The prevalence and incidence of heart failure (HF) and diabetes mellitus (DM) have progressively increased in the last decade.1,2 Type 2 DM is considered an independent factor for the development of HF3 and is in turn associated with a worse HF prognosis.4,5 In patients hospitalized due to HF, DM is associated with a longer hospital stay, higher readmission rate,6,7 and greater associated comorbidity.7,8 However, the role of DM in in-hospital and long-term mortality in patients with HF remains controversial.9,10 Data from the National Registry of Patients with Heart Failure (RICA, for its initials in Spanish) show that patients with type 2 DM have higher readmission rates due to HF and higher long-term mortality compared to those who do not have the disease, though in-hospital mortality appears to be equal in both groups.4

The neurohormonal treatment that have become in recent years has improved the prognosis of patients with HF.11 Two groups of hypoglycemic drugs—sodium-glucose cotransporter 2 inhibitors (SGLT2i) and glucagon-like peptide-1 agonists (GLP-1ra)—have demonstrated cardiovascular benefits in patients with and without type 2 DM.12–27 The most recent European guidelines on DM28 recommend these drugs as initial treatment in patients with high or very high cardiovascular risk who are not yet receiving metformin.

Metformin is a classic drug in type 2 DM treatment.29,30 Given the evidence that it decreases macrovascular complications,31 the guidelines recommend it as the first treatment option in most patients. However, its effect on the prognosis of patients with HF has yet to be investigated. The principal aim of this study was to analyze whether patients with DM and acute HF who receive treatment with metformin at discharge have a better prognosis at one year of follow-up.

Of the 10,122 patients included by combining the EAHFE and RICA databases, 4403 patients had type 2 DM, 43.5% of the total. The subgroup analysis was conducted based on whether or not they received treatment with metformin at discharge from the index decompensation episode (Fig. 1).

Figure 1.

Patient flowchart.

Acronyms: Epidemiology of Acute Heart Failure in Emergency Departments (EAHFE) Registry; National Registry of Patients with Heart Failure (RICA, for its initials in Spanish) Registry; Diabetes Mellitus (DM).

(0.11MB).

Table 1 describes the characteristics of the group of patients with type 2 DM and both subgroups. The 1453 patients who received treatment with metformin after hospital discharge represent 33% of all patients with type 2 DM. These patients were younger (77 vs 79 years, P < 0.001), had less comorbidity, and the prevalence of ischemic heart disease (29% versus 34%, P < 0.011) and chronic kidney disease (24% versus 50%; P < 0.01) was significantly lower. In addition, they have a better functional status evidenced by a lower percentage of NYHA III-IV functional class (35% versus 29%; P < 0.001) and higher Barthel Index scores (from 84 versus 79; P < 0.01). The drugs that both groups received for HF treatment are shown in the table. In regard to DM treatment, the metformin group received associated oral antidiabetics more often and there was a lower percentage of insulin use (Table 1).

After 12 months of follow-up (Fig. 2), it was observed that patients with metformin upon discharge had significantly lower mortality (22% versus 32%, log-rank test P < 0.001).

Figure 2.

Ratio of metformin in different patient subpopulations.

(0.14MB).

Univariate and multivariate analyses were performed on mortality at one-year of follow-up (Table 2). In the adjusted analysis, the patients who had greater mortality at one-year of follow-up were older (RR 1.025; 95% CI 1.018–1.033; P < 0.001), more often had associated comorbidity such as ischemic heart disease (RR 1.155; 95% CI 1.020–1.307; P = 0.023) or chronic kidney disease (RR 1.360; 95% CI 1.210–1.529; P < 0.001), and had worse functional class (NYHA functional class III–IV) (RR 1.280; 95% CI 1.139–1.438; P < 0.001). In this analysis, treatment with ACEi (RR 0.839; 95%CI 0.736–0.956; P < 0.01), ARB (RR 0.766; 95%CI 0.668–0.878; P < 0.001), beta blockers (RR 0.822; 95% CI 0.731–0.925; P < 0.01), and metformin were protective factors.

The results of the forest plot are shown in Fig. 2. As can be seen, metformin maintains its protective effect on mortality in most subgroups except for the SGLT2i, sulfonylureas, and medical history of hypertension groups.

Patients who receive treatment with metformin have significantly lower mortality at one year of follow-up (RR 0.814; 95% CI 0.712–0.930; P < 0.01) regardless of the rest of variables (Fig. 3).

Figure 3.

Differences in all-cause mortality at one year in patients with type 2 DM and heart failure who are prescribed metformin at discharge compared to those who do not receive treatment with this drug.

(0.31MB).

First, it is noteworthy that metformin was prescribed at discharge in just one-third of patients with DM, less than what is to be expected according to clinical practice guidelines (CPG) recommendations. Patients from the EAHFE and RICA registries are cared for in HED or the internal medicine hospitalization ward and tend to have a more geriatric profile than the population included in randomized clinical trials (RCT); namely, they are older, frailer, and have more comorbidities. This difference in the population is a possible explanation for the low prescribing rate in our study. Other population studies have also reported a significant percentage of underuse of drugs such as metformin, GLP-1ra, and SGLT2i.39 They attribute it to the significant percentage of older adult patients in the Spanish population, whose particular characteristics may limit the use of some drugs.

However, the underuse of metformin may also be explained by adherence to CPG or due to fear of adverse effects (AE). This drug is contraindicated when the glomerular filtration rate (GFR) is <30 mL/min/1.73 m2 and it is recommended to suspend it in severe patients with risk of acute renal failure or metabolic acidosis. The fear of developing AE, in particular lactic acidosis, is the main reason for the tendency toward systematically suspending this drug in hospitalized patients regardless of their clinical situation, despite the fact that scientific evidence in this regard has progressed in recent years. In the past, although some studies already suggested that this relationship could be casual,40 the use of metformin in patients with decompensated or advanced HF (NYHA III-IV) was not recommended in the CPGs. Later, it was demonstrated in registries and observational studies that the risk of this complication is regardless of whether the patient takes metformin or not.41,42 The most recent CPGs on DM28 establish that metformin is safe in all phases of HF with conserved or moderately reduced renal function (GFR > 30 mL/min/1.73 m2) and entails a lower risk of mortality or admission due to HF compared to insulin and sulfonylureas. This document also rejects the risk of lactic acidosis related to this drug.41,43,44

As stated above, metformin was demonstrated to reduce cardiovascular complications in DM, decreasing total and cardiovascular-related mortality by 30%, acute myocardial infarction by 39%, and cerebrovascular accident by 41% in the first ten years following diagnosis.31 However, to date, no RCTs have been conducted that evaluate the cardiovascular effects of metformin. After pivotal studies on GLP-1ra and SGLT2i, these drugs are now included in DM guidelines28 as drugs of choice in patients with DM and very high cardiovascular risk; they are even recommended as initial treatment in some cases. Nevertheless, when these RCTs were conducted, most received concomitant treatment with metformin and its effects together with those of this drug were evaluated. The DANIsh RCT with the Met-HeFT substudy is designed to evaluate the efficacy and safety of metformin in patients with chronic HF; it is currently in phase 4. For the first time, it will provide data from an RCT on the risk of lactic acidosis.45 While awaiting its publication, the currently available scientific evidence allows for both HF11 and DM CPGs28 to recommend the use of metformin in this patient profile. The results of this study strengthen this recommendation, as patients with type 2 DM and HF in treatment with metformin had lower mortality at one year of follow-up regardless of the rest of factors. Therefore, the authors believe that prescribing this drug in this group of patients must continue to be a fundamental pillar of treatment and underprescribing due to fear of potential AE must be avoided.

Second, it should be noted that in this study, the percentage of patients in treatment with drugs from the GLP-1ra and SGLT2i groups is very small. This is due to the fact that patient inclusion ended in 2018, before their use became widespread in this patient profile. Although a larger percentage of patients with type 2 DM treated with metformin had associated SGLT2i use than the group without metformin in this study, they were still just 40 patients out of a total of 1453 patients with DM in treatment with metformin. Therefore, it is not believed that the results have been biased due to the influence of this drug group. It would be of interest to conduct this analysis again on the EAHFE and RICA cohorts after 2018, in which the percentage of patients treated with these drug groups would presumably be greater.

This study has several limitations. First, as it is a retrospective analysis, the results are correlational and do not imply causality. Therefore, they only allow for formulating hypotheses. Second, there may be selection bias because patients come from centers which voluntarily joined and because the study period was very extensive. The third limitation is related to a high degree of variability of the participating centers in terms of both structure and management. Indeed, the considerable heterogeneity of management and follow-up strategies for HF in Spain is well-known.46,47 Fourth, the patients’ diagnosis was established according to clinical criteria. Although it was confirmed with natriuretic peptides or echocardiography in most cases, there still is a possibility of diagnostic error. Fifth, assigning patients to groups with and without treatment with metformin was done based on prescribing at the time of discharge, but this aspect was not monitored during follow-up. For this reason, there may be patients who changed group.

Nevertheless, given the experience there is with this drug, we believe that the percentage of patients who may have changed group is small. Sixth, although the patient’s degree of dependence (measured using the Barthel Index) was taken into account in the adjusted models, frailty was not. This aspect has an important impact on older patients and specifically in patients with HF.48,49 Lastly, a sample size estimation was not performed in this analysis. Therefore, it is possible that beta statistical error may have occurred in some estimates. However, given that the results are based on information obtained from the two main multicentric registries of patients with HF in Spain, it is believed they are representative and potentially able to be extrapolated to the whole of the Spanish population.

In conclusion, patients with DM and HF in this study who receive treatment with metformin had lower mortality at one year of follow-up. Therefore, we believe that prescribing this drug in this patient profile must continue to be a fundamental treatment pillar and underprescribing due to fear of potential AE must be avoided.

This research has not received specific grants from agencies in the public, commercial, or non-profit sectors.

The authors declare that they do not have any conflicts of interest.