Emulation of the REWIND Cardiovascular Outcomes Trial in Healthcare Claims Data.

This is a non-randomized, non-interventional study that is part of the Randomized Controlled Trials Duplicated Using Prospective Longitudinal Insurance Claims: Applying Techniques of Epidemiology (RCT-DUPLICATE) initiative (www.rctduplicate.org) of the Brigham and Women's Hospital, Harvard Medical School. It is intended to emulate, as closely as possible in healthcare insurance claims data, the REWIND trial described below. Although many features of the trial cannot be directly replicated in healthcare claims, key design features, including outcomes, exposures, and inclusion/exclusion criteria, were selected to proxy those features from the trial. In addition to closely emulating the trial population, this study also evaluates outcomes in an expanded cohort following the eligibility criteria outlined in a set of completed emulation studies (NCT06659744, NCT07088718, NCT07096063) to enhance generalizability to patients typically encountered in clinical practice. Randomization cannot be emulated in healthcare claims data but was proxied through a statistical balancing of measured covariates according to standard practice. Investigators assume that the RCT provides the reference standard treatment effect estimate and that failure to replicate RCT findings is indicative of the inadequacy of the healthcare claims data for emulation for a range of possible reasons and does not provide information on the validity of the original RCT finding.

The REWIND (NCT01394952) trial is a superiority trial that evaluated the effect of dulaglutide, a glucagon-like peptide-1 receptor agonist (GLP-1-RA), vs placebo on time to first occurrence of any major adverse cardiovascular event (MACE), defined as cardiovascular death, myocardial infarction, or stroke among patients with type 2 diabetes mellitus (T2DM) with and without previous cardiovascular disease (CVD).

The database study designed to emulate the REWIND trial will be a new-user active comparative study, where we compare the effect of dulaglutide vs sitagliptin, a dipeptidyl peptidase-4 inhibitor (DPP4i), on MACE among patients with T2DM with and without previous CVD. While the REWIND trial compared dulaglutide vs placebo, we chose to use sitagliptin as an active-comparator proxy for placebo. Sitagliptin was specifically chosen because a major randomized controlled trial on cardiovascular outcomes demonstrated that the drug does not affect the cardiovascular outcomes under investigation. Furthermore, clinical guidelines during the study period recommended both drug classes under investigation as second- or third-line options for glucose lowering and were similarly costly.