Comparative Effectiveness of SGLT-2 Inhibitors versus GLP-1 Receptor Agonists on Recurrent Cardiovascular Events in Patients with Type 2 Diabetes After Acute Ischemic Stroke: A Target Trial Emulation Using Chinese Insurance Claims Data

Read full text Summarise with VIRAT
Authors: Yuliang Xiang
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Background: Among patients with type 2 diabetes mellitus (T2DM) hospitalized for acute ischemic stroke (AIS), the comparative effectiveness of SGLT-2 inhibitors (SGLT-2i) versus GLP-1 receptor agonists (GLP-1RA) on recurrent cardiovascular events remains unknown. No randomized trial or target trial emulation has been conducted in this high-risk, post-stroke population. Objective: To estimate the comparative effectiveness of initiating SGLT-2i versus GLP-1RA after discharge from an index AIS hospitalization on major adverse cardiovascular events (MACE-H: recurrent ischemic stroke, acute myocardial infarction, or hospitalization for heart failure) and recurrent ischemic stroke in patients with T2DM. Design: Target trial emulation using a retrospective active-comparator new-user (ACNU) design. Primary estimand: intention-to-treat (ITT) effect. Secondary estimand: per-protocol (PP) effect via clone-censor-weighting (CCW) with inverse probability of censoring weighting (IPCW). Data source: A Chinese municipal medical insurance claims database covering inpatient and outpatient records from 2021 to 2024 (n = 60,432 first AIS hospitalizations; n ≈ 32,165 with T2DM). Population: Adults with confirmed T2DM who initiated either an SGLT-2i or GLP-1RA within 90 days after discharge from a first AIS hospitalization (ICD-10: I63.x), with no prior use of the initiated drug class in the preceding 12 months. Exposure: Initiation of any SGLT-2i (SGLT-2i arm, n = 2,005) versus any GLP-1RA (GLP-1RA arm, n = 358) as the first post-discharge dispensing. Primary outcome: MACE-H composite (first occurrence of recurrent ischemic stroke, acute MI, or hospitalization for heart failure). Secondary: recurrent ischemic stroke; safety outcomes including hemorrhagic stroke, hypoglycemia, DKA, and malnutrition. Statistical methods: Propensity score-based inverse probability of treatment weighting (IPTW) using stabilized weights. Primary analysis: IPTW-weighted Cox proportional hazards model with robust sandwich standard errors. Per-protocol analysis: weighted pooled logistic regression (PLR). Balance assessed by standardized mean differences (SMD < 0.10). Sensitivity analyses: double-robust estimation, grace period sensitivity (120d/180d), landmark Cox analysis for proportional hazards assumption violations, negative control outcome analysis (infection hospitalization), E-values for unmeasured confounding, and pre-specified subgroup analyses (7 subgroups). Methodologic validation via Protocol A: emulation of EMPA-REG OUTCOME design (SGLT-2i vs. DPP-4i in T2DM with established CVD) to confirm data and method credibility prior to interpreting primary results. Registration note: The study uses a pre-existing administrative claims database; data collection was complete prior to protocol drafting. This registration is filed retrospectively for transparency. The protocol (v1.0, 2026-03-24) was finalized prior to submission.

Fields of Study & Subjects

Diabetes Treatment and Management Acute Ischemic Stroke Management Diabetes Cardiovascular Risks and Lipoproteins

Publication details

Publication date
1 January 2026
Language
EN
Document type
Preprint