Rise in Cost for Treating Type 2 Diabetes Medicare Beneficiaries

A data-driven-approach study was conducted to identify the cost of spending for long-term care patients with diabetes by accessing data from Medicare Fee-for-Service beneficiaries, including Medicare parts A, B, and D for two years.  

Diabetes is an expensive health condition that requires regular doctor visits, medication therapy, and various other healthcare services and tools depending on the degree of disease. According to an American Diabetes Association report, diabetes has become the most expensive chronic disease in the United States, at an estimated dollar amount of $327 billion. Prescription medications are among the top reasons for the continuous increase in diabetes management costs in the United States. To further analyze and assess the cost of long-term care of patients with diabetes, Dr. Julie Lauffenburger and colleagues at Brigham and Woman’s Hospital and Harvard Medical School conducted a study using data-driven methods to classify Medicare claims for patients with type 2 diabetes. In the study, low utilizers were categorized based on their diabetes care spending patterns over a two-year follow-up period. According to Dr. Lauffenburger, predicting which individuals are likely to be costly is essential for targeting interventions. Current approaches generally focus on composite measures, short time horizons, or patients who are already high utilizers, whose costs may be harder to modify. Implementing a preventative treatment intervention is essential to improve overall health outcomes and manage diabetes-related costs. The objective of Dr. Lauffenburger’s and team’s study was to identify diabetes-related spending patterns for Medicare beneficiaries and use the findings to inform and improve the way future costs are predicted. 

The study used data from a 1-million-member sample of Medicare fee-for-service beneficiaries, including Medicare part A (inpatient services), B (outpatient services), and D (prescription services). Data collected included around 20,000 nationwide quality improvement program beneficiaries and 980,000 randomly selected patients nationally. The majority of the patient data came from Medicare Fee-for-service insurance coverage covering 65 years and older patients in the United States. Specific beneficiaries were identified as “low-cost,” denoting a baseline of previously having lower spending levels or the bottom 90% of diabetes care spending. Their spending patterns were analyzed using administrator claims documented from the insurance company. Group-based trajectory models were used to record diabetes-related expenditures over two years.  

Prediction models were used to assess the following factors: generalized boosted regression, a machine learning method, baseline predictors, diabetes predictors, predictors potentially modified through interventions. These models were furtherly evaluated using C-statics and five cross-validations. The eligibility criteria resulted in the inclusion of 33,789 Medicare beneficiaries with type 2 diabetes. The patients’ mean age was 75.9 years, with 55.8% women who maintained their insurance coverage Medicare A, B, or D from 2011 to 2013. The baseline median diabetes spending was $4,153 and included the “low-cost” group of people in the bottom 90% of the expenditure for 2011. Five distinct spending patterns were identified among the beneficiaries that could be predicted: 68.1% of patients had consistent spending, 25.3% had spending that increased rapidly, and 6.6% of patients had a gradual rise in spending. The prediction ability was moderate, with the validated C-statistics 0.63 to 0.87. Factors contributing to the group of patients that showed a progressive rising in spending were age, generosity of coverage, medication adherence, and prior spending. Each person has a unique total cost, including inpatient visits, outpatient medical visits, physician office visits, and outpatient medications. The results of the two-year spending patterns were measured to be the minimal user group at $4,907, the low-cost group at $8,525, rising cost group where cost continuously increased in the first year at $2,215, a moderate cost group at $9,594, and high-cost group at $8,548.   

This study highlights the essential need to keep costs low by using target interventions to benefit patients. The cost can increase over time due to uncontrolled diabetes, and it is necessary to keep patients continually well controlled on their diabetes medication. Identifying high utilizers can help better understand diabetes management’s needs, which can better inform effective intervention and treatment plans. Diabetes-related costs can range from medications, testing supplies, doctor visits, and increase if diabetes is not adequately treated with additional hospitalizations. 

Practice Pearls: 

• The data-driven approach is beneficial for determining future costs and implementing cost-containment interventions for managing type 2 diabetes.  
• Cost for inpatient visits, outpatient medical, physician visits, and outpatient medications can be used to target interventions earlier to decrease future costs.  
• Approaches to decrease costs need to focus on longer time horizons to have a significant amount of time to implement a targeted treatment plan. 

Garber, MJ. Abrahamson et al. Not There Yet: Using Data-Driven Methods to Predict Who Becomes Costly among Low-Cost Patients with Type 2 Diabetes. January 1, 1970, bmcendocrdisord.biomedcentral.com/articles/10.1186/s12902-020-00609-1. 

American Diabetes Association. “The Cost of Diabetes.” The Cost of Diabetes | ADA, 2020, www.diabetes.org/resources/statistics/cost-diabetes.  

Aleksandra Kusic, PharmD Candidate, Florida A&M University, College of Pharmacy and Pharmaceutical Sciences