all-cause mortality against 1.6% and 0.9%. The confidence intervals overlap, so this is a shift in the central estimate rather than a demonstrated difference.
Why it is interesting: It puts a number on how far a GLP-1 trial's effect moves when carried to a higher-risk population the trial under-represented.