Abstract
Objective
This study examined decade-long trends and differences in sodium-glucose cotransporter 2 inhibitor (SGLT2i) and glucagon-like peptide-1 receptor agonist (GLP-1 RA) prescriptions among adults with type 2 diabetes using real-world data from Epic Cosmos.
Research Design and Methods
We analyzed electronic health records of 1 517 594 adults with type 2 diabetes without end-stage renal disease from 2014 to 2024. Annual prescribing trends were evaluated by patient race and insurance type using negative binomial regression. Medication exposure was defined using active prescriptions/orders recorded in Epic Cosmos during the calendar year. In pooled descriptive analyses, we also characterized patients prescribed these medications by clinical characteristics, neighborhood-level social vulnerability, and prescriber specialty.
Results
From 2014 to 2024, SGLT2i use rose from 0.5% to 12.1% and GLP-1 RA use increased from 0.9% to 15.9%. Black patients had consistently lower prescription rates than White patients across insurance groups. Primary care physicians prescribed about one-third of these medications. In pooled descriptive analyses, endocrinology was associated with higher observed prescribing rates than primary care or cardiology. Patients from neighborhoods with lower social vulnerability were more likely to receive these therapies.
Conclusions
Use of SGLT2i and GLP-1 RA increased substantially over the past decade, significant racial, socioeconomic, and insurance-related differences persist in prescribing these therapies.
Key words: electronic health records, GLP-1 receptor agonists, prescription trends, SGLT2 inhibitors, Social Vulnerability Index, type 2 diabetes
Highlights
-
•Why did we undertake this study?
-
-We assessed real-world prescribing of SGLT2 inhibitors and GLP-1 receptor agonists for type 2 diabetes.
-
-
-
•What is the specific question(s) we wanted to answer?
-
-We assessed the trend of therapy use from 2014 to 2024 and whether differences exist by race, insurance, or other factors over the study period.
-
-
-
•What did we find?
-
-We found SGLT2i use rose from 0.5% to 12.1% and GLP-1 RA from 0.9% to 15.9%. Black and publicly insured patients were consistently less likely to receive them, while patients from less socially vulnerable neighborhoods and those treated by specialists were more likely to.
-
-
-
•What are the implications of our findings?
-
-Reducing disparities will require targeted health system efforts (e.g., EHR prompts, provider communication) and policy changes to lower costs and expand coverage.
-
-
Clinical Relevance
Although use of SGLT2 inhibitors and GLP-1 receptor agonists increased substantially from 2014 to 2024, Black patients, publicly insured patients, and those living in more socially vulnerable neighborhoods remained less likely to receive these guideline-recommended therapies. Given the established cardiovascular and renal benefits of these agents, targeted health-system and policy interventions to reduce prescribing disparities are needed to improve equitable outcomes in type 2 diabetes care.
Introduction
Sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1 RA) have fundamentally changed type 2 diabetes (T2D) management since their U.S. Food and Drug Administration approvals.1, 2, 3 Beyond their glucose-lowering effects, these agents have demonstrated significant benefits in reducing the risk of heart failure, chronic kidney disease progression, and major cardiovascular events.3, 4, 5, 6 Guidelines from the American Diabetes Association and other professional societies recommend their use as second-line or adjunctive therapies, particularly for patients with these conditions.7, 8, 9
Despite the clear clinical benefits, differences in the prescription of these guideline-recommended therapies persist. Prior studies show that racial/ethnic minorities, lower-income individuals, and those with limited insurance are less likely to receive these medications,10, 11, 12, 13 yet few studies have leveraged large, diverse, real-world data to analyze prescription trends across multiple health care organizations, including variation by comorbidities, social vulnerability, and prescriber specialty.14,15
In this study, we utilize Epic Cosmos, a large, federated, de-identified electronic health record (EHR) research network containing over 300 million patients, to describe decade-long prescribing trends and patterns for SGLT2i and GLP-1 RA through 2024.16 Our analysis incorporates neighborhood-level social vulnerability and prescriber specialty and models trends in active prescriptions over a decade, stratified by race and insurance.
Research Design and Methods
Study Design and Subjects
This retrospective cohort study utilized de-identified and aggregated EHR data from over 1700 participating healthcare organizations using the Epic system. Our analysis included data from January 1, 2014, to December 31, 2024,16 representing the most recent full calendar year available at the time of extraction (We accessed the Epic Cosmos aggregate data in 2025). Initial cohort identification and data extraction were performed using SlicerDicer, an interactive reporting tool within Epic.
We included adult patients aged 19 years and older with at least one encounter diagnosis of T2D, identified using ICD-9/10-CM codes.17 To ensure a cohort eligible for SGLT2i prescription, individuals with a diagnosis of end-stage renal disease were excluded.18 The study was deemed exempt from Institutional Review Board review due to the use of aggregated, de-identified data. The final cohort consisted of 1 517 594 eligible patients.
Outcomes and Independent Variables
We examined the trend in prescription rates by measuring the number of eligible patients with any active prescription/order for GLP-1 RA or SGLT2i recorded in Epic Cosmos during each calendar year from 2014 to 2024. This measure reflects EHR-documented prescribing rather than confirmed dispensing, or medication fills. Because Because Cosmos/SlicerDicer returned annual aggregate counts rather than patient-level medication histories, we could not apply an additional patient-level lookback period beyond the calendar-year extract. The GLP-1 RA category was based on the Epic Cosmos medication grouping used for this study and included dual GIP/GLP-1 receptor agonists captured within that grouping in later study years. The “other diabetes medications” group included patients with active prescriptions for medications other than GLP-1 RA or SGLT2i, which could include one or multiple drugs such as metformin, sulfonylureas, DPP-4 inhibitors, or insulin. We did not further differentiate insulin users from noninsulin users because such stratification was beyond the primary goal of the study. Specifically, we compared rates of GLP-1 RA and SGLT2i prescription by subgroups (race and insurance type). Since cells with fewer than 10 prescriptions could not be reported due to privacy concerns, we treated values < 10 as 10 prescriptions and tested alternative approaches in sensitivity analyses.
Independent variables included patient age, sex, race, and insurance type. Race/ethnicity was summarized using more granular categories in pooled descriptive analyses, but for the year-by-year regression models race was collapsed into Non-Hispanic White, Non-Hispanic Black, and Other because further stratification produced sparse cells in the aggregate Epic Cosmos data. Insurance type was grouped as Private, Medicare, Medicaid, or Other Payer. Other characteristics of patients on these drugs included rural residence, neighborhood-level SVI, body mass index (BMI), diabetes control status, comorbidities, and provider specialty. Neighborhood-level SVI was defined by linking patient ZIP codes to the CDC/ATSDR Social Vulnerability Index via a ZIP–tract crosswalk,19 and population-weighted mean SVI per ZIP was computed in Epic Cosmos. For analysis, SVI was categorized into quartiles (<25%, 25–<50%, 50–<75%, ≥75%). Diabetes control status was categorized as follows: HbA1C <7%, HbA1C 7% to 10%, HbA1C ≥ 10%, and unknown. Comorbidities, including hypertension, heart failure, cardiovascular disease, and chronic kidney disease, were identified using ICD-9/10 codes and computable phenotypes embedded in the tool.
Statistical Analysis
Patient sociodemographic and clinical characteristics were described based on whether they were prescribed SGLT2i, GLP-1 RA or other diabetes medications. Statistical significance was assessed using Chi-square tests for categorical variables and Wilcoxon rank-sum tests for continuous variables.
We fit 2 negative binomial models with the log of the eligible population as an offset. In the first model, annual class-specific counts were stacked across medication class, and a medication-type indicator was included so that fitted estimates could be generated separately for SGLT2i and GLP-1 RA. In the second model, the outcome was receipt of either medication class combined. Both models included year, age group, sex, race, insurance type, and all 2-way interaction terms among these variables. Likelihood ratio tests comparing models with interaction terms to simpler models without interactions indicated that the complex models provided a significantly better fit (P < 0.001). We outputted the fitted estimates from the first negative binomial model with medication separated in to Figures 1 and 2, and outputted the fitted estimates from the second jointly fitted negative binomial model into Figure 3. Plotted trend estimates are presented with 95% confidence intervals, and corresponding regression estimates are provided in the Supplementary Tables. All analyses were performed using R (version 4.3.1). A two-sided P value of 0.05 was used as the threshold for statistical significance. We also assessed covariate balance using standardized mean differences (SMDs), with an absolute SMD greater than 0.10 considered indicative of a meaningful difference, based on established guidelines.20
Fig. 1.
Trends in GLP-1 RA prescription rates by race, sex, and insurance across age groups. GLP-1 RA, glucagon-like peptide-1 receptor agonist.
Fig. 2.
Trends in SGLT2i prescription rates by race, sex, and insurance across age groups. SGLT2i, sodium-glucose cotransporter 2 inhibitor.
Fig. 3.
Trends in combined GLP-1 RA and SGLT2i Prescription rates by race, sex, and insurance across age groups. GLP-1 RA, glucagon-like peptide-1 receptor agonist; SGLT2i, sodium-glucose cotransporter 2 inhibitor.
Results
Among the 1 517 594 eligible patients, 145 955 were prescribed SGLT2i, 164 337 were prescribed GLP-1 RA and 1 207 302 were prescribed other diabetes medications during the study period. Missing/unknown values were low for race/ethnicity, insurance, and ZIP-level SVI, but were more frequent for HbA1C and BMI across medication groups (unknown race: 4.7% to 6.9%; unknown insurance: 1.2% to 4.9%; unknown SVI: 1.1% to 1.7%; unknown HbA1C: 40.9% to 50.2%; BMI no value: 9.4% to 18.6%). As detailed in Table, the demographic and clinical characteristics of patients prescribed these medications differed from those prescribed other diabetes medications. Patients prescribed SGLT2i and GLP-1 RA had a higher mean BMI (SMD = 6.6% for SGLT2i vs other; 15.8% for GLP-1RA vs other). The GLP-1 RA comparison exceeds the prespecified 10% threshold for a meaningful difference.20 For most other characteristics, differences were negligible (SMD <10%). For example, the prevalence of comorbidities such as heart failure (SMD = 0.8% and 0.3%) and chronic kidney disease (SMD = 0.2% and 1.8%) showed no meaningful imbalance. Patients receiving these medications were somewhat more likely to be Non-Hispanic White (SMD = 6.9% and 4.6%), privately insured (SMD = 5.4% and 4.8%), and from neighborhoods with lower SVI scores (SMD = 2.6% and 2.1% for lowest quartile). SGLT2i and GLP-1 RA were most commonly prescribed by primary care physicians. In pooled descriptive analyses, endocrinology was associated with higher observed prescribing rates than primary care or cardiology, although differences in prescriber specialty by medication group were small (SMD = 0.4% and 3.4%) and these comparisons were not adjusted for case mix or comorbidity burden.
Table.
Sociodemographic Characteristics of the Study Sample from the Epic Cosmos Database (January 1, 2014 – December 31, 2024)
| Characteristics | SGLT2i N = 145 955 |
GLP-1 RA N = 164 337 |
Other diabetes medications N = 1 207 302 |
SGLT2i vs other difference |
GLP-1 RA vs other difference |
|---|---|---|---|---|---|
| Mean (SD)/No (%) | Mean (SD)/No (%) | Mean (SD)/No (%) | SMD | SMD | |
| Age (y) | 68 (12) | 65 (12) | 69 (14) | −0.077 | −0.307 |
| Sex | |||||
| Female | 59 142 (40.5%) | 77 546 (47.2%) | 419 993 (43.8%) | 3.3% | 3.4% |
| Male | 85 621 (58.7%) | 85 463 (52.0%) | 532 890 (55.5%) | 3.2% | 3.5% |
| None of the above | 1192 (0.8%) | 1328 (0.8%) | 6830 (0.7%) | 0.1% | 0.1% |
| Race/Ethnicity | |||||
| Non-Hispanic White | 84 648 (51.6%) | 91 998 (49.3%) | 470 498 (44.7%) | 6.9% | 4.6% |
| Non-Hispanic Black | 36 348 (22.2%) | 43 498 (23.3%) | 279 800 (26.6%) | 4.4% | 3.3% |
| Hispanic | 19 030 (11.6%) | 22 956 (12.3%) | 137 295 (13.1%) | 1.5% | 0.8% |
| Non-Hispanic Asian | 5302 (3.2%) | 5390 (2.9%) | 34 899 (3.3%) | 0.1% | 0.4% |
| Native Hawaians and Pacific Islanders | 783 (0.5%) | 1136 (0.6%) | 7043 (0.7%) | 0.2% | 0.1% |
| Other races | 10 212 (6.2%) | 12 329 (6.6%) | 49 419 (4.7%) | 1.5% | 1.9% |
| Unknown | 7637 (4.7%) | 9390 (5.0%) | 72 692 (6.9%) | 2.2% | 1.9% |
| Insurance | |||||
| Private | 116 397 (37.2%) | 132 778 (36.6%) | 593 863 (31.8%) | 5.4% | 4.8% |
| Medicare | 88 413 (28.3%) | 103 245 (28.5%) | 595 030 (31.8%) | 3.5% | 3.3% |
| Medicaid | 41 806 (13.4%) | 50 974 (14.1%) | 285 843 (15.3%) | 1.9% | 1.2% |
| Self-pay | 14 214 (4.5%) | 15 423 (4.3%) | 57 114 (3.1%) | 1.4% | 1.2% |
| Veterans Affair or Tricare | 4223 (1.4%) | 3931 (1.1%) | 26 853 (1.4%) | 0.0% | 0.3% |
| Other insurance | 43 980 (14.1%) | 51 898 (14.3%) | 220 569 (11.8%) | 2.3% | 2.5% |
| Unknown | 3731 (1.2%) | 4410 (1.2%) | 91 451 (4.9%) | 3.7% | 3.7% |
| Rural residence | 3891 (2.7%) | 4622 (2.8%) | 27 525 (2.3%) | 0.4% | 0.5% |
| SVI at the zip level | |||||
| <25% | 19 619 (13.4%) | 21 229 (12.9%) | 103 868 (10.8%) | 2.6% | 2.1% |
| ≥25% & < 50% | 27 425 (18.8%) | 30 859 (18.8%) | 159 714 (16.6%) | 2.2% | 2.2% |
| ≥50% & < 75% | 35 874 (24.6%) | 40 899 (24.9%) | 226 018 (23.6%) | 1.0% | 1.3% |
| ≥75% | 61 349 (42.0%) | 69 503 (42.3%) | 453 577 (47.3%) | 5.3% | 5.0% |
| Unknown | 1688 (1.2%) | 1847 (1.1%) | 16 536 (1.7%) | 0.5% | 0.6% |
| Diabetes control status | |||||
| HbA1C < 7% | 39 802 (25.1%) | 53 992 (28.3%) | 350 640 (34.0%) | 8.9% | 5.7% |
| HbA1C 7% to 10% | 32 107 (20.2%) | 46 883 (24.6%) | 168 515 (16.3%) | 3.9% | 8.3% |
| HbA1C ≥ 10% | 7218 (4.6%) | 11 980 (6.3%) | 36 579 (3.6%) | 1.0% | 2.7% |
| Unknown | 79 622 (50.2%) | 77 939 (40.9%) | 475 571 (46.1%) | 4.1% | 5.2% |
| Weight status | |||||
| BMI ≤18.5 | 3601 (1.9%) | 3026 (1.4%) | 67 956 (4.7%) | 2.8% | 3.3% |
| BMI >18.5 & < 25 | 28 604 (15.0%) | 24 724 (11.4%) | 338 479 (23.5%) | 8.5% | 12.1% |
| BMI ≥25 & < 30 | 49 338 (25.9%) | 52 374 (24.1%) | 436 311 (30.3%) | 4.4% | 6.2% |
| BMI ≥30 | 73 743 (38.7%) | 103 963 (47.9%) | 462 425 (32.1) | 6.6% | 15.8% |
| No value | 35 470 (18.6%) | 32 833 (15.1%) | 134 909 (9.4%) | 9.2% | 5.7% |
| Comorbidities | |||||
| Hypertension | 15 062 (10.3%) | 25 000 (15.2%) | 218 752 (18.1%) | 7.8% | 2.9% |
| Heart failure | 21 952 (15.0%) | 23 883 (14.5%) | 171 796 (14.2%) | 0.8% | 0.3% |
| CVD | 10 672 (7.3%) | 15 917 (9.7%) | 129 649 (10.7%) | 3.4% | 1.0% |
| CKD | 14 547 (10.0%) | 19 779 (12.0%) | 123 216 (10.2%) | 0.2% | 1.8% |
| Provider specialty | |||||
| Primary care | 66 497 (32.3%) | 89 962 (33.4%) | 516 497 (34.8%) | 2.5% | 1.4% |
| Endocrinology | 10 856 (5.3%) | 22 469 (8.3%) | 72 109 (4.9%) | 0.4% | 3.4% |
| Cardiology | 28 701 (13.9%) | 41 983 (15.6%) | 240 781 (16.2%) | 2.3% | 0.6% |
| Nephrology | 39 291 (19.1%) | 60 995 (22.7%) | 346 977 (23.4%) | 4.3% | 0.7% |
| Other specialty | 60 539 (29.4%) | 53 865 (20.0%) | 308 200 (20.8%) | 8.6% | 0.8% |
Abbreviations: BMI, body mass index; CKD, chronic kidney disease; CVD, cardiovascular disease; SMDs, standardized mean differences.
Prescription rates for both medication classes showed a consistent and significant increase over the study period. As shown in Figures 1 and 2, GLP-1 RA prescriptions increased from 0.89% in 2014% to 15.85% in 2024, and SGLT2i prescriptions rose steadily from 0.50% in 2014% to 12.11% in 2024. Figure 3 illustrate similar upward trends for combined GLP-1 RA and SGLT2i prescriptions. White patients had consistently higher rates of prescription for both SGLT2i and GLP-1 RA compared to Black/African American patients. Similarly, patients with private insurance had the highest rates compared to those with Medicare or Medicaid. These racial differences in prescription rates persisted within each insurance category, particularly among older patients aged 65 and over. A sensitivity analysis treating cells with fewer than 10 prescriptions as missing values confirmed that these findings were consistent and robust (Supplementary Fig. 1).
To account for potential confounding and interaction effects, we fit negative binomial models including all 2-way interactions. The main effect for Black/African American patients was no longer significant overall in the combined medication model; however, age-by-race interactions remained significant. Among patients aged 45–64 years, Black patients had an IRR of 0.85 (95% CI: 0.77–0.95, P = 0.004), indicating they were 15% less likely to receive these medications compared to White patients in the same age group. For patients aged ≥65 years, the IRR was 0.71 (95% CI: 0.64–0.79, P < 0.001), meaning they were 29% less likely to receive prescriptions (Supplementary Table 1). Insurance type also showed strong associations: compared to commercially insured patients, those with Medicaid (IRR = 0.65, 95% CI: 0.55–0.77, P < 0.001) or Medicare (IRR = 0.59, 95% CI: 0.50–0.70, P < 0.001) were significantly less likely to receive these medications. Similar patterns were observed in separate models for GLP-1 RA and SGLT2i (Supplementary Table 2).
Conclusions
In this retrospective cohort study from 2014 to 2024, we observed a significant and sustained increase in the prescription rates of both SGLT2i and GLP-1 RA among patients with T2D. This trend aligns with the growing evidence base for these medications and their increasing prominence in clinical guidelines. However, despite this overall rise, significant differences in prescription rates persist, particularly for Black patients and those with public insurance.
Our findings are consistent with existing literature highlighting disparities in the use of these guideline-recommended therapies.10, 11, 12, 13 Similar to other studies, we found that Black patients were less likely to be prescribed these medications compared to their White counterparts. Our results also demonstrate that patients with private insurance had consistently higher prescription rates than those with Medicare or Medicaid, suggesting that financial barriers, such as out-of-pocket costs and formulary coverage, may be significant determinants of access. The observed differences are also present when stratified by insurance type, with White patients consistently receiving a higher proportion of prescriptions within each insurance category. Furthermore, our findings indicate that patients receiving SGLT2i or GLP-1 RA prescriptions had a greater number of comorbidities and higher average BMI, which aligns with clinical practice guidelines recommending these medications for patients with established cardiovascular disease, heart failure, or chronic kidney disease. Primary care physicians prescribed about one-third of SGLT2i and GLP-1 RA drugs. In pooled descriptive analyses, higher observed prescribing rates were seen in endocrinology than in some other specialties; however, these comparisons were unadjusted and should be interpreted descriptively.
These findings have important clinical implications for T2D management. The observed differences suggest that factors beyond clinical eligibility, such as socioeconomic status, race, and insurance coverage, influence prescribing patterns.21 Addressing these differences is crucial to ensure equitable access to therapies that have proven benefits for cardiovascular and renal outcomes. Interventions within healthcare systems, such as real-time electronic health record prompts for eligible patients and enhanced patient-provider communication regarding medication benefits and costs, may help reduce these differences.22 Additionally, policy-level changes, such as adjustments to reimbursement strategies and expanding insurance coverage, are needed to support more equitable access.
Strengths and Limitations
A key strength of our study is the use of Epic Cosmos, which provides a large and diverse patient population, enhancing the generalizability of our findings. The use of EHR data allowed us to include laboratory values and to define the eligibility based on clinical characteristics, providing a more robust and clinically relevant cohort.
However, several limitations should be considered. First, EHR data may not capture all confounding variables, such as disease duration or patient preference, which could influence prescribing decisions. Because medication exposure was derived from annual aggregate Cosmos/SlicerDicer counts rather than patient-level medication histories, we could not apply an additional lookback period or verify dispensing, fills, or actual medication use. We also did not account for site- or region-level variation in prescribing. Also, we could not provide further breakdowns by Hispanic ethnicity or comorbidities for certain subgroups because of small cell sizes. When a cell count was below 10, Cosmos reported it as “<10.” We treated these cells as equal to 10. We still saw small cell sizes in early years of data from 2014 to 2016. As a result, the fitted estimates from the early years of data may be less accurate or interpretable. In addition, temporal prescribing trends within guideline-relevant subgroups (established ASCVD, HF, and CKD) could not be examined because stratification of the aggregate Cosmos data produced extensive small-cell suppression. And the increase in GLP-1 RA prescribing after 2022 may partly reflect inclusion of tirzepatide within the broader GLP-1 RA category, which could not be separated reliably in the aggregate Cosmos data. Further, we were limited to using the aggregated computable phenotypes of type 2 diabetes built into the Epic Cosmos SlicerDicer tool rather than more robust phenotypes described in the literature, including our own prior work.23 In addition, we could not restrict the cohort to patients with longer-term relationships with the health system because such criteria were not supported by the aggregated data available. Also, we could not account for combination therapy or overlapping prescriptions. Moreover, while the Epic Cosmos network is vast, it does not represent all health care systems in the United States, and the findings may not be generalizable to smaller, non–Epic-affiliated practices. Despite these limitations, our study provides valuable insights into real-world prescribing trends and highlights critical areas for targeted interventions to reduce health disparities.
In summary, these findings highlight the need for continued efforts to understand the root causes of these differences and to implement strategies, both at the health system and policy levels, to ensure that all eligible patients with T2D have equitable access to these life-saving medications.
Disclosure
The authors have no conflicts of interest to disclose.
Acknowledgment
This work was supported by the Doucette Innovation Fund (PI: Dr Divers). This research was also supported, in part, by the National Institutes of Health (NIH) under Agreement No. 1OT2OD032581 (PI: Dr Zhang). The views and conclusions expressed in this document are those of the authors and do not necessarily represent the official policies, either expressed or implied, of the NIH.
Drs Jasmin Drivers and Donglan Stacy Zhang are the guarantor of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. We also thank the AIM-AHEAD Coordinating Center for supporting this research through the FAIR‑MED Award.
Author Contributions
D.S.Z. and A.R. are co-first authors and contributed equally to the manuscript. D.S.Z., A.R., S.I., D.M.C., A.J., D.R.W., J.W., and J.D. designed the research and drafted the manuscript. A.R. and S.I. conducted the statistical analysis. D.S.Z. and J.D. had primary responsibility for the final content. All authors contributed to data interpretation, critically reviewed and edited the manuscript, and approved the final version.
Declaration of Generative AI and AI-Assisted Technologies in the Writing Process
During the course of preparing this work, the author(s) used ChatGPT 5.0 to check for grammatical errors, improve flow, and design the graphic abstract. Following the use of this tool/service, the author(s) formally reviewed the content for its accuracy and edited it as necessary. The author(s) take full responsibility for all the content of this publication.
Supplementary Material
References
- 1.McCoy R.G., Dykhoff H.J., Sangaralingham L., et al. Adoption of new glucose-lowering medications in the US—The case of SGLT2 inhibitors: Nationwide cohort study. Diabetes Technol Ther. 2019;21(12):702–712. doi: 10.1089/dia.2019.0213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Burke K.R., Schumacher C.A., Harpe S.E. SGLT 2 inhibitors: a systematic review of diabetic ketoacidosis and related risk factors in the primary literature. Pharmacother J Hum Pharmacol Drug Ther. 2017;37(2):187–194. doi: 10.1002/phar.1881. [DOI] [PubMed] [Google Scholar]
- 3.Drucker D.J. Efficacy and safety of GLP-1 medicines for type 2 diabetes and obesity. Diabetes Care. 2024;47(11):1873–1888. doi: 10.2337/dci24-0003. [DOI] [PubMed] [Google Scholar]
- 4.Zelniker T.A., Wiviott S.D., Raz I., et al. SGLT2 inhibitors for primary and secondary prevention of cardiovascular and renal outcomes in type 2 diabetes: a systematic review and meta-analysis of cardiovascular outcome trials. Lancet. 2019;393(10166):31–39. doi: 10.1016/S0140-6736(18)32590-X. [DOI] [PubMed] [Google Scholar]
- 5.Ryan P.B., Buse J.B., Schuemie M.J., et al. Comparative effectiveness of canagliflozin, SGLT2 inhibitors and non-SGLT2 inhibitors on the risk of hospitalization for heart failure and amputation in patients with type 2 diabetes mellitus: a real-world meta-analysis of 4 observational databases (OBSERVE-4D) Diabetes Obes Metab. 2018;20(11):2585–2597. doi: 10.1111/dom.13424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.McGuire D.K., Shih W.J., Cosentino F., et al. Association of SGLT2 inhibitors with cardiovascular and kidney outcomes in patients with type 2 diabetes: a meta-analysis. JAMA Cardiol. 2021;6(2):148–158. doi: 10.1001/jamacardio.2020.4511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Yau K., Dharia A., Alrowiyti I., Cherney D.Z. Prescribing SGLT2 inhibitors in patients with chronic kidney disease: expanding indications and practical considerations. Kidney Int Rep. 2022;7(7):1463–1476. doi: 10.1016/j.ekir.2022.04.094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Choi J.G., Winn A.N., Skandari M.R., et al. First-line therapy for type 2 diabetes with sodium–glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists: a cost-effectiveness study. Ann Intern Med. 2022;175(10):1392–1400. doi: 10.7326/M21-2941. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.O’Brien M.J., Karam S.L., Wallia A., et al. Association of second-line antidiabetic medications with cardiovascular events among insured adults with type 2 diabetes. JAMA Network Open. 2018;1(8) doi: 10.1001/jamanetworkopen.2018.6125. e186125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gregg L.P., Ramsey D.J., Akeroyd J.M., et al. Predictors, disparities, and facility-level variation: SGLT2 inhibitor prescription among US veterans with CKD. Am J Kidney Dis. 2023;82(1):53–62.e1. doi: 10.1053/j.ajkd.2022.11.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Zhao J.Z., Weinhandl E.D., Carlson A.M., Peter W.L.S. Disparities in SGLT2 inhibitor or glucagon-like peptide 1 receptor agonist initiation among medicare-insured adults with CKD in the United States. Kidney Med. 2023;5(1) doi: 10.1016/j.xkme.2022.100564. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Lamprea-Montealegre J.A., Madden E., Tummalapalli S.L., et al. Association of race and ethnicity with prescription of SGLT2 inhibitors and GLP1 receptor agonists among patients with type 2 diabetes in the Veterans Health Administration System. JAMA. 2022;328(9):861–871. doi: 10.1001/jama.2022.13885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gay H.C., Yu J., Persell S.D., et al. Comparison of sodium-glucose Cotransporter-2 inhibitor and glucagon-like Peptide-1 receptor agonist prescribing in patients with diabetes mellitus with and without cardiovascular disease. Am J Cardiol. 2023;189:121–130. doi: 10.1016/j.amjcard.2022.10.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Adhikari R., Jha K., Dardari Z., et al. National trends in use of sodium-glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists by cardiologists and other specialties, 2015 to 2020. J Am Heart Assoc. 2022;11(9) doi: 10.1161/JAHA.121.023811. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Mahtta D., Ramsey D.J., Lee M.T., et al. Utilization rates of SGLT2 inhibitors and GLP-1 receptor agonists and their facility-level variation among patients with atherosclerotic cardiovascular disease and type 2 diabetes: insights from the Department of Veterans Affairs. Diabetes Care. 2022;45(2):372–380. doi: 10.2337/dc21-1815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.About Cosmos. https://cosmos.epic.com/about/ Accessed August 6, 2025.
- 17.Chi G.C., Li X., Tartof S.Y., Slezak J.M., Koebnick C., Lawrence J.M. Validity of ICD-10-CM codes for determination of diabetes type for persons with youth-onset type 1 and type 2 diabetes. BMJ Open Diabetes Res Care. 2019;7(1) doi: 10.1136/bmjdrc-2018-000547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Zoungas S., de Boer I.H. SGLT2 inhibitors in diabetic kidney disease. Clin Journal Am Soc Nephrol. 2021;16(4):631–633. doi: 10.2215/CJN.18881220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Carlson S.A., Watson K.B., Rockhill S., Wang Y., Pankowska M.M., Greenlund K.J. Linking local-level chronic disease and social vulnerability measures to inform planning efforts: a COPD example. Preventing Chronic Dis. 2023;20:E76. doi: 10.5888/pcd20.230025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Bakbergenuly I., Hoaglin D.C., Kulinskaya E. Estimation in meta-analyses of mean difference and standardized mean difference. Stat Med. 2020;39(2):171–191. doi: 10.1002/sim.8422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Luo J., Feldman R., Kim K.C., et al. Evaluation of out-of-pocket costs and treatment intensification with an SGLT2 inhibitor or GLP-1 RA in patients with type 2 diabetes and cardiovascular disease. JAMA Netw Open. 2023;6(6) doi: 10.1001/jamanetworkopen.2023.17886. e2317886. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Okoro O.N., Hillman L.A., Cernasev A. “We get double slammed!”: healthcare experiences of perceived discrimination among low-income African-American women. Women's Health. 2020;16 doi: 10.1177/1745506520953348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Shao H., Thorpe L.E., Islam S., et al. Developing a computable phenotype for identifying children, adolescents, and young adults with diabetes using electronic health records in the DiCAYA network. Diabetes Care. 2025;48(6):914–921. doi: 10.2337/dc24-1972. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.