Background: Diabetes is a significant cause of cardiovascular disease (CVD), kidney and eye diseases, and mortality. Diabetes and its complications also generate substantial healthcare utilization. Medications contribute to both the benefits and costs of diabetes care. Newer diabetes medications have been developed and marketed, but there remains a mismatch in our understanding of their net benefits, risks, and cost- effectiveness. This is particularly relevant for older adults (≥65 years), where benefits and risks are more finely balanced. Clinical practice guidelines recommend initial treatment with metformin across all age groups but are unclear about next steps when metformin is ineffective. This is a frequent treatment transition. Medications such as glucagon-like peptide-1 receptor agonists (GLP1) and sodium-glucose co-transporter-2 inhibitors (SGLT2) have favorable effects on CVD and renal disease but also carry greater costs and side-effects. Within VA, their usage is accelerating. It remains uncertain if these medications have specific advantages over the frequently used generic medications, sulfonylureas (SU). Thus, we hypothesize that diabetes medication usage is influenced by a complex interplay of patient factors and provider practice patterns. We further hypothesize that when studied using real-world data, GLP1 and SGLT2 medications will reduce major adverse outcomes and be cost-effective in comparison to SU when used as add-on treatment to metformin. Significance: There are major gaps in the evidence base that informs clinical and cost-effective diabetes treatment decisions among older Veterans (≥65 years), which comprise >70% of the VA diabetes cohort. Innovation and Impact: We will employ a large nationwide sample of Veterans with diabetes who are dually enrolled in VA/Medicare and apply advanced methods to reduce the impact of unmeasured factors. By using real-world data, our study will add significant new information that informs the care of older Veterans, VA’s diabetes and pharmacy programs, and clinical guidelines. Specific Aims: Aim 1. Determine prescribing patterns for SU, GLP1 and SGLT2. We will examine the extent to which patient characteristics and clinician prescribing patterns influence initiating an SU, GLP1 or SGLT2 as add-on to metformin. Aim 2. Estimate the relationships between SU, GLP1, and SGLT2 usage as add-on to metformin with healthcare utilization and adverse outcomes. Using clinician prescribing patterns as an instrumental variable, we will measure the effects on emergency department (ED) visits, hospitalizations, healthcare costs, hypoglycemia events, medication adherence, new diabetes complications and mortality. Aim 3. Investigate the cost-effectiveness of initiating an SU, GLP1 or SGLT2 as add-on to metformin. Methodology: This is a retrospective observational study of secondary data from patient-level administrative and claims data from VA and Medicare, and uses a quasi-experimental design involving i...