Translating trial results into interpretable risk estimates: Systematic analysis of cardiorenal outcome trials of glucagon-like peptide-1 receptor agonists and sodium-glucose cotransporter-2 inhibitors

Abstract

Background and aimsIn a randomised controlled trial (RCT), the between-arm difference in the average probability of an event per unit of time (i.e., yearly incidence risk difference, YIRD) is an easy-to-interpret treatment effect metric. We aimed to quantify the YIRD in cardiorenal RCTs of GLP-1RAs or SGLT-2is. Methods and resultsWe digitally searched for RCTs published up to March 1st, 2023, including subjects with type 2 diabetes randomised to GLP-1RAs or SGLT-2is and investigating cardiorenal outcomes or death. We extracted information from Kaplan-Meier (KM) plots to obtain time-to-event individual data and estimate within-arm yearly incidence risk and YIRD. Data from 19 RCTs (28 kM plots) were analysed: comparing treatment to placebo, in GLP-1RA RCTs the YIRD ranged from 0.2 % (95 % CI: −0.7 %, 1.1 %) to −1.9 % (−3.1, −0.7), for primary outcome; and from −0.2 % (−0.5, 0.2) to −0.4 % (−0.7 %, −0.0 %), for mortality. With the exception of SOLOIST-WHF (YIRD 11.9 % for primary outcome), corresponding estimates in SGLT-2is RCTs were: from −0.1 % (−0.4, 0.1) to −5.0 % (−7.7, −2.6), for primary outcome; and from −0.1 % (−0.2, 0.1) to −1.9 % (−4.4 %, 0.6 %), for mortality. ConclusionThe YIRD metric complements other relative treatment effect estimates and helps quantify the absolute benefit of GLP-1RAs and SGLT-2is.