Target trial emulation of sodium glucose cotransporter 2 inhibitors and clinical outcomes in diabetes and end stage kidney disease

Abstract

Abstract Patients with end-stage kidney disease (ESKD) have been largely excluded from randomized trials of sodium–glucose cotransporter-2 inhibitors (SGLT2is). Despite the lack of guideline recommendations, SGLT2i prescriptions occur in real-world clinical practice. We aimed to describe real-world associations between SGLT2i exposure and clinical outcomes among patients with type 2 diabetes mellitus (T2DM) coded with ESKD. We conducted a target trail emulation with retrospective, new-user, active-comparator cohort study using the TriNetX US Collaborative Network (2016–2023). Adults with T2DM and ESKD who initiated an SGLT2i or a dipeptidyl peptidase-4 inhibitor (DPP4i) were included. Propensity score matching (1:1) was used to balance baseline characteristics. The primary outcome was all-cause mortality; secondary outcomes included sepsis, pneumonia, major adverse cardiovascular events (MACE), all-cause hospitalization, and emergency department visits. Subgroup analyses were exploratory, and heterogeneity was assessed using Cochran’s statistics. After matching, 5295 SGLT2i users were compared with 5295 DPP4i users. Over a follow-up of up to 4 years, SGLT2i exposure was associated with lower all-cause mortality (hazard ratio [HR] 0.90, 95% confidence interval [CI] 0.84–0.97), sepsis (HR 0.87, 95% CI 0.79–0.95), and all-cause hospitalization (HR 0.93, 95% CI 0.89–0.97). No significant associations were observed for MACE, pneumonia, or emergency department visits. Subgroup-specific estimates varied in magnitude, with no consistent evidence of heterogeneity. In this large real-world cohort of patients coded with ESKD, SGLT2i exposure was associated with favorable outcome patterns compared with DPP4i. Given the observational design, potential misclassification of kidney disease status, and off-label drug use, these findings should be interpreted as hypothesis-generating and do not establish causality.