Second-line Therapies for Patients With Type 2 Diabetes and Moderate Cardiovascular Disease Risk

Overview

We will use the target trial framework for causal inference to conduct this observational retrospective cohort study that uses claims data of adults with type 2 diabetes (T2D) included in the de-identified datasets of OptumLabs Data Warehouse (OLDW) and Medicare fee-for-service. In Aim 1, we will emulate a target trial comparing the effectiveness of glucagon-like peptide-1 receptor agonists (GLP-1RA), sodium-glucose cotransporter 2 inhibitors (SGLT2i), dipeptidyl peptidase-4 inhibitors (DPP-4i), and sulfonylureas (SU) in adults with T2D at moderate risk of cardiovascular disease (CVD) with regard to major adverse cardiovascular events (MACE), expanded MACE, microvascular complications, severe hypoglycemia, and other adverse events. In Aim 2, we will compare these four drug classes in the same population of adults with T2D included in OLDW and Medicare fee-for-service data with respect to a set of composite outcomes identified by a group of patients with T2D as being most important to them. Specifically, in Aim 2A, we will prospectively elicit patient preferences toward various treatment outcomes (e.g., hospitalization, kidney disease) using a participatory ranking exercise, then use these rankings to generate individually weighted composite outcomes. Then, in Aim 2B, we will estimate patient-centered treatment effects of four different second-line T2D medications that reflect the patient's value for each outcome. In Aim 3, we will compare different medications within each of the four therapeutic classes with respect to MACE.

Study Design: We will use the target trial framework for causal inference to conduct this observational cohort study. Comparators: Aims 1-2 compare the GLP-1RA, SGLT2i, DPP-4i, and SU classes, while Aim 3 compares the individual drugs within each therapeutic class. Population: Using data from OptumLabs Data Warehouse linked to 100% Medicare FFS claims, we will identify adults (≥21 years) with T2D at moderate risk for CVD who started a GLP-1RA, SGLT2i, DPP-4i, or SU Outcomes: In AIMs 1 and 3, the primary outcome will be time to MACE (non-fatal MI, non-fatal stroke, all-cause mortality). Secondary outcomes will include times to expanded MACE (MACE, HF hospitalizations, revascularization procedures) and its components, lower extremity complications, severe hypoglycemia, microvascular complications, and other significant adverse events. In AIM 2A, we will elicit patient preferences toward various treatment outcomes using a participatory ranking exercise, use these rankings to generate individually weighted composite outcomes, and then estimate patient-centered treatment effects of GLP-1RA, SGLT2i, DPP4i, and SU reflecting the patient values for each of the outcomes. Timeframe: January 1, 2014 to December 31, 2021. Methods: Inverse probability weighting will be used to emulate baseline randomization for pairwise comparisons between the drug classes (AIMs 1-2) and individual drugs within each class (AIM 3). Causal cumulative incidence rates will be estimated in the weighted sample using the targeted maximum likelihood estimator adjusting for time-dependent confounding and loss-to-follow-up.