Ex) Article Title, Author, Keywords
Ex) Article Title, Author, Keywords
DTT 2024; 3(2): 177-184
Published online September 30, 2024
https://doi.org/10.58502/DTT.24.0001
Copyright © The Pharmaceutical Society of Korea.
Sohee Park1* , Sungho Bea1* , Yunha Noh1,2 , Gregory Y. H. Lip3,4 , Seng Chan You5 , Eue-Keun Choi6,7 , Han Eol Jeong1,8 , Ju-Young Shin1,8,9
Correspondence to:Han Eol Jeong, haneoljeong@hotmail.com; Ju-Young Shin, shin.jy@skku.edu
*These authors contributed equally to this work.
Sohee Park’s current affiliation: Aston Pharmacy School, Aston University, Birmingham, United Kingdom
Sungho Bea’s current affiliation: Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Little is known on the real-world comparative effectiveness of sodium-glucose cotransporter-2 inhibitors (SGLT2i) versus angiotensin receptor-neprilysin inhibitor (ARNi) used for heart failure (HF) management. This study used South Korea’s nationwide claims data from 2015 to 2020 to construct a population-based cohort of new users of SGLT2is or ARNi. Individuals were followed from the first prescription date of SGLT2is or ARNi until outcome occurrence, treatment switch or discontinuation, death, or end of the study period. Within the 1:1 propensity score-matched cohort, we estimated hazard ratios (HR) with 95% confidence intervals (CI) for the risk of HF admission with SGLT2is compared with ARNi using proportional subdistribution hazards model of Fine and Gray. We identified 496 propensity-score matched patient-pairs of SGLT2is and ARNi; with a mean age of 72.5 years and a male representation of 57.6%. Incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04). Effect modifications were observed by history of hospitalization for HF (p-for-interaction=0.002) and by recent use of renin-angiotensin-system inhibitors (p-for-interaction= 0.005). With future studies using more recent data warranted to corroborate our study results, these preliminary findings support current guideline recommendations for HF management and further, suggest similar effectiveness between SGLT2is and ARNi in routine care settings.
KeywordsSGLT2i, ARNi, heart failure, target trial emulation, real-world evidence
Sodium-glucose cotransporter-2 inhibitors (SGLT2is) are the recent addition to guideline-directed medical therapy in heart failure (HF) (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). SGLT2is are recommended for managing comorbid conditions and addressing the varying severity levels of HF alongside another relatively novel class of HF medication, angiotensin receptor-neprilysin inhibitor (ARNi), along with conventional HF medications such as renin-angiotensin-system inhibitors and beta-blockers (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). Although the absolute usage remains lower than other conventional medications, the utilization of SGLT2is and ARNis are increasing rapidly. Notably, SGLT2is primarily target renal reabsorption, reducing fluid accumulation and promoting beneficial cardiac structural changes, while ARNis aim to improve neurohormonal balance and alleviate symptoms. This divergence in mechanisms raises questions about potential differences in their effectiveness in preventing HF admission. However, despite these guideline updates and the distinct mechanisms of action, the direct comparative effectiveness of SGLT2is versus ARNi in routine clinical practice remains unexplored.
Previous data comparing their effectiveness are primarily based on indirect comparisons of randomized controlled trials through network meta-analysis, which have reported inconsistent results. Two studies showed non-differential risk of HF admission (hazard ratio [HR] 0.87, 95% CI 0.75-1.02) (Aimo et al. 2021) or composite of cardiovascular death or HF admission (HR 0.93, 95% CI 0.82-1.06) (Yan et al. 2021) associated with SGLT2is compared to ARNi, whereas one study found modest lower risk of composite of cardiovascular death or HF admission associated with ARNi compared to SGLT2is (HR 0.86, 95% CI 0.75-0.98) (Teo et al. 2022). Uncertainty remains on the comparative effectiveness of SGLT2is versus ARNi in preventing HF admission. Challenges also exist in generalizing and translating results from ideal trial settings to real-world settings (Dhruva and Redberg 2008). This study aimed to explore the potential effectiveness of SGLT2 inhibitors compared to ARNi in decreasing the risk of HF admission, providing insights into these new treatment options for managing HF in real-world settings. To focus our investigation while addressing potential confounding by indication, we have restricted our study population to patients with both type 2 diabetes and HF, as SGLT2is are indicated for both conditions, whereas ARNi is indicated exclusively for HF.
We emulated a hypothetical target trial using South Korea’s nationwide health insurance claims data between 1 January 2015 and 31 December 2020 (Supplementary Table 1) (Kim et al. 2017; Matthews et al. 2022). Individuals newly prescribed a SGLT2i or an ARNi from 1 January 2020 to 31 December 2020 were eligible for entry into the study cohort, given the publication of DAPA-HF (dapagliflozin and prevention of adverse outcomes in HF) trial results in late 2019; we assumed that SGLT2is were likely to have been considered for HF management after this landmark trial.
Cohort entry was defined as the date of the first prescription for either a SGLT2i (dapagliflozin, empagliflozin, ipragliflozin, ertugliflozin) or an ARNi (sacubitril/valsartan) during this period. Of these eligible patients, we excluded patients aged <18 years or prescribed both study drugs of interest at cohort entry, and those meeting any of the following criteria before cohort entry: not diagnosed with HF or type 2 diabetes; not prescribed conventional HF medications (e.g., renin-angiotensin-system inhibitors, beta-blockers, mineralocorticoid receptor antagonists); diagnosed with type 1 diabetes; prescribed insulin as monotherapy; had contraindication to SGLT2is, which is end-stage renal disease or received dialysis; or received cardiac surgery (Supplementary Fig. 1).
The outcome of interest was time to first HF admission, defined by hospital admission with primary or secondary diagnosis of HF, which has shown positive predictive value of 82.1%. Patient were followed from the date of cohort entry until the earliest of outcome occurrence, switch to a comparator drug, treatment discontinuation (no successive prescription within 30 days from the end of the day supply), death, or end of the study (31 December 2020).
To address the potential confounding between groups and obtain comparability, we conducted 1:1 greedy nearest neighborhood matching without replacement within caliper width of 0.05 in propensity scores. The propensity score was generated by including all baseline characteristics in the multivariable logistic regression model: age at cohort entry date, sex, history of hospitalization of HF, comorbidities, use of medications, baseline diabetes treatment (drug classes and the number of classes), baseline HF treatment (drug classes and the number of classes), and healthcare utilization (Table 1). Baseline comorbidities, use of medications, treatment for diabetes and HF, and healthcare utilization were assessed within a year before cohort entry. We also assessed recent diabetes and HF treatment within a month before and at the date of cohort entry and included in the propensity score model to minimize confounding by indication.
Table 1 Baseline characteristics of patients initiating a SGLT2 inhibitor or an ARN inhibitor in 2020
Characteristic | Before PS* matching | After PS matching | |||||
---|---|---|---|---|---|---|---|
SGLT2i | ARNi | SMD | SGLT2i | ARNi | SMD | ||
Total | 7,562 | 793 | 496 | 496 | |||
Age, years | |||||||
Mean (SD) | 67.0 (12.5) | 72.0 (11.1) | –0.426 | 73.1 (11.3) | 72.0 (11.3) | 0.098 | |
Sex, n (%) | |||||||
Male | 4,129 (54.6) | 483 (60.9) | –0.128 | 283 (57.1) | 288 (58.1) | –0.020 | |
Female | 3,433 (45.4) | 310 (39.1) | 0.128 | 213 (42.9) | 208 (41.9) | 0.020 | |
Comorbidities†, n (%) | |||||||
Hospitalization for heart failure | 551 (7.3) | 268 (33.8) | –0.695 | 150 (30.2) | 138 (27.8) | 0.053 | |
Hypertension | 4,954 (65.5) | 420 (53.0) | 0.257 | 273 (55.0) | 277 (55.8) | –0.016 | |
Dyslipidemia | 2,949 (39.0) | 327 (41.2) | –0.046 | 219 (44.2) | 199 (40.1) | 0.082 | |
Atrial fibrillation or flutter | 1,055 (14.0) | 156 (19.7) | –0.153 | 112 (22.6) | 102 (20.6) | 0.049 | |
Myocardial infarction | 412 (5.4) | 98 (12.4) | –0.244 | 49 (9.9) | 52 (10.5) | –0.020 | |
Coronary artery disease | 1,813 (24.0) | 224 (28.2) | –0.097 | 126 (25.4) | 134 (27.0) | –0.037 | |
Coronary revascularization | 241 (3.2) | 58 (7.3) | –0.186 | 38 (7.7) | 37 (7.5) | 0.008 | |
Aortic valve surgery | 7 (0.1) | 3 (0.4) | –0.059 | 2 (0.4) | 3 (0.6) | –0.028 | |
Pacemaker or ICD | 19 (0.3) | 9 (1.1) | –0.107 | 3 (0.6) | 2 (0.4) | 0.028 | |
Stroke | 591 (7.8) | 86 (10.8) | –0.104 | 51 (10.3) | 52 (10.5) | –0.007 | |
Transient ischemic attack | 101 (1.3) | 6 (0.8) | 0.057 | 6 (1.2) | 4 (0.8) | 0.040 | |
Peripheral arterial disease | 38 (0.5) | 8 (1.0) | –0.058 | 10 (2.0) | 7 (1.4) | 0.047 | |
Cancer | 491 (6.5) | 79 (10.0) | –0.127 | 47 (9.5) | 48 (9.7) | –0.007 | |
Chronic kidney disease | 396 (5.2) | 136 (17.2) | –0.385 | 72 (14.5) | 70 (14.1) | 0.012 | |
Chronic pulmonary disease | 1,334 (17.6) | 197 (24.8) | –0.177 | 128 (25.8) | 125 (25.2) | 0.014 | |
Chronic liver disease | 1,070 (14.1) | 103 (13.0) | 0.034 | 66 (13.3) | 68 (13.7) | –0.012 | |
Comedications†, n (%) | |||||||
Statin | 5,988 (79.2) | 655 (82.6) | –0.087 | 393 (79.2) | 401 (80.8) | –0.040 | |
Calcium-channel blocker | 4,481 (59.3) | 337 (42.5) | 0.340 | 246 (49.6) | 239 (48.2) | 0.028 | |
Nonsteroidal anti-inflammatory drug | 5,049 (66.8) | 482 (60.8) | 0.125 | 319 (64.3) | 311 (62.7) | 0.034 | |
Systemic corticosteroid | 4,169 (55.1) | 414 (52.2) | 0.059 | 272 (54.8) | 269 (54.2) | 0.012 | |
Antidepressants | 1,540 (20.4) | 154 (19.4) | 0.024 | 104 (21.0) | 99 (20.0) | 0.025 | |
Antipsychotics | 1,821 (24.1) | 214 (27.0) | –0.067 | 142 (28.6) | 143 (28.8) | –0.004 | |
Baseline diabetes treatment†, n (%) | |||||||
Metformin | 6,045 (79.9) | 556 (70.1) | 0.228 | 374 (75.4) | 368 (74.2) | 0.028 | |
Sulfonylurea | 3,541 (46.8) | 400 (50.4) | –0.072 | 252 (50.8) | 245 (49.4) | 0.028 | |
Meglitinide | 34 (0.4) | 4 (0.5) | –0.008 | 2 (0.4) | 3 (0.6) | –0.028 | |
α-glucosidase inhibitor | 147 (1.9) | 21 (2.6) | –0.047 | 9 (1.8) | 9 (1.8) | 0.000 | |
Thiazolidinedione | 952 (12.6) | 65 (8.2) | 0.144 | 51 (10.3) | 52 (10.5) | –0.007 | |
Dipeptidyl peptidase-4 inhibitor | 5,036 (66.6) | 680 (85.8) | –0.461 | 417 (84.1) | 411 (82.9) | 0.033 | |
GLP-1 receptor agonist | 95 (1.3) | 14 (1.8) | –0.042 | 8 (1.6) | 8 (1.6) | 0.000 | |
Insulin | 1,377 (18.2) | 305 (38.5) | –0.461 | 188 (37.9) | 183 (36.9) | 0.021 | |
No. of diabetes medications | |||||||
Without medications | 819 (10.8) | 33 (4.2) | 0.255 | 20 (4.0) | 26 (5.2) | –0.058 | |
1 class | 1,107 (14.6) | 85 (10.7) | 0.118 | 47 (9.5) | 50 (10.1) | –0.020 | |
2 class | 2,018 (26.7) | 237 (29.9) | –0.071 | 148 (29.8) | 144 (29.0) | 0.018 | |
3+ class | 3,618 (47.8) | 438 (55.2) | –0.148 | 281 (56.7) | 276 (55.6) | 0.020 | |
Baseline heart failure treatment†, n (%) | |||||||
RAS inhibitor without neprilysin inhibitor | 6,232 (82.4) | 748 (94.3) | –0.378 | 447 (90.1) | 454 (91.5) | –0.049 | |
Beta-blocker | 4,706 (62.2) | 682 (86.0) | –0.564 | 386 (77.8) | 412 (83.1) | –0.132 | |
Mineralocorticoid antagonist | 1,492 (19.7) | 498 (62.8) | –0.973 | 290 (58.5) | 270 (54.4) | 0.081 | |
Diuretic | 4,199 (55.5) | 700 (88.3) | –0.782 | 438 (88.3) | 424 (85.5) | 0.084 | |
Digitalis | 721 (9.5) | 169 (21.3) | –0.331 | 113 (22.8) | 100 (20.2) | 0.064 | |
No. of heart failure medications | |||||||
Without medications | 175 (2.3) | 5 (0.6) | 0.140 | 9 (1.8) | 5 (1.0) | 0.068 | |
1 class | 1,730 (22.9) | 22 (2.8) | 0.630 | 14 (2.8) | 18 (3.6) | –0.046 | |
2 class | 2,604 (34.4) | 91 (11.5) | 0.568 | 70 (14.1) | 73 (14.7) | –0.017 | |
3+ class | 3,053 (40.4) | 675 (85.1) | –1.044 | 403 (81.3) | 400 (80.6) | 0.015 | |
Recent diabetes treatment‡, n (%) | |||||||
Metformin | 6,466 (85.5) | 363 (45.8) | 0.921 | 277 (55.8) | 273 (55.0) | 0.016 | |
Sulfonylurea | 3,156 (41.7) | 285 (35.9) | 0.119 | 187 (37.7) | 183 (36.9) | 0.017 | |
Meglitinide | 10 (0.1) | 2 (0.3) | –0.027 | 1 (0.2) | 1 (0.2) | 0.000 | |
α-glucosidase inhibitor | 55 (0.7) | 7 (0.9) | –0.017 | 4 (0.8) | 5 (1.0) | –0.021 | |
Thiazolidinedione | 391 (5.2) | 23 (2.9) | 0.116 | 18 (3.6) | 20 (4.0) | –0.021 | |
Dipeptidyl peptidase-4 inhibitor | 2,312 (30.6) | 547 (69.0) | –0.832 | 332 (66.9) | 330 (66.5) | 0.009 | |
GLP-1 receptor agonist | 40 (0.5) | 10 (1.3) | –0.078 | 6 (1.2) | 7 (1.4) | –0.018 | |
Insulin | 1,111 (14.7) | 186 (23.5) | –0.224 | 132 (26.6) | 128 (25.8) | 0.018 | |
No. of diabetes medications | |||||||
Without medications | 570 (7.5) | 164 (20.7) | –0.384 | 89 (17.9) | 93 (18.8) | –0.021 | |
1 class | 2,675 (35.4) | 136 (17.2) | 0.423 | 78 (15.7) | 78 (15.7) | 0.000 | |
2 class | 2,553 (33.8) | 233 (29.4) | 0.094 | 141 (28.4) | 142 (28.6) | –0.004 | |
3+ class | 1,764 (23.3) | 260 (32.8) | –0.212 | 188 (37.9) | 183 (36.9) | 0.021 | |
Recent heart failure treatment‡, n (%) | |||||||
RAS inhibitor without neprilysin inhibitor | 6,253 (82.7) | 364 (45.9) | 0.832 | 310 (62.5) | 305 (61.5) | 0.021 | |
Beta-blocker | 4,552 (60.2) | 703 (88.7) | –0.690 | 410 (82.7) | 423 (85.3) | –0.071 | |
Mineralocorticoid antagonist | 1,490 (19.7) | 483 (60.9) | –0.926 | 298 (60.1) | 288 (58.1) | 0.041 | |
Diuretic | 3,656 (48.3) | 686 (86.5) | –0.891 | 428 (86.3) | 420 (84.7) | 0.046 | |
Digitalis | 649 (8.6) | 125 (15.8) | –0.221 | 99 (20.0) | 81 (16.3) | 0.094 | |
No. of heart failure medications | |||||||
1 class | 2,266 (30.0) | 49 (6.2) | 0.650 | 35 (7.1) | 33 (6.7) | 0.016 | |
2 class | 2,675 (35.4) | 206 (26.0) | 0.205 | 104 (21.0) | 117 (23.6) | –0.063 | |
3+ class | 2,621 (34.7) | 538 (67.8) | –0.704 | 357 (72.0) | 346 (69.8) | 0.049 | |
Healthcare use†, n (%) | |||||||
Hospitalizations | |||||||
0 | 4,611 (61.0) | 263 (33.2) | 0.580 | 155 (31.3) | 174 (35.1) | –0.081 | |
1-2 | 2,397 (31.7) | 401 (50.6) | –0.391 | 247 (49.8) | 238 (48.0) | 0.036 | |
3+ | 554 (7.3) | 129 (16.3) | –0.280 | 94 (19.0) | 84 (16.9) | 0.053 | |
Outpatient visits | |||||||
0-2 | 73 (1.0) | 4 (0.5) | 0.054 | 3 (0.6) | 4 (0.8) | –0.024 | |
3-5 | 202 (2.7) | 16 (2.0) | 0.043 | 10 (2.0) | 8 (1.6) | 0.030 | |
6+ | 7,287 (96.4) | 773 (97.5) | –0.065 | 483 (97.4) | 484 (97.6) | –0.013 |
*Propensity score was estimated using a multivariable logistic regression model, which included all potential confounders shown in the Table 1 as independent variables. †Assessed during the 365-day period before cohort entry. ‡Assessed during the 30-day period before and at cohort entry. ARNi, angiotensin receptor/neprilysin inhibitor; ICD, intra-cardiac defibrillation; PS, propensity score; RAS, renin–angiotensin system; SD, standard deviation; SMD, standardized mean difference; SGLT2i, sodium-glucose cotransporter 2 inhibitor.
Within the propensity score matched cohort, we estimated the incidence rate of HF admission per 100 person-years, and HRs with 95% CIs using proportional subdistribution hazards model of Fine and Gray that treated death as a competing event (Fine and Gray 1999). Schoenfeld residual test was used to assess the proportional hazards assumptions before the survival analysis. As a sensitivity analysis, we adopted an intention-to-treat approach that did not censor follow-up at treatment interruption (switch or discontinuation). We performed subgroup analysis to investigate potential effect modification by sex, history of hospitalization for HF, chronic kidney disease, recent HF treatments. Interactions were tested using Wald test for heterogeneity. Propensity scores were re-estimated within each subgroup and the matching was reapplied for each comparison.
All analyses were conducted using SAS Enterprise Guide version 7.1 (SAS Institute Inc., USA). This study was approved by the institutional review board of Sungkyunkwan University, with a waiver of informed consent (SKKU 2022-04-015).
We identified 8,355 eligible new users of SGLT2is (7,562, 90.5%) and ARNi (793, 9.5%) in our study cohort (Supplementary Fig. 1). During the follow-up period, there were 17 deaths in the SGLT2is group and 3 in the ARNi group. After propensity score matching, 496 matched pairs were included in the main analysis, with a mean age of 72.5 years and a male representation of 57.6%. Among these, there were 6 deaths in the SGLT2is group and 1 in the ARNi group during the follow-up period. All baseline covariates, except for baseline use of beta-blockers, achieved balance between groups after matching, with absolute standardized mean differences <0.1 (Table 1).
In the propensity score matched cohort, incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04) In the intention-to-treat analysis, HR was 0.71 (95% CI 0.49-1.02); the point estimate was nearly identical with main analysis, indicating that the effect of treatment initiation was similar to the effect of adhering to the treatment in this study.
We observed effect modification by history of hospitalization for HF (p-for-interaction = 0.002) and recent use of renin-angiotensin-system inhibitors (p-for-interaction = 0.005). Among patients with a history of hospitalization for HF, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 1.37 (95% CI 0.67-2.77), while among those without such history, it was 0.58 (95% CI 0.39-0.91). Regarding recent use of renin-angiotensin-system inhibitors, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 0.55 (95% CI 0.33-0.92) among users and 0.77 (95% CI 0.37-1.63) among non-users (Fig. 1 and 2).
This is the nationwide population-based active comparator new-user cohort study that applied a target trial emulation framework to investigate the association between the risk of HF admission and the treatment with SGLT2is compared to ARNi. We have shown that there is no statistically significant difference in the risk of HF admission between SGLT2is- and ARNi-treated patients who have both type 2 diabetes and HF. Our findings from subgroup analysis suggests that the use of SGLT2is is associated with a reduced risk of HF admissions compared with ARNi, particularly in patients with a history of hospitalization for HF. Operating through mechanisms that involve the reduction of fluid accumulation and the promotion of favorable structural changes in the heart, SGLT2is may exhibit enhanced efficacy in advanced HF. Further research and clinical consideration are warranted to validate our findings.
This study has several limitations that should be considered when interpreting the findings. First, we acknowledge the substantial reduction in sample size resulting from our propensity score matching. This decision was driven by several considerations, including the relatively small comparator group of ARNi users within the larger cohort and significant differences in patient characteristics between treatment groups before matching. We applied matching to address these issues by creating a balanced cohort, enabling a more robust assessment of treatment effects while mitigating potential confounding. Second, it’s important to note that our study is based on data available up to December 31, 2020. This timeframe limits our ability to extrapolate findings beyond the guideline updates made around late 2021 or early 2022. Additionally, our findings are specific to patients with HF with type 2 diabetes and may not fully reflect the landscape of SGLT2i and ARNi use. Third, our study could not differentiate between specific types of HF as information on ejection fraction was unavailable for assessment in the data. Fourth, the potential for differences in censoring patterns between the treatment groups should be acknowledged. Last, despite our efforts in propensity score matching, the presence of more serious health conditions among ARNi group before matching indicates that imbalances may not have been fully addressed, raising the possibility of residual unmeasured confounding.
These preliminary real-world findings align with current guidelines for HF management (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022), and previous network meta-analyses of trials (Aimo et al. 2021; Yan et al. 2021). However, a meta-analysis of 25 RCTs, including 47,275 individuals, reported a comparable risk of hospitalization for HF between SGLT2is and ARNi. Although these differences in point estimates are not statistically significant, they could be due to unmeasured confounding factors inherent in observational studies, which lack randomization (Teo et al. 2022). Future studies should utilize more recent data and improve control over residual confounders to validate our findings.
These results indicate that SGLT2is may provide comparable benefits to ARNi in the context of routine clinical care. This can potentially inform healthcare in clinical decision-making regarding HF management.
Supplementary materials can be found via https://doi.org/10.58502/DTT.24.0001.
dtt-3-2-177-supple.pdfJYS received grants from the Ministry of Food and Drug Safety, the Ministry of Health and Welfare, the National Research Foundation of Korea, and pharmaceutical companies, including Pfizer, Celltrion, and SK Bioscience outside the submitted work. HEJ is employed by the Lunit Inc.
This study used data from the Health Insurance Review & Assessment Service of South Korea (M20210607316). This study was supported by a grant (No. 22213MFDS486) from the Ministry of Food and Drug Safety, South Korea, in 2022-2023. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the manuscript. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
DTT 2024; 3(2): 177-184
Published online September 30, 2024 https://doi.org/10.58502/DTT.24.0001
Copyright © The Pharmaceutical Society of Korea.
Sohee Park1* , Sungho Bea1* , Yunha Noh1,2 , Gregory Y. H. Lip3,4 , Seng Chan You5 , Eue-Keun Choi6,7 , Han Eol Jeong1,8 , Ju-Young Shin1,8,9
1School of Pharmacy, Sungkyunkwan University, Suwon, Korea
2Centre for Clinical Epidemiology, Lady Davis Institute, Jewish General Hospital, Montreal, Canada
3Liverpool Centre for Cardiovascular Science, University of Liverpool and Liverpool Heart & Chest Hospital, Liverpool, United Kingdom
4Department of Clinical Medicine, Aalborg University, Aalborg, Denmark
5Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Korea
6Department of Internal Medicine, Seoul National University Hospital, Seoul, Korea
7Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Korea
8Department of Biohealth Regulatory Science, Sungkyunkwan University, Suwon, Korea
9Department of Clinical Research Design & Evaluation, Samsung Advanced Institute for Health Sciences & Technology, Sungkyunkwan University, Seoul, Korea
Correspondence to:Han Eol Jeong, haneoljeong@hotmail.com; Ju-Young Shin, shin.jy@skku.edu
*These authors contributed equally to this work.
Sohee Park’s current affiliation: Aston Pharmacy School, Aston University, Birmingham, United Kingdom
Sungho Bea’s current affiliation: Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Little is known on the real-world comparative effectiveness of sodium-glucose cotransporter-2 inhibitors (SGLT2i) versus angiotensin receptor-neprilysin inhibitor (ARNi) used for heart failure (HF) management. This study used South Korea’s nationwide claims data from 2015 to 2020 to construct a population-based cohort of new users of SGLT2is or ARNi. Individuals were followed from the first prescription date of SGLT2is or ARNi until outcome occurrence, treatment switch or discontinuation, death, or end of the study period. Within the 1:1 propensity score-matched cohort, we estimated hazard ratios (HR) with 95% confidence intervals (CI) for the risk of HF admission with SGLT2is compared with ARNi using proportional subdistribution hazards model of Fine and Gray. We identified 496 propensity-score matched patient-pairs of SGLT2is and ARNi; with a mean age of 72.5 years and a male representation of 57.6%. Incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04). Effect modifications were observed by history of hospitalization for HF (p-for-interaction=0.002) and by recent use of renin-angiotensin-system inhibitors (p-for-interaction= 0.005). With future studies using more recent data warranted to corroborate our study results, these preliminary findings support current guideline recommendations for HF management and further, suggest similar effectiveness between SGLT2is and ARNi in routine care settings.
Keywords: SGLT2i, ARNi, heart failure, target trial emulation, real-world evidence
Sodium-glucose cotransporter-2 inhibitors (SGLT2is) are the recent addition to guideline-directed medical therapy in heart failure (HF) (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). SGLT2is are recommended for managing comorbid conditions and addressing the varying severity levels of HF alongside another relatively novel class of HF medication, angiotensin receptor-neprilysin inhibitor (ARNi), along with conventional HF medications such as renin-angiotensin-system inhibitors and beta-blockers (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). Although the absolute usage remains lower than other conventional medications, the utilization of SGLT2is and ARNis are increasing rapidly. Notably, SGLT2is primarily target renal reabsorption, reducing fluid accumulation and promoting beneficial cardiac structural changes, while ARNis aim to improve neurohormonal balance and alleviate symptoms. This divergence in mechanisms raises questions about potential differences in their effectiveness in preventing HF admission. However, despite these guideline updates and the distinct mechanisms of action, the direct comparative effectiveness of SGLT2is versus ARNi in routine clinical practice remains unexplored.
Previous data comparing their effectiveness are primarily based on indirect comparisons of randomized controlled trials through network meta-analysis, which have reported inconsistent results. Two studies showed non-differential risk of HF admission (hazard ratio [HR] 0.87, 95% CI 0.75-1.02) (Aimo et al. 2021) or composite of cardiovascular death or HF admission (HR 0.93, 95% CI 0.82-1.06) (Yan et al. 2021) associated with SGLT2is compared to ARNi, whereas one study found modest lower risk of composite of cardiovascular death or HF admission associated with ARNi compared to SGLT2is (HR 0.86, 95% CI 0.75-0.98) (Teo et al. 2022). Uncertainty remains on the comparative effectiveness of SGLT2is versus ARNi in preventing HF admission. Challenges also exist in generalizing and translating results from ideal trial settings to real-world settings (Dhruva and Redberg 2008). This study aimed to explore the potential effectiveness of SGLT2 inhibitors compared to ARNi in decreasing the risk of HF admission, providing insights into these new treatment options for managing HF in real-world settings. To focus our investigation while addressing potential confounding by indication, we have restricted our study population to patients with both type 2 diabetes and HF, as SGLT2is are indicated for both conditions, whereas ARNi is indicated exclusively for HF.
We emulated a hypothetical target trial using South Korea’s nationwide health insurance claims data between 1 January 2015 and 31 December 2020 (Supplementary Table 1) (Kim et al. 2017; Matthews et al. 2022). Individuals newly prescribed a SGLT2i or an ARNi from 1 January 2020 to 31 December 2020 were eligible for entry into the study cohort, given the publication of DAPA-HF (dapagliflozin and prevention of adverse outcomes in HF) trial results in late 2019; we assumed that SGLT2is were likely to have been considered for HF management after this landmark trial.
Cohort entry was defined as the date of the first prescription for either a SGLT2i (dapagliflozin, empagliflozin, ipragliflozin, ertugliflozin) or an ARNi (sacubitril/valsartan) during this period. Of these eligible patients, we excluded patients aged <18 years or prescribed both study drugs of interest at cohort entry, and those meeting any of the following criteria before cohort entry: not diagnosed with HF or type 2 diabetes; not prescribed conventional HF medications (e.g., renin-angiotensin-system inhibitors, beta-blockers, mineralocorticoid receptor antagonists); diagnosed with type 1 diabetes; prescribed insulin as monotherapy; had contraindication to SGLT2is, which is end-stage renal disease or received dialysis; or received cardiac surgery (Supplementary Fig. 1).
The outcome of interest was time to first HF admission, defined by hospital admission with primary or secondary diagnosis of HF, which has shown positive predictive value of 82.1%. Patient were followed from the date of cohort entry until the earliest of outcome occurrence, switch to a comparator drug, treatment discontinuation (no successive prescription within 30 days from the end of the day supply), death, or end of the study (31 December 2020).
To address the potential confounding between groups and obtain comparability, we conducted 1:1 greedy nearest neighborhood matching without replacement within caliper width of 0.05 in propensity scores. The propensity score was generated by including all baseline characteristics in the multivariable logistic regression model: age at cohort entry date, sex, history of hospitalization of HF, comorbidities, use of medications, baseline diabetes treatment (drug classes and the number of classes), baseline HF treatment (drug classes and the number of classes), and healthcare utilization (Table 1). Baseline comorbidities, use of medications, treatment for diabetes and HF, and healthcare utilization were assessed within a year before cohort entry. We also assessed recent diabetes and HF treatment within a month before and at the date of cohort entry and included in the propensity score model to minimize confounding by indication.
Table 1 . Baseline characteristics of patients initiating a SGLT2 inhibitor or an ARN inhibitor in 2020.
Characteristic | Before PS* matching | After PS matching | |||||
---|---|---|---|---|---|---|---|
SGLT2i | ARNi | SMD | SGLT2i | ARNi | SMD | ||
Total | 7,562 | 793 | 496 | 496 | |||
Age, years | |||||||
Mean (SD) | 67.0 (12.5) | 72.0 (11.1) | –0.426 | 73.1 (11.3) | 72.0 (11.3) | 0.098 | |
Sex, n (%) | |||||||
Male | 4,129 (54.6) | 483 (60.9) | –0.128 | 283 (57.1) | 288 (58.1) | –0.020 | |
Female | 3,433 (45.4) | 310 (39.1) | 0.128 | 213 (42.9) | 208 (41.9) | 0.020 | |
Comorbidities†, n (%) | |||||||
Hospitalization for heart failure | 551 (7.3) | 268 (33.8) | –0.695 | 150 (30.2) | 138 (27.8) | 0.053 | |
Hypertension | 4,954 (65.5) | 420 (53.0) | 0.257 | 273 (55.0) | 277 (55.8) | –0.016 | |
Dyslipidemia | 2,949 (39.0) | 327 (41.2) | –0.046 | 219 (44.2) | 199 (40.1) | 0.082 | |
Atrial fibrillation or flutter | 1,055 (14.0) | 156 (19.7) | –0.153 | 112 (22.6) | 102 (20.6) | 0.049 | |
Myocardial infarction | 412 (5.4) | 98 (12.4) | –0.244 | 49 (9.9) | 52 (10.5) | –0.020 | |
Coronary artery disease | 1,813 (24.0) | 224 (28.2) | –0.097 | 126 (25.4) | 134 (27.0) | –0.037 | |
Coronary revascularization | 241 (3.2) | 58 (7.3) | –0.186 | 38 (7.7) | 37 (7.5) | 0.008 | |
Aortic valve surgery | 7 (0.1) | 3 (0.4) | –0.059 | 2 (0.4) | 3 (0.6) | –0.028 | |
Pacemaker or ICD | 19 (0.3) | 9 (1.1) | –0.107 | 3 (0.6) | 2 (0.4) | 0.028 | |
Stroke | 591 (7.8) | 86 (10.8) | –0.104 | 51 (10.3) | 52 (10.5) | –0.007 | |
Transient ischemic attack | 101 (1.3) | 6 (0.8) | 0.057 | 6 (1.2) | 4 (0.8) | 0.040 | |
Peripheral arterial disease | 38 (0.5) | 8 (1.0) | –0.058 | 10 (2.0) | 7 (1.4) | 0.047 | |
Cancer | 491 (6.5) | 79 (10.0) | –0.127 | 47 (9.5) | 48 (9.7) | –0.007 | |
Chronic kidney disease | 396 (5.2) | 136 (17.2) | –0.385 | 72 (14.5) | 70 (14.1) | 0.012 | |
Chronic pulmonary disease | 1,334 (17.6) | 197 (24.8) | –0.177 | 128 (25.8) | 125 (25.2) | 0.014 | |
Chronic liver disease | 1,070 (14.1) | 103 (13.0) | 0.034 | 66 (13.3) | 68 (13.7) | –0.012 | |
Comedications†, n (%) | |||||||
Statin | 5,988 (79.2) | 655 (82.6) | –0.087 | 393 (79.2) | 401 (80.8) | –0.040 | |
Calcium-channel blocker | 4,481 (59.3) | 337 (42.5) | 0.340 | 246 (49.6) | 239 (48.2) | 0.028 | |
Nonsteroidal anti-inflammatory drug | 5,049 (66.8) | 482 (60.8) | 0.125 | 319 (64.3) | 311 (62.7) | 0.034 | |
Systemic corticosteroid | 4,169 (55.1) | 414 (52.2) | 0.059 | 272 (54.8) | 269 (54.2) | 0.012 | |
Antidepressants | 1,540 (20.4) | 154 (19.4) | 0.024 | 104 (21.0) | 99 (20.0) | 0.025 | |
Antipsychotics | 1,821 (24.1) | 214 (27.0) | –0.067 | 142 (28.6) | 143 (28.8) | –0.004 | |
Baseline diabetes treatment†, n (%) | |||||||
Metformin | 6,045 (79.9) | 556 (70.1) | 0.228 | 374 (75.4) | 368 (74.2) | 0.028 | |
Sulfonylurea | 3,541 (46.8) | 400 (50.4) | –0.072 | 252 (50.8) | 245 (49.4) | 0.028 | |
Meglitinide | 34 (0.4) | 4 (0.5) | –0.008 | 2 (0.4) | 3 (0.6) | –0.028 | |
α-glucosidase inhibitor | 147 (1.9) | 21 (2.6) | –0.047 | 9 (1.8) | 9 (1.8) | 0.000 | |
Thiazolidinedione | 952 (12.6) | 65 (8.2) | 0.144 | 51 (10.3) | 52 (10.5) | –0.007 | |
Dipeptidyl peptidase-4 inhibitor | 5,036 (66.6) | 680 (85.8) | –0.461 | 417 (84.1) | 411 (82.9) | 0.033 | |
GLP-1 receptor agonist | 95 (1.3) | 14 (1.8) | –0.042 | 8 (1.6) | 8 (1.6) | 0.000 | |
Insulin | 1,377 (18.2) | 305 (38.5) | –0.461 | 188 (37.9) | 183 (36.9) | 0.021 | |
No. of diabetes medications | |||||||
Without medications | 819 (10.8) | 33 (4.2) | 0.255 | 20 (4.0) | 26 (5.2) | –0.058 | |
1 class | 1,107 (14.6) | 85 (10.7) | 0.118 | 47 (9.5) | 50 (10.1) | –0.020 | |
2 class | 2,018 (26.7) | 237 (29.9) | –0.071 | 148 (29.8) | 144 (29.0) | 0.018 | |
3+ class | 3,618 (47.8) | 438 (55.2) | –0.148 | 281 (56.7) | 276 (55.6) | 0.020 | |
Baseline heart failure treatment†, n (%) | |||||||
RAS inhibitor without neprilysin inhibitor | 6,232 (82.4) | 748 (94.3) | –0.378 | 447 (90.1) | 454 (91.5) | –0.049 | |
Beta-blocker | 4,706 (62.2) | 682 (86.0) | –0.564 | 386 (77.8) | 412 (83.1) | –0.132 | |
Mineralocorticoid antagonist | 1,492 (19.7) | 498 (62.8) | –0.973 | 290 (58.5) | 270 (54.4) | 0.081 | |
Diuretic | 4,199 (55.5) | 700 (88.3) | –0.782 | 438 (88.3) | 424 (85.5) | 0.084 | |
Digitalis | 721 (9.5) | 169 (21.3) | –0.331 | 113 (22.8) | 100 (20.2) | 0.064 | |
No. of heart failure medications | |||||||
Without medications | 175 (2.3) | 5 (0.6) | 0.140 | 9 (1.8) | 5 (1.0) | 0.068 | |
1 class | 1,730 (22.9) | 22 (2.8) | 0.630 | 14 (2.8) | 18 (3.6) | –0.046 | |
2 class | 2,604 (34.4) | 91 (11.5) | 0.568 | 70 (14.1) | 73 (14.7) | –0.017 | |
3+ class | 3,053 (40.4) | 675 (85.1) | –1.044 | 403 (81.3) | 400 (80.6) | 0.015 | |
Recent diabetes treatment‡, n (%) | |||||||
Metformin | 6,466 (85.5) | 363 (45.8) | 0.921 | 277 (55.8) | 273 (55.0) | 0.016 | |
Sulfonylurea | 3,156 (41.7) | 285 (35.9) | 0.119 | 187 (37.7) | 183 (36.9) | 0.017 | |
Meglitinide | 10 (0.1) | 2 (0.3) | –0.027 | 1 (0.2) | 1 (0.2) | 0.000 | |
α-glucosidase inhibitor | 55 (0.7) | 7 (0.9) | –0.017 | 4 (0.8) | 5 (1.0) | –0.021 | |
Thiazolidinedione | 391 (5.2) | 23 (2.9) | 0.116 | 18 (3.6) | 20 (4.0) | –0.021 | |
Dipeptidyl peptidase-4 inhibitor | 2,312 (30.6) | 547 (69.0) | –0.832 | 332 (66.9) | 330 (66.5) | 0.009 | |
GLP-1 receptor agonist | 40 (0.5) | 10 (1.3) | –0.078 | 6 (1.2) | 7 (1.4) | –0.018 | |
Insulin | 1,111 (14.7) | 186 (23.5) | –0.224 | 132 (26.6) | 128 (25.8) | 0.018 | |
No. of diabetes medications | |||||||
Without medications | 570 (7.5) | 164 (20.7) | –0.384 | 89 (17.9) | 93 (18.8) | –0.021 | |
1 class | 2,675 (35.4) | 136 (17.2) | 0.423 | 78 (15.7) | 78 (15.7) | 0.000 | |
2 class | 2,553 (33.8) | 233 (29.4) | 0.094 | 141 (28.4) | 142 (28.6) | –0.004 | |
3+ class | 1,764 (23.3) | 260 (32.8) | –0.212 | 188 (37.9) | 183 (36.9) | 0.021 | |
Recent heart failure treatment‡, n (%) | |||||||
RAS inhibitor without neprilysin inhibitor | 6,253 (82.7) | 364 (45.9) | 0.832 | 310 (62.5) | 305 (61.5) | 0.021 | |
Beta-blocker | 4,552 (60.2) | 703 (88.7) | –0.690 | 410 (82.7) | 423 (85.3) | –0.071 | |
Mineralocorticoid antagonist | 1,490 (19.7) | 483 (60.9) | –0.926 | 298 (60.1) | 288 (58.1) | 0.041 | |
Diuretic | 3,656 (48.3) | 686 (86.5) | –0.891 | 428 (86.3) | 420 (84.7) | 0.046 | |
Digitalis | 649 (8.6) | 125 (15.8) | –0.221 | 99 (20.0) | 81 (16.3) | 0.094 | |
No. of heart failure medications | |||||||
1 class | 2,266 (30.0) | 49 (6.2) | 0.650 | 35 (7.1) | 33 (6.7) | 0.016 | |
2 class | 2,675 (35.4) | 206 (26.0) | 0.205 | 104 (21.0) | 117 (23.6) | –0.063 | |
3+ class | 2,621 (34.7) | 538 (67.8) | –0.704 | 357 (72.0) | 346 (69.8) | 0.049 | |
Healthcare use†, n (%) | |||||||
Hospitalizations | |||||||
0 | 4,611 (61.0) | 263 (33.2) | 0.580 | 155 (31.3) | 174 (35.1) | –0.081 | |
1-2 | 2,397 (31.7) | 401 (50.6) | –0.391 | 247 (49.8) | 238 (48.0) | 0.036 | |
3+ | 554 (7.3) | 129 (16.3) | –0.280 | 94 (19.0) | 84 (16.9) | 0.053 | |
Outpatient visits | |||||||
0-2 | 73 (1.0) | 4 (0.5) | 0.054 | 3 (0.6) | 4 (0.8) | –0.024 | |
3-5 | 202 (2.7) | 16 (2.0) | 0.043 | 10 (2.0) | 8 (1.6) | 0.030 | |
6+ | 7,287 (96.4) | 773 (97.5) | –0.065 | 483 (97.4) | 484 (97.6) | –0.013 |
*Propensity score was estimated using a multivariable logistic regression model, which included all potential confounders shown in the Table 1 as independent variables. †Assessed during the 365-day period before cohort entry. ‡Assessed during the 30-day period before and at cohort entry. ARNi, angiotensin receptor/neprilysin inhibitor; ICD, intra-cardiac defibrillation; PS, propensity score; RAS, renin–angiotensin system; SD, standard deviation; SMD, standardized mean difference; SGLT2i, sodium-glucose cotransporter 2 inhibitor..
Within the propensity score matched cohort, we estimated the incidence rate of HF admission per 100 person-years, and HRs with 95% CIs using proportional subdistribution hazards model of Fine and Gray that treated death as a competing event (Fine and Gray 1999). Schoenfeld residual test was used to assess the proportional hazards assumptions before the survival analysis. As a sensitivity analysis, we adopted an intention-to-treat approach that did not censor follow-up at treatment interruption (switch or discontinuation). We performed subgroup analysis to investigate potential effect modification by sex, history of hospitalization for HF, chronic kidney disease, recent HF treatments. Interactions were tested using Wald test for heterogeneity. Propensity scores were re-estimated within each subgroup and the matching was reapplied for each comparison.
All analyses were conducted using SAS Enterprise Guide version 7.1 (SAS Institute Inc., USA). This study was approved by the institutional review board of Sungkyunkwan University, with a waiver of informed consent (SKKU 2022-04-015).
We identified 8,355 eligible new users of SGLT2is (7,562, 90.5%) and ARNi (793, 9.5%) in our study cohort (Supplementary Fig. 1). During the follow-up period, there were 17 deaths in the SGLT2is group and 3 in the ARNi group. After propensity score matching, 496 matched pairs were included in the main analysis, with a mean age of 72.5 years and a male representation of 57.6%. Among these, there were 6 deaths in the SGLT2is group and 1 in the ARNi group during the follow-up period. All baseline covariates, except for baseline use of beta-blockers, achieved balance between groups after matching, with absolute standardized mean differences <0.1 (Table 1).
In the propensity score matched cohort, incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04) In the intention-to-treat analysis, HR was 0.71 (95% CI 0.49-1.02); the point estimate was nearly identical with main analysis, indicating that the effect of treatment initiation was similar to the effect of adhering to the treatment in this study.
We observed effect modification by history of hospitalization for HF (p-for-interaction = 0.002) and recent use of renin-angiotensin-system inhibitors (p-for-interaction = 0.005). Among patients with a history of hospitalization for HF, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 1.37 (95% CI 0.67-2.77), while among those without such history, it was 0.58 (95% CI 0.39-0.91). Regarding recent use of renin-angiotensin-system inhibitors, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 0.55 (95% CI 0.33-0.92) among users and 0.77 (95% CI 0.37-1.63) among non-users (Fig. 1 and 2).
This is the nationwide population-based active comparator new-user cohort study that applied a target trial emulation framework to investigate the association between the risk of HF admission and the treatment with SGLT2is compared to ARNi. We have shown that there is no statistically significant difference in the risk of HF admission between SGLT2is- and ARNi-treated patients who have both type 2 diabetes and HF. Our findings from subgroup analysis suggests that the use of SGLT2is is associated with a reduced risk of HF admissions compared with ARNi, particularly in patients with a history of hospitalization for HF. Operating through mechanisms that involve the reduction of fluid accumulation and the promotion of favorable structural changes in the heart, SGLT2is may exhibit enhanced efficacy in advanced HF. Further research and clinical consideration are warranted to validate our findings.
This study has several limitations that should be considered when interpreting the findings. First, we acknowledge the substantial reduction in sample size resulting from our propensity score matching. This decision was driven by several considerations, including the relatively small comparator group of ARNi users within the larger cohort and significant differences in patient characteristics between treatment groups before matching. We applied matching to address these issues by creating a balanced cohort, enabling a more robust assessment of treatment effects while mitigating potential confounding. Second, it’s important to note that our study is based on data available up to December 31, 2020. This timeframe limits our ability to extrapolate findings beyond the guideline updates made around late 2021 or early 2022. Additionally, our findings are specific to patients with HF with type 2 diabetes and may not fully reflect the landscape of SGLT2i and ARNi use. Third, our study could not differentiate between specific types of HF as information on ejection fraction was unavailable for assessment in the data. Fourth, the potential for differences in censoring patterns between the treatment groups should be acknowledged. Last, despite our efforts in propensity score matching, the presence of more serious health conditions among ARNi group before matching indicates that imbalances may not have been fully addressed, raising the possibility of residual unmeasured confounding.
These preliminary real-world findings align with current guidelines for HF management (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022), and previous network meta-analyses of trials (Aimo et al. 2021; Yan et al. 2021). However, a meta-analysis of 25 RCTs, including 47,275 individuals, reported a comparable risk of hospitalization for HF between SGLT2is and ARNi. Although these differences in point estimates are not statistically significant, they could be due to unmeasured confounding factors inherent in observational studies, which lack randomization (Teo et al. 2022). Future studies should utilize more recent data and improve control over residual confounders to validate our findings.
These results indicate that SGLT2is may provide comparable benefits to ARNi in the context of routine clinical care. This can potentially inform healthcare in clinical decision-making regarding HF management.
Supplementary materials can be found via https://doi.org/10.58502/DTT.24.0001.
dtt-3-2-177-supple.pdfJYS received grants from the Ministry of Food and Drug Safety, the Ministry of Health and Welfare, the National Research Foundation of Korea, and pharmaceutical companies, including Pfizer, Celltrion, and SK Bioscience outside the submitted work. HEJ is employed by the Lunit Inc.
This study used data from the Health Insurance Review & Assessment Service of South Korea (M20210607316). This study was supported by a grant (No. 22213MFDS486) from the Ministry of Food and Drug Safety, South Korea, in 2022-2023. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the manuscript. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Table 1 Baseline characteristics of patients initiating a SGLT2 inhibitor or an ARN inhibitor in 2020
Characteristic | Before PS* matching | After PS matching | |||||
---|---|---|---|---|---|---|---|
SGLT2i | ARNi | SMD | SGLT2i | ARNi | SMD | ||
Total | 7,562 | 793 | 496 | 496 | |||
Age, years | |||||||
Mean (SD) | 67.0 (12.5) | 72.0 (11.1) | –0.426 | 73.1 (11.3) | 72.0 (11.3) | 0.098 | |
Sex, n (%) | |||||||
Male | 4,129 (54.6) | 483 (60.9) | –0.128 | 283 (57.1) | 288 (58.1) | –0.020 | |
Female | 3,433 (45.4) | 310 (39.1) | 0.128 | 213 (42.9) | 208 (41.9) | 0.020 | |
Comorbidities†, n (%) | |||||||
Hospitalization for heart failure | 551 (7.3) | 268 (33.8) | –0.695 | 150 (30.2) | 138 (27.8) | 0.053 | |
Hypertension | 4,954 (65.5) | 420 (53.0) | 0.257 | 273 (55.0) | 277 (55.8) | –0.016 | |
Dyslipidemia | 2,949 (39.0) | 327 (41.2) | –0.046 | 219 (44.2) | 199 (40.1) | 0.082 | |
Atrial fibrillation or flutter | 1,055 (14.0) | 156 (19.7) | –0.153 | 112 (22.6) | 102 (20.6) | 0.049 | |
Myocardial infarction | 412 (5.4) | 98 (12.4) | –0.244 | 49 (9.9) | 52 (10.5) | –0.020 | |
Coronary artery disease | 1,813 (24.0) | 224 (28.2) | –0.097 | 126 (25.4) | 134 (27.0) | –0.037 | |
Coronary revascularization | 241 (3.2) | 58 (7.3) | –0.186 | 38 (7.7) | 37 (7.5) | 0.008 | |
Aortic valve surgery | 7 (0.1) | 3 (0.4) | –0.059 | 2 (0.4) | 3 (0.6) | –0.028 | |
Pacemaker or ICD | 19 (0.3) | 9 (1.1) | –0.107 | 3 (0.6) | 2 (0.4) | 0.028 | |
Stroke | 591 (7.8) | 86 (10.8) | –0.104 | 51 (10.3) | 52 (10.5) | –0.007 | |
Transient ischemic attack | 101 (1.3) | 6 (0.8) | 0.057 | 6 (1.2) | 4 (0.8) | 0.040 | |
Peripheral arterial disease | 38 (0.5) | 8 (1.0) | –0.058 | 10 (2.0) | 7 (1.4) | 0.047 | |
Cancer | 491 (6.5) | 79 (10.0) | –0.127 | 47 (9.5) | 48 (9.7) | –0.007 | |
Chronic kidney disease | 396 (5.2) | 136 (17.2) | –0.385 | 72 (14.5) | 70 (14.1) | 0.012 | |
Chronic pulmonary disease | 1,334 (17.6) | 197 (24.8) | –0.177 | 128 (25.8) | 125 (25.2) | 0.014 | |
Chronic liver disease | 1,070 (14.1) | 103 (13.0) | 0.034 | 66 (13.3) | 68 (13.7) | –0.012 | |
Comedications†, n (%) | |||||||
Statin | 5,988 (79.2) | 655 (82.6) | –0.087 | 393 (79.2) | 401 (80.8) | –0.040 | |
Calcium-channel blocker | 4,481 (59.3) | 337 (42.5) | 0.340 | 246 (49.6) | 239 (48.2) | 0.028 | |
Nonsteroidal anti-inflammatory drug | 5,049 (66.8) | 482 (60.8) | 0.125 | 319 (64.3) | 311 (62.7) | 0.034 | |
Systemic corticosteroid | 4,169 (55.1) | 414 (52.2) | 0.059 | 272 (54.8) | 269 (54.2) | 0.012 | |
Antidepressants | 1,540 (20.4) | 154 (19.4) | 0.024 | 104 (21.0) | 99 (20.0) | 0.025 | |
Antipsychotics | 1,821 (24.1) | 214 (27.0) | –0.067 | 142 (28.6) | 143 (28.8) | –0.004 | |
Baseline diabetes treatment†, n (%) | |||||||
Metformin | 6,045 (79.9) | 556 (70.1) | 0.228 | 374 (75.4) | 368 (74.2) | 0.028 | |
Sulfonylurea | 3,541 (46.8) | 400 (50.4) | –0.072 | 252 (50.8) | 245 (49.4) | 0.028 | |
Meglitinide | 34 (0.4) | 4 (0.5) | –0.008 | 2 (0.4) | 3 (0.6) | –0.028 | |
α-glucosidase inhibitor | 147 (1.9) | 21 (2.6) | –0.047 | 9 (1.8) | 9 (1.8) | 0.000 | |
Thiazolidinedione | 952 (12.6) | 65 (8.2) | 0.144 | 51 (10.3) | 52 (10.5) | –0.007 | |
Dipeptidyl peptidase-4 inhibitor | 5,036 (66.6) | 680 (85.8) | –0.461 | 417 (84.1) | 411 (82.9) | 0.033 | |
GLP-1 receptor agonist | 95 (1.3) | 14 (1.8) | –0.042 | 8 (1.6) | 8 (1.6) | 0.000 | |
Insulin | 1,377 (18.2) | 305 (38.5) | –0.461 | 188 (37.9) | 183 (36.9) | 0.021 | |
No. of diabetes medications | |||||||
Without medications | 819 (10.8) | 33 (4.2) | 0.255 | 20 (4.0) | 26 (5.2) | –0.058 | |
1 class | 1,107 (14.6) | 85 (10.7) | 0.118 | 47 (9.5) | 50 (10.1) | –0.020 | |
2 class | 2,018 (26.7) | 237 (29.9) | –0.071 | 148 (29.8) | 144 (29.0) | 0.018 | |
3+ class | 3,618 (47.8) | 438 (55.2) | –0.148 | 281 (56.7) | 276 (55.6) | 0.020 | |
Baseline heart failure treatment†, n (%) | |||||||
RAS inhibitor without neprilysin inhibitor | 6,232 (82.4) | 748 (94.3) | –0.378 | 447 (90.1) | 454 (91.5) | –0.049 | |
Beta-blocker | 4,706 (62.2) | 682 (86.0) | –0.564 | 386 (77.8) | 412 (83.1) | –0.132 | |
Mineralocorticoid antagonist | 1,492 (19.7) | 498 (62.8) | –0.973 | 290 (58.5) | 270 (54.4) | 0.081 | |
Diuretic | 4,199 (55.5) | 700 (88.3) | –0.782 | 438 (88.3) | 424 (85.5) | 0.084 | |
Digitalis | 721 (9.5) | 169 (21.3) | –0.331 | 113 (22.8) | 100 (20.2) | 0.064 | |
No. of heart failure medications | |||||||
Without medications | 175 (2.3) | 5 (0.6) | 0.140 | 9 (1.8) | 5 (1.0) | 0.068 | |
1 class | 1,730 (22.9) | 22 (2.8) | 0.630 | 14 (2.8) | 18 (3.6) | –0.046 | |
2 class | 2,604 (34.4) | 91 (11.5) | 0.568 | 70 (14.1) | 73 (14.7) | –0.017 | |
3+ class | 3,053 (40.4) | 675 (85.1) | –1.044 | 403 (81.3) | 400 (80.6) | 0.015 | |
Recent diabetes treatment‡, n (%) | |||||||
Metformin | 6,466 (85.5) | 363 (45.8) | 0.921 | 277 (55.8) | 273 (55.0) | 0.016 | |
Sulfonylurea | 3,156 (41.7) | 285 (35.9) | 0.119 | 187 (37.7) | 183 (36.9) | 0.017 | |
Meglitinide | 10 (0.1) | 2 (0.3) | –0.027 | 1 (0.2) | 1 (0.2) | 0.000 | |
α-glucosidase inhibitor | 55 (0.7) | 7 (0.9) | –0.017 | 4 (0.8) | 5 (1.0) | –0.021 | |
Thiazolidinedione | 391 (5.2) | 23 (2.9) | 0.116 | 18 (3.6) | 20 (4.0) | –0.021 | |
Dipeptidyl peptidase-4 inhibitor | 2,312 (30.6) | 547 (69.0) | –0.832 | 332 (66.9) | 330 (66.5) | 0.009 | |
GLP-1 receptor agonist | 40 (0.5) | 10 (1.3) | –0.078 | 6 (1.2) | 7 (1.4) | –0.018 | |
Insulin | 1,111 (14.7) | 186 (23.5) | –0.224 | 132 (26.6) | 128 (25.8) | 0.018 | |
No. of diabetes medications | |||||||
Without medications | 570 (7.5) | 164 (20.7) | –0.384 | 89 (17.9) | 93 (18.8) | –0.021 | |
1 class | 2,675 (35.4) | 136 (17.2) | 0.423 | 78 (15.7) | 78 (15.7) | 0.000 | |
2 class | 2,553 (33.8) | 233 (29.4) | 0.094 | 141 (28.4) | 142 (28.6) | –0.004 | |
3+ class | 1,764 (23.3) | 260 (32.8) | –0.212 | 188 (37.9) | 183 (36.9) | 0.021 | |
Recent heart failure treatment‡, n (%) | |||||||
RAS inhibitor without neprilysin inhibitor | 6,253 (82.7) | 364 (45.9) | 0.832 | 310 (62.5) | 305 (61.5) | 0.021 | |
Beta-blocker | 4,552 (60.2) | 703 (88.7) | –0.690 | 410 (82.7) | 423 (85.3) | –0.071 | |
Mineralocorticoid antagonist | 1,490 (19.7) | 483 (60.9) | –0.926 | 298 (60.1) | 288 (58.1) | 0.041 | |
Diuretic | 3,656 (48.3) | 686 (86.5) | –0.891 | 428 (86.3) | 420 (84.7) | 0.046 | |
Digitalis | 649 (8.6) | 125 (15.8) | –0.221 | 99 (20.0) | 81 (16.3) | 0.094 | |
No. of heart failure medications | |||||||
1 class | 2,266 (30.0) | 49 (6.2) | 0.650 | 35 (7.1) | 33 (6.7) | 0.016 | |
2 class | 2,675 (35.4) | 206 (26.0) | 0.205 | 104 (21.0) | 117 (23.6) | –0.063 | |
3+ class | 2,621 (34.7) | 538 (67.8) | –0.704 | 357 (72.0) | 346 (69.8) | 0.049 | |
Healthcare use†, n (%) | |||||||
Hospitalizations | |||||||
0 | 4,611 (61.0) | 263 (33.2) | 0.580 | 155 (31.3) | 174 (35.1) | –0.081 | |
1-2 | 2,397 (31.7) | 401 (50.6) | –0.391 | 247 (49.8) | 238 (48.0) | 0.036 | |
3+ | 554 (7.3) | 129 (16.3) | –0.280 | 94 (19.0) | 84 (16.9) | 0.053 | |
Outpatient visits | |||||||
0-2 | 73 (1.0) | 4 (0.5) | 0.054 | 3 (0.6) | 4 (0.8) | –0.024 | |
3-5 | 202 (2.7) | 16 (2.0) | 0.043 | 10 (2.0) | 8 (1.6) | 0.030 | |
6+ | 7,287 (96.4) | 773 (97.5) | –0.065 | 483 (97.4) | 484 (97.6) | –0.013 |
*Propensity score was estimated using a multivariable logistic regression model, which included all potential confounders shown in the Table 1 as independent variables. †Assessed during the 365-day period before cohort entry. ‡Assessed during the 30-day period before and at cohort entry. ARNi, angiotensin receptor/neprilysin inhibitor; ICD, intra-cardiac defibrillation; PS, propensity score; RAS, renin–angiotensin system; SD, standard deviation; SMD, standardized mean difference; SGLT2i, sodium-glucose cotransporter 2 inhibitor.