New research co-authored by Providence CARDS found striking disparities in off-label prescribing of GLP1-RA medications such as semaglutide and tirzepatide between different areas of the US, and by race, ethnicity, income, and social vulnerability.
The past few years have seen a massive surge in prescriptions for glucagon-like peptide 1 receptor agonists (GLP-1 RAs), a type of medication initially approved for the treatment of type 2 diabetes and more recently for weight loss. Media hype and the weight loss effects have made these drugs, such as semaglutide and tirzepatide, appealing to the average American regardless of medical need. Yet in the face of drug supply shortages and varying insurance coverage, this has resulted in GLP-1 RAs being less available to patient populations that are most in need.
A new study, co-authored by researchers from the Providence Center for Cardiovascular Analytics, Research + Data Science (CARDS), found that nearly 40 percent of GLP-1 RAs approved by the FDA for diabetes are being prescribed off-label, meaning prescribed to patients who are not diagnosed with diabetes. The researchers also found striking disparities in off-label prescribing between different areas of the U.S., and by race, ethnicity, income and social vulnerability. Areas with higher incomes were associated with increased off-label prescribing, while areas with lower incomes and higher social risk had fewer off-label prescriptions. The findings were published in the American Journal of Medicine Open.
According to the researchers, this study highlights the need for further steps to address barriers to accessing these life-changing medications.
Read on for highlights, or check out the full paper by clicking here.
Findings highlight variations in GLP-1 RA prescribing linked to location, income and social vulnerability
The research shows how off-label use of GLP-1 RAs varies depending on factors like location, income and social vulnerability. Key findings include:
- Higher household income was modestly correlated with a higher off-label prescribing rate.
- The highest off-label prescribing rate (51.6%) occurred in a cluster of counties in Hawaii with a high median income ($92,124).
- Areas with predominantly White residents, high median household income and low social vulnerability had the second highest off-label prescription rate (42.2%).
- The lowest off-label prescribing rate (31.2%) occurred in a cluster of counties that included American Indian Tribal reservation lands, with low median income ($52,437) and high social vulnerability (0.88).
What can we learn from this research? Kateri Spinelli, PhD, research manager and clinical researcher at CARDS explains "these findings show that access to these potentially life-changing medications is heavily influenced by income and location. This study points to the need for regional approaches that consider both the clinical and social needs of underserved patients to help overcome barriers to access for these treatments,” she adds.
About the study’s methods: Uncovering new insights with unsupervised machine learning
To conduct the study, researchers from CARDS and the healthcare analytics and research company Trilliant Health analyzed over 3.6 million GLP-1 RA prescriptions given from January 1, 2022, to December 31, 2022, using Trilliant’s all-payer claims database. The database includes prescription drug claims from all 50 U.S. states and the District of Columbia.
They also drew on data from the U.S. Census Bureau and the Centers for Disease Control and Prevention (CDC) to evaluate how prescription patterns might vary based on race, ethnicity, household income and communities’ SVI (social vulnerability index). SVI is a measurement that evaluates communities’ ability to respond to natural disasters and public health crises.
The researchers used several methods to evaluate the data, including examining the relationship between off-label prescribing rates in different counties and measures of health disparities. The team started with traditional statistical analysis approaches, but it wasn’t until they employed an unsupervised machine learning model that the connections between location, income, race, ethnicity and social vulnerability were discovered. Unsupervised machine learning is a cutting-edge statistical approach similar to Artificial Intelligence (AI) that is one of the many skills of the Providence Biostatistical Hub, a collaboration between CARDS and the Providence Health Research Accelerator. Using this approach, the researchers were able to group similar areas of the U.S. together based on shared characteristics to reveal patterns that were not clear with standard approaches.
“Collaboration was key to the success of this project,” says Dr. Spinelli. “Our partners at Trilliant Health brought invaluable market insights and provided the all-payer claims data, our CARDS team brought our research expertise and passion for health equity and the Biostatistical Hub brought creative, cutting-edge methods. Together, we uncovered important findings about equitable access to GLP-1 RAs that we may not have discovered on our own.”
Read more about the findings, methods and prior research at https://www.sciencedirect.com/science/article/pii/S2667036425000147
Related resources
