Author + information
- Received May 20, 2016
- Revision received June 17, 2016
- Accepted June 21, 2016
- Published online September 27, 2016.
- Isabelle Johansson, MD, PhD Fellowa,∗ (, )
- Ulf Dahlström, MD, PhDb,
- Magnus Edner, MD, PhDa,
- Per Näsman, PhDc,
- Lars Rydén, MD, PhDa and
- Anna Norhammar, MD, PhDa
- aCardiology Unit, Department of Medicine K2, Karolinska Institutet, Stockholm, Sweden
- bDepartment of Cardiology and Department of Medical and Health Sciences, Linköping University, Linköping, Sweden
- cCenter for Safety Research, KTH Royal Institute of Technology, Stockholm, Sweden
- ↵∗Reprint requests and correspondence:
Dr. Isabelle Johansson, Cardiology Unit, Department of Medicine, N3:06, Karolinska University Hospital Solna, SE-171 76 Stockholm, Sweden.
Background Heart failure (HF) is a common and serious complication in type 2 diabetes mellitus (T2DM). The prognosis of ischemic HF and impact of revascularization in such patients have not been investigated fully in a patient population representing everyday practice.
Objectives This study examined the impact of ischemic versus nonischemic HF and previous revascularization on long-term prognosis in an unselected population of patients with and without T2DM.
Methods Patients stratified by diabetes status and ischemic or nonischemic HF and history of revascularization in the Swedish Heart Failure Registry (SwedeHF) from 2003 to 2011 were followed up for mortality predictors and longevity. A propensity score analysis was applied to evaluate the impact of previous revascularization.
Results Among 35,163 HF patients, those with T2DM were younger, and 90% had 1 or more associated comorbidities. Ischemic heart disease (IHD) occurred in 62% of patients with T2DM and 47% of those without T2DM, of whom 53% and 48%, respectively, had previously undergone revascularization. T2DM predicted mortality regardless of the presence of IHD, with adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) of 1.40 (1.33 to 1.46) and 1.30 (1.22 to 1.39) in those with and without IHD, respectively. Patients with both T2DM and IHD had the highest mortality, which was further accentuated by the absence of previous revascularization (adjusted HR: 0.82 in favor of such treatment; 95% CI: 0.75 to 0.91). Propensity score adjustment did not change these results (HR: 0.87; 95% CI: 0.78 to 0.96). Revascularization did not abolish the impact of T2DM, which predicted mortality in those with (HR: 1.36; 95% CI: 1.24 to 1.48) and without (HR: 1.45; 95% CI: 1.33 to 1.56) a history of revascularization.
Conclusions Ninety percent of HF patients with T2DM have preventable comorbidities. IHD in patients with T2DM had an especially negative influence on mortality, an impact that was beneficially influenced by previous revascularization.
Diabetes mellitus, predominantly type 2 (T2DM), affects 390 million people globally, a number expected to increase to 600 million in 2030 (1). Cardiovascular disease is the most common complication and mortality cause related to T2DM. Recent evidence favors increasing life expectancy in patients with T2DM because of improved risk factor management and post–myocardial infarction (MI) survival (2,3). Unfortunately, these improvements are expected to increase the prevalence of chronic complications, including heart failure (HF) (4). Besides enhancing the risk of HF, T2DM has an adverse impact on the prognosis of HF (5). Consequently, an increasing proportion of patients with T2DM are expected to develop ischemic HF in the future. Diabetes is over-represented in HF populations, even in the absence of ischemic heart disease (IHD). The presence of a specific diabetes-induced cardiomyopathy unrelated to hypertension or IHD but rather caused by deranged myocardial metabolism has been proposed as an explanation (6). However, the actual prevalence and role of this condition in terms of morbidity and mortality remain to be fully understood (7–9). Moreover, existing evidence related to the true impact of T2DM on HF survival, often based on subgroup analyses of either clinical trials or registry reports, is conflicting. Studies have indicated better (9), equal (7,8), and compromised (9) survival in ischemic and nonischemic HF (10,11). These discrepant results probably reflect the recruitment of different and selected patient populations and often vaguely defined glucometabolic states.
The aim of this study was to investigate the prevalence and impact of ischemic and nonischemic HF and the role of previous revascularization in patients with T2DM in a contemporary unselected HF population that reflects everyday clinical practice.
The Swedish Heart Failure Registry (SwedeHF), introduced throughout Sweden in 2003, has been described in detail elsewhere (12). Registry information is available on the SwedHF website (13). Participating centers (65 hospitals, 113 outpatient clinics) report to the registry, which contained information on 47,000 patients in 2011. The primary inclusion criterion is a physician-judged diagnosis of HF; 76 variables are recorded via an Internet-based case report form at hospital discharge or outpatient visits. Data are entered into a database managed by the Uppsala Clinical Research Center (Uppsala, Sweden) and run monthly against the Swedish Population Registry.
Study population and design
Between January 2003 and September 2011, 36,595 patients with HF managed in a specialist setting were included in SwedeHF. The present cohort of 35,163 patients (68% from hospitals, 32% from outpatient clinics) was established after the exclusion of patients with no information on sex or glucometabolic state (n = 123) or who had type 1 diabetes mellitus (n = 198) or IHD (n = 1,111). The date of the first registration was assigned as the index date, and all descriptive data in the present analysis were extracted from this single occasion. The primary endpoint was all-cause mortality, acquired by merging the SwedeHF database with the Swedish Population Registry using the unique 10-digit personal identification number of Swedish citizens. Follow-up ended on September 13, 2011.
Definitions are based on the pre-defined definitions used in SwedeHF (Online Appendix). HF was diagnosed by the attending physician based on guideline recommendations at the time of inclusion. New York Heart Association (NYHA) functional classes I to IV were used to define HF severity. An adapted definition was used for IHD, defined as present or absent based on the case history from patient records. Patients reported as without IHD but with a confirmed coronary revascularization procedure (n = 739) or a history of previous MI (n = 218) were reclassified as having IHD (Figure 1). Revascularization was defined in SwedeHF as a history of coronary artery bypass surgery (CABG) or percutaneous coronary intervention (PCI). This procedure is freely available in Sweden, and patients are generally offered this treatment according to the current European guidelines (14). MI was based on information from patient records. T2DM was defined as a confirmed history of this diagnosis or the prescription of glucose-lowering drugs. Comorbidities were defined as the presence of any or several of the following: hypertension, atrial fibrillation, pulmonary disease, valvular heart disease, or idiopathic dilated cardiomyopathy, all of which were classified as “yes” or “no” based on patient records. Left ventricular ejection fraction (LVEF) was the most recently estimated, grouped into 4 classes: ≥50%, 40% to 50%, 30% to 39% and <30%. Estimated glomerular filtration rate (eGFR) was calculated with the MDRD formula, and an eGFR <60 ml/min1/1.73 m2 was considered the cutoff for chronic kidney disease (CKD).
Statistical comparisons of differences between independent groups were performed with the Student t test for continuous variables. Quantitative normally distributed variables are presented as the mean (standard deviation) and 95% confidence intervals (CIs) or, when appropriate, median, range, and 95% CI. Natriuretic peptides were logarithmically transformed before Student t testing. Categorical variables were compared by chi-square tests and are presented as counts and proportions (%). Analyses of differences between those with and without T2DM were performed in subgroups by IHD status and further in the IHD population in those with and without previous revascularization. Statistical differences for all-cause mortality by T2DM, IHD status, and revascularization were estimated with Cox proportional hazard regression and depicted as crude and age-adjusted estimated survival curves. Univariate and multivariate models were used to evaluate the predictive value of T2DM, as well as revascularization, in all-cause mortality. Adjustments for potentially important covariates were performed in 2 models. Model 1 included age, sex, HF duration, level of care, weight, systolic and diastolic blood pressure, LVEF class, eGFR class, hemoglobin class, T2DM, hypertension, atrial fibrillation, pulmonary disease, and use of angiotensin-converting enzyme inhibitors, angiotensin receptor II blockers, beta blockers, mineralocorticoid receptor antagonists, diuretic agents, digitalis, nitrates, statins, and antithrombotic agents. Model 2 included the same variables as model 1 plus 2 additional variables (heart rate and NYHA functional class) that were excluded from model 1 because of a large number of missing values. In addition, because of the observational character of the investigation, a propensity score model was applied to avoid potential bias regarding the impact of previous revascularization on all-cause mortality. The propensity score, which expressed the probability of an assigned treatment, in this case revascularization, given a set of known baseline characteristics, was used to balance the study population with regard to a chosen dependent variable (15). Logistic regression was used to estimate individual propensity scores for the history of revascularization in patients with IHD and T2DM (n = 5,182) with a good fit (Hosmer and Lemeshow test; p = 0.27 [p > 0.05 is considered a good fit] and c-statistic of 0.7) based on 26 baseline variables (including demographics, medical history, and reported pharmacological treatment) (15). The selected variables were those that affected all-cause mortality in univariate logistic regression given they had a reasonably low amount of missing data. An individual propensity score was estimated for 3,467 patients with complete information on all variables. The impact of previous revascularization on all-cause mortality was thereafter determined by means of Cox regression adjusted for the propensity score.
A 2-sided probability value of p < 0.05 was considered significant and is reported with 95% CI. Analyses were performed with SAS statistical software (version 9.4).
The Swedish National Board of Health and Welfare and the Swedish Data Inspection Board approved the establishment of SwedeHF and subsequent patient registration and data analyses. The registry and this study conform to the Declaration of Helsinki. Individual patient consent was not required or obtained, but patients were informed about registry entry with permission to opt out. The Regional Ethical Review Board at Linköping University approved the merging of SwedeHF with the Swedish Population Registry.
Important clinical characteristics are presented in Table 1. Among 35,163 patients, IHD was reported in 51%. The proportion of IHD in the T2DM cohort was 62% compared with 47% in those without T2DM, of whom 53% and 48%, respectively, had undergone previous revascularization.
Patients with T2DM with and without IHD
When we compared T2DM patients with and without IHD, the former were slightly older (75 years vs. 74 years), more often men (66% vs. 57%), and more frequently had reduced LVEF (<50%: 83% vs. 69%) and CKD (63% vs. 55%). Smoking (current/former) was more common in T2DM patients with IHD (62% vs. 53%), whereas associated comorbidities were less common: atrial fibrillation (40% vs. 51%), idiopathic dilated cardiomyopathy (7% vs. 12%), hypertension (59% vs. 60%), and pulmonary disease (18% vs. 20%). Eighty-eight percent of the patients with T2DM without IHD had at least 1 comorbidity. This increased to 90% when idiopathic dilated cardiomyopathy was added (Table 2). Patients with and without IHD underwent similar pharmacological treatment except for a more frequent use of statins and acetylsalicylic acid among those with IHD.
Patients with IHD in relation to revascularization
Baseline data for patients with IHD according to previous revascularization and the presence or absence of T2DM are outlined in Table 3. When we focused on patients with T2DM and compared those with and without previous revascularization, the latter were older (mean age 77 years vs. 73 years) and more often women (43% vs. 26%). Comorbidities, such as atrial fibrillation and pulmonary disease and preserved LVEF (16% vs. 19%), were less common in T2DM patients who had been revascularized than in those who had not. Patients without previous revascularization were more often prescribed diuretic agents and digitalis but less often given renin-angiotensin-aldosterone system inhibitors, beta blockers, and statins.
A corresponding comparison of baseline data between patients without T2DM revealed a similar pattern, although in slightly different proportions from the T2DM cohort.
Patients with and without T2DM
Among patients with IHD (Table 1), those with T2DM were younger (75 years vs. 77 years) and had a higher prevalence of hypertension (59% vs. 45%) and CKD (63% vs. 56%). The proportion with preserved LVEF (≥50%) was similar in both groups (17% vs. 18%), but the IHD patients with T2DM were more symptomatic (NYHA functional class III or IV; 54% vs. 46%). In general, patients with T2DM had more extensive pharmacological treatment with renin-angiotensin-aldosterone system inhibitors (84% vs. 79%), diuretic agents (88% vs. 79%), and statins (67 vs. 59%). About 50% of the IHD patients had a history of revascularization, which was more common in patients with T2DM than without (53% vs. 48%).
In patients without IHD (Table 1), the mean age was 73 years regardless of T2DM state. The comorbidity pattern between those with and without T2DM corresponded with that in patients with IHD, except that preserved LVEF (≥50%) was more common in patients without IHD than among those with this disease and more common in those with T2DM than in those without (31% vs. 26%).
The time of follow-up ranged between 0 and 8.7 years (median 1.9 years). By the end of the study period, 14,144 patients (40%) had died, 3,950 (46%) with T2DM and 10,194 (38%) without. Estimated survival curves among patients with and without T2DM and with and without IHD, before and after age adjustment, are shown in Figure 2 and Central Illustration (left), whereas crude and adjusted survival rates in patients with IHD by previous revascularization are shown in Figure 3 and Central Illustration (right). The most serious prognosis was seen in patients with IHD and T2DM (Figure 2), especially those without previous revascularization (Figure 3).
Hazard ratios (HRs) of mortality from univariate and multivariate analyses are presented in Table 4. T2DM predicted mortality regardless of the presence or absence of IHD (unadjusted HR: 1.20 [95% CI: 1.14 to 1.25] and 1.21 [95% CI: 1.14 to 1.29], respectively) and remained a mortality predictor in patients with and without IHD after adjustment (HR: 1.40 [95% CI: 1.33 to 1.46] and 1.30 [95% CI: 1.22 to 1.39], respectively). This pattern persisted in the different adjustment models, with a slight weakening of the HRs when the number of covariates increased (Table 4). T2DM remained an independent mortality predictor among revascularized IHD patients (adjusted HR: 1.36; 95% CI: 1.24 to 1.48), as well as in IHD patients without previous revascularization (HR: 1.45; 95% CI: 1.33 to 1.56).
Among patients with T2DM and IHD, previous revascularization was associated with a decreased mortality risk compared with T2DM patients without a history of any coronary intervention (adjusted HR: 0.82; 95% CI: 0.75 to 0.91). This association remained after adjustment for the propensity score of previous revascularization (HR: 0.87; 95% CI: 0.78 to 0.96). Likewise, previous revascularization was associated with better prognosis within the non-T2DM cohort (adjusted HR: 0.89; 95% CI: 0.83 to 0.96).
There are 3 findings of major importance in this study of a large, contemporary HF population representing patients receiving typical (“everyday”) clinical care. First, T2DM is associated with higher all-cause mortality, regardless of whether HF is of ischemic or nonischemic origin. In addition, previous revascularization was associated with improved survival, which highlights how important it is to always consider the possibility of a coronary intervention in patients with T2DM and IHD. Finally, the vast majority of patients with T2DM and HF of nonischemic origin have 1 or more manageable comorbidities known to cause or influence HF.
There are 2 reasonable explanations for the low use of revascularization in patients with T2DM. One is that the risk of this intervention might have been considered high because of the large comorbidity burden, and another is that T2DM often causes a more diffuse atherosclerosis with challenging coronary artery stenoses, which lessens the opportunity or even willingness to perform a PCI or CABG. Still, it is unlikely that these conditions can explain a proportion as large as 50% of eligible patients, that is, patients with T2DM with HF of ischemic origin. Some discrepancy exists in the evidence that supports revascularization as a tool to improve prognosis in T2DM and coronary artery disease. The BARI 2D (Bypass Angioplasty Revascularization Investigation 2 Diabetes) trial failed to show any survival benefit through revascularization compared with optimal medical therapy in patients with T2DM and coronary artery disease (16). These findings, however, should be interpreted bearing in mind that all patients had undergone coronary angiography before inclusion. Those who fulfilled prevailing criteria for immediate revascularization were excluded, which left only those with mild to moderate disease eligible for BARI 2D. In addition, the neutral results might have been driven by a 49% crossover frequency from medical therapy to revascularization during the study period. Convincing evidence favoring revascularization to improve prognosis in patients with T2DM and acute coronary syndromes and multivessel disease is provided by the FRISC II (Fragmin and Fast Revascularization During Instability in Coronary Artery Disease) and FREEDOM (Future Revascularization Evaluation in Patients With Diabetes Mellitus: Optional Management of Multivessel Disease) trials (17,18), although neither of these trials addressed the outcome of revascularization specifically in relation to the presence of HF. With regard to HF in general, a meta-analysis of 3,088 patients with left ventricular dysfunction and signs of preserved myocardial viability found a strong association between revascularization and survival (19). Unfortunately, this analysis did not report on the proportion of patients with diabetes. On the basis of general knowledge of T2DM prevalence in a population with coronary artery disease, it can be assumed that at least 20% to 30% of the studied population could have this disease (20,21). The concept of revascularization in patients with established HF has been challenged by the results of the STICH (Surgical Treatment for Ischemic Heart Failure) trial, which randomized 1,212 patients with IHD and left ventricular dysfunction to optimal medical therapy alone or optimal medical therapy plus CABG. Unexpectedly, there was no difference in all-cause mortality (22), but as emphasized, there are several critical issues related to this trial (23). One important limitation is the considerable proportion of medically treated patients who crossed over to revascularization during the 5 years of follow-up. That this undermined the power of the observations is highlighted by a renewed as-treated rather than intention-to-treat analysis of the STICH database, which showed that CABG reduced mortality compared with medical therapy alone (24). Indeed, a recently published 10-year follow-up of the STICH study (STICHES [STICH Extension Study]) found CABG to be significantly superior to medical therapy alone when analyzed according to intention to treat, which provides a strong evidence base for the benefits of such intervention in HF (25). In addition, the results of 2 studies that applied propensity score analyses strengthened the benefit of CABG over medical therapy in patients with IHD and left ventricular dysfunction (26,27). Nonetheless, there is a gap in knowledge concerning demand for prospective clinical trials to assess myocardial viability with cardiac magnetic resonance imaging or positron emission tomography and the actual impact of revascularization in patients with T2DM and ischemic HF (14,28). It is hoped that the ongoing REVIVED (Revascularization for Ischemic Ventricular Dysfunction) trial studying the efficacy and safety of PCI in systolic HF will provide further insights in this field. Until results from this and similar trials are available, even a cautious interpretation of the present observational analysis of a large HF population underscores the potential benefit of revascularization and the need to at least offer all patients a thorough investigation by means of coronary angiography or computed tomography angiography.
The finding that the vast majority (≈90%) of patients with T2DM and HF of nonischemic origin had at least 1 comorbidity known to cause HF has important clinical implications. These comorbidities included hypertension, atrial fibrillation, pulmonary disease, and valvular heart disease, all conditions largely manageable with well-established therapeutic or preventive strategies. Taking into consideration all T2DM patients in the present contemporary HF population leaves only 7% to 10% of them without a reasonable pathogenesis. This makes it tempting to question the existence of a pure diabetes cardiomyopathy that by definition requires signs of left ventricular dysfunction in patients with T2DM without any obvious cause of the condition, such as coronary artery disease, hypertension, or valvular heart disease (29). Previous data reported on deranged myocardial metabolism in HF related to T2DM with increased myocardial energy production via beta oxidation of free fatty acids (6). More current studies, using modern imaging methods, have reported on structural changes in the myocardial tissue of HF patients, with increased diffuse fibrosis enhancing myocardial stiffness and compromising relaxation (6,30). These changes might be more frequent in patients with diabetes as a consequence of high glucose and possibly also high insulin levels; however, they could also be explained by long-standing, poorly controlled hypertension or merely a result of aging. Another assumption behind the increased HF prevalence and poor prognosis in patients with T2DM is that the group judged clinically to have nonischemic HF might have a compromised myocardial function secondary to silent ischemia or atherosclerosis in small vessels, causing hibernation or stunning, as well as scar tissue resulting from silent MIs. Some observational studies report that undiagnosed ischemia is more common among patients with diabetes than among those without and that it has adverse prognostic implications (31,32). Further evidence for these pathogenic conditions is not provided by the present investigation, which is only able to indicate opportunities for further studies. These investigations need to apply advanced imaging methods to well-defined diabetes populations to further enhance potential mechanisms for preventing or treating HF, thereby improving the dismal prognosis in patients with the combination of T2DM and HF.
Lastly, to put our observations in the context of large clinical trials, compared with the placebo-treated group in a recent outcome trial focused on ischemic HF in patients with T2DM, the EMPA-REG trial (Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients) (33), mortality in patients with T2DM and ischemic HF throughout the trial period and median follow-up of almost 2 years was considerably higher in the present study (50%) than in the EMPA-REG trial population (15%). Although such comparison must be made with great caution because of obvious selection biases in the trial participants (e.g., younger, fewer comorbidities), it underlines the serious prognosis in an everyday population with T2DM and ischemic HF.
Study strengths and limitations
The major strength of the present report is the size and unselected nature of the population, which reflects contemporary, everyday clinical practice. One limitation with registry data is the possibility of varying diagnostic criteria. The registry defines the different variables, but they have not been validated in any detail. The proportion with T2DM is probably an underestimation, because undiagnosed diabetes is common in populations with cardiovascular diseases (20,34). With regard to IHD in SwedeHF, the definition was expanded in an attempt to cover as many patients as possible (Figure 1), but information on patient history at the time of inclusion was sometimes incomplete. In addition, no information was available regarding silent angina, which raises the possibility of misclassification of IHD. This also applies to revascularization, and because it was retrospectively collected and not randomized, the assumptions made regarding this treatment should be seen as hypothesis generating rather than representing facts. However, the large database allowed for extensive adjustment of clinically relevant confounders when evaluating the outcome. Further on the adjustment, the application of a propensity score for treatment bias (which to some degree mimics a randomization) strengthened the assumptions of the benefits of revascularization. It should be acknowledged that small differences detected in the descriptive analyses between different groups become statistically significant because of the large sample size. These statistical significances might not always be of clinical relevance, and this necessitates a cautious interpretation of the data. Because of missing data on important diabetes variables, including hemoglobin A1c, it was not possible to extend prognostic analyses related to different levels of glucose control. Moreover, we did not have details about diabetes duration or classes of glucose-lowering therapy, factors with known prognostic implications. A median follow-up of 1.9 years might appear short, but it is mainly an effect of a larger proportion being included during the most recent period and that the follow-up period was generally fairly long.
In a contemporary setting, almost 90% of HF patients with T2DM have an accompanying risk factor that can contribute to HF development, among which IHD is the most common. The presence of IHD was associated with the worst prognosis, which was, however, less severe if revascularization had been performed previously.
COMPETENCY IN MEDICAL KNOWLEDGE: T2DM is associated with a higher risk of mortality in patients with HF, whether of ischemic or nonischemic origin. In patients with T2DM and ischemic HF, previous revascularization is associated with improved prognosis compared with patients without revascularization. Most patients with nonischemic HF have other comorbidities associated with HF.
TRANSLATIONAL OUTLOOK: Randomized trials are needed to validate these observational data and confirm the impact of revascularization in patients with T2DM and ischemic HF.
For an expanded Methods section, please see the online version of this article.
This work was supported by unrestricted grants from the Swedish Heart-Lung Foundation and the Regional Agreement on Medical Training and Clinical Research (ALF) between Stockholm County Council and the Karolinska Institute. Dr. Dahlström has received research grants from Linkoping University and AstraZeneca Inc.; and honoraria from expert group participation organized by different pharmaceutical companies, none of which were directly related to the present publication. Dr. Rydén has received research grants from the Swedish Heart-Lung Foundation, AFA Insurance, and the Swedish Diabetes Foundation; honoraria from expert group participation; and personal fees for delivering educational lectures organized by pharmaceutical and societal organizations, none of which were directly related to this publication. Dr. Norhammar has received funding from the Swedish Heart-Lung Foundation and the Swedish Diabetes Foundation for the present report; and has received honoraria for advisory boards and lectures from Eli Lilly, AstraZeneca, MSD, and Boehringer Ingelheim. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery bypass surgery
- confidence interval
- chronic kidney disease
- estimated glomerular filtration rate
- heart failure
- hazard ratio
- ischemic heart disease
- left ventricular ejection fraction
- myocardial infarction
- New York Heart Association
- percutaneous coronary intervention
- Swedish Heart Failure Registry
- type 2 diabetes mellitus
- Received May 20, 2016.
- Revision received June 17, 2016.
- Accepted June 21, 2016.
- 2016 American College of Cardiology Foundation
- ↵International Diabetes Federation. IDF Diabetes Atlas. Brussels, Belgium; 2013. Available at: http://www.idf.org/diabetesatlas. Accessed May 15, 2016.
- Dries D.L.,
- Sweitzer N.K.,
- Drazner M.H.,
- Stevenson L.W.,
- Gersh B.J.
- Domanski M.,
- Krause-Steinrauf H.,
- Deedwania P.,
- et al.,
- BEST Investigators
- De Groote P.,
- Lamblin N.,
- Mouquet F.,
- et al.
- ↵Swedish Heart Failure Registry (RiksSvikt) website. Available at: http://www.ucr.uu.se/rikssvikt-en/. Accessed January 11, 2016.
- Kolh P.,
- Windecker S.,
- Alfonso F.,
- et al.
- Rosenbaum P.R.,
- Rubin D.B.
- Norhammar A.,
- Malmberg K.,
- Diderholm E.,
- et al.
- Allman K.C.,
- Shaw L.J.,
- Hachamovitch R.,
- Udelson J.E.
- Gyberg V.,
- De Bacquer D.,
- Kotseva K.,
- et al.,
- EUROASPIRE IV Investigators
- Gurunathan S.,
- Ahmed A.,
- Senior R.
- Doenst T.,
- Cleland J.G.,
- Rouleau J.L.,
- et al.,
- STICH Investigators
- Ryden L.,
- Grant P.J.,
- Anker S.D.,
- et al.
- Ugander M.,
- Oki A.J.,
- Hsu L.Y.,
- et al.
- Zethelius B.,
- Johnston N.,
- Venge P.
- Fitchett D.,
- Zinman B.,
- Wanner C.,
- et al.,
- EMPA-REG OUTCOME Trial Investigators
- Bartnik M.,
- Ryden L.,
- Ferrari R.,
- et al.,
- Euro Heart Survey Investigators