Patterns and costs associated with glucagon-like peptide-1 receptor agonist use in US adults with type 2 diabetes

Abstract

BACKGROUND:

Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) offer cardiorenal benefits in diabetes management. Since 2020, public awareness of GLP-1 RAs for diabetes, weight loss, and the prevention of cardiovascular disease has led to a surge in their utilization. However, the high cost of GLP-1 RAs and limitations in insurance coverage have been considered significant barriers to access. Current knowledge regarding how GLP-1 RA use affects total health care costs in diabetes care after 2020 in the United States remains limited. Consequently, further research is needed to examine the financial burden of GLP-1 RA use on patients and payers, as well as its overall impact on total health care costs at the national level.

OBJECTIVE:

To examine GLP-1 RA utilization and association with health care costs among US adults with type 2 diabetes.

METHODS:

Using data from the 2021-2022 Medical Expenditure Panel Survey, the study sample included individuals (aged ≥18 years) with a diagnosis of type 2 diabetes. Outcomes included GLP-1 RA use and all-cause and diabetes-related health care costs, including medical and prescription drug costs paid by patients and insurers. Generalized linear regression with a log link and gamma distribution was used to assess the effect of GLP-1 RA use on health care costs, adjusting for sociodemographic and health-related characteristics.

RESULTS:

Among 3,587 eligible adults with type 2 diabetes, 637 (18.8%) used GLP-1 RAs, representing an estimated 3.66 million US adults—a marked increase compared with pre-2020 estimates of less than 10%. Fewer older adults (aged ≥65 years) used GLP-1 RAs (35.1%) compared with adults aged 45-64 years (50.6%). The average annual per-person cost of GLP-1 RA was $6,947. Although insurance covered more than 95% of GLP-1 RA cost, these medications represented a substantial proportion of diabetes care costs: 63.3% of antidiabetic drug costs and 55.7% of total diabetes-related costs among GLP-1 RA users. After adjustment, GLP-1 RA use was associated with a 219% increase in diabetes-related costs and a 55.3% increase in total all-cause health care costs.

CONCLUSIONS:

GLP-1 RA utilization among US adults with type 2 diabetes has substantially increased, with use in 2021-2022 nearly double that of the period prior to 2020. The higher health care costs associated with taking GLP-1 RAs were largely attributable to high drug costs, of which over 95% were covered by insurance.

Plain language summary

Our study found a substantial increase in glucagon-like peptide-1 receptor agonist (GLP-1 RA) utilization among US adults with type 2 diabetes, rising from pre-2020 levels of less than 10% to nearly 20% in 2021-2022. Older adults used them less. Although insurance covered about 95% of GLP-1 RA costs, these medications were associated with significantly higher diabetes-related and total health care expenditures.

Implications for managed care pharmacy

Our study demonstrated a significant increase in total health care costs associated with GLP-1 RA use in diabetes care. The high overall cost also places a substantial financial burden on health insurance plans, as the majority of GLP-1 RA costs are covered by insurance. The increasing use of GLP-1 RAs highlights the crucial role of prescription drug coverage in ensuring equitable and affordable access.

Glucagon-like peptide-1 receptor agonists (GLP-1 RAs), such as semaglutide, liraglutide, and dulaglutide, are a class of antidiabetic medications that mimic the action of the naturally occurring hormone GLP-1, effectively managing type 2 diabetes.^1,2 These agents offer multiple clinical benefits, including improved glycemic control, weight loss through delayed gastric emptying and appetite reduction, and mitigation of cardiovascular risk.^1–4 The role of GLP-1 RAs in diabetes treatment has evolved over time since their initial US Food and Drug Administration (FDA) approval in 2005. Initially introduced as a second-line therapy for type 2 diabetes, GLP-1 RAs were used in combination with metformin to improve glycemic control.⁵ In 2016, the American Diabetes Association treatment guidelines added a new section on obesity management and recommended GLP-1 RAs for their weight loss benefits.⁵ In 2020, recommendations were extended to treat diabetes-related comorbidities, regardless of hemoglobin A1c levels, particularly for patients with predominant atherosclerotic cardiovascular disease (ASCVD), heart failure, and chronic kidney disease.⁶ Consequently, GLP-1 RAs are recommended not only for glycemic and weight management but also for reducing cardiovascular events.⁷

Despite the established benefits, prior to 2020, GLP-1 RA utilization among eligible patients was low, ranging from 3.4% to 9.2% in the United States.^8–13 The low prevalence was interpreted as a “lag time” in translating novel therapies into widespread clinical practice.^8,14 However, since 2020, increased public awareness of GLP-1 RAs for diabetes, weight loss, and cardiovascular disease prevention has led to a surge in their use.^15,16 Among the new GLP-1 RA users, the proportion of patients without diabetes but with obesity and associated comorbidities increased from about 10% to 20% in the United States from 2019 to 2023.¹⁵ This shift in use patterns necessitates reevaluating GLP-1 RA utilization in type 2 diabetes using more recent data.

Beyond clinical benefits, GLP-1 RAs have the potential to reduce or offset non–drug-related diabetes care costs through their positive effects on weight loss and cardiorenal risk reduction.^2,17,18 Despite these clinical and potential economic benefits, the high cost of GLP-1 RAs and limitations in insurance coverage have been considered significant barriers to access.¹⁹ High out-of-pocket drug costs have been identified as a factor limiting treatment initiation and reducing medication adherence in diabetes care.^20–22 For example, among patients with Medicare Advantage and private insurance, those in the highest quartile of out-of-pocket costs (75th percentile or above) were 13% less likely to initiate GLP-1 RA therapy than those in the lowest quartile (25th percentile or below).²³ Although a previous study indicated generally high coverage for GLP-1 RAs in Medicare Part D plans, beneficiaries may still face formulary restrictions and higher tiers associated with substantial cost sharing, potentially rendering these medications unaffordable for older populations.²⁴ There is a lack of studies from a nationally representative perspective using data after 2020 to evaluate the costs associated with GLP-1 RA use for patients and payers in the United States. Consequently, further research is needed to examine the financial burden of GLP-1 RA use and its overall effect on total health care costs at the national level.

This study used the most recent US Medical Expenditure Panel Survey (MEPS) data to examine GLP-1 RA use patterns in diabetes management following increased public awareness and to assess the effect of GLP-1 RA use on health care costs.

Methods

DATA SOURCE

This study used data from the MEPS Household Component, a large-scale set of surveys conducted by the Agency for Healthcare Research and Quality. MEPS is a primary data source on how people in the United States access and finance health care, including data on health care utilization, prescription drug use, payment sources, medical conditions, and insurance coverage.²⁵ The health care utilization and expenditure data reported by households are supplemented and verified by medical providers through the Medical Providers Component. The final reconciled and provider-validated estimates provided in the Household Component were used in this study.²⁶ Traditionally, MEPS uses a 2-year overlapping panel design involving 5 rounds of household interviews. However, because of the pandemic, panels 23 and 24 were extended to 9 rounds (4 years) of data collection. This study included individuals from MEPS 2021 (panels 23 and 25) and 2022 (panels 24, 26, and 27) to maintain data independence. This study was determined to be exempt from institutional review board oversight.

STUDY SAMPLE

Eligible respondents in MEPS 2021 and 2022 Household Component data met the following inclusion criteria: (1) assigned positive person-level weight, indicating an individual with valid and representative data that can be used in the derivation of national estimates from the complex survey design of MEPS; (2) aged 18 years or older; (3) a diagnosis of type 2 diabetes; and (4) current use of GLP-1 RAs or other antidiabetic medications. Type 2 diabetes diagnoses were identified using Clinical Classification Software Refined (CCSR) codes derived from medical condition files (END002, END003, and END005). Developed by the Healthcare Cost and Utilization Project (HCUP), CCSR codes aggregate International Classification of Diseases, Tenth Revision, Clinical Modification diagnosis codes into clinically meaningful categories. After applying the inclusion criteria, eligible respondents with diabetes were divided into two groups: GLP-1 RA users and non–GLP-1 RA users.

Antidiabetic medication use was identified from the prescribed medicines files using Multum Lexicon Variables from Cerner Multum, including Multum Therapeutic Sub-Class 1(99) for antidiabetic agents and sub-subclass for various types of antidiabetic agents, including sulfonylureas (213), metformin (214), insulin (215), thiazolidinediones (271), dipeptidyl peptidase-4 inhibitors (371), GLP-1 RAs (373), and sodium-glucose cotransporter-2 (SGLT2) inhibitors (458). Supplementary Table 1 ^{(128.6KB, pdf)} (available in online article) shows the full list of antidiabetic drugs used in the study sample.

OUTCOMES

This study primarily assessed annual per-person direct costs associated with GLP-1 RAs, diabetes-related health care utilization, and all-cause health care utilization, including total payments, out-of-pocket payments, and the amounts paid by insurance, as reported in the Household Component. Diabetes-related health care costs included medical care and pharmacy costs associated with diabetes diagnoses. Diabetes-related medical care costs were identified by CCSR codes from linked medical event and condition files. Medical care costs included costs associated with office visits, hospital outpatient care, emergency department visits, inpatient stays, and home health care services. Diabetes-related pharmacy costs included antidiabetic drug costs identified by Multum therapeutic subclass variables in prescribed medicines files. All-cause health care costs included medical care and pharmacy costs for all conditions.

SOCIODEMOGRAPHIC AND HEALTH-RELATED CHARACTERISTICS

The following sociodemographic variables were selected from MEPS consolidated data files to describe the characteristics of the study sample: age (18-44, 45-64, and ≥65 years), sex, race and ethnicity (Hispanic, non-Hispanic White, non-Hispanic Black, and other), region (Northeast, Midwest, South, and West), health insurance (private, public, and uninsured), and income level defined by the percentage of the federal poverty line (negative/poor: <125%; low: 125%-199%; middle: 200%-399%; high: ≥400%).

Comorbid ASCVD and body mass index (BMI) were used to reflect health characteristics. ASCVD was identified by CCSR codes from the medical condition files, including coronary heart disease, cerebrovascular disease, peripheral artery disease, and aortic atherosclerotic disease. The BMI was used to categorize individuals as underweight/normal (<25), overweight (25-29.9), and obesity (≥30). The underweight and normal weight categories were combined because of the small proportion of the underweight group.

STATISTICAL ANALYSIS

The person-level weights were applied to the analysis to obtain national estimates. The weighted relative frequencies and 95% CIs were reported for the study sample characteristics. Means and 95% CIs were reported for unadjusted cost outcomes, and t-tests were used to compare the unadjusted cost outcomes between GLP-1 RA and non–GLP-1 RA groups. The proportions of prescription drug costs in the total health care costs were described in the GLP-1 RA and non–GLP-1 RA groups, respectively. Generalized linear regression models (log link and gamma distribution) were used to examine the associations between GLP-1 RA use (independent variable) and 4 cost outcomes (dependent variables), respectively, including antidiabetic drug costs, total diabetes-related health care costs, all-cause prescription drug costs, and total all-cause health care costs. The regression models were adjusted for sociodemographic and health-related characteristics, as detailed in the methods section. The sum of out-of-pocket and insurance payments was used in the regression analysis. The log coefficients from the regression models were exponentiated and interpreted as percentage changes to reflect the effect of GLP-1 RA use on cost outcomes. The statistical significance was set at P < 0.05. The statistical analysis was performed by using SAS 9.4.

Results

Among 3,587 eligible respondents with type 2 diabetes, 637 (18.8%) used GLP-1 RAs, representing an estimated 3.66 million US adults (Figure 1). Table 1 describes the characteristics of the study sample. Compared with the non–GLP-1 RA group, the GLP-1 RA group had significantly higher proportions of individuals reporting age 45-64 years (50.6% vs 38.1%, P < 0.001), non-Hispanic White race (64.8% vs 55.5%, P = 0.002), high income (41.9% vs 34.1%, P = 0.015), private insurance coverage (64.6% vs 51.4%, P < 0.001), obesity (73.1% vs 50.3%, P < 0.001), insulin use (38.5% vs 25.4%, P < 0.001), and SGLT2 inhibitor use (26.9% vs 11.8%, P < 0.001).

FIGURE 1.

Study Sample

GLP-1 RA = glucagon-like peptide-1 receptor agonist; MEPS = Medical Expenditure Panel Survey.

TABLE 1.

Characteristics of Study Sample by GLP-1 RA Use (n=3,587, Weighted n=19,421,687)

Characteristics	GLP-1 RA users, weighted % (95% CI) (n=637)	Non−GLP-1 RA users, weighted % (95% CI) (n = 2,950)	P value
Age			<0.001
18-44 years	14.4 (10.6-18.1)	10.7 (9.2-12.3)
45-64 years	50.6 (46.1-55.1)	38.1 (35.8-40.5)
65+ years	35.1 (30.6-39.5)	51.1 (48.7-53.5)
Sex			0.675
Male	49.7 (44.7-54.8)	50.9 (48.5-53.3)
Female	50.3 (45.2-55.5)	49.1 (46.7-51.5)
Race and ethnicity			0.002
Hispanic	13.2 (9.4-16.9)	18.2 (15.5-21.0)
Non-Hispanic White	64.8 (59.6-69.9)	55.5 (52.2-58.9)
Non-Hispanic Black	15.6 (11.4-19.7)	15.3 (13.1-17.4)
Other	6.5 (4.1-8.9)	11.0 (8.7-13.3)
Income			0.015
Negative/poor	19.6 (15.6-23.5)	20.7 (18.7-22.8)
low	10.7 (8.1-13.2)	14.4 (12.8-16.0)
Middle	27.8 (23.6-32.1)	30.7 (28.4-32.9)
High	41.9 (36.6-47.3)	34.1 (31.7-36.6)
Region			0.093
Northeast	17.1 (12.0- 22.1)	15.6 (12.7-18.5)
Midwest	22.6 (17.9-27.3)	21.1 (17.6-24.6)
South	44.8 (38.4-51.2)	41.6 (37.4-46.2)
West	15.5 (11.5-19.6)	21.5 (18.0-25.1)
Insurance			<0.001
Private	64.6 (59.8-69.4)	51.4 (48.9-54.0)
Publica	33.2 (28.5-38.0)	45.7 (43.2-48.1)
Uninsured	2.2 (0.5-3.9)	2.9 (2.0-3.8)
BMI			<0.001
Underweight/normal	8.5 (5.3-11.7)	16.1 (13.9-18.3)
Overweight	18.3 (14.0-22.6)	33.6 (30.9-36.3)
Obesity	73.1 (68.3-78.0)	50.3 (47.4-53.3)
ASCVD	13.3 (10.2-16.3)	14.7 (13.1-16.4)	0.385
Other antidiabetic medication use
Insulin	38.5 (33.5-43.4)	25.4 (23.4-27.4)	<0.001
Metformin	66.0 (61.3-70.7)	65.4 (63.2-67.5)	0.818
SLGT2	26.9 (22.8-31.0)	11.8 (10.3-13.3)	<0.001
Sulfonylurea	21.0 (17.2-24.9)	22.1 (20.1-24.0)	0.621
DPP-4 inhibitor	6.7 (4.1-9.3)	8.9 (7.5-10.3)	0.181
TZD	5.7 (3.3-8.1)	3.8 (2.9-4.6)	0.083

^a Public insurance included Medicare, Medicaid, Veterans Affairs, and other public hospital/physician coverage.

ASCVD = atherosclerotic cardiovascular disease; BMI = body mass index; DPP-4 = dipeptidyl peptidase; GLP-1 RA = glucagon-like peptide-1 receptor agonist; TZD = thiazolidinedione; SLGT = sodium-glucose cotransporter.

Table 2 presents unadjusted annual per-person costs for diabetes in GLP-1 RA vs non–GLP-1 RA groups. First, in the GLP-1 RA group, total GLP-1 RA costs ($6,974) represented 63.3% of total antidiabetic drug costs ($10,974) and 55.7% of total diabetes-related health care costs ($12,478). Insurance covered 96.3% of the total costs of GLP-1 RA. Antidiabetic drug costs accounted for a higher proportion of total diabetes-related health care costs in the GLP-1 RA group (87.9%) than in the non–GLP-1 RA group (68.4%). Second, compared with the non–GLP-1 RA group, the GLP-1-RA group showed 312% higher antidiabetic drug costs ($10,974 vs $2,665, P < 0.001) and 220% higher total diabetes-related health care costs ($12,478 vs $3,899, P < 0.001). Finally, insurance covered the majority of antidiabetic drug costs in both groups (96% GLP-1 RA group vs 93% non–GLP-1 RA group). Both out-of-pocket ($411 vs $178, P < 0.001) and insurance payments ($10,563 vs $2,478, P < 0.001) for antidiabetic drugs were significantly higher in the GLP-1 RA group than in the non–GLP-1 RA group.

TABLE 2.

Unadjusted Annual Per-Person Diabetes-Related Health Care Costs and GLP-1 RA Cost in US Dollars by GLP-1 RA Use (n = 3,587, Weighted n = 19,421,687)

Direct cost category	Group	Out-of-pocket (95% CI), $	Paid by insurance (95% CI), $	Total costsa (95% CI), $
Medical care	GLP-1 RA	79 (59-99)	1,425 (1,043-1,807)	1,504 (1,118-1,891)
	Non−GLP-1 RA	113 (47-179)	1,121 (882-1,360)	1,234 (979-1,489)
	P value	0.326	0.204	0.269
Antidiabetic drugs	GLP-1 RA	411 (343-480)	10,563 (9,689-11,437)	10,974 (10,088-11,861)
	Non−GLP-1 RA	178 (123-234)	2,487 (2,050-2,924)	2,665 (2,215-3,116)
	P value	<0.001	<0.001	<0.001
Total health careb	GLP-1 RA	490 (418-563)	11,988 (11,026-12,950)	12,478 (11,500-13,458)
	Non−GLP-1 RA	291 (204-380)	3,608 (3,102-4,114)	3,899 (3,372-4,427)
	P value	<0.001	<0.001	<0.001
GLP-1 RA		260 (204-317)	6,687 (6,105-7,269)	6,947 (6,357-7,537)

^a Total costs = out-of-pocket payment + insurance payment.

^b Total direct health care costs = medical care costs + antidiabetic drug costs.

GLP-1 RA = glucagon-like peptide-1 receptor agonist.

Table 3 compares unadjusted annual per-person all-cause health care costs in GLP-1 RA vs non–GLP-1 RA groups. First, in the GLP-1 RA group, total prescription drug costs represented 58.6% of total health care costs, compared with 33.3% in the non–GLP-1 RA group. This translates to 182% greater prescription drug costs in the GLP-1 RA group ($14,232 vs $5,054, P < 0.001). Second, although the total health care costs in the GLP-1 RA group ($24,276) were 60% higher than that in the non–GLP-1 group ($15,184, P < 0.001), medical care costs showed no significant difference ($10,043 vs $10,130, P = 0.927). Finally, insurance covered a substantial proportion of prescription drug costs in both groups: 95.3% in the GLP-1 RA group and 91.6% in the non–GLP-1 RA group. Compared with the non–GLP-1 RA group, out-of-pocket ($669 vs $423, P < 0.001) and insurance payments ($13,563 vs $4,631, P < 0.001) for prescription drugs were significantly higher in the GLP-1 group.

TABLE 3.

Unadjusted Annual Per-Person All-Cause Health Care Costs in US Dollars by GLP-1 RA Use (n=3,587, Weighted n=19,421,687)

Direct cost category	Group	Out-of-pocket (95% CI), $	Paid by insurance (95% CI), $	Total costsa (95% CI), $
Medical care	GLP-1 RA	1,116 (852-1,381)	8,927 (7,280-10,575)	10,043 (8,382-11,704)
	Non−GLP-1 RA	1,309 (1,014-1,604)	8,821 (8,163-9,479)	10,130 (9,400-10,860)
	P value	0.318	0.908	0.927
Prescription drugs	GLP-1 RA	669 (582-756)	13,563 (12,131-14,996)	14,232 (12,772-15,694)
	Non−GLP-1 RA	423 (357-488)	4,631 (4,082-5,181)	5,054 (4,485-5,623)
	P value	<0.001	<0.001	<0.001
Total health careb	GLP-1 RA	1,785 (1,510-2,061)	22,491 (20,322-24,659)	24,276 (22,106-26,447)
	Non−GLP-1 RA	1,732 (1,423-2,040)	13,452 (12,547-14,357)	15,184 (14,196-16,172)
	P value	0.785	<0.001	<0.001

^a Total costs = out-of-pocket payment + insurance payment.

^b Total direct health care costs = medical care costs + prescription drug costs.

GLP-1 RA = glucagon-like peptide-1 receptor agonist.

Table 4 shows the effect of GLP-1 RA use on all-cause and diabetes-related costs after adjusting for covariates. GLP-1 RA use significantly contributed to diabetes-related costs, with total antidiabetic drug costs for the GLP-1 RA group being 286% greater than that for the non–GLP-1 RA group (P < 0.001) and total diabetes-related health care costs being 219% greater than that for the non–GLP-1 RA group (P < 0.001). Adjusted antidiabetic drug costs accounted for 87.4% of the total diabetes-related costs in the GLP-1 RA group compared with 71.2% in the non–GLP-1 RA group. GLP-1 RA use also contributed to all-cause costs, with prescription drug costs for the GLP-1 RA group being 172% greater than that for the non–GLP-1 RA group (P < 0.001), and total health care costs being 55.3% greater than that for the non–GLP-1 RA group (P < 0.001). Adjusted all-cause prescription drug costs accounted for 62% of total health care costs in the GLP-1 RA group compared with 34.9% in the non–GLP-1 RA group. Supplementary Table 2 ^{(128.6KB, pdf)} shows the full results of the regression analysis

TABLE 4.

Associations Between GLP-1 RA Use and All-Cause and Diabetes-Related Prescription Drug and Total Health Care Costs After Adjusting for Covariates

Direct cost category	Group	Log coefficient (95% CI)	Change, %a	Adjusted costsb (95% CI), $
Diabetes-related
Antidiabetic drugs	GLP-1 RA	1.35 (1.11-1.59)	286 (203-390)	11,265 (10,863-11,665)
	Non−GLP-1 RA	Reference		2,892 (2,838-2,946)
	P value	<0.001
Total health carec	GLP-1 RA	1.16 (0.97-1.36)	219 (163-290)	12,893 (12,412-13,374)
	Non−GLP-1 RA	Reference		4,061 (3,987-4,134)
	P value	<0.001
All-cause
Prescription drugs	GLP-1 RA	1.00 (0.82-1.18)	172 (127-225)	14,794 (14,285-15,301)
	Non-GLP-1 RA	Reference		5,374 (5,278-5,470)
	P value	<0.001
Total health carec	GLP-1 RA	0.44 (0.30-0.58)	55.3 (35.0-78.6)	23,857 (22,997-24,716)
	Non-GLP-1 RA	Reference		15,391 (15,117-15,665)
	P value	<0.001

Log coefficients and adjusted costs were estimated by generalized linear regression with a log link and gamma distribution, adjusting for age, sex, race and ethnicity, region, health insurance, income level, and comorbidities.

^a Percentage of change in cost = [exp (log coefficient) – 1] x 100%.

^b Adjusted costs = out-of-pocket payment + insurance payment.

^c Total direct health care costs = medical care costs + prescription drug (or antidiabetic drug) costs.

GLP-1 RA = glucagon-like peptide-1 receptor agonist.

Discussion

This study, using nationally representative data from MEPS, examined recent GLP-1 RA use patterns and associated costs among US adults with type 2 diabetes. Our analysis revealed a substantial increase in GLP-1 RA utilization, with nearly 1 in 5 US adults with diabetes using these medications in 2021 and 2022, a marked increase compared with estimates prior to 2020. Individuals with obesity accounted for more than 70% of GLP-1-RA users. Older people (aged ≥65 years) were less likely to use GLP-1 RAs. Although insurance covered the vast majority (95%) of GLP-1 RA costs, these medications represented a substantial proportion of overall health care expenditures, accounting for 28.6% of total all-cause health care costs and 55.7% of diabetes-related costs among GLP-1 RA users. After adjusting for sociodemographic characteristics, we observed significantly higher health care costs in the GLP-1 RA group, with total all-cause costs 55% higher and diabetes-related costs 219% higher than the non–GLP-1 RA group. These findings highlight the increasing prevalence of GLP-1 RA use since 2020 and its significant impact on health care expenditures in diabetes care in the United States.

Our finding of nearly 20% GLP-1 RA use among US adults with type 2 diabetes in 2021-2022 represents a dramatic increase compared with previous estimates. This trend is further corroborated by another study, which, using claims data, observed a notable increase in the first-line use of GLP-1RAs from 6% to 18% from January 2022 to May 2023.²⁷ In 2016 and 2017, only 3.4% of patients with diabetes were prescribed a GLP-1 RA based on a study using the electronic health records of a large health system.⁸ Similarly, an analysis of National Health and Nutrition Examination Survey data from 2017 to 2020 found that only 4.4% of US adults with type 2 diabetes used GLP-1 RAs.⁹ The increase in GLP-1 RA use aligns with the growing body of clinical evidence demonstrating their effectiveness in glycemic control and in promoting significant weight loss.²⁸ Furthermore, the FDA approval of several GLP-1 RAs specifically for weight management has expanded their use beyond diabetes treatment. This shift began around 2020, with observed increases in GLP-1 RA use among individuals with obesity and related comorbidities, even in the absence of diabetes.¹⁵ This expanding use of GLP-1 RAs in both diabetes and nondiabetes populations has prompted calls for strategies to ensure equitable access and universal affordability.^15,29 In an effort to improve access, the FDA has prioritized the review of generic drug applications for GLP-1 RAs, recently approving generic liraglutide injection and exenatide in 2024.³⁰

Our results found that the increases in diabetes-specific and all-cause costs shown for the group taking GLP-1 RAs were largely attributable to the higher drug acquisition costs of the GLP-1 RAs. The rising demand for these drugs, particularly for weight management, coupled with recent drug shortages, has contributed to escalating prices. Despite the high cost of GLP-1 RAs, our findings further showed that nationwide only less than 5% of the drug cost of GLP-1 RAs was paid by the patient, and more than 95% was paid by insurance. This disparity underscores the critical role of prescription drug coverage in ensuring affordable access to GLP-1 RA treatment. However, it is unknown whether manufacturer coupons and discount programs could reduce out-of-pocket expenditures. Our study sample demonstrated that individuals with higher income and private insurance were more likely to use GLP-1 RAs. This suggests that these individuals are better positioned to be able to pay the out-of-pocket cost of GLP-1 RAs and are more likely to have comprehensive insurance plans.

Given that insurance covered more than 95% of GLP-1 RA costs in our study, the high cost of GLP-1 RAs placed a substantial financial burden on health insurance plans. For example, the Congressional Research Service reported that Medicare alone spent $5.7 billion on selected GLP-1 RAs (including semaglutide and tirzepatide) for diabetes treatment in 2022, with average spending per claim ranging from $1,189 to $1,429.³¹ Even with high insurance coverage, the long-term impact of out-of-pocket payments on medication adherence, particularly among older patients managing multiple chronic conditions with polypharmacy, remains an important area for future research.²³ In an effort to mitigate these costs, the Department of Health and Human Services has selected semaglutide and 14 other drugs for Medicare Drug Price Negotiations in 2025. This measure is intended to lower prescription drug costs for Medicare beneficiaries and potentially reduce the financial burden on both Medicare and patients.³² Furthermore, the recent FDA approvals of generic liraglutide and exenatide offer the potential for increased affordability and access to GLP-1 RA therapies.³⁰

Our study revealed a significantly lower proportion of older adults (aged ≥65 years) using GLP-1 RAs (35.1%) compared with both the non–GLP-1 RA group (51.1%) and the 45-64 age group within the GLP-1 RA users (50.6%). Clinical evidence suggests that GLP-1 RAs reduce cardiovascular event risk in older patients, consistent with the benefits observed in overall populations.³³ However, American Diabetes Association guidelines highlight practical challenges that likely contribute to the lower utilization of GLP-1 RAs in older adults. First, the injectable nature of many GLP-1 RAs poses administration challenges for some older patients, requiring adequate visual, motor, and cognitive function.³⁴ Second, although generally considered beneficial, the potential for weight loss and gastrointestinal side effects necessitates careful consideration in older patients, particularly those with frailty. Older patients are more likely to discontinue GLP-1 RA treatment because of a higher risk of gastrointestinal adverse events.^34–36 Consequently, these practical administration challenges and potential for adverse events likely lead to greater caution among health care providers when prescribing GLP-1 RAs to older patients.³⁵

LIMITATIONS

Several limitations should be considered when interpreting the results. First, this study provides a cross-sectional snapshot of GLP-1 RA use among US adults with type 2 diabetes during 2021 and 2022 and does not distinguish between new and existing users. Therefore, our results may not fully capture the dynamic trends in GLP-1 RA utilization since 2020. Additionally, MEPS data does not provide specific dates of prescriptions filled. Consequently, our study was not able to determine the exact start dates for medications or the temporal sequence of concomitant antidiabetic medication use for individual respondents. Second, our analysis examined the impact of GLP-1 RA use on 1-year health care costs. This short time frame may not fully reflect whether the potential long-term cost savings associated with the clinical benefits of GLP-1 RAs could be achieved. Although increased adherence to GLP-1 RAs may increase prescription drug costs in the short term, it could also lead to long-term medical cost savings by preventing costly complications. However, the long-term trajectory of GLP-1 RA–related costs, including potential sustained increases in drug costs and any unforeseen long-term adverse event management, remains an area requiring further investigation. The precise extent to which these potential long-term costs may be offset by avoiding complication-related costs is yet to be fully elucidated. Therefore, future studies using longitudinal data are needed to assess the net economic impact of GLP-1 RA use over extended periods. Third, although our study provides valuable insights into the costs associated with GLP-1 RA use, future studies focused specifically on health care resource utilization may further improve our understanding of the sources and dynamics of the cost outcomes presented herein. Fourth, the lack of severity of diabetes and comorbidities in the MEPS data prevented us from adjusting for these clinical characteristics in the regression analysis, which could potentially affect the observed differences in health care utilization and costs. Additionally, prescriber information is not available in the MEPS data. Although prior studies using pre-2020 data indicated that primary care physicians and endocrinologists accounted for approximately 90% of GLP-1 RA prescriptions, with cardiologists representing a smaller proportion,³⁵ it is unknown whether these prescribing patterns have changed with the increased awareness, broader recognition, and expanded indications of GLP-1 RAs beyond diabetes. Future research is needed to investigate these evolving prescribing patterns among different health care provider specialties. Finally, a consideration for our study using MEPS data is the potential for recall bias. Respondents in MEPS participate in 5 rounds of interviews spanning a 2-year period, during which they are asked to recall health care events and expenditures. This potential for recall bias could influence the accuracy of the self-reported cost and utilization data and, consequently, impact our findings on GLP-1 RA use and associated costs.

Conclusions

Our study demonstrates a substantial increase in GLP-1 RA utilization among US adults with type 2 diabetes, rising from pre-2020 levels of less than 10% to nearly 20% in 2021-2022. Because of the high cost of GLP-1 RAs, patients taking them had significantly higher diabetes-related and total health care expenditures. Further research using longitudinal data is crucial to comprehensively assess whether long-term GLP-1 RA use could potentially yield cost savings through its clinical benefits in diabetes management.