Non-Insulin Antidiabetic Agents & Utilization

The U.S. Food and Drug Administration (FDA) initiated this study in the Sentinel System to explore whether aggregated electronic health record (EHR) system data could be used to complement administrative claims data in evaluating the safety of non-insulin antidiabetic medications. The study utilized the TriNetX Live™ platform to evaluate the utilization patterns and user demographics for incident new users of specific non-insulin antidiabetic medications, including second-generation sulfonylurea, dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, metformin, and sodium-glucose co-transporter 2 (SGLT2) inhibitors. Additionally, the study assessed the capture of obesity/overweight diagnoses and body mass index (BMI) for incident new users of GLP-1 receptor agonists, regardless of drug indications. 

The results showed utilization patterns for various non-insulin antidiabetic medications. Notably, the results showed that the number of incident new users of GLP-1 receptor agonists and SGLT2 inhibitors increased over time. In the EHR data, race information was available for over 86% of the non-insulin antidiabetic drug users studied. Among all users of GLP-1 receptor agonists in the study, approximately 46% had a recorded Body Mass Index (BMI), while 24% had a documented overweight/obesity diagnosis, both within 183 days prior to their index prescription. These findings indicate that aggregated EHR data captured certain patient demographics, such as race, for a high proportion of patients in our study, suggesting potential for improved capture compared to claims data. However, important characteristics like BMI remained incompletely available in EHR data. Based on the insights gained from this study, future safety evaluations for non-insulin antidiabetic medications could consider integrating EHR data with claims data, particularly for demographics like race. However, for variables like BMI, a careful assessment of EHR data quality and completeness should precede any integration to ensure improved feasibility.