Racial, Ethnic, and Socioeconomic Inequities in Glucagon-Like Peptide-1 Receptor Agonist Use Among Patients With Diabetes in the US

Abstract

Importance:

Glucagon-like peptide-1 receptor agonists (GLP-1 RA) cause significant weight loss and reduce cardiovascular (CV) events among patients with diabetes type 2 (T2D). Black patients have a disproportionate burden of obesity and CV disease, with higher CV mortality. Racial and ethnic disparities in health outcomes are largely due to the pervasiveness of structural racism and marginalized patients have less access to novel therapeutics.

Objective:

To assess for inequitable utilization in GLP-1 RA utilization between racial, ethnic, gender, and socioeconomic groups in the United States (US).

Design:

Retrospective cohort analysis performed 07/2020 with data from 10/2015 to 06/2019 using the Optum Clinformatics Data Mart. We estimated multivariable logistic regression models to identify the association of race, ethnicity, gender, and socioeconomic status with use of GLP-1 RA.

Setting:

Commercially insured patients in the US.

Participants:

Adult patients with a diagnosis of T2D, with or without atherosclerotic cardiovascular disease (ASCVD)

Main outcome and Measure:

Prescription of an GLP-1 RA

Results:

Of 1,180,260 patients with T2D (median [IQR] age, 69 [59, 76] years; 50.3% female; 57.7% White), 90,934 (7.7%) were treated with GLP-1 RA during the study period. Between 2015 and 2019, the percentage of T2D patients treated with an GLP-1 RA increased from 3.2% to 10.7%. Among patients with T2D and ASCVD, use also increased but remained low (2.8 to 9.4%).

In multivariable analyses, Black race (adjusted odds ratio [aOR] 0.81; 95% confidence interval [CI] 0.79–0.83), Asian race (aOR 0.59; 95% CI 0.56–0.62), Hispanic ethnicity (aOR 0.91; 95% CI 0.88–0.93) were all associated with less GLP-1 RA use among patients with T2D. Female gender (aOR 1.22 [95% CI 1.20–1.24]) and higher zip-code linked median household income (>$100,000, OR 1.13; 95% CI 1.11–1.16; $50–99.999K, OR 1.07, 95% CI 1.05–1.09 vs. <$50K) were associated with higher GLP-1 RA use. These results were similar among patients with ASCVD.

Conclusions and Relevance:

In this cohort study of US patients with T2D, GLP-1 RA use increased, but remained low overall for treatment of T2D, particularly among patients with ASCVD who are likely to derive the most benefit. Black, Asian, Hispanic and low-income patients were less likely to receive treatment with a GLP-1 RA. Strategies to lower barriers to utilization, including increasing affordability, are needed to prevent widening of well-documented inequities in cardiovascular disease outcomes in the US.

Introduction

Cardiovascular death is the leading cause of mortality and morbidity among patients with type 2 diabetes (T2D) [1]. GLP-1 receptor agonists (GLP-1 RA), a recommended treatment option for glycemic control in diabetes, more recently have emerged as a cardioprotective therapy [2]. Multiple large randomized trials have shown GLP-1 RA prevent cardiovascular (CV) events among patients with T2D, particularly patients with established atherosclerotic cardiovascular disease (ASCVD) [3–5]. These data have prompted a paradigm shift to utilize these agents not only for glycemic control, but for CV risk reduction. The updated American Diabetes Association (ADA) guidelines and American College of Cardiology (ACC) expert consensus statements now recommend a GLP-1 RA with demonstrated CV benefit for diabetic patients with established or at very high risk for ASCVD [6].

There are significant disparities in diabetes prevalence, complications, and death rates.

Black and Latinx patients have a disproportionate burden of T2D [7]. Black patients have higher diabetes complication rates such as CV disease, and CV mortality continues to be highest among Black patients [8–10]. However, inequitable quality of care by race and ethnicity is a well-documented phenomenon in the United States [11].

Cardiovascular therapeutics with proven benefit are underutilized among Black/Latinx patients, even among those who are commercially insured [12–15]. In addition, there is decreased adoption of novel cardiovascular therapies among female patients and those with low socioeconomic status. Given this, the objective of this study was to evaluate the uptake of GLP-1 RA, with specific attention to the association between GLP-1RA use and race, ethnicity, gender, and socioeconomic status among a commercially-insured patient population in the US with T2D, including separately those with ASCVD given the known benefit in this population.

Methods

Study Data

Data were obtained from the OptumInsight Clinformatics Data Mart database, a large administrative private payer claims database from recipients of commercial health insurance and Medicare Advantage health plans. This database consists of inpatient, outpatient, and pharmacy claims of over 17 million patients annually from all 50 states. Patient demographic variables, including age, sex, and race and ethnicity are collected by Optum for each individual member at enrollment. Race and ethnicity categories in the database are Asian, Black, Hispanic (all races), White, and Other/Unknown. Socioeconomic data, including median household income, are available through ZIP code–linked enrollment data from the US Census Bureau. Mean number of outpatient cardiology visits and endocrinology visits per 12 months after cohort entry until the end of available data (06/2019) were determined based on having a visit with a cardiology/endocrinology provider with CPT codes 99201–99205 or 99211–99215. All prescription claims for available GLP-1 RA during the study period, including all formulations of dulaglutide, abiglutide, exenatide, exenatide extended release, liraglutide, lixisenatide, or semaglutide were extracted. This study was classified as exempt by the University of Pennsylvania Institutional Review Board. This study follows the STROBE reporting guidelines for cohort studies.

Study Cohort

We identified adult (age ≥18 years) patients with a diagnosis of T2D based on ICD-10-CM codes (ICD-10-CM: E11.0, E11.1, E11.9) between October 2015 and December 2018 to allow for 6 months of continuous enrollment and prescription of therapy after diagnosis given data are available through June 2019. This time period encompasses a period when CV benefits were clearly known and there was level 1 evidence available for CV benefits [3–5]. Patients were required to have diabetes diagnosis coded at least twice, on separate dates, either in the inpatient or outpatient settings to improve diagnostic accuracy. Patients entered the cohort on the date of second diagnostic code for diabetes and then were evaluated for a prescription claim filled for dulaglutide, albiglutide, exenatide, exenatide extended release, liraglutide, lixisenatide, or semaglutide through June 2019. The primary outcome of interest was a prescription for one of the aforementioned GLP1–1 RA filled at any point during the study period for each individual patient (from the second coded diagnosis of diabetes to end of available data [June 2019]).

Patients without continuous insurance enrollment for at least 1 year before and at least 6 months after study entry were excluded to ensure comorbidities, clinical data, and prescription claims could be accurately captured. In addition, patients without any pharmacy claims for medication for 1 year prior to the study period were excluded to ensure that medication use for each patient was accurately being captured in our data. Comorbidities were evaluated from earliest available data to cohort entry.

Statistical Analysis

We divided patients into those treated and those not treated with a GLP-1 RA during the study period. For each group, summary statistics for patient characteristics are presented as medians with interquartile ranges or means with standard deviations for continuous data, and as total number and percentages for categorical data. Continuous variables were compared using the Student’s t-test, and categorical variables were compared using the chi square test. We described the proportion of patients using GLP-1 RA over the entirety of the study period and each year. We repeated this analysis in Asian, Black, and Hispanic patients, and those with ASCVD. For those that filled a prescription for GLP-1 RA, we determined the median 30-day copayment for the first filled prescription.

To assess the relationship of race and ethnicity with the use of GLP-1 RA we estimated multivariable logistic regression models with filled prescription for GLP-1 RA as the dependent variable and independent variables including age, gender, race and ethnicity (Black, Hispanic, White, Asian), region of residence, zip code–linked household income, health insurance subset (commercial-only or Medicare Advantage, which provides Medicare benefits through commercial insurers), hyperlipidemia, coronary artery disease, cerebrovascular disease, chronic kidney disease, kidney failure/end-stage renal disease, hypertension, obesity, peripheral vascular disease, heart failure with reduced ejection fraction (HFrEF), heart failure with preserved ejection fraction (HFpEF), number of Elixhauser comorbidities [16], number of cardiology visits per 12 months, number of endocrinology visits per 12 months, insulin use, and metformin use. We repeated this analysis in the subgroup of patients with a diagnosis of ASCVD based on ICD 10-CM codes (Supplemental Table 1). Patients with missing data for one of the aforementioned covariates were not included in the multivariable analyses.

Estimated adjusted odds ratios (aORs) are reported with 95% confidence intervals (CI). Statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC). All statistical testing was 2-tailed, with P values <0.05 designated statistically significant.

Results

A total of 1,180,260 patients with T2D met inclusion criteria. CONSORT diagram, detailing applied exclusion criteria and final cohort, is shown in Figure 1. Overall, 90,934 (7.7%) were treated with a GLP-1 RA during the study period and 1,089,326 (92.3%) were not. The median (IQR) age of patients was 69 (59,76) years and 594,088 were female (50.3%) patients. The cohort included 681,579 White (57.7%), 146,861 Black (12.4%), 173, 561 Hispanic (14.7%) and 52, 349 Asian (4.4%) patients. The zip code–linked median household income for 369,474 patients (31.3%) was less than $50 000 and for 209,200 patients (17.7%) was $100 000 or greater. Baseline demographic, socioeconomic, and clinical differences between those who were prescribed GLP-1 RA vs. those who were not are summarized in Table 1.

Figure 1.

Selection of study population

Table 1.

Baseline characteristics of patients with Diabetes with and without Filled Prescription for Glucagon-like Peptide 1 Receptor Agonists

Characteristic	No GLP-1RA (n=1,089,326)	GLP-1RA adopter (n=90,934)	P-value
Age, median(IQR)	70(60,77)	59(51,68)	<.0001
Gender, n(%)			<.0001
Female	547,292(50.2)	46,796(51.5)
Male	541,903(49.7)	44,130(48.5)
Race, n(%)			<.0001
Asian	49,995(4.6)	2,354(2.6)
Black	135,962(12.5)	10,899(12.0)
Hispanic	160,191(14.7)	13,370(14.7)
White	624,643(57.3)	56,936(62.6)
Other*/Unknown	118,535(10.9)	7,375(8.1)
Region of Residence, n(%)			<.0001
Midwest	220,717(20.3)	20,693(22.8)
Northeast	141,677(13.0)	7,634(8.4)
South	470,746(43.2)	45,610(50.2)
West	253,585(23.3)	16,856(18.5)
Unknown	2,601(0.2)	141(0.2)
Zip-code Linked Median Household Income, n(%)			<.0001
<$50K	344,405(31.6)	25,069(27.6)
$50–$99,999K	324,033(29.7)	27,673(30.4)
$≥100K	188,165(17.3)	21,035(23.1)
Unknown	232,723(21.4)	17,157(18.9)
Insurance Subtype, n(%)			<.0001
Commercial	341,441(31.3)	53,682(59.0)
Medicare Advantage	747,885(68.7)	37,252(41.0)
Comorbidities, n(%)
Dyslipidemia	950,226(87.2)	80,589(88.6)	<.0001
Cerebrovascular disease	247,473(22.7)	13,986(15.4)	<.0001
Coronary artery disease	114,750(10.5)	6,987(7.7)	<.0001
Coronary artery bypass graft surgery	17,355(1.6)	1,218(1.3)	<.0001
Chronic kidney disease	270,757(24.9)	17,326(19.1)	<.0001
End stage Renal disease	18,368(1.7)	668(0.7)	<.0001
Obesity	344,707(31.6)	43,639(48.0)	<.0001
Hypertension	926,911(85.1)	76,906(84.6)	<.0001
Peripheral vascular disease	240,657(22.1)	12,489(13.7)	<.0001
HFrEF	57,971(5.3)	2,801(3.1)	<.0001
HFpEF	55,496(5.1)	2,693(3.0)	<.0001
No. of Elixhauser Comorbidities, n(%)			<.0001
0–1	216,669(19.9)	18,833(20.7)
2–3	375,720(34.5)	35,847(39.4)
4–6	316,300(29.0)	26,202(28.8)
≥7	180,637(16.6)	10,052(11.1)
Medications.* n(%)
Metformin	508,259(46.7)	58,602(64.4)	<.0001
Insulin	174,724(16.0)	35,624(39.2)	<.0001
# of Endocrinology visits per 12 months, n(%)			<.0001
0	992,831(91.1)	63,760(70.1)
1	43,234(4.0)	9,616(10.6)
>1	53,261(4.9)	17,558(19.3)
# of Cardiology visits per 12 months, n(%)			<.0001
0	778,972(71.5)	64,124(70.5)
1	143,662(13.2)	13,873(15.3)
>1	166,692(15.3)	12,937(14.2)

HFrEF-heart failure with reduced ejection fraction; HFpEF- heart failure with preserved ejection fraction.

^*Racial and ethnic category of ‘Other’ includes patients who identify as ‘Other’, which includes categories other than Asian, Black, White, Hispanic; biracial or multiracial patients.

The proportion of patients with diabetes using GLP-1 RA increased from 3.2% in 2015 to 10.7% in 2019 (Figure 2A). The proportion of patients using GLP-1 RA increased from 2.0% to 6.4% among Asian patients, 2.9% to 10.4% among Black patients, 2.9% to 10.8% among Hispanic patients, and 3.6% to 11.7% among white patients (Figure 2B). Among those with ASCVD (n=815,309), the use of GLP-1RA inhibitors increased from 2.8% to 9.4%.

Figure 2A.

Rates of Treatment with GLP-1RA Over Time in the United States 2B.

Rates of Treatment with GLP-1RA over Time in the United States by Race and ethnicity

For those that filled a prescription for GLP-1 RA, the median 30-day copayment was $40.00 [IQR 8.35, 60].

In multivariable analyses (Table 2), female gender was associated with higher odds of GLP-1 RA use (aOR 1.22 [95% CI 1.20–1.24]). Compared with white race, Black race (aOR 0.81 [95% CI 0.79–0.83]), Hispanic ethnicity (aOR 0.91 [95% CI 0.88–0.93]) and Asian race (aOR 0.59 [95% CI 0.56–0.62]) were each associated with lower GLP-1 RA use. Higher median household income was associated with higher GLP-1 RA use (income >$100K: aOR 1.13 [95% CI 1.11–1.16]); income $50–99,999K: aOR 1.07 [95% CI 1.05–1.09] compared with income <$50K). Coronary artery disease (aOR 0.95 [95% CI 0.92–0.98]) and cerebrovascular disease (aOR 0.96 [95% CI 0.93–0.98]) were both independently associated with lower GLP-1 RA use.

Table 2.

Factors associated with GLP-1 RA Use Among All Patients on Multivariable Analysis

Characteristic	Odds Ratio	95% CI	p-value
Age	0.97	0.97–0.97	<.0001
Female Gender (Male as Ref)	1.22	1.20–1.24	<.0001
Race (White as Ref)
Asian	0.59	0.56–0.62	<.0001
Black	0.81	0.79–0.83	<.0001
Hispanic	0.91	0.88–0.93	<.0001
Region (West as Ref)
Midwest	1.01	0.98–1.04	0.44
Northeast	0.79	0.76–0.81	<.0001
South	1.17	1.14–1.20	<.0001
Zip-Code Linked Median Household Income (<$50K as Ref)
$50K–$99,99K	1.07	1.05–1.09	<.0001
$>100K	1.13	1.11–1.16	<.0001
Insurance subtype (Medicare Advantage as Ref)
Commercial	1.53	1.50–1.57	<.0001
No. of Elixhauser comorbidities	0.93	0.93–0.93	<.0001
Dyslipidemia	1.55	1.51–1.59	<.0001
Coronary artery disease	0.95	0.92–0.98	0.001
Cerebrovascular disease	0.96	0.93–0.98	0.0002
Chronic Kidney Disease	1.26	1.23–1.29	<.0001
End-stage Renal Disease	0.49	0.44–0.54	<.0001
Obesity	1.72	1.69–1.75	<.0001
Hypertension	1.49	1.45–1.53	<.0001
Peripheral vascular disease	1.00	0.97–1.03	0.95
HFrEF	0.83	0.79–0.88	<.0001
HFpEF	0.87	0.83–0.92	<.0001
Metformin	1.88	1.85–1.91	<.0001
Insulin	2.65	2.60–2.69	<.0001
Cardiology Visits per 12 months (0 as Ref)
1	1.19	1.16–1.22	<.0001
>1	1.16	1.13–1.19	<.0001
Endocrine visits per 12 months (0 as Ref)
1	2.26	2.20–2.33	<.0001
>1	3.14	3.07–3.22	<.0001

HFrEF-heart failure with reduced ejection fraction; HFpEF- heart failure with preserved ejection fraction.

More Elixhauser comorbidities were associated with lower GLP-1 RA use (aOR 0.93, [95% CI 0.93–0.93). Having more endocrinology visits per 12 months (aOR 2.26 [95% CI 2.20–2.33] for 1 endocrinology visit; aOR of 3.14 [95% CI 3.07–3.22] for >1 endocrinology visit vs 0 visits). Similarly, having >1 cardiology visit per 12 months (aOR of 1.16 [95% CI 1.13–1.19] or 1 cardiology visit (aOR of 1.19 [95% CI 1.16–1.22] were independently associated with increased use of GLP-1 RA as compared to no cardiology visits.

Subgroup Analysis

Factors associated with GLP-1 RA use among patients with ASCVD on multivariable analyses are shown in Table 3. Female gender was associated with higher GLP-1 RA use with aOR of 1.18 [95% 1.15–1.20]. Black race (aOR 0.82 [95% 0.79–0.85]), Hispanic ethnicity (aOR 0.94 [95% CI 0.91–0.96]), and Asian race (aOR 0.69 [95% CI 0.65–0.73]) were associated with less GLP-1 RA use compared to white patients. Higher median household income was associated with more GLP-1 RA use (aOR 1.15 [95% CI 1.11–1.18] for income $50K-99,999K and aOR 1.06 [95% CI 1.03–1.08] for income ≥$100K vs. income <$50K).

Table 3.

Factors associated with GLP-1 RA Use Among Patients with ASCVD on Multivariable Analysis

Characteristic	Odds Ratio	95% CI	p-value
Age	0.96	0.96–0.96	<.0001
Female Gender (Male as Ref)	1.18	1.15–1.20	<.0001
Race (White as Ref)
Asian	0.69	0.65–0.73	<.0001
Black	0.82	0.79–0.85	<.0001
Hispanic	0.94	0.91–0.96	<.0001
Region (West as Ref)
Midwest	0.97	0.93–1.00	0.04
Northeast	0.78	0.75–0.82	<.0001
South	1.13	1.10–1.17	<.0001
Zip-Code Linked Median Household Income (<$50K as Ref)
$50K–$99,999K	1.06	1.03–1.08	<.0001
$>100K	1.15	1.11–1.18	<.0001
Insurance subtype (Medicare Advantage as Ref)
Commercial	1.42	1.38–1.46	<.0001
No. of Elixhauser comorbidities	0.92	0.92–0.93	<.0001
Dyslipidemia	1.57	1.51–1.63	<.0001
Coronary artery disease	0.93	0.90–0.96	0.001
Cerebrovascular disease	0.96	0.93–0.98	0.0002
Chronic Kidney Disease	1.29	1.26–1.33	<.0001
End-Stage Renal Disease	0.45	0.41–0.50	<.0001
Obesity	1.73	1.70–1.77	<.0001
Hypertension	1.43	1.38–1.49	<.0001
Peripheral vascular disease	0.99	0.96–1.02	0.57
HFrEF	0.83	0.79–0.87	<.0001
HFpEF	0.88	0.84–0.93	<.0001
Metformin	1.92	1.88–1.96	<.0001
Insulin	2.81	2.74–2.87	<.0001
Cardiology Visits per 12 months (0 as Ref)
1	1.15	1.12–1.18	<.0001
>1	1.15	1.12–1.18	<.0001
Endocrine visits per 12 months (0 as Ref)
1	2.13	2.06–2.21	<.0001
>1	3.08	2.99–3.17	<.0001

HFrEF-heart failure with reduced ejection fraction; HFpEF- heart failure with preserved ejection fraction.

Discussion

In this study, we found low utilization of GLP-1 RA, including among patients with ASCVD. Despite this population being 100% commercially insured, this is the first study, to our knowledge, to demonstrate notable racial, ethnic, and socioeconomic inequities in GLP-1 RA utilization. Black and Asian race and Hispanic ethnicity were associated with less GLP-1 RA use, while higher household income was independently associated with higher use. These inequities were also present among diabetic patients with ASCVD.

Structural racism, defined as the differential distribution of goods, services, and opportunities of society based on race [17], is a major barrier to achieving health equity [18]. Similar to sodium-glucose cotransporter 2 (SGLT2) inhibitor utilization [12], Black and Asian race were independently associated with less GLP-1 RA use. Additionally, Hispanic ethnicity was also associated with lower GLP-1 RA utilization. Racial and ethnic minoritized patients consistently have inequitable access to guideline-based therapeutics that improve cardiovascular disease burden and outcomes, despite often experiencing a disproportionately higher rate of these conditions [10][19]. Often socioeconomic status and lack of health insurance are blamed for racial/ethnic inequities in healthcare and health outcomes [20][21]. However, the racial and ethnic disparities in GLP-1RA use demonstrated in our study persisted after adjustment for not only clinical factors, but engagement with specialty care, and socioeconomic status, and was in the setting of a 100% commercial insured population. Therefore, these results are concerning for biases in care delivery, which must be rectified.

While the diabetes-related risk for coronary heart disease has declined among white patients since 1990, it has doubled among Black patients [22]. Therefore, better understanding barriers to utilization of GLP-1 RA among this population and other marginalized populations are needed. Inequitable uptake must be addressed to prevent widening of racial/ethnic disparities in CV disease and outcomes in the United States [23].

Similar to other novel cardiovascular therapies [12][13][15], including SGLT2 inhibitors, socioeconomic status was also independently associated with lower GLP-1RA use. Despite adequate prescription drug insurance, among Medicare part D patients, median estimated annual out-of-pocket costs are $2447 for liraglutide [24]. We found that for those that filled a prescription for GLP-1RA, the median 30-day copayment was $40; our findings suggest that cost is likely still prohibitive, and may be worsening inequities by potentially forcing utilization of inexpensive medications with no cardiac benefit among marginalized patient groups. In fact, primary care physicians and endocrinologists cite cost and unapproved prior authorizations as the major barriers for prescription of these agents [25]. Additionally, while we adjusted for median household income, it is likely that cost also may contribute to the racial and ethnic inequities given differences in wealth and ability to afford out of pocket costs [26].

Interestingly, unlike SGLT2 inhibitor use, and other cardiovascular therapeutics [12][27] female gender was associated with higher GLP-1 RA use. Previous studies have shown that, among diabetic patients, men might be more comprehensively treated in regard to coronary heart disease risk [28]. It is unclear why GLP-1RA use was higher among females in our study, but may reflect the well-documented higher rates of contact with the healthcare system among female patients [29]. In addition to the cardiovascular benefits, GLP-1RA have also been shown to cause significant and sustained weight loss [30]. It is possible that female patients with T2D may engage more with nutritionist and may be more likely to seek pharmacotherapy for medical management of obesity, which may include counseling for GLP-1 RA use [31].

Notably, Asian patients had the lowest rates of GLP-1 RA use and >40% lower odds of GLP-1RA prescription. Barriers to accessing care, less patient-centered interactions by providers, and biases in care delivery have been well documented among Asian patients and likely play a role in the inequitable use of GLP-1 RA in this population [32][33]. Among patients with T2D, Asian patients have the lowest BMI compared to other races/ethnicities, [34] which may influence prescription of this therapy. However, these findings are consistent with previously documented inequitable SLT2i use among Asian and Black patients [12] and reflect the pervasive inequity of the U.S. healthcare system for non-White patients.

Similar to prior outpatient registries of T2D patients [35–39], our results confirm low utilization of GLP-1 RA, with only slightly higher use among a commercially-insured cohort. Despite multiple clinical trials demonstrating improved CV outcomes with the use of certain GLP-1 RA among patients with ASCVD starting in 2015, [3–5] the rate of GLP-1RA use among patients with ASCVD remained low through 2019 (9.4%), and in fact lower than the overall diabetes cohort (10.7%). Updated guidelines by the ACC and ADA strongly recommend a GLP-1 RA in patients with T2D who already have or are at high risk for CVD [2,6]. Our results suggest that these agents have not yet been adopted as a strategy for broader CV risk reduction. In fact, coronary artery disease and cerebrovascular disease were associated with lower GLP-1 RA use. Cardiology prescription of GLP-1 RA has previously been shown to be minimal, even among patients with cardiovascular conditions [38]. In our cohort, a visit with a cardiologist was associated with higher GLP-1 RA use and having a visit with an endocrinologist was one of the strongest predictors of GLP-1 RA use (>3 times the odds of prescription). Yet, only a small percentage of patients with T2D receive care from an endocrinologist [40] and patients with cardiovascular disease are much more likely to see a cardiologist as part of their care team than an endocrinologist [41]. Therefore, along with other traditional agents such as statins, cardiologists must start viewing prescription of these agents as part of their cardiac risk reduction armamentarium [42]. In addition, given demonstrated inequities in accessing specialty cardiology care [43][44], it is essential that barriers are decreased and knowledge/comfort increased to facilitate primary care provider prescription of these agents for all patients with T2D and ASCVD risk factors, with special attention to marginalized groups of patients.

Our study has several limitations. The factors and clinical decision making that drive the decision to initiate a certain therapy, such as GLP-1 agonist can be complex and not well characterized in an administrative database. Given the majority of GLP-1 RA agonists are injectable, there may be residual confounding by patient preference. Patient self-advocacy might also impact treatment decisions, if white patients more frequently advocate for this therapy given increased awareness of its benefits. Racial differences in self-advocacy have been previously observed and must be contextualized in a healthcare system with historical and ongoing discrimination against people of color in the US [45]. Additionally, provider preference, knowledge of guidelines and benefits, as well as comfort with this therapy influences its use; disparities may be driven, in part, by where a patient seeks care. This database captures prescriptions that are filled at the pharmacy. Thus, we are unable to distinguish between barriers in provider prescription pattern versus barriers in prescription fullfillment. The demonstrated differences may be due in part to prescription abandonment at the pharmacy due to high co-payments but we are unable to characterize the degree to which this or provider bias, or other barriers explain our findings. Additional aspects of benefit design may mediate some of our findings, but we are unable to characterize the extent to which as copayment data is not available for patients who did not fill their prescription for this therapy. There are contraindications to the use of and side effects of GLP-1 RA therapy, which may have contributed to the results. We adjusted for endocrinology visits and cardiology visits, but this database did not allow us to evaluate the type of practice or health care professional prescribing therapy, limiting our complete understanding of how differences in access to specialty care contributed to our findings. There were missing data at baseline, including data regarding race, ethnicity, and median household income. More granular socioeconomic status data are missing from this database, which may also influence a patient’s ability to fill a prescription. Additionally, this database is unable to fully capture how race and ethnicity operate in a broader socioeconomic and sociopolitical context, the effects of structural racism, a long and continued history of mistreatment, and levels of discrimination, which all impact how patients of color navigate the healthcare system and receive care [11][18].

Conclusion

In this cohort study of US patients with T2D, we found low rates of GLP-1RA use, even among patients with ASCVD. We found that Black and Asian race, Hispanic ethnicity, and lower-zip code linked household income were independently associated with less GLP-1RA use, with a similar pattern of inequitable utilization among patients with ASCVD. Implementation of strategies to ensure more equitable utilization of this therapy is warranted.

Key Points.

Question:

Are there inequities in Glucagon-like peptide-1 receptor agonist (GLP-1 RA) utilization based on race, sex, and socioeconomic status among patients with diabetes in the United States (US)?

Findings:

In a 5-year cohort study of 1,180,260 commercially insured patients with diabetes in the US, we found that GLP-1 RA use increased, but utilization remained low. Black race, Hispanic ethnicity, Asian race, and lower household income were associated with lower rates of GLP-1 RA use; results were similar among diabetic patients with cardiovascular disease.

Meaning:

Racial/ethnic and socioeconomic inequities are present in access to GLP-1 RA a class of medication which improves cardiovascular outcomes in patients with diabetes.