Emulating Six Cardiovascular Outcome Trials of Antidiabetic Drugs in Individuals with Type 2 Diabetes: Analyses Based on Real-World Databases from Korea and Taiwan.

Abstract

With the increasing volume of clinical evidence derived from large-scale Asian real-world data (RWD) and the growing interest in its use in regulatory decision-making, it is important to conduct benchmarking exercises that evaluate whether studies using Asian RWD can generate similar conclusions as randomized controlled trials (RCTs). We aimed to assess whether observational studies based on Korea and Taiwan RWD can yield comparable results with trials by emulating six cardiovascular outcome trials (CVOTs) of antidiabetic drugs in individuals with type 2 diabetes (T2D). We emulated six CVOTs using nationwide claims of Korea and Taiwan. An active comparator, new-user design was applied, and observational analogues to the eligibility criteria and outcomes of the corresponding RCT were implemented. Propensity score matching was utilized to balance the treatment groups. Hazard ratios and 95% confidence intervals (CI) were estimated and compared with corresponding RCT estimates. We used three pre-specified binary agreement metrics that have been used in prior benchmarking studies to define concordance in results. Results from each of the six emulations were concordant with the corresponding CVOT on ≥ 1 binary agreement metric. Five out of six emulations indicated superiority when the corresponding CVOT only demonstrated non-inferiority. Results from emulations were more concordant with Asian-specific results from RCT, with four emulations meeting all agreement metrics. In this binational study using two Asian healthcare claims data, emulations yielded comparable clinical conclusions with the corresponding RCT, increasing the confidence in the validity of RWE studies in patients with T2D using these databases.