SGLT2 inhibitors are associated with reduced cardiovascular disease in patients with type 2 diabetes: An analysis of real-world data

Abstract

Objective:

To assess the association between sodium-glucose cotransporter-2 (SGLT2) inhibitors and other 2^nd line diabetes therapies with risk of cardiovascular disease (CVD), as well as conduct head-to-head comparisons between SGLT2 inhibitors.

Patients and Methods:

Using data from the MarketScan databases (January 1, 2013-December 31, 2019), SGLT2 inhibitor users were matched with up to 5 other 2^nd line therapy users by age, sex, date of enrollment, and date of 2^nd line therapy initiation. The primary composite outcome included stroke, atrial fibrillation, myocardial infarction, and heart failure. Hazard ratios (HR) were estimated, adjusting for demographics and a propensity score reflecting comorbidities and medications.

Results:

In this study population of 313,396 patients (mean age 53±10 years; 47% female), 9,787 incident CVD events occurred over a median follow-up of 1.36 years. After multivariable adjustments, SGLT2 inhibitor users had a lower risk of CVD than other 2^nd line therapy users (HR [95% CI]: 0.66 [0.62–0.71]). Significant associations were also observed when each CVD outcome was assessed separately. No differences were noted when comparing individual SGLT2 inhibitors.

Conclusion:

SGLT2 inhibitors were associated with a clinically meaningfully lower CVD risk in the real-world setting. In head-to-head comparisons, the different SGLT2 inhibitors were consistent in their protective associations with CVD. This suggests that as a class, SGLT2 inhibitors may have widespread benefit in preventing CVD among patients with type 2 diabetes.

INTRODUCTION

Type 2 diabetes increases the risk of cardiovascular events. The American Diabetes Association has recommended that among patients with type 2 diabetes and established cardiovascular disease (CVD), treatment should include a glucose-lowering medication that also has cardiovascular benefits.¹ Sodium-glucose cotransporter-2 (SGLT2) inhibitors have been advocated as a key 2^nd line therapy, used after metformin, to improve glycemic control while conferring cardioprotective properties. SGLT2 inhibitors block the reabsorption of glucose in the proximal tubules of the kidney.² The first SGLT2 inhibitor, canagliflozin, was approved by the US Food and Drug Association in March 2013. Since then, three other SGLT2 inhibitors (empagliflozin, dapagliflozin, and ertugliflozin) have been approved. Recent data suggests that use of newer 2^nd line therapies, such as SGLT2 inhibitors, has increased, although the use of newer medications continues to remain low.³

Data from several randomized clinical trials (RCTs) have suggested that among patients with diabetes at high-risk or with established CVD, SGLT2 inhibitors reduce the risk for cardiovascular events compared to placebo.^4–6 More recently, studies using real-world data have found that SGLT2 inhibitor use is associated with a lower risk of composite cardiovascular events compared to use of other antihyperglycemic medications.^7–10 However, although metformin is the recommended 1^st line therapy for type 2 diabetes treatment,¹¹ prior studies have not typically restricted to metformin users. In addition, among studies that utilized real-world data to evaluate cardiovascular outcomes separately, results have been mixed. Although SGLT2 inhibitor use has been consistently associated with lower risk of heart failure (HF),^{8, 9, 12–14} their effects on other outcomes, such as myocardial infarction (MI), stroke and atrial fibrillation (AF), have been inconsistent.^{7–10, 15, 16}

Existing evidence suggests that SGLT2 inhibitor use may reduce the risk of CVD, but additional research using real-world data is warranted to better understand the relationship of SGLT2 inhibitors and CVD outcomes (in aggregate and individually), particularly in assessing the association among patients without CVD, exploring outcomes individually, and evaluating the association among subgroups of interest. Furthermore, head-to-head comparisons of the different SGLT2 inhibitors (i.e., canagliflozin, dapagliflozin, empagliflozin) have not yet been reported. Therefore, utilizing data from the U.S. MarketScan administrative databases, we tested the hypothesis that among patients with type 2 diabetes taking metformin who were free of established CVD (stroke, AF, MI, or HF), addition of SGLT2 inhibitors as a 2^nd line therapy have a lower risk of CVD compared to addition other 2^nd line therapies (sulfonylureas, dipeptidyl peptidase-4 [DPP-4] inhibitors, glucagon-like peptide-1 [GLP-1] receptor agonists, thiazolidinediones, and insulin). We also performed head-to-head comparisons between SGLT2 inhibitors.

PATIENTS AND METHODS

Study Population

We used data from the MarketScan Commercial Claims and Encounters and Medicare Supplemental and Coordination of Benefits databases (IBM Corporation) from January 1, 2013 to December 31, 2019.¹⁷ These databases include enrollment records, as well as billing claims for inpatient, outpatient, and pharmacy services. Because the data were deidentified, the University of Minnesota institutional review board deemed this study exempt from review and informed consent was not required.

Our analytic approach aimed to emulate the target trial, and therefore minimize potential biases by using an active comparator, new user study design.^{18, 19} To identify patients with type 2 diabetes, we required at least one inpatient, or two outpatient claims for type 2 diabetes (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 250.x0 and 250.x2 or Tenth Revision [ICD-10-CM] codes E11.xx in any position). Patients with type 1 diabetes (ICD-9-CM codes 250.x1 and 250.x3 or ICD-10-CM codes E10.xx) were excluded. In a prior validation study, this definition of diabetes had a sensitivity of 92%, while positive predictive values ranged from 72% to 77%.²⁰ In addition, we required all patients to have at least one prescription of metformin, which is the recommended 1^st line therapy for type 2 diabetes treatment per the American Diabetes Association guidelines.¹¹

The initial sample included 4,944,269 patients with type 2 diabetes. Of these patients, 1,565,715 were taking both metformin and a 2^nd line therapy. Exclusion criteria included being aged <18 years, <90 days enrollment, taking a 2^nd line antihyperglycemic combination drug therapy, initiation of >1 2^nd line therapies on the same day, and having established CVD (stroke, AF, MI, or HF). CVD was defined using the same ICD-9-CM and ICD-10-CM criteria as the outcome ascertainment, described below.^21–24 After exclusions, there were 535,279 patients eligible for matching. The final analytic sample after matching was 313,396 patients (Figure 1).

Figure 1.

Study sample exclusion flowchart

2^nd Line Diabetes Therapy Use

2^nd line diabetes therapies included in this study were sulfonylureas, DPP-4 inhibitors, SGLT2 inhibitors, GLP-1 receptor agonists, thiazolidinediones, or insulin as these are the six preferred treatment options that can be combined with metformin.¹¹ New users of a 2^nd line diabetes therapy were categorized based on the first 2^nd line therapy they were taking following their type 2 diabetes diagnosis. Patients remained in their initial category for the entire analysis in order to mimic an intention-to-treat analysis of RCTs.²⁵ SGLT2 inhibitors included canagliflozin, dapagliflozin, empagliflozin, and ertugliflozin, while other 2^nd line diabetes therapy included sulfonylureas, DPP-4 inhibitors, GLP-1 receptor agonists, thiazolidinediones, and insulin.

Outcome Ascertainment

The primary outcome of interest was a cardiovascular event (stroke, AF, MI, or HF) as a composite outcome. Stroke was defined as ICD-9-CM codes 430, 431, 434, and 436 or ICD-10-CM codes I60-I69 as the primary diagnosis in an inpatient claim.²¹ AF was defined as ICD-9-CM codes 427.31, 427.32 or ICD-10-CM codes I48 in any position in one inpatient claim or 2 separate outpatient claims.²² MI was defined as ICD-9-CM code 410 or ICD-10-CM codes I21-I22 as the primary diagnosis in an inpatient claim,²³ while HF was defined as ICD-9-CM codes 402.x1, 404.x1, 404.x3, and 428 or ICD-10-CM codes I11.0, I13.0, I13.2, I50 as the primary diagnosis in an inpatient claim.²⁴

Covariate Ascertainment

Covariates were defined using inpatient, outpatient, and pharmacy claims prior to or at the time of initiation of 2^nd line diabetes therapy. Prescription medications included angiotensin converting enzyme inhibitors, alpha-blockers, angiotensin receptor blockers, beta-blockers, calcium channel blockers, loop diuretics, potassium-sparing diuretics, thiazide diuretics, direct vasodilators. Comorbidities included hypertension, kidney disease, chronic pulmonary disease, depression, metastatic cancer, and asthma. Supplemental Table 1 provides the ICD-9-CM and ICD-10-CM codes used to define comorbidities.

Matching

Each SGLT2 inhibitor new user was matched with up to 5 other 2^nd line therapy new users by age (±3 years), sex, date of enrollment (±90 days), and date of 2^nd line therapy initiation (±90 days). Matching was done using an automated greedy matching algorithm.²⁶ For each head-to-head comparison of SGLT2 inhibitors, a 1:1 matching process was performed. Head-to-head comparisons with ertugliflozin were not performed given the small sample size of patients taking this medication.

Statistical Analysis

Baseline characteristics, stratified by 2^nd line therapy, were described using means and standard deviations for continuous variables and count and percentages for categorical variables. Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for incident CVD. Follow-up time was defined as time from first 2^nd line therapy prescription to the occurrence of incident cardiovascular event, health plan disenrollment, or December 31, 2019, whichever occurred first. Information on the reason a patient disenrolled from their health plan, such as death or change in insurance plan, was not available. Models adjusted for age, sex, ICD 9/10 status, and a propensity score reflecting chronic health conditions and CVD medications. Propensity scores were calculated from logistic regression models that predicted likelihood of 2^nd line therapy based on comorbidities and medications (age, sex, hypertension, kidney disease, chronic pulmonary disease, depression, metastatic cancer, asthma, ACE inhibitors/ARBs, alpha blockers, beta blockers, calcium channel blockers, loop diuretics, potassium, thiazide, and vasodilators). For each comparison, separate propensity scores were calculated. Head-to-head comparisons of SGLT2 inhibitors (i.e., canagliflozin, dapagliflozin, empagliflozin) were also performed.

Secondary analyses were performed 1) comparing each SGLT2 inhibitor (i.e., canagliflozin, dapagliflozin, empagliflozin) to other 2^nd line therapies and 2) comparing SGLT2 inhibitors to other 2^nd line therapies individually (i.e., sulfonylureas, DPP-4 inhibitors, GLP-1 receptor agonists, thiazolidinediones, and insulin). All analyses were performed using SAS, version 9.4 (SAS Institute Inc., Cary, NC).

RESULTS

Overall, there were 53,704 patients with type 2 diabetes on metformin who were taking a SGLT2 inhibitor, and 481,575 who were taking a different class of 2^nd line therapy. Descriptive characteristics for the full sample are provided in Supplemental Table 2.

Our primary analysis was matched and included 313,396 patients with type 2 diabetes taking metformin. These patients were on average (SD) 52.9 (9.5) years old and 47.1% were female. Baseline characteristics of matched patients are shown in Table 1, stratified by initial 2^nd line therapy. Among those who were taking an SGLT2 inhibitor, canagliflozin was the most common (37.9%), followed by empagliflozin (32.0%), dapagliflozin (29.6%), and ertugliflozin (0.5%). In those taking a non-SGLT2 inhibitor as 2^nd line therapy, sulfonylureas were the most common (45.2%), followed by DPP-4 inhibitors (18.9%).

Table 1.

Baseline Characteristics of Patients with Type 2 Diabetes by New Users of SGLT2 Inhibitor and Other 2^nd Line Therapies, MarketScan Databases^a

	SGLT2 Inhibitor New Users (n=53,691)	Matched Other 2nd Line Therapy New Users (n=259,705)
Age, years	52.9±9.5	52.9±9.6
Female sex	46.6	47.2
2nd line diabetes therapy b
Canagliflozin	37.9	--
Dapagliflozin	29.6	--
Empagliflozin	32.0	--
Ertugliflozin	0.5	--
Sulfonylurea	--	45.2
DPP-4 inhibitors	--	18.9
GLP-1 receptor agonist	--	14.7
Thiazolidinediones	--	4.3
Insulin	--	16.9
Comorbidities
Hypertension	65.2	61.9
Kidney disease	1.8	3.1
Chronic pulmonary disease	2.9	3.5
Depression	7.7	8.3
Metastatic cancer	0.3	0.6
Asthma	2.9	3.0
Other medications
ACE inhibitors	41.9	42.9
Angiotensin receptor blockers	26.3	22.5
Alpha-blockers	0.5	0.7
Beta-blockers	23.5	23.7
Calcium channel blockers	20.0	20.3
Loop diuretics	5.5	6.1
Potassium-sparing diuretics	5.2	5.3
Thiazide diuretics	12.9	13.3
Direct vasodilators	2.8	2.9

^aData are expressed as mean±SD or %.

^bFirst 2^nd line therapy patients were prescribed.

Abbreviations: SGLT2 = sodium-glucose cotransporter-2; DPP-4 = dipeptidyl peptidase-4; GLP-1 = glucagon-like peptide-1; ACE = angiotensin-converting enzyme

Over a median follow-up of 1.36 years, 9,787 incident cardiovascular events occurred. Incident rate per 1,000 person-years was 17.64 for composite cardiovascular events. When assessed individually, incident rates for stroke, AF, MI, and HF were 3.86, 8.50, 4.73, and 2.52 per 1,000 person-years, respectively. The primary Cox analysis was adjusted for age, sex, ICD-9/10 status, and a propensity score derived using information on numerous comorbidities and medications. Compared to other 2^nd line therapies, patients taking a SGLT2 inhibitor had a lower risk of cardiovascular events (Table 2; HR [95% CI]: 0.66 [0.62, 0.71]). When each outcome was assessed individually, risk reductions were observed for all outcomes: HRs [95% CIs] for stroke = 0.52 [0.45, 0.60], AF = 0.79 [0.73, 0.86], MI = 0.65 [0.58, 0.73] and HF = 0.45 [0.37, 0.54], respectively. Head-to-head comparison of SGLT2 inhibitors are shown in Table 3. Overall, risk of cardiovascular events did not differ meaningfully based on type of SGLT2 inhibitor. For the composite CVD outcome, the HRs (95% CIs) were 0.91 (0.77, 1.08) for dapagliflozin vs. canagliflozin, 0.85 (0.70, 1.03) for dapagliflozin vs. empagliflozin, and 0.98 (0.79, 1.22) for empagliflozin vs. canagliflozin. Furthermore, when stratified by subgroups of interests, results remained relatively consistent (Figure 2).

Table 2.

Hazard Ratios (95% Confidence Intervals)^a for Incident Cardiovascular Disease Comparing SGLT2 Inhibitor Users vs. Users of Other 2^nd Line Therapies, MarketScan Databases, 2013–2019

	SGLT2 Inhibitor Users (n=53,691)	Other 2nd Line Therapy Users (n=259,705)
Composite CVD Outcome
N, events	1,178	8,609
Incidence Rateb	12.07	18.83
Hazard Ratio (95% CI)	0.66 (0.62, 0.71)	Reference
Stroke
N, events	210	1,934
Incidence Rateb	2.15	4.23
Hazard Ratio (95% CI)	0.52 (0.45, 0.60)	Reference
Atrial Fibrillation
N, events	653	4,060
Incidence Rateb	6.69	8.88
Hazard Ratio (95% CI)	0.79 (0.73, 0.86)	Reference
Myocardial Infarction
N, events	313	2,311
Incidence Rateb	3.21	5.06
Hazard Ratio (95% CI)	0.65 (0.58, 0.73)	Reference
Heart Failure
N, events	112	1,287
Incidence Rateb	1.15	2.82
Hazard Ratio (95% CI)	0.45 (0.37, 0.54)	Reference

^aAdjusted for age, sex, ICD-9/10 status, propensity score (hypertension, kidney disease, chronic pulmonary disease, depression, metastatic cancer, asthma, ACE inhibitors, alpha-blockers, angiotensin receptor blockers, beta-blockers, calcium channel blockers, loop diuretics, potassium-sparing diuretics, thiazide diuretics, direct vasodilators)

^bPer 1,000 person-years

Abbreviations: SGLT2 = sodium-glucose cotransporter-2

Table 3.

Hazard Ratios (95% Confidence Intervals)^a for Incident Cardiovascular Disease Comparing Types of SGLT2 Inhibitors, MarketScan Databases, 2013–2019

	Dapagliflozin vs. Matched Canagliflozin Users		Dapagliflozin vs. Matched Empagliflozin Users		Empagliflozin vs. Matched Canagliflozin Users
	Dapagliflozin (n=11,149)	Canagliflozin (n=11,149)	Dapagliflozin (n=11,680)	Empagliflozin (n=11,680)	Empagliflozin (n=7,689)	Canagliflozin (n=7,689)
Composite CVD Outcome
N, events	247	276	190	225	164	168
Hazard Ratio (95% CI)	0.91 (0.77, 1.08)	Reference	0.85 (0.70, 1.03)	Reference	0.98 (0.79, 1.22)	Reference
Stroke
N, events	43	51	33	36	26	31
Hazard Ratio (95% CI)	0.88 (0.59, 1.33)	Reference	0.90 (0.56, 1.45)	Reference	0.87 (0.51, 1.46)	Reference
Atrial Fibrillation
N, events	135	144	100	121	89	82
Hazard Ratio (95% CI)	0.96 (0.76, 1.21)	Reference	0.84 (0.65, 1.10)	Reference	1.07 (0.80, 1.45)	Reference
Myocardial Infarction
N, events	60	80	45	63	48	52
Hazard Ratio (95% CI)	0.75 (0.54, 1.05)	Reference	0.73 (0.50, 1.07)	Reference	0.92 (0.62, 1.37)	Reference
Heart Failure
N, events	27	32	19	18	13	18
Hazard Ratio (95% CI)	0.90 (0.54, 1.50)	Reference	1.06 (0.56, 2.02)	Reference	0.74 (0.36, 1.51)	Reference

^aAdjusted for age, sex, ICD-9/10 status, propensity score (hypertension, kidney disease, chronic pulmonary disease, depression, metastatic cancer, asthma, ACE inhibitors, alpha-blockers, angiotensin receptor blockers, beta-blockers, calcium channel blockers, loop diuretics, potassium-sparing diuretics, thiazide diuretics, direct vasodilators)

Abbreviations: SGLT2 = sodium-glucose cotransporter-2

Figure 2.

Hazard Ratios (95% Confidence Intervals)^a for Incident Cardiovascular Disease^b Comparing SGLT2 Inhibitor Users vs. Users of Other 2^nd Line Therapies, Stratified by Subgroups, MarketScan Databases, 2013–2019

^aAdjusted for age, sex, ICD-9/10 status, propensity score (hypertension, kidney disease, chronic pulmonary disease, depression, metastatic cancer, asthma, ACE inhibitors, alpha-blockers, angiotensin receptor blockers, beta-blockers, calcium channel blockers, loop diuretics, potassium-sparing diuretics, thiazide diuretics, direct vasodilators)

^bIncludes stroke, atrial fibrillation, myocardial infarction, heart failure

Secondary analyses

Results continued to remain consistent when comparing each SGLT2 inhibitor to other 2^nd line therapies. Results for dapagliflozin, empagliflozin and canagliflozin are provided in Supplemental Tables 3–5, respectively. SGLT2 inhibitor users, regardless of type of SGLT2 inhibitor, had lower risk of cardiovascular events than users of other 2^nd line therapies.

Next, we compared SGLT2 inhibitors with each 2^nd line therapy individually (Supplemental Tables 6–10). When compared to sulfonylurea or insulin users, SGLT2 inhibitor users had a lower risk of all cardiovascular events (as a composite outcome and individually). In addition, SGLT2 inhibitor users had a lower risk of composite cardiovascular events, as well as stroke and HF than DPP-4 inhibitor users. SGLT2 inhibitor users also had a lower risk of HF when compared to thiazolidinedione users. There were no differences in the composite or individual component cardiovascular risks between SGLT2 inhibitor users and GLP-1 receptor agonist users.

DISCUSSION

Using a large claims database, we found that among patients with type 2 diabetes taking metformin without established CVD, the addition of a SGLT2 inhibitor as a 2^nd line therapy was associated with 34% reduced risk of cardiovascular events compared to other 2^nd line therapies. When assessed separately, risk was reduced among each cardiovascular event type, with stroke and HF showing the strongest reduced associations. Results were robust across patient characteristics of age, sex, and prevalence of hypertension and kidney disease. In secondary analyses comparing SGLT2 inhibitors to individual 2^nd line therapies, SGLT2 inhibitors were beneficial as compared to sulfonylureas, DPP-4 inhibitors, thiazolidinediones, and insulin, but for GLP-1 receptor agonists, no differences were observed. A novel aspect of this analysis is the head-to-head comparison of SGLT2 inhibitors, which revealed that the different drugs were consistent in their protective associations with risk of CVD. These findings complement results from clinical trials suggesting that SGLT2 inhibitor use, regardless of specific SGLT2 inhibitor, may be a powerful tool for reducing CVD risk among a broad array of patients with type 2 diabetes.

As noted previously, several RCTs have demonstrated SGLT2 inhibitors versus placebo to be protective against CVD.^4–6 However, understanding the effect of these medications in real-world patients is of value, since real-world patients often have more comorbidities than those enrolled in RCTs, RCTs often are based in academic medical centers, often have relatively few events for subgroup analyses, and in many instances do not conduct head-to-head comparisons for novel treatments. Real-world data also allow for testing of hypotheses in the context of current clinical guidelines. SGLT2 inhibitors are presently recommended by the American Diabetes Association as a 2^nd line therapy after metformin,¹ but SGLT2 inhibitor RCTs allowed variable use of background glucose-lowering therapies (at the discretion of the treating clinician).^4–6 Furthermore, in 2020, the U.S. Food and Drug Administration published a statement supporting the increasing role of real-world evidence in healthcare decisions.²⁷

Prior studies using real-world data have evaluated SGLT2 inhibitor use and risk of CVD, but when assessing outcomes separately, results have been mixed. The CVD-REAL 2 study and a study using MarketScan data from 2013–2015 both reported that SGLT2 inhibitor users had a lower risk of MI and stroke compared to users of other glucose-lowering medications.^{7, 9} On the other hand, the CVD-REAL Nordic study⁸ and the EASEL study¹⁰ found no difference in risk of MI or stroke when comparing SGLT2 inhibitor users to users of other glucose-lowering medications. Similarly contrasting results have been reported when analyzing AF. A Taiwanese study reported that those taking SGLT2 inhibitors had a lower risk of new-onset AF compared to those taking DPP-4 inhibitors.¹⁵ In contrast, the CVD-REAL Nordic did not find an association between SGLT2 inhibitor use (vs. other glucose-lowering medications) and AF.⁸ Furthermore, another Taiwanese study did not find an association between SGLT2 inhibitor users (compared to non-users) and AF, though an association was noted when assessing new-onset arrhythmias, which included several atrial and ventricular arrhythmias.¹⁶

Our study advances the field by showing that not only does SGLT2 inhibitor use reduce the risk of cardiovascular events as a composite outcome, but also reduces the risk of stroke, AF, MI, and HF when compared to other 2^nd line therapies. In addition, we aimed to follow clinical guidelines more closely than several of the prior analyses by requiring metformin use among all patients included in our study, given that metformin is the recommended 1^st line therapy for diabetes.¹¹ In contrast, prior studies often did not require patients to be taking metformin.^{7–10, 15} We also aimed to assess the potential of CVD prevention by excluding patients who had established CVD. Current clinical guidelines recommend SGLT2 inhibitors as a 2^nd line therapy for those with established CVD,¹¹ but additional evidence is needed to determine whether there is a clinical benefit for CVD prevention among patients with type 2 diabetes regardless of established CVD status. Other real-world studies typically did not exclude individuals with established CVD,^{7, 8, 10} and the SGLT2 inhibitor RCTs recruited participants with type 2 diabetes who had established CVD or were at high cardiovascular risk.^4–6 Furthermore, we conducted novel head-to-head comparisons between SGLT2 inhibitors.

Current clinical guidelines by the American Diabetes Association indicate that there are six preferred treatment options that can be combined with metformin (sulfonylureas, DPP-4 inhibitors, SGLT2 inhibitors, GLP-1 receptor agonists, thiazolidinediones, or insulin) as a 2^nd line therapy. The choice of which 2^nd line therapy to initiate after metformin is generally decided based on patient clinical characteristics, drug side effects, cost, and patient preference.¹¹ Results from our analysis in which we compared SGLT2 inhibitors with each 2^nd line therapy individually suggested that SGLT2 inhibitor users had lower risk of composite cardiovascular events than users of sulfonylurea, insulin, and DPP-4 inhibitors, but no difference was noted with GLP-1 receptor agonist users. In prior trials, SGLT2 inhibitors and GLP-1 receptor agonists both have shown cardiovascular benefits when compared to placebo.^{4–6, 28, 29} However, in observational data comparing SGLT2 inhibitor and GLP-1 receptor agonist use among individuals free of established CVD, no difference in risk of composite cardiovascular events was observed in our study, as well as in others.^{30, 31} As both of these therapies have shown cardiovascular benefits, clinical guidelines recommend patients with established or at a high risk for CVD initiate either SGLT2 inhibitors or GLP-1 receptor agonists as their 2^nd line therapy, independent of metformin use.^{11, 32} Guidelines also suggest that among patients who have kidney disease or HF, SGLT2 inhibitors should be initiated.¹¹ However, results from our study indicate that it may be beneficial to initiate SGLT2 inhibitors as a 2^nd line therapy regardless of established CVD status.

The adoption of SGLT2 inhibitor use is perhaps lower than would be expected given their beneficial effects in RCTs.³ SGLT2 inhibitors are more expensive than other 2^nd line therapies,¹¹ and it has been speculated that cost is a probable barrier for many patients.³³ Other therapies, such as sulfonylureas, are less expensive. However, risk of cardiovascular events are significantly reduced in SGLT2 inhibitors users compared to sulfonylurea users, as evidenced in our results, as well as in prior studies.³⁴ It is likely that generic versions of newer 2^nd line therapies, including SGLT2 inhibitors, will eventually become available.³³ This could possibly decrease the financial burden and improve access to SGLT2 inhibitors for patients with type 2 diabetes.³³ In addition, other benefits of SGLT2 inhibitors include evidence that use of SGLT2 inhibitors can result in weight loss and a reduction in systolic blood pressure,³⁵ and SGLT2 inhibitors may be more cost-effective than other diabetes medications, including DPP4-inhibitors, sulfonylureas, insulin, and thiazolidinediones.³⁶ This, in combination with our results, suggest that SGLT2 inhibitors may be the preferred 2^nd line treatment in patients with type 2 diabetes.

Several strengths are present in this analysis. First, we conducted novel head-to-head comparison of SGLT2 inhibitors. Second, the large sample size of patients yielded a sizeable number of cardiovascular events which we were able to assess as a composite outcome, as well as individually. Third, we required all patients to be on metformin, to conform to current clinical guidelines. Fourth, we conducted our analyses using data in a real-world population, which allows for our results to be more generalizable than those of RCTs. However, limitations also exist. Although we initially matched patients based on their baseline characteristics and adjusted for a propensity score of chronic health conditions and medications, uncontrolled confounding may exist. Next, misclassification is possible with the use of administrative data; therefore, we utilized validated algorithms in defining type 2 diabetes, cardiovascular outcomes, and comorbidities. Additionally, we cannot take into account other variables, such as HbA1c or socioeconomic status, as they are not included in the MarketScan databases. Furthermore, we were unable to account for death given that death information is not included in the MarketScan databases. Finally, as the MarketScan database only includes those who are insured, our results may not be generalizable to individuals who do not have insurance.

CONCLUSION

Results from this analysis using real-world data indicates that SGLT2 inhibitor use was associated with a clinically meaningful reduction in risk for CVD, including stroke, AF, MI, and HF, compared to use of other 2^nd line diabetes therapies. These findings persisted across subgroups of interest, including sex, kidney disease, hypertension status, and year of medication initiation. In addition, head-to-head comparisons suggest there is no difference in risk based on type of SGLT2 inhibitor. Results from this study complement findings from clinical trials and may be useful in further informing clinical guidelines.

ABBREVIATIONS

AF

atrial fibrillation

CVD

cardiovascular disease

DPP-4

dipeptidyl peptidase-4

GLP-1

glucagon-like peptide

HF

heart failure

ICD

international classification of diseases

MI

myocardial infarction

RCT

randomized clinical trials

SGLT2

sodium-glucose cotransporter-2