Background: The high cost and widespread use of glucagon-like peptide 1 receptor agonists (GLP-1RAs) are a concern for healthcare budgets. Whether GLP-1RA use reduces other health care spending is unclear. Methods: We conducted a cohort study using insurance claims data for United States adults aged 18-64 from 2016-2024, matching GLP-1RA treated members with untreated members (controls) on baseline demographics, clinical conditions, hospitalization, and medication use. Primary outcomes were per member per month (PMPM) healthcare costs overall and by service type, analyzed separately for members with and without diabetes. Results: Among 742,824 matched treated and control members, 55.6% had diabetes. In year 1 following GLP-1RA initiation, total costs were 68.7% higher in treated members (95% CI, 68.0%- 69.4%, $743 PMPM difference); in years 2-6 costs were 44.8% higher (95% CI, 43.7%- 45.9%; $530 PMPM difference). Excluding GLP-1RA costs, treated members had 5.8% higher costs in year 1 (95% CI 5.1%- 6.5%) and 4.1% higher costs (95% CI 3.0% – 5.2%) in years 2-6. Among treated members with diabetes, cost increases were modest: 3.8% (95% CI 2.8% – 4.8%) in year 1 and 0.8% in years 2-6 (95% CI 0.6%- 2.2%), with non-GLP-1RA pharmacy and provider visits offset by reduced admissions and dialysis. Treated members without diabetes had more substantial cost increases: 8.9% in year 1 (95% CI 7.7% – 10.1%) and 9.7% in years 2-6 (95% CI 8.0% – 11.4%). Conclusions: GLP-1RA treatment was associated with increases in spending on healthcare net of the GLP-1RA cost, particularly in members without diabetes.
The Hidden Power of Normalization: Exponential Capacity Control in Deep Neural Networks
arXiv:2511.00958v1 Announce Type: cross Abstract: Normalization methods are fundamental components of modern deep neural networks (DNNs). Empirically, they are known to stabilize optimization dynamics and