Trends in Prescriptions of Cardioprotective Diabetic Agents After Coronary Artery Bypass Grafting Among U.S. Veterans

Coronary artery bypass grafting (CABG) is the preferred treatment for patients with type 2 diabetes (T2D) and multivessel coronary artery disease (1,2). Sodium–glucose cotransporter 2 receptor inhibitors (SGLT2i) and glucagon-like peptide 1 receptor agonists (GLP-1RA) reduce cardiovascular risk in patients with T2D and are, therefore, recommended for all T2D patients with atherosclerotic vascular disease (3,4). However, their adoption after CABG is uncertain. We therefore analyzed patterns and trends for SGLT2i/GLP-1RA prescription in patients receiving CABG at Veterans Affairs (VA) medical centers nationwide. We evaluated longitudinal trends in SGLT2i/GLP-1RA use during the first postoperative year and studied clinical and socioeconomic factors associated with the use of these cardioprotective medications.

Our aims in this study were 1) to evaluate the overall use of SGLT2i/GLP-1RA after CABG and explore longitudinal trends and 2) to examine patient-related factors associated with the use of SGLT2i or GLP-1RA.

We analyzed patients with T2D who underwent CABG (1 January 2016 through 31 December 2019) at VA medical centers nationwide (Supplementary Fig. 1). We excluded patients with unknown vital status during the first postoperative year and used outpatient pharmacy fill records to determine which patients filled prescriptions for either an SGLT2i (empagliflozin, canagliflozin, dapagliflozin, or ertugliflozin) or a GLP-1RA (semaglutide, liraglutide, exenatide, or dulaglutide). We obtained the preoperative clinical, laboratory, and socioeconomic characteristics (community deprivation index and zip code–derived median household income) for all patients (Supplementary Table 1).

The study was approved by the institutional review board of the Cleveland VA Medical Center (CY 040).

We present data as counts (percentages) or medians (interquartile range [IQR]). We calculated proportions (with 95% CI) for SGLT2i and GLP-1RA use 1) for the entire study period, 2) for every quarter year, and 3) for each VA medical center. We compared the use of SGLT2i and GLP-1RA for each VA medical center using a correlation test. To study the association between baseline characteristics and SGLT2i/GLP-1RA use, we fit a multivariable binomial logistic regression model with patient demographics (age, sex, race, and ethnicity), clinical characteristics (heart failure [HF], heart failure with reduced ejection fraction [HFrEF], peripheral arterial disease [PAD], cerebrovascular disease, chronic kidney disease [CKD], etc.), and socioeconomic data (community deprivation index and zip code–derived median household income) as covariates and SGLT2i or GLP-1RA use as the end point. We report results as adjusted odds ratios (OR) with 95% CI. To evaluate whether prescription rates for SGLT2i and GLP-1RA changed over time, we initially fit a separate generalized additive model (GAM) of the quarterly prescription rates for each drug against time. On observing a substantial nonlinear increase in SGLT2i prescription rates over time, we performed a breakpoint analysis to identify the time point beyond which prescription rates increased.

From 2016 to 2019, 5,109 patients with T2D underwent CABG (median age 68 [IQR 63, 71] years, 98.6% male, 77.8% White, 11.6% Hispanic) at 40 VA medical centers. CKD, PAD, HFrEF, and cerebrovascular disease were present in 38.6%, 27.9%, 9%, and 6.8%, respectively (Supplementary Table 2). Overall, 10.4% (535 of 5,109; 95% CI 9.6, 11.4) and 6.8% (352 of 5,109; 95% CI 6.2, 7.6) received a prescription for SGLT2i and GLP-1RA, respectively, and 1.8% (94 of 5,109; 95% CI 1.5, 2.2) received both. Variation in prescription rates between VA medical centers was large (Supplementary Fig. 2), with poor correlation between SGLT2i and GLP-1RA prescription rates in each VA medical center (correlation coefficient 0.08 [95% CI −0.23, 0.38]) (Supplementary Fig. 3). A higher median income (OR 1.08 [1.03, 1.13] per USD 5,000 increase in median income), living in less deprived neighborhoods (OR 1.19 [0.98, 1.44]), and obesity (OR 1.39 [1.15, 1.69]) were factors associated with receiving SGLT2i, while patients with preexisting PAD (OR 0.75 [0.60, 0.94]) or CKD (OR 0.72 [0.58, 0.88]) were less likely to receive SGLT2i. We did not observe any association between preexisting HF (OR 1.10 [0.85, 1.40]) or HFrEF (OR 0.86 [0.59, 1.27]) and SGLT2i prescription. With Black race as a reference, White individuals (OR 1.64 [1.11, 2.51]) were more likely to receive GLP-1RA therapy. Obesity (OR 1.91 [1.50, 2.46]) was associated with GLP-1RA prescription, while patients with cerebrovascular disease (OR 0.59 [0.32, 0.99]) were less likely to receive GLP-1RA prescription (Supplementary Table 3).

From quarter 1 of 2016 (2016Q1) to 2019Q4, prescription rates per quarter increased for both medications; however, this effect was far greater for SGLT2i (20-fold increase, from 1.6% to 33%) than GLP-1RA (14-fold increase, from 0.8% to 11.2%) (Fig. 1A and Supplementary Table 4). Since 2018Q1, we observed a nonlinear increase in prescription rates for SGLT2i (GAM smoothing parameter P value of 0.03). While GLP-1RA prescriptions also increased over time, they did not increase at the same rate, and prescription rates appear to stabilize from 2019Q1 (Fig. 1B). The exploratory breakpoint analysis model further supports the GAM by defining 2018Q1 as the breakpoint for SGLT2i prescriptions (Fig. 1C).

From over 5,000 patients receiving CABG at 40 different VA medical centers, we observed low utilization of SGLT2i or GLP-1RA therapy after surgery. SGLT2i were more likely to be prescribed than GLP-1RA, and prescription rates for SGLT2i have rapidly increased since 2018Q1.

Current guidelines recommend SGLT2i or GLP-1RA for all patients with T2D and established atherosclerotic vascular disease (3,4). In CABG patients, T2D is independently associated with future cardiovascular risk (5); thus, surgery provides an opportunity to initiate guideline-directed care. Therefore, it is unfortunate that the overall prescription rates among CABG patients are not higher than those for patients with T2D and stable coronary artery disease (6). While the observed 20-fold increase in SGLT2i use over a 4-year study period is encouraging, two-thirds of patients with T2D undergoing CABG remained untreated with either recommended drug, suggesting a significant opportunity to improve cardiovascular outcomes in this high-risk group. As in a prior study from Denmark, we observed higher prescription rates for SGLT2i than GLP1-RA (7). Possible reasons for preferring SGLT2i in our study are the advantage of oral therapy and more familiarity among cardiologists with SGLT2i. Drug cost is unlikely to play a role, as veterans have the same copay for either medication.

Increased median household income was associated with SGLT2i use, while obesity was associated with GLP-1RA use. Compared with Black patients, White patients were also more likely to receive GLP-1RA therapy. These findings support prior observations from commercially insured patients (8). That these socioeconomic disparities should exist among veterans is perhaps surprising, as copay amounts are highly subsidized. However, data from Denmark, a country with a universal health care system, also reported similar observations (9). SGLT2i use was lower among patients with CKD and PAD, both high-risk subgroups that derive high absolute benefits from SGLT2i therapy (10,11). While the low use of SGLT2i in PAD patients is likely related to concerns of increased amputation rates observed in the Canagliflozin Cardiovascular Assessment Study (CANVAS) trial, neither the Canagliflozin and Renal Events in Diabetes with Established Nephropathy Clinical Evaluation (CREDENCE) trial nor results from large retrospective data searches support these findings (12,13).

Retrospective data analysis, a predominantly male cohort, use of ICD codes to define T2D, and the reliance on pharmacy fill data are limitations of our study. Our study is, however, likely the first to evaluate the use of SGLT2i/GLP-1RA among patients with T2D following CABG, a high-risk cohort with considerable potential to benefit from receiving either cardioprotective agent.

In conclusion, between 2016 and 2019, SGLT2i use and, to a lesser extent, GLP1-RA use increased substantially among U.S. veterans undergoing CABG, with SGLT2i use accelerating rapidly since 2018. However, socioeconomic disparities exist and opportunities for improvement remain.

Funding and Duality of Interest. This material is the result of work supported with resources and use of facilities at the Louis Stokes Cleveland VA Medical Center, VA Northeast Ohio Healthcare System. N.S. reports personal fees from Afimmune, Amgen, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Hanmi Pharmaceuticals, Merck Sharp & Dohme, Novartis, Novo Nordisk, Pfizer, and Sanofi and grant funding paid to his university from AstraZeneca, Boehringer Ingelheim, Novartis, and Roche Diagnostics outside the submitted work. No other potential conflicts of interest relevant to this article were reported.

The views expressed in this article are those of the authors. They do not represent the position or policy of the Department of Veterans Affairs or the U.S. government.

Author Contributions. S.V.D. designed the study, did the statistical analysis, and wrote the manuscript. D.A.M. designed the study and reviewed and edited the manuscript. S.A.-K. wrote the manuscript. Y.E. reviewed and edited the manuscript. D.C. reviewed and edited the manuscript. J.P. designed the study and reviewed and edited the manuscript. N.S. designed the study and reviewed and edited the manuscript. S.V.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.