Real et al. Cardiovasc Diabetol (2021) 20:139 https://doi.org/10.1186/s12933-021-01323-5 Cardiovascular Diabetology ORIGINAL INVESTIGATION Open Access Cardiovascular and mortality benefits of sodium–glucose co‑transporter‑2 inhibitors in patients with type 2 diabetes mellitus: CVD‑Real Catalonia Jordi Real1,2†, Bogdan Vlacho1†, Emilio Ortega3,4, Joan Antoni Vallés1,5, Manel Mata‑Cases1,2,6, Esmeralda Castelblanco1,2, Eric T. Wittbrodt7, Peter Fenici8, Mikhail Kosiborod9, Dídac Mauricio1,2,10,11* and Josep Franch‑Nadal1,2,12* Abstract Background: Evidence from prospective cardiovascular (CV) outcome trials in type 2 diabetes (T2DM) patients sup‑ ports the use of sodium–glucose co-transporter-2 inhibitors (SGLT2i) to reduce the risk of CV events. In this study, we compared the risk of several CV outcomes between new users of SGLT2i and other glucose-lowering drugs (oGLDs) in Catalonia, Spain. Methods: CVD-REAL Catalonia was a retrospective cohort study using real-world data routinely collected between 2013 and 2016. The cohorts of new users of SGLT2i and oGLDs were matched by propensity score on a 1:1 ratio. We compared the incidence rates and hazard ratio (HR) for all-cause death, hospitalization for heart failure, chronic kidney disease, and modified major adverse CV event (MACE; all-cause mortality, myocardial infarction, or stroke). Results: After propensity score matching, 12,917 new users were included in each group. About 27% of users had a previous history of CV disease. In the SGLT2i group, the exposure time was 60% for dapagliflozin, 26% for empa‑ gliflozin and 14% for canagliflozin. The use of SGLT2i was associated with a lower risk of heart failure (HR: 0.59; 95% confidence interval [CI] 0.47–0.74; p < 0.001), all-cause death (HR = 0.41; 95% CI 0.31–0.54; p < 0.001), all-cause death or heart failure (HR = 0.55; 95% CI 0.47–0.63; p < 0.001), modified MACE (HR = 0.62; 95% CI 0.52–0.74; p < 0.001), and chronic kidney disease (HR = 0.66; 95% CI 0.54–0.80; p < 0.001). Conclusions: In this large, retrospective observational study of patients with T2DM from a Catalonia, initiation of SGLT-2i was associated with lower risk of mortality, as well as heart failure and CKD. Keywords: SGLT2i, Heart failure, All-cause mortality, Type 2 diabetes mellitus Introduction Cardiovascular disease (CVD) is the leading cause of *Correspondence: didacmauricio@gmail.com; josep.franch@gmail.com morbidity and mortality among individuals with type † Jordi Real, and Bogdan Vlacho contributed equally to this work 2 diabetes mellitus (T2DM) [1, 2]. Indeed, people with 10 Department of Endocrinology and Nutrition, Hospital Universitari de la Santa Creu i Sant Pau, Autonomous Universtity of Barcelona, Sant Quintí, T2DM have a two- to fourfold increased risk for coro- 89, 08041 Barcelona, Spain nary heart disease compared to subjects without dia- 12 Primary Health Care Center Raval Sud, Gerència d’Atenció Primaria, betes [3, 4], and heart failure (HF) is another common Institut Català de la Salut, Av. Drassanes, 17‑21, 08001 Barcelona, Spain Full list of author information is available at the end of the article complication of T2DM associated with high risk of CV © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​ mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​mmons.​org/​publi​cdoma​in/​ zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Real et al. Cardiovasc Diabetol (2021) 20:139 Page 2 of 11 death [5]. Moreover, different studies have shown that [27, 28]. As such, SGLT2i are nowadays endorsed as glycaemic control does not significantly reduce CV risk, second line therapy by European and American clini- stressing the need for novel treatments that can prevent cal guidelines for patients with T2DM and CV risk fac- the development of CVD complications independent of tors, history of HF, or chronic kidney disease (CKD) [29, glucose lowering [6–8]. 30]. However, it is important to examine whether these So far, meta-analyses of prospective CV outcome tri- results can be replicated in Southern Europe areas other als (CVOTs) have consistently shown that treatment with than Italy, the only countries of this region included and sodium–glucose co-transporter-2 inhibitors (SGLT2i) only regarding the assessment of renal outcomes in the modestly reduces the incidence of major adverse cardi- CVD REAL available studies [19]. It is also important to ovascular outcomes (MACE) and have a consistent and consider that Fadini et al. reported that there are coun- robust beneficial effect on hospitalizations for heart fail- try-specific differences in the profile and characteristics ure and progression of kidney disease regardless of prior of patients initiating dapagliflozin in Southern Europe history of CVD or established CKD [9–12]. Besides, trials [31]. In their real-world practice study, including patients studying kidney function as a secondary outcome or sub- from Italy, Spain, and Greece, the authors found that the group analysis have reported additional positive benefits, SGLT2i dapagliflozin was initiated at different stages including reduced risk of end-stage renal disease and depending on the country, with significant differences renal failure death [13–17]. regarding age, mean T2DM duration, and the presence In the real-world setting, several large multina- of comorbidities (e.g., retinopathy, prior stroke, MI, or tional, observational studies assessed the comparative CVD) [31]. The authors concluded that the observed geo- effectiveness of initiating treatment with a SGLT2i vs. graphic heterogeneity might not only have an impact in other glucose-lowering drugs (oGLDs) (CVD-REAL; the glucose-lowering effectiveness of SGLT2i but also in NCT02993614). The results of the CVD-REAL 1 and 2 the protective CV and renal outcomes. Moreover, the CV studies showed that the use of SGLT2i was associated risk factors and the prevalence of CVD in persons with with decreased risk of hospitalisation for HF and all- T2DM differ across regions in Europe [32, 33]. For exam- cause mortality in subjects with a broad range of CV risk ple, Spain has a lower prevalence of CVD than Northern [18, 19]. Moreover, the CVD-REAL Nordic (conducted European countries but higher than Italy [33]. Consider- in Denmark, Norway, and Sweden) reported an associa- ing these potential differences, we hypothesised that the tion between SGLT2i initiation and a decreased risk of CVD-Real outcomes could be different when the general CV mortality and MACE compared with oGLDs [20]. T2D population is estimated in a Southern European Regarding dapagliflozin in particular, treatment initiation country—region such as Catalonia (Spain). was associated with a lower risk of nonfatal myocardial Based on the need to confirm the applicability of results infarction (MI), nonfatal stroke or CV death, and all- from other observational, real-world studies, the primary cause mortality compared with dipeptidyl peptidase-4 aim of the present study was to investigate, in a primary inhibitors (DPP-4i) [21]. The results from the CVD-REAL care population database in Catalonia (Spain), whether 2 Study (conducted in Asia Pacific, the Middle East, the risk for HF among users initiating SGLT2i was differ- North America and also Spain) confirmed that the use of ent from those initiating oGLDs. Secondarily, we aimed SGLT2i was associated with a lower risk of death, death to compare all-cause mortality, modified MACE (i.e., all- or hospitalisation for HF, MI, and stroke in a broad range cause mortality, MI, or stroke), all-cause mortality or HF, of T2DM subjects with or without CVD [19]. Lastly, the MI, stroke, and CKD between these two groups of new recent CVD-REAL 3 Study (conducted in Israel, Italy, users. Japan, Taiwan, and the UK) showed that new users of an SGLT2i had a slower rate of kidney function decline and Methods a lower risk of clinically meaningful kidney events com- Study design and data sources pared with those on oGLDs [22]. Other retrospective This was a retrospective cohort study of adult subjects observational cohort studies conducted by independent with T2DM between January 2013 and December 2016. groups in different populations and applying different Data from patients were extracted from the electronic methodologies have produced consistent results to what medical records of the SIDIAP database (Information already observed in the CVD REAL studies [23–26]. System for the development of Primary Care Research) Available observational, real-world studies suggested and the CMDB register (Hospital Discharge Records in that there might be a class effect of SGLT2i on CV and the National Health System) from the Catalan Institute of renal outcomes, and that the benefits could be extended Health (ICS) [34]. The assigned population of the ICS is to a broader population with diabetes (i.e., with a broad around 5.6 million individuals (approximately 74% of the CV risk profile and different stages of renal disease) total Catalan population) organised in 288 primary care Real et al. Cardiovasc Diabetol (2021) 20:139 Page 3 of 11 teams. These electronic datasets include anonymised healthcare region not served by ICS (thus withdrawn demographic data, visits to primary care services, spe- from the database), last date of data collection, or to the cialist referrals, diagnoses and procedure codes (Interna- death date in the on treatment approach. In the intention tional Classification of Diseases [ICD]-9 and 10 systems), to treat approach, subjects were followed until the last clinical data, laboratory test results, all-cause mortal- date of data collection independently if they discontinued ity, information on drugs prescriptions, and pharmacy their index treatment or switched to another treatment. dispensations. Variables Subjects Variables were captured for all patients at the index date To identify patients initiating glucose-lowering treat- inclusive or before (1 year prior to the index date). At ment with either an SGLT2i (canagliflozin, dapagliflozin, baseline, we collected data for social-demographic char- or empagliflozin) or oGLDs in our databases we used the acteristics, including age, gender, toxic habits (smoking), codes of the Anatomical Therapeutic Chemical (ACT) and deprivation index (DI). The DI assesses five socioeco- classification system from the World Health Organiza- nomic indicators related with work and education (i.e., tion (WHO) [35]. New users were defined as individu- unemployment, manual and eventual workers, and insuf- als with a registered prescription or drug dispensation ficient education overall and in young people) extracted (either as initial or add-on therapy) for any SGLT-2i (i.e., from census tracks [37]. It detects small areas of large cit- canagliflozin, dapagliflozin or empagliflozin) or oGLDs, ies in Spain with unfavorable socioeconomic characteris- including fixed-dose combinations, with no prior pre- tics and is associated with overall mortality. The higher is scriptions of an SGLT-2i during the preceding year [18]. the DI, the worse the social deprivation is. If the subject in the oGLDs group started with more than Clinical characteristics included diabetes-related one drug, one of these drugs was randomly selected as parameters [e.g., glicated haemoglobin (HbA1c) and dia- the index drug. betes duration], concomitant medications, comorbidi- New users were eligible if they initiated treatment with ties, and laboratory parameters (with the baseline value either an SGLT2i or oGLD between the 1st December defined as the last available value during the prior year 2013 and the 31st December 2016. In our approach, all including the index date). During the follow-up, we col- episodes of SGLT-2i and oGLD initiation were eligible lected data (ICD10 diagnostic codes) on the follow- to be included, thus one patient might be included more ing events as single outcomes: all-cause mortality, HF than once and might have contributed with more than [hospitalisation for HF or diagnostic code (ICD10: I50), one episode of new glucose-lowering medication initia- atrial fibrillation (ICD10: I48), stroke (ICD10: I60–I62), tion for different drug classes (e.g., SGLT-2i and various ischemic stroke (ICD10: I63–I64), MI (ICD10: I21–I22), classes of oGLDs) and at different time points. Further- and CKD (ICD10: N18, N08.3, E11.2]. In addition, we more, subjects were not selected hierarchically; there- also analysed the following composite outcomes: (a) all- fore, the potential risk period from the oGLD group cause mortality or HF (hospitalisation for HF or diagnos- was removed, which could lead to mortality bias [36]. tic code), and (b) modified MACE, defined as all-cause This methodological approach allowed all oGLD expo- mortality, MI or stroke. sure episodes to contribute to oGLD estimates and all SGLT2i exposure episodes to contribute to SGLT2i esti- Statistical methods mates. Additionally, the groups were built with 1:1 pro- The SGLT-2i group was matched to the oGLDs group pensity matching for new glucose-lowering agent (oGLD (comparator) based on propensity score and calendar or SGLT2i) episodes rather than individual subjects, tak- period of study entry. For matching, we used the nearest ing into account each drug episode within subjects. This neighbour calliper width of 0.25 multiplied by the stand- methodologic approach was made in order to minimize ard deviation (SD) of the propensity score distribution immortal time bias. Moreover, subjects were included in [38]. The variables considered for the estimation of the the study by the index date (defined as the prescription/ propensity score included age at study date entry (index dispensation date for the new SGLT2i or oGLD) were date), gender, CV risk factors, indicators of diabetes ≥ 18 years and had more than 1-year medical history in severity, and use of concomitant medications (Additional the database. Subjects with a diagnosis of type 1 diabe- file 1: Table S1). Once the propensity score was per- tes (ICD10 code E10), gestational diabetes (ICD10 code formed, new treatment episodes of initiators of SGLT2i O24.4), or on dialysis treatment (ICD10 code Z49) were and oGLDs were matched 1:1. excluded from the study. All subjects were followed from The baseline characteristics for each cohort were sum- the index date to the earliest end of the use of the given marised by frequencies and percentages for categorial treatment or the date that the subject moved to another variable, and as mean (± SD or median and quartiles) for Real et al. Cardiovasc Diabetol (2021) 20:139 Page 4 of 11 continuous and count variables. For the main on treat- 10%, which resulted in a total of 25,834 new user epi- ment analysis, we calculated the person-time at risk for sodes, 12,917 in each drug cohort (Additional file 1: Fig- each patient as the length of the index exposure episode, ures S1 and S2; http://​dapcat.​shiny​apps.​io/​C VD_​REAL). defined as the number of days from the day after the In the overall population, more than two-thirds of users index prescription start date to the last day of follow-up. at the moment of initiation with a new glucose-lowering For each outcome of interest, the crude incidence rate drug had a diabetes duration of ≥ 5 years (77.7%), subop- (IR) in each index exposure group was estimated as the timal glycemic control (mean HbA1c = 8.69%; SD = 4.8), number of incident events divided by the total number of and nearly two-thirds (65.1%) were on statins treatment patient-years (PY) at risk and was expressed as per 100 (Table 1). PY. The IRs for the SGLT2i group and oGLD group were In the SGLT2i group, the overall exposure to the drug then compared using a hazard ratio (HR) and the corre- class was 9484 years. The dapagliflozin, empagliflozin sponding 95% confidence interval (CI). This analysis was and canagliflozin accounted for 60%, 26%, and 14% of the performed using Cox proportional hazards regression total exposure time, respectively. The overall exposure by clusters (patient ID) with robust estimation of stand- time for oGLDs was 10,012 years (see Additional file 1: ard errors. Both crude and adjusted HR were estimated Table S4 for detailed information on exposure time for for all endpoints. As covariates for adjustment, we used the different drugs). New users of dipeptidyl peptidase-4 gender, age, T2DM duration, hypertension, body mass inhibitors (DPP4i) had the highest proportion of expo- index (BMI), use of antihypertensive drugs, and HF or MI sure time (20.8%), followed by metformin (20.7%), insulin (Additional file 1: Table S2). In addition, Kaplan–Meier (20.6%), sulphonylurea (15.5%), glucagon-like peptide-1 plots were are also generated for each of the analysed receptor agonists (GLP1RA; 10.3%), meglitinide (9.1%), outcomes. A sensitivity analysis was further conducted and other drugs (3.4%). with an intent-to-treat approach with those subjects who continued in the study even after they discontinued Incidence rate and risk of CV events and renal impairment their index treatment or switched to another treatment. At baseline, 27% of patients had established CVD, and Finally, a subgroup analysis was performed within the fol- 6.6% CKD. The incident rate (i.e., new events per 100 lowing pre-specified subgroups: prior CV disease, prior patient-years) of HF, all-cause death, the combination heart failure, ≥ 65 years of age at the index date, gender, of both outcomes, and modified MACE was lower in prior kidney disease, and baseline treatment with anti- patients with T2DM receiving SGLT2i than in patients hypertensive drugs, insulin, sulfonylureas, GLP1-RA, initiating oGLDs (IR = 1.24 vs. 2.16, 0.72 vs. 1.97, 1.88 thiazolidinedione and statins. All statistical analyses were vs. 3.90, and 1.96 vs. 3.31, respectively; Fig. 1). However, performed using the free R statistical software, version the IRs of the other studied CV events (i.e., non-fatal MI, 3.6.1 (https://​www.r-​proje​ct.​org/). The source code is non-fatal stroke, ischemic stroke, and atrial fibrillation) available at https://​github.​com/​jreal​gatius/​C VD_​REAL_​ were similar between the two groups (Fig. 1). Finally, OP2, and a dashboard with interactive Additional mate- the occurrence of CKD during the follow-up period was rial at https://​dapcat.​shiny​apps.​io/​C VD_​REAL. lower after initiation of SGLT2i compared with initiation of oGLDs (IR 1.65 vs. 2.58). Results The initiation of an SGLT2i was associated with 41% A total of 239,733 subjects with T2DM were identified lower adjusted risk of incident HF, 59% lower risk of as new user episodes of glucose-lowering drugs during all-cause death, 54% lower risk of the composite out- the observational period in the SIDIAP database, with come, and 38% lower risk of modified MACE compared 226,452 of them (94.5%) initiating treatment with and the with matched patients on oGLDs (all p < 0.001; Fig. 2). remaining 13,281 with an SGLT2i (5.5%). Baseline char- In contrast, no statistically significant difference was acteristics before propensity-score matching are shown detected in the risk for incident non-fatal MI, non-fatal in Additional file 1: Table S3. Briefly, we observed that stroke, ischemic stroke, and atrial fibrillation between the SGLT2i new users were younger, with a longer duration two treatment groups. Lastly, we observed that patients of T2DM, higher BMI, and poorer glycemic control than receiving SGLT2i had significantly reduced incident CKD oGLDs new users. Moreover, the proportion of patients compared with those on oGLDs (p < 0.001; Fig. 2). with microvascular disease, hypertension, previous MI, and peripheral artery disease (PAD) was higher among Sensitivity and subgroup analyses those initiating an SGLT2i. Similar results regarding the number of events and IRs After propensity-score matching, we obtained well-bal- were observed for all of the study outcomes in the sen- anced cohorts for patients’ baseline characteristics with sitivity analysis using the intention-to-treat approach standardised differences for all of the variables less than population (Additional file 1: Table S5 and http://​ Real et al. Cardiovasc Diabetol (2021) 20:139 Page 5 of 11 Table 1 Baseline characteristics of the study participants after propensity score matching Total population oGLD SGLT2i Standardized N = 25,834 N = 12,917 N = 12,917 mean differences Age, mean (SD), years 62.9 (11.1) 62.8 (11.8) 62.9 (10.4) 0.007 Gender, n (%), females 11,331 (43.9) 5682 (44.0) 5649 (43.7) 0.005 Smoking, n (%) 4145 (16.3) 2081 (16.3) 2064 (16.2) 0.004 Diabetes duration ≥ 5 years, n (%) 20,065 (77.7) 10,037 (77.7) 10,028 (77.6) 0.002 HbA1c (%), mean (SD) 8.69 (1.57) 8.79 (1.64) 8.59 (1.48) 0.128 Missing’s, n (%) 5980 (23.1) 2948 (22.8) 3032 (23.5) Deprivation index (medium–high), n (%) 12,356 (47.9) 6156 (47.6) 6200 (48.0) 0.016 Comorbidities, n (%) Cardiovascular disease 7019 (27.2) 3466 (26.8) 3553 (27.5) 0.015 Heart failure 1421 (5.5) 692 (5.4) 729 (5.6) 0.013 Myocardial infarction 1624 (6.3) 800 (6.2) 824 (6.4) 0.008 Unstable angina 383 (1.5) 197 (1.5) 186 (1.4) 0.007 Atrial fibrillation 1343 (5.2) 668 (5.2) 675 (5.2) 0.002 Stroke 1488 (5.8) 752 (5.8) 736 (5.7) 0.005 Chronic kidney disease 1697 (6.6) 1021 (7.9) 676 (5.2) 0.108 Peripheral artery disease 1895 (7.3) 942 (7.3) 953 (7.4) 0.003 Microvascular ­diseasea 5749 (22.3) 2884 (22.3) 2865 (22.2) 0.004 Cancer 2975 (11.5) 1589 (12.3) 1386 (10.7) 0.049 eGFR, mL/min/1.73 ­m2, mean (SD) 58.8 (4.82) 58.5 (5.5) 59.0 (3.98) 0.100 Missing’s, n (%) 6058 (23.4) 3003 (23.2) 3055 (23.7) Concomitant medications, n (%) Anti-hypertensives 19,453 (75.3) 9645 (74.7) 9808 (75.9) 0.029 ACE inhibitors 9891 (38.3) 4950 (38.3) 4941 (38.3) 0.001 Angiotensin II receptor blockers 8577 (33.2) 4272 (33.1) 4305 (33.3) 0.005 Beta blockers 6926 (26.8) 3441 (26.6) 3485 (27.0) 0.008 Calcium channel blockers 540 (2.1) 290 (2.3) 250 (1.9) 0.022 Thiazides 2431 (9.4) 1219 (9.4) 1212 (9.4) 0.002 Loop-diuretics 3008 (11.6) 1492 (11.6) 1516 (11.7) 0.006 Statins 16,812 (65.1) 8381 (64.9) 8431 (65.3) 0.008 a Defined as diabetic neuropathy, retinopathy, or nephropathy eGFR estimated glomerular filtration rate by Chronic Kidney Disease Epidemiology formula, oGLD other glucose-lowering drugs, SGLT2i sodium–glucose co-transporter inhibitors dapcat.​shiny​apps.​io/​C VD_​REAL). Moreover, crude and the association between SGLT2i use and reduced risk adjusted Cox regression analyses (i.e., by gender, age, of HF, all-cause death, all-cause death or HF, MACE, T2DM duration, hypertension, BMI, and antihyperten- and CKD (Additional file 1: Figures S3 to S11 or http://​ sive drugs) showed magnitude and effects similar to the dapcat.​shiny​apps.​io/​C VD_​REAL). ones detected in the on-treatment analysis approach (Additional file 1: Table S6 and http://​dapcat.​shiny​apps.​ io/​C VD_​REAL). Discussion The subgroup analyses showed no statistically sig- In the present study involving a primary care popula- nificant interactions with users’ characteristics, includ- tion with T2DM from Catalonia starting new antidia- ing age (≥ 65 years), gender, comorbidity profile (i.e., betic treatment we observed a significantly lower risk of prior CVD, HF, or CKD), or CV and baseline medica- HF associated with initiation of SGLT2i compared with tion (i.e., treatment with antihypertensive drugs, insu- matched patients initiating oGLDs. We also detected lin, sulfonylureas, GLP1-RA thiazolidinedione, and lower risk of all-cause mortality, CKD, modified MACE, statins). As such, none of the these variables modified and all-cause death or HF. In contrast, we did not identify Real et al. Cardiovasc Diabetol (2021) 20:139 Page 6 of 11 Fig. 1 Incidence rates of the different events per 100 person-years by treatment. CKD chronic kidney disease, HF heart failure, MACE major adverse cardiovascular events, MI myocardial infarction, oGLDs other glucose-lowering drugs, PY patient-years of exposure, SGLT2i sodium–glucose co-transporter inhibitors any significant differences between groups regarding MI, studies (26% vs. 5–9%) [22, 27, 28]. The users who initi- non-fatal stroke, ischemic stroke and atrial fibrillation. ated SGLT2i in our study reported 68 events related to Several real-world observational studies have been all-cause mortality compared with 197 events in the published with routinely collected data from different oGLD group, corresponding to a risk reduction of 59%. countries [18–22, 39–41]. The baseline characteristics These results were slightly higher, but in line with the of the subjects included in our study are similar to those other real-world studies, in which the risk reduction reported regarding age, prevalence of CVD, microvascu- of all-cause mortality ranged from 46 to 56% [18–22, lar complications, and prior CV therapies. Moreover, the 37–39]. It could be hypothesized that the differences highest proportion of exposure time was to dapagliflozin, between our results and those of other RCTs or RW stud- which was also the case in other RW studies (60% in our ies could be related to different exposure to the individual population, and between 51 and 94% in other observa- SGLT2i across studies. However, the relative effective- tional studies) [18–22, 41]. ness of each product on the different outcomes remains The main outcome of our study was the reduction in to be elucidated, because study populations and designs the incidence rate of HF, which is a common complica- are different. Recent attempts to indirectly address this tion associated with CV death in T2DM [5]. The rela- issue through network meta-analyses led to differing tive risk reduction of HF was 41%, which is in line with conclusions regarding the differential effect of individual the 30–39% reduction observed in other RW studies for SGLT2i on outcomes such as HF and all-cause mortal- hospitalizations for HF [18–22, 41]. In addition, the risk ity [44, 45]. Therefore, in the absence of prospective or reduction for HF or all-cause mortality was 45% in our retrospective head-to-head trials comparing individual analysis, which is as well in agreement with the 31–46% SGLT2i, it cannot be ascertained to what extent the expo- range reported in observational studies [18–21, 40]. It is sure to the individual SGLT2i in CVD-REAL Catalonia possible that the greater risk reduction observed in our vs. other studies could have contributed to the observed study was due to the fact that we had no data on hospi- differences. tal events for HF admissions. Indeed, the need for hos- New use of SGLT2i was associated with a 38% lower pitalization among patients with HF is frequent due to risk of MACE compared to a new use of oGLDs, which recurrent decompensation, with an estimated 12–45% is higher than the 21–22% reported by two other obser- incidence of hospitalizations for HF at 1 year [42, 43]. vational studies [19, 20]. In the absence of data on CV In addition, we had a higher exposition to empagliflo- death, we used a modified MACE and replaced CV death zin in our study compared with the other observational by all-cause death. It is then conceivable that we included Real et al. Cardiovasc Diabetol (2021) 20:139 Page 7 of 11 HR (95% CI); P-value All-cause death 0.41 (0.31-0.54); p<0.001 HF 0.59 (0.47-0.74); p<0.001 Non-fatal MI 1.11 (0.75-1.63); p=0.608 Non-fatal stroke 0.84 (0.63-1.12); p=0.238 Ischaemic stroke 0.81 (0.57-1.13); p=0.213 Atrial fibrillation 0.93 (0.67-1.24); p=0.618 CDK 0.66 (0.54-0.80); p<0.001 All-cause death or HF 0.55 (0.47-0.63); p<0.001 Modified MACE 0.62 (0.52-0.74); p<0.001 0.0 0.5 1.0 1.5 2.0 Hazard Ratio Favours SGLT2i Favours oGLDs Fig. 2 Forest plot of the adjusted incident hazard ratios (HR) and 95% CI for the different outcomes. CKD chronic kidney disease, CI confidence interval, HF heart failure, MACE major adverse cardiovascular events, HR hazard ratio, MI myocardial infarction, oGLDs other glucose-lowering drugs, PY patient-years of exposure, SGLT2i sodium–glucose co-transporter inhibitors a greater number of events, thus resulting in a higher- (ESRD) was 50% lower in the CVD REAL 3 cohort [22]. than-expected reduction in this particular outcome. On the one hand, the mean eGFR value in our study was In the moment of realization of the study SGLT2i were 58.8 mL/min/1.73 ­m2, indicating a mild to moderate loss not recommended in severe renal impairment, namely of kidney function, while in CVD REAL 3 the mean value < 45 mL/min/1.73 ­m2 for empagliflozin and canagliflo- was above 90 mL/min/1.73 ­m2 (normal range) and only zin, and < 60 mL/min/1.73 ­m2 for dapagliflozin [46, 47]. 8% of initiators had values below 60 mL/min/1.73 ­m2. In Regarding the assessed renal outcome, we found that that study, the annual rate of change in eGFR from base- the risk of CKD after initiation with SGLT2i was 34% line showed the benefit of initiating an SGLT2i regard- less than in the oGLDs group, while the risk of the com- less the eGFR subgroup, although the magnitude of the posite outcome eGFR decline or end stage renal disease change was lower among those with compromised kidney Real et al. Cardiovasc Diabetol (2021) 20:139 Page 8 of 11 function (30% and 79% in those with < 60 and 60–90 mL/ observational studies based on health care records of real min/1.73 ­m2, respectively). Another RW study conducted clinical practice, the possibility of residual and unmeas- in Italy, the DARWIN-T2D, assessed the albumin excre- ured confounding factors cannot be ruled out. However, tion rate (AER) as a surrogate of kidney outcome in we used robust statistical techniques, including well-bal- T2DM patients treated with dapagliflozin vs. active com- anced groups created by propensity-matching and sensi- parators (i.e., GLPR-1a, DPP-4i, or glicazide) [48]. The tivity analyses. Another limitation is that we did not have mean eGFR was 83 mL/min/1.73 ­m2 at baseline and the information on CV mortality, which could have impacted authors reported a decline of 37% in the AER irrespec- the estimation of the incident rates of MACE if defined tive of baseline eGFR, while there was no change among as including CV death. Moreover, our average follow-up those who received a comparator. Lastly, a systematic time was relatively limited, as SGLT2i use in real-world review and meta-analysis of RCTs in patients with T2DM settings is still relatively recent; thus, longer period anal- and CKD (defined as eGFR < 60 mL/min/1.73 ­m2) treated yses will be needed to evaluate if the positive effects of with SGLT2i found that these agents slowed the annual SGLT2i are sustained over time. Since the comparator loss in kidney function (eGFR slope) and led to a 29% group was oGLD’s in our study, there was a possibility reduction in the risk of the composite outcome doubling of “immortal time bias,” which occurs when two patient serum creatinine, ESRD, or renal death [15]. A reduc- groups are hierarchically formed within a time interval. tion of 34% vs. placebo in the composite outcome was However, the new user episodes design combined with as well reported in the CREDENCE trial, conducted in propensity score matching for index date and time since patients with T2DM and albuminuric CKD receiving initiation eliminate this bias. Furthermore, if the compar- canagliflozin [17]. Besides, in the DAPA-CKD trial with ator group was DPP4i instead of OGLD´s, a similar ten- dapagliflozin, which included 68% of patients with CKD dency in outcomes favouring SGLT2i users was observed and T2DM, reductions of 39% and 31% vs. placebo were and published [55]. Lastly, we had limited socio- eco- observed in the primary composite outcome and in all- nomic data and no information on lifestyle variables for cause mortality, respectively [16]. patients. For instance, dietary differences between Euro- The incident rates of non-fatal MI, stroke (non-fatal pean countries may modulate CV outcomes, since it is or ischemic), and atrial fibrillation were not different well known that subjects highly adherent to the Medi- between patients initiating SGLT2i or other oGLDs in terranean diet have about 30% lower risk of CVD mor- our study. A neutral or small effect of SGLT2i on all of bidity and mortality, and this raises up to 40–45% in these particular outcomes has been as well observed in randomized clinical trials (RCTs) with patients at high previous RW studies [20, 21, 39, 49]. There is also a pos- CVD risk [56–58]. sibility that the diagnostic codes used and proposed by the global CVD Real study protocol for these events had low prevalence in our database. Moreover, this modest or Conclusion even nonsignificant association of SGLT2i with athero- In this real-world study of patients with T2DM attended thrombotic events has been as well observed in several in routine clinical practice from a South European region, meta-analyses of RCTs comparing this drug class against new use of SGLT2i was associated with a lower risk of placebo or active comparators [10, 50–52]. HF, mortality and renal events compared with the use The results of our study should be considered in the of oGLDs. These results expand previous observational context of several potential strengths and limitations. The studies supporting the use of SGLT2i in patients with main strength of our study is a large number of patients a broad CV risk profile in real-world clinical practice. included the representativeness of the diabetic popula- Moreover, they show that the magnitudes of the associ- tion [19]. Moreover, our study involves real-world data ated CV and renal benefits are similar to those observed from a South European region where the prevalence of in other European geographical regions. CV risk factors and CV disease in patients with T2DM is expected to be different from that in Northern Europe Abbreviations or the US [53]. Indeed, our results go in the same direc- CVD: Cardiovascular disease; CKD: Chronic kidney disease; HF: Heart failure; tion as other CVD-Real studies, besides differences in MACE: Major adverse cardiovascular outcomes; MI: Myocardial infarction; the studied populations. Moreover, the SIDIAP database oGLDs: Other glucose-lowering drugs; RW: Real world; RCT​: Randomized clinical trials; SGLT2i: Sodium–glucose co-transporter-2 inhibitors; T2DM: Type has been extensively used for different epidemiologic 2 diabetes mellitus. and pharmacoepidemiologic national and international research, and it is established as the well-validated pri- Supplementary Information mary care Spanish database for the study of diabetes [33, The online version contains supplementary material available at https://​doi.​ 54]. One limitation of the study is that inherent to all org/​10.​1186/​s12933-​021-​01323-5. Real et al. Cardiovasc Diabetol (2021) 20:139 Page 9 of 11 Competing interests Additional file 1: Figure S1. Flow chart of the patients included in the The CVD REAL study was conceived and funded by AstraZeneca, and led by its study. Figure S2. Pre and post propensity score matching and stand‑ independent Executive Scientific Board. The CVD REAL Catalonia study design ardised differences. Table S1. Variables included in the propensity score. adaptation, data collection and analyses have been conducted indepen‑ Table S2. Variables included in the adjustments of Cox regression analysis. dently from AstraZeneca. The manuscript writing, results interpretation and Table S4. Number of patients and follow-up time (years). Table S5. conclusions are those of the authors alone. M. M.-C. has received advisory Sensitivity crude analysis of intention-to-treat. Table S6. Sensitivity honorarium from Astra-Zeneca, Bayer, Boehringer Ingelheim, GSK, Lilly, MSD, analysis of intention-to-treat output from Cox regression models. Figure Novartis, Novo Nordisk, and Sanofi; he has received speaker honorarium from S3. Subgroup analysis for heart failure outcome, including gender, age, Astra-Zeneca, Bayer, Boehringer Ingelheim, GSK, Lilly, Menarini, MSD, Novartis, chronic kidney disease, history of cardiovascular disease, cardiovascular Novo Nordisk, and Sanofi; he has received research grants to the institution and diabetes medications at baseline (ITT approach). Figure S4. Subgroup from Astra-Zeneca, GSK, Lilly, MSD, Novartis, Novo Nordisk, and Sanofi. J. F.-N. analysis for all-cause death or hearth failure outcome, including gender, has received advisory and or speaking fees from Astra-Zeneca, Ascensia, age, chronic kidney disease, history of cardiovascular disease, cardiovas‑ Boehringer Ingelheim, GSK, Lilly, MSD, Novartis, Novo Nordisk, and Sanofi; he cular and diabetes medications at baseline (ITT approach). Figure S5. has received research grants to the institution from Astra-Zeneca, GSK, Lilly, Subgroup analysis for modified MACE outcome, including gender, age, MSD, Novartis, Novo Nordisk, Sanofi, and Boehringer. D. M. has received advi‑ chronic kidney disease, history of cardiovascular disease, cardiovascular sory and/or speaking fees from Astra-Zeneca. Ascensia, Boehringer Ingelheim, and diabetes medications at baseline (ITT approach). Figure S6. Subgroup GSK, Lilly, MSD, Novartis, Novo Nordisk, and Sanofi; he has received research analysis for all-cause death outcome, including gender, age, chronic kid‑ grants to the institution from Astra-Zeneca, GSK, Lilly, MSD, Novartis, Novo ney disease, history of cardiovascular disease, cardiovascular and diabetes Nordisk, Sanofi, and Boehringer. M. K. reports research grants fees from Astra medications at baseline (ITT approach). Figure S7. Subgroup analysis for Zeneca, Boehringer Ingelheim; consultant/Advisory Board fees from Amgen, nonfatal myocardial infarction outcome, including gender, age, chronic Applied Therapeutics, Astra Zeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Jans‑ kidney disease, history of cardiovascular disease, cardiovascular and sen, Merck (Diabetes), Novo Nordisk, Sanofi and Vifor Pharma; other research diabetes medications at baseline (ITT approach). Figure S8. Subgroup support fees from Astra Zeneca, and Honorarium fees from Astra Zeneca, analysis for non-fatal stroke outcome, including gender, age, chronic kid‑ Boehringer Ingelheim, Novo Nordisk. P. F. and E. T. W. are full time employees ney disease, history of cardiovascular disease, cardiovascular and diabetes of AstraZeneca. P. F., M. K. and E. T. W. are members of the CVD REAL Executive medications at baseline (ITT approach). Figure S9. Subgroup analysis for Scientific Committee, conceived and led the initial CVD REAL global studies ischemic stroke outcome, including gender, age, chronic kidney disease, and analyses. J. R. B. V. and E. C. have no conflict of interest to declare. history of cardiovascular disease, cardiovascular and diabetes medica‑ tions at baseline (ITT approach). Figure S10. Subgroup analysis for atrial Author details fibrillation outcome, including gender, age, chronic kidney disease, history 1 DAP‑Cat Group, Unitat de Suport a la Recerca Barcelona, Fundació Institut of cardiovascular disease, cardiovascular and diabetes medications at Universitari per a la recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDI‑ baseline (ITT approach). Figure S11. Subgroup analysis for chronic kidney APJGol), Barcelona, Spain. 2 CIBER of Diabetes and Associated Metabolic Dis‑ disease outcome, including gender, age, chronic kidney disease, history eases (CIBERDEM), Instituto de Salud Carlos III (ISCIII), Madrid, Spain. 3 Depart‑ of cardiovascular disease, cardiovascular and diabetes medications at ment of Endocrinology and Nutrition, Institut d’Investigacions Biomèdiques baseline (ITT approach). Table S7. Specification of diseases codes used. August Pi i Suñer, Hospital Clinic, Barcelona, Spain. 4 CIBER of Physiopathology Table S8. Specification of ATC/DDD codes used. of Obesity and Nutrition (CIBEROBN), Instituto de Salud Carlos III (ISCIII), Madrid, Spain. 5 Drug Area, Gerència d’Atenció Primaria, Institut Català de la Salut, Barcelona, Spain. 6 Primary Health Care Center La Mina, Gerència d’Àmbit Acknowledgements d’Atenció Primària Barcelona Ciutat, Institut Català de la Salut, Sant Adrià de We thank Mònica Gratacòs for the editorial support of this manuscript. CIBER Besòs, Spain. 7 AstraZeneca, Gaithersburg, MD, USA. 8 Cardiovascular Renal of Diabetes and Associated Metabolic Diseases (CIBERDEM), from Instituto de Metabolisms, BioPharmaceuticals Global Medical, AstraZeneca, Cambridge, Salud Carlos III. Madrid. Spain. This work has been carried out within the frame‑ UK. 9 Saint Luke’s Mid America Heart Institute and University of Missouri-Kansas work of the Bogdan Vlacho Doctorate in Pharmacology of the Department of City, Kansas City, MO, USA. 10 Department of Endocrinology and Nutrition, Hos‑ Pharmacology, Therapeutics and Toxicology, Faculty of Medicine, Autonomous pital Universitari de la Santa Creu i Sant Pau, Autonomous Universtity of Bar‑ University of Barcelona. celona, Sant Quintí, 89, 08041 Barcelona, Spain. 11 Departament of Medicine, University of Vic-Central University of Catalonia, Vic, Barcelona, Spain. 12 Primary Authors’ contributions Health Care Center Raval Sud, Gerència d’Atenció Primaria, Institut Català de la Conceptualisation: JFN, MMC, DM and JR; methodology: JR; formal analysis: JR; Salut, Av. Drassanes, 17‑21, 08001 Barcelona, Spain. resources and data curation: BV and JR; writing—original draft preparation. EC and BV; writing—review and editing: JFN, EC, BV, EO, DM, MMC, MK, PF, ETW Received: 10 March 2021 Accepted: 23 June 2021 and JR; supervision: JFN and DM; project administration. BV and JFN; funding acquisition: JFN. All authors read and approved the final manuscript. Funding This study was sponsored by AstraZeneca, Spain [Grant Number 4R17/071-3]. References Availability of data and materials 1. Leon BM. Diabetes and cardiovascular disease: epidemiology, biological The data controller for SIDIAP does not allow the sharing of raw data. Statisti‑ mechanisms, treatment recommendations and future research. World J cal codes are available upon request from the corresponding authors (J.F.-N Diabetes. 2015;6:1246. https://​doi.​org/​10.​4239/​wjd.​v6.​i13.​1246. or D.M). 2. Tancredi M, Rosengren A, Svensson A-M, et al. Excess mortality among persons with type 2 diabetes. N Engl J Med. 2015;373:1720–32. https://​ doi.​org/​10.​1056/​NEJMo​a1504​347. Declarations 3. Shah AD, Langenberg C, Rapsomaniki E, et al. Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1.9 million people. Ethics approval and consent to participate Lancet Diabetes Endocrinol. 2015;3:105–13. https://​doi.​org/​10.​1016/​ The study was approved by the Ethics Committee of the Primary Health Care S2213-​8587(14)​70219-0. University Research Institute (IDIAP) Jordi Gol following the Spanish regula‑ 4. The Emerging Risk Factors Collaboration. Diabetes mellitus, fasting blood tions on observational studies (Approval Number: P17/153). glucose concentration, and risk of vascular disease: a collaborative meta- analysis of 102 prospective studies. Lancet. 2010;375:2215–22. https://​doi.​ Consent for publication org/​10.​1016/​S0140-​6736(10)​60484-9. Not applicable. 5. Cavender MA, Steg PG, Smith SC, et al. Impact of diabetes mellitus on hospitalization for heart failure, cardiovascular events, and death. Real et al. Cardiovasc Diabetol (2021) 20:139 Page 10 of 11 Circulation. 2015;132:923–31. https://​doi.​org/​10.​1161/​CIRCU​LATIO​NAHA.​ 23. Udell JA, Yuan Z, Rush T, et al. Cardiovascular outcomes and risks after 114.​014796. initiation of a sodium glucose cotransporter 2 inhibitor. Circulation. 6. Scirica BM, Bhatt DL, Braunwald E, et al. Saxagliptin and cardiovascu‑ 2018;137:1450–9. https://​doi.​org/​10.​1161/​CIRCU​LATIO​NAHA.​117.​031227. lar outcomes in patients with type 2 diabetes mellitus. N Engl J Med. 24. Udell JA, Yuan Z, Ryan P, et al. Cardiovascular outcomes and mortality 2013;369:1317–26. https://​doi.​org/​10.​1056/​NEJMo​a1307​684. after initiation of canagliflozin: analyses from the EASEL study. Endocrinol 7. Action to Control Cardiovascular Risk in Diabetes Study Group. Effects Diabetes Metab. 2020;3:e00096. https://​doi.​org/​10.​1002/​edm2.​96. of intensive glucose lowering in type 2 diabetes. N Engl J Med. 25. Patorno E, Pawar A, Franklin JM, et al. Empagliflozin and the risk of heart 2008;358:2545–59. https://​doi.​org/​10.​1056/​NEJMo​a0802​743. failure hospitalization in routine clinical care. Circulation. 2019;139:2822– 8. ADVANCE Collaborative Group. Intensive blood glucose control and 30. https://​doi.​org/​10.​1161/​CIRCU​LATIO​NAHA.​118.​039177. vascular outcomes in patients with type 2 diabetes. N Engl J Med. 26. Pasternak B, Wintzell V, Melbye M, et al. Use of sodium–glucose co- 2008;358:2560–72. https://​doi.​org/​10.​1056/​NEJMo​a0802​987. transporter 2 inhibitors and risk of serious renal events: Scandinavian 9. Zelniker TA, Wiviott SD, Raz I, et al. SGLT2 inhibitors for primary and sec‑ cohort study. BMJ. 2020;369: m1186. https://​doi.​org/​10.​1136/​bmj.​m1186. ondary prevention of cardiovascular and renal outcomes in type 2 dia‑ 27. Gallwitz B. The cardiovascular benefits associated with the use of betes: a systematic review and meta-analysis of cardiovascular outcome sodium–glucose cotransporter 2 inhibitors—real-world data. Eur Endo‑ trials. Lancet. 2019;393:31–9. https://​doi.​org/​10.​1016/​S0140-​6736(18)​ crinol. 2018;14:17. https://​doi.​org/​10.​17925/​EE.​2018.​14.1.​17. 32590-X. 28. Kluger AY. Class effects of SGLT2 inhibitors on cardiorenal outcomes. In: 10. Wu JHY, Foote C, Blomster J, et al. Effects of sodium–glucose cotrans‑ Textbook of cardiorenal medicine. Cham: Springer International Publish‑ porter-2 inhibitors on cardiovascular events, death, and major safety ing; 2021. p. 279–94. https://​doi.​org/​10.​1007/​978-3-​030-​57460-4_​22. outcomes in adults with type 2 diabetes: a systematic review and meta- 29. American diabetes association. 9. Pharmacologic approaches to glycemic analysis. Lancet Diabetes Endocrinol. 2016;4:411–9. https://​doi.​org/​10.​ treatment: standards of medical care in diabetes—2020. Diabetes Care. 1016/​S2213-​8587(16)​00052-8. 2020;43:S98–110. https://​doi.​org/​10.​2337/​dc20-​S009. 11. Sonesson C, Johansson PA, Johnsson E, et al. Cardiovascular effects of 30. Davies MJ, D’Alessio DA, Fradkin J, et al. Management of hyperglycaemia dapagliflozin in patients with type 2 diabetes and different risk catego‑ in type 2 diabetes, 2018. A consensus report by the American Diabe‑ ries: a meta-analysis. Cardiovasc Diabetol. 2016;15:37. https://​doi.​org/​10.​ tes Association (ADA) and the European Association for the Study of 1186/​s12933-​016-​0356-y. Diabetes (EASD). Diabetologia. 2018;61:2461–98. https://​doi.​org/​10.​1007/​ 12. Monami M, Dicembrini I, Mannucci E. Effects of SGLT-2 inhibitors on s00125-​018-​4729-5. mortality and cardiovascular events: a comprehensive meta-analysis of 31. Fadini GP, Tentolouris N, Caballero Mateos I, et al. A multinational real- randomized controlled trials. Acta Diabetol. 2017;54:19–36. https://​doi.​ world study on the clinical characteristics of patients with type 2 diabetes org/​10.​1007/​s00592-​016-​0892-7. initiating dapagliflozin in southern Europe. Diabetes Ther. 2020;11:423– 13. Mosenzon O, Wiviott SD, Cahn A, et al. Effects of dapagliflozin on 36. https://​doi.​org/​10.​1007/​s13300-​019-​00744-6. development and progression of kidney disease in patients with type 2 32. Tamayo T, Rosenbauer J, Wild SH, Spijkerman AMW, Baan C, Forouhi diabetes: an analysis from the DECLARE–TIMI 58 randomised trial. Lancet NG, et al. Diabetes in Europe: an update. Diabetes Res Clin Pract. Diabetes Endocrinol. 2019;7:606–17. https://​doi.​org/​10.​1016/​S2213-​ 2014;103:206–17. 8587(19)​30180-9. 33. Einarson TR, Acs A, Ludwig C, et al. Prevalence of cardiovascular disease in 14. Kadowaki T, Nangaku M, Hantel S, et al. Empagliflozin and kidney out‑ type 2 diabetes: a systematic literature review of scientific evidence from cular disease: results from the EMPA-REG OUTCOME ® trial. J Diabetes comes in Asian patients with type 2 diabetes and established cardiovas‑ across the world in 2007–2017. Cardiovasc Diabetol. 2018;17:83. https://​ doi.​org/​10.​1186/​s12933-​018-​0728-6. Investig. 2019;10:760–70. https://​doi.​org/​10.​1111/​jdi.​12971. 34. Bolíbar B, Fina Avilés F, Morros R, Del Mar G-G, Hermosilla E, Ramos R, et al. 15. Toyama T, Neuen BL, Jun M, et al. Effect of SGLT2 inhibitors on cardiovas‑ Base de datos SIDIAP: La historia clínica informatizada de Atención Pri‑ cular, renal and safety outcomes in patients with type 2 diabetes mellitus maria como fuente de información para la investigación epidemiológica. and chronic kidney disease: a systematic review and meta-analysis. Med Clin. 2012;138(14):617–21. Diabetes Obes Metab. 2019;21:1237–50. https://​doi.​org/​10.​1111/​dom.​ 35. WHO Collaborating Centre for Drug Statistics Methodology. Purpose of 13648​16. the ATC/DDD system. Norwegian Institute for Public Health; 2018 16. Heerspink HJL, Stefánsson BV, Correa-Rotter R, et al. Dapagliflozin in 36. Suissa S, Moodie EE, Dell’Aniello S. Prevalent new-user cohort designs for patients with chronic kidney disease. N Engl J Med. 2020;383:1436–46. comparative drug effect studies by time-conditional propensity scores. 17. Perkovic V, Jardine MJ, Neal B, et al. Canagliflozin and renal outcomes Pharmacoepidemiol Drug Saf. 2017;26:459–68. in type 2 diabetes and nephropathy. N Engl J Med. 2019;380:2295–306. 37. Felícitas Domínguez-Berjón M, Borrell C, Cano-Serral G, et al. Construc‑ https://​doi.​org/​10.​1056/​NEJMo​a1811​744. ción de un índice de privación a partir de datos censales en grandes ciu‑ 18. Kosiborod M, Cavender MA, Fu AZ, et al. Lower risk of heart failure and dades españolas (Proyecto MEDEA). Gac Sanit. 2008;22:179–87. https://​ death in patients initiated on sodium–glucose cotransporter-2 inhibitors doi.​org/​10.​1157/​13123​961. versus other glucose-lowering drugs. Circulation. 2017;136:249–59. 38. Ho D, Imai K, King G, Stuart E. Matching as nonparametric preprocessing https://​doi.​org/​10.​1161/​CIRCU​LATIO​NAHA.​117.​029190. for reducing model dependence in parametric causal inference. Political 19. Kosiborod M, Lam CSP, Kohsaka S, et al. Cardiovascular events associated Anal. 2007;15(3):199–236. with SGLT-2 inhibitors versus other glucose-lowering drugs. J Am Coll 39. Toulis KA, Willis BH, Marshall T, et al. All-cause mortality in patients Cardiol. 2018;71:2628–39. https://​doi.​org/​10.​1016/j.​jacc.​2018.​03.​009. with diabetes under treatment with dapagliflozin: a population-based, 20. Birkeland KI, Jørgensen ME, Carstensen B, et al. Cardiovascular mortal‑ open-cohort study in the health improvement network database. J Clin ity and morbidity in patients with type 2 diabetes following initiation Endocrinol Metab. 2017;102:1719–25. of sodium–glucose co-transporter-2 inhibitors versus other glucose- 40. Nyström T, Bodegard J, Nathanson D, et al. Novel oral glucose-lowering lowering drugs (CVD-REAL Nordic): a multinational observational analysis. drugs are associated with lower risk of all-cause mortality, cardiovascular Lancet Diabetes Endocrinol. 2017;5:709–17. https://​doi.​org/​10.​1016/​ events and severe hypoglycaemia compared with insulin in patients with S2213-​8587(17)​30258-9. type 2 diabetes. Diabetes Obes Metab. 2017;19:831–41. https://​doi.​org/​ 21. Persson F, Nyström T, Jørgensen ME, et al. Dapagliflozin is associated 10.​1111/​dom.​12889. with lower risk of cardiovascular events and all-cause mortality in people 41. Norhammar A, Bodegård J, Nyström T, et al. Dapagliflozin and cardio‑ with type 2 diabetes (CVD-REAL Nordic) when compared with dipeptidyl vascular mortality and disease outcomes in a population with type peptidase-4 inhibitor therapy: a multinational observational study. Diabe‑ 2 diabetes similar to that of the DECLARE-TIMI 58 trial: a nationwide tes Obes Metab. 2018;20:344–51. https://​doi.​org/​10.​1111/​dom.​13077. observational study. Diabetes Obes Metab. 2019;21:1136–45. https://​doi.​ 22. Heerspink HJL, Karasik A, Thuresson M, et al. Kidney outcomes associated org/​10.​1111/​dom.​13627. with use of SGLT2 inhibitors in real-world clinical practice (CVD-REAL 3): 42. Chioncel O, Lainscak M, Seferovic PM, et al. Epidemiology and one-year a multinational observational cohort study. Lancet Diabetes Endocrinol. outcomes in patients with chronic heart failure and preserved, mid- 2020;8:27–35. https://​doi.​org/​10.​1016/​S2213-​8587(19)​30384-5. range and reduced ejection fraction: an analysis of the ESC Heart Failure Real et al. Cardiovasc Diabetol (2021) 20:139 Page 11 of 11 Long-Term Registry. Eur J Heart Fail. 2017;19:1574–85. https://​doi.​org/​10.​ myocardial infarction, stroke and hospitalization for heart failure. Diabetes 1002/​ejhf.​813. Res Clin Pract. 2019;150:8–16. https://​doi.​org/​10.​1016/j.​diabr​es.​2019.​02.​ 43. Tuppin P, Cuerq A, de Peretti C, et al. Two-year outcome of patients after 014. a first hospitalization for heart failure: a national observational study. Arch 52. Zheng SL, Roddick AJ, Aghar-Jaffar R, et al. Association between use of Cardiovasc Dis. 2014;107:158–68. https://​doi.​org/​10.​1016/j.​acvd.​2014.​01.​ sodium–glucose cotransporter 2 inhibitors, glucagon-like peptide 1 012. agonists, and dipeptidyl peptidase 4 inhibitors with all-cause mortality in 44. Zheng C, Lin M, Chen Y, Xu H, Yan L, Dai H. Effects of sodium–glucose patients with type 2 diabetes. JAMA. 2018;319:1580. https://​doi.​org/​10.​ cotransporter type 2 inhibitors on cardiovascular, renal, and safety 1001/​jama.​2018.​3024. outcomes in patients with cardiovascular disease: a meta-analysis of 53. Mata-Cases M, Franch-Nadal J, Mauricio D, et al. Investigar en diabetes randomized controlled trials. Cardiovasc Diabetol. 2021;20:83. https://​doi.​ desde una base de datos de atención primaria: la experiencia del Sistema org/​10.​1186/​s12933-​021-​01272-z. de Información para el Desarrollo de la Investigación en Atención Prima‑ 45. Täger T, Atar D, Agewall S, Katus HA, Grundtvig M, Cleland JG, Clark AL, ria (SIDIAP). Av en Diabetol. 2013;29:169–74. https://​doi.​org/​10.​1016/j.​ Fröhlich H, Frankenstein L. Comparative efficacy of sodium–glucose avdiab.​2013.​09.​002. cotransporter-2 inhibitors (SGLT2i) for cardiovascular outcomes in type 54. Mata-Cases M, Mauricio D, Real J, et al. Is diabetes mellitus correctly 2 diabetes: a systematic review and network meta-analysis of ran‑ registered and classified in primary care? A population-based study in domised controlled trials. Heart Fail Rev. 2020. https://​doi.​org/​10.​1007/​ Catalonia, Spain. Endocrinol Nutr. 2016;63:440–8. https://​doi.​org/​10.​ s10741-​020-​09954-8. 1016/j.​endonu.​2016.​07.​0048. 46. Hinnen D. Glucuretic effects and renal safety of dapagliflozin in patients 55. Kohsaka S, Lam CSP, Kim DJ, Cavender MA, Norhammar A, Jørgensen ME, with type 2 diabetes. Ther Adv Endocrinol Metab. 2015;6:92–102. https://​ et al. Risk of cardiovascular events and death associated with initiation doi.​org/​10.​1177/​20420​18815​575273. of SGLT2 inhibitors compared with DPP-4 inhibitors: an analysis from 47. Scheen AJ. Pharmacokinetics, pharmacodynamics and clinical use of the CVD-REAL 2 multinational cohort study. Lancet Diabetes Endocrinol. SGLT2 inhibitors in patients with type 2 diabetes mellitus and chronic 2020;8:606–15. kidney disease. Clin Pharmacokinet. 2015;54:691–708. https://​doi.​org/​10.​ 56. Martínez-González MA, Gea A, Ruiz-Canela M. The Mediterranean diet 1007/​s40262-​015-​0264-4. and cardiovascular health. Circ Res. 2019;124:779–98. https://​doi.​org/​10.​ 48. Fadini GP, Solini A, Manca ML, et al. Effectiveness of dapagliflozin versus 1161/​CIRCR​ESAHA.​118.​313348. comparators on renal endpoints in the real world: a multicentre retro‑ 57. Grosso G, Marventano S, Yang J, et al. A comprehensive meta-analysis on spective study. Diabetes Obes Metab. 2019;21:252–60. https://​doi.​org/​10.​ evidence of Mediterranean diet and cardiovascular disease: are individual 1111/​dom.​13508. components equal? Crit Rev Food Sci Nutr. 2017;57:3218–32. https://​doi.​ 49. Kosiborod M, Birkeland KI, Cavender MA, et al. Rates of myocardial infarc‑ org/​10.​1080/​10408​398.​2015.​11070​21. tion and stroke in patients initiating treatment with SGLT2-inhibitors 58. Razquin C, Martinez-Gonzalez MA. A traditional mediterranean diet versus other glucose-lowering agents in real-world clinical practice: effectively reduces inflammation and improves cardiovascular health. results from the CVD-REAL study. Diabetes Obes Metab. 2018;20:1983–7. Nutrients. 2019;11:1842. https://​doi.​org/​10.​3390/​nu110​81842. https://​doi.​org/​10.​1111/​dom.​13299. 50. Giugliano D, Maiorino MI, Longo M, et al. Type 2 diabetes and risk of heart failure: a systematic review and meta-analysis from cardiovascu‑ Publisher’s Note lar outcome trials. Endocrine. 2019;65:15–24. https://​doi.​org/​10.​1007/​ Springer Nature remains neutral with regard to jurisdictional claims in pub‑ s12020-​019-​01931-y. lished maps and institutional affiliations. 51. Sinha B, Ghosal S. Meta-analyses of the effects of DPP-4 inhibitors, SGLT2 inhibitors and GLP1 receptor analogues on cardiovascular death, Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions