EVALUATING GLP-1 RECEPTOR AGONIST (RA) UTILIZATION AND HEALTHCARE RESOURCE IMPACT IN ADULTS WITH TYPE 2 DIABETES, PRE-DIABETES, OR OBESITY USING PRIVACY-PRESERVING SYNTHETIC EHR DATA
Author(s)
Margaret Hogan, MSN RN1, Julie Johnson, PhD MPH RN1, Marcel Miron-Celis, MSc2, Emelly Rusli, MPH2, Katherine Jean, MS, RN, NI-BC2, Aaron Galaznik, MBA, MD2.
1University of Chicago, Chicago, IL, USA, 2MDClone, San Jose, CA, USA.
1University of Chicago, Chicago, IL, USA, 2MDClone, San Jose, CA, USA.
OBJECTIVES: To characterize longitudinal patterns of GLP-1 RA utilization and associated outcomes and healthcare resource utilization (HRU) in adults with type 2 diabetes, pre-diabetes, or obesity, while demonstrating the feasibility of conducting real-world outcomes research using synthetic EHR data that preserves population representativeness without accessing patient identifiers.
METHODS: Two linked retrospective observational studies analyzed a synthetic derivative of University of Chicago Medicine EHR data (January 2022-December 2025). Study population: adults ≥18 years with ≥2 encounters (12 months apart) documenting pre-diabetes, type 2 diabetes, or obesity (BMI >30); N 1,595. Study 1 examined GLP-1 RA initiation prevalence, medication types, and specialty prescriber patterns stratified by indication and year. Study 2 evaluated 409 patients with documented GLP-1 RA initiation, measuring HbA1c and BMI at 6 and 12 months post-initiation, and ED visits and hospitalizations pre/post-initiation, stratified by indication and race. Synthetic methodology enabled analysis while eliminating exposure to protected health information.
RESULTS: In Study 1, 21% of eligible patients initiated GLP-1 RA, with Semaglutide (63%) and Dulaglutide (32%) predominating. Initiation varied by indication (T2DM 19.25% vs. obesity 1.07%), and 39% were prescribed by endocrinology specialists. In Study 2, mean HbA1c decreased 0.45 percentage points and BMI decreased 0.8 kg/m² at 12 months. Unique patients visiting the ED decreased by 31.9%, with hospitalization rates decreasing by 25.7%. These outcomes varied significantly by race and GLP-1 RA indication.
CONCLUSIONS: Synthetic EHR derivatives enable rigorous, privacy-preserving outcomes research achieving population-level fidelity without identifier exposure. GLP-1 RA utilization patterns and healthcare impact are reliably characterizable through this approach, supporting scaled adoption of synthetic data methods in clinical research infrastructure. Next steps include further exploration of post-GLP-1 RA initiation resource utilization changes by race and GLP-1 RA indication.
METHODS: Two linked retrospective observational studies analyzed a synthetic derivative of University of Chicago Medicine EHR data (January 2022-December 2025). Study population: adults ≥18 years with ≥2 encounters (12 months apart) documenting pre-diabetes, type 2 diabetes, or obesity (BMI >30); N 1,595. Study 1 examined GLP-1 RA initiation prevalence, medication types, and specialty prescriber patterns stratified by indication and year. Study 2 evaluated 409 patients with documented GLP-1 RA initiation, measuring HbA1c and BMI at 6 and 12 months post-initiation, and ED visits and hospitalizations pre/post-initiation, stratified by indication and race. Synthetic methodology enabled analysis while eliminating exposure to protected health information.
RESULTS: In Study 1, 21% of eligible patients initiated GLP-1 RA, with Semaglutide (63%) and Dulaglutide (32%) predominating. Initiation varied by indication (T2DM 19.25% vs. obesity 1.07%), and 39% were prescribed by endocrinology specialists. In Study 2, mean HbA1c decreased 0.45 percentage points and BMI decreased 0.8 kg/m² at 12 months. Unique patients visiting the ED decreased by 31.9%, with hospitalization rates decreasing by 25.7%. These outcomes varied significantly by race and GLP-1 RA indication.
CONCLUSIONS: Synthetic EHR derivatives enable rigorous, privacy-preserving outcomes research achieving population-level fidelity without identifier exposure. GLP-1 RA utilization patterns and healthcare impact are reliably characterizable through this approach, supporting scaled adoption of synthetic data methods in clinical research infrastructure. Next steps include further exploration of post-GLP-1 RA initiation resource utilization changes by race and GLP-1 RA indication.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
HSD4
Topic
Clinical Outcomes, Health Service Delivery & Process of Care, Real World Data & Information Systems
Disease
Diabetes/Endocrine/Metabolic Disorders (including obesity)