Target Trial Emulation: A Rigorous Framework for Causal Inference in Real-World Data Studies (Motivated by the OneFlorida+ Study on GLP-1RA and SGLT2i Use and Dementia Risk in Type 2 Diabetes by Tang et al.)

Target trial emulation offers a rigorous methodological framework for drawing causal inferences from real-world clinical data, particularly in scenarios where randomized controlled trials (RCTs) are infeasible. This report illustrates the design of emulated trials using the OneFlorida+ study on type 2 diabetes treatments and dementia risk as a motivating example. The study emulated hypothetical RCTs comparing GLP-1 receptor agonists (GLP-1RA), SGLT2 inhibitors (SGLT2i), and other second-line glucose-lowering drugs (GLDs), using longitudinal electronic health record (EHR) data from over 21 million individuals. By specifying eligibility criteria, treatment strategies, follow-up periods, and outcome definitions, and applying inverse probability of treatment weighting (IPTW), the investigators successfully balanced covariates across treatment groups. Results showed significantly lower incidence rates of Alzheimer’s disease and related dementias (ADRD) in patients initiating GLP-1RA and SGLT2i therapies compared to other agents, with no significant difference between the two newer drug classes. Supporting visuals—such as standardized mean difference plots, Kaplan-Meier cumulative incidence curves, and forest plots for subgroup analysis—demonstrate the transparency and robustness of the emulation process. While powerful, target trial emulation depends on data completeness and is limited by potential unmeasured confounding.