Heart failure (HF) is a prevalent condition associated with high mortality and morbidity and reduced quality of life.1 Comorbidities are frequent among HF patients, leading to increased hospitalization rates and mortality. The prevalence of comorbidities is higher in patients with more severe signs of HF.2 According to international reports, one-year mortality and re-hospitalization rates for HF patients are 17% and 44%, respectively.3,4 HF is the most common reason for hospitalization in people >65 years in economically developed countries.5

Little is known about the long-term prognosis and risk factors for a fatal outcome in patients hospitalized with acute HF (AHF).6 The increasing incidence and associated morbidity and mortality of AHF mean there is an urgent need to better understand this population.7 Risk stratification of HF patients is important to identify high-risk patients who may benefit from advanced treatments. Besides improving risk stratification, identifying new prognostic markers may also provide information about the pathophysiology of HF and support the clinical management of HF patients and decision-making as to the best therapeutic strategy.

This study reports the impact on one-year mortality of comorbidities and biochemical parameters of patients hospitalized for HF during one-year at an Internal Medicine Department of a large tertiary university hospital in Porto, Portugal.

This study reports the impact of multiple comorbidities and biochemical parameters on the one-year mortality rate in patients hospitalized for HF in an Internal Medicine Department. After adjusting the multivariable analysis model for all clinically relevant characteristics, the independent predictors of higher one-year mortality risk at admission were: age ≥80 years, active cancer, dementia, high urea levels, high RDW levels, and low PDW levels.

The characteristics of this study population were different from those of clinical studies addressing angiotensin-neprilysin inhibition in stabilized hospitalized patients with HF with reduced ejection fraction (HFrEF).10,11 Our study population was older and predominantly female, with a higher prevalence of hypertension and a lower prevalence of ischemic etiology. This may be explained by the fact that our study included HF patients with and without reduced ejection fraction, which may preclude direct comparison with other studies. As previously discussed, the sex distribution of this study population reflects the fact that elderly patients hospitalized with HF are mainly women.8

The in-hospital and one-year all-cause mortality rates found in this study, 7.9% and 34.3%, respectively, are in accordance with those reported previously in the European database and registry studies,4,5,12 as well as in a Portuguese study.13 This observation confirms the general burden of HF, as expressed by the high one-year mortality rate, ruling out the hypothesis of poor clinical management of HF at local or national levels. Of note, we observed relatively low mortality in the first 30 days after discharge (3.5%). This may be due to the longer hospitalization stays for AHF patients (mean of 10 days in Europe vs. five days in USA), which gives the clinical teams more time to stabilize patients and thereby optimize the therapeutic strategy before hospital discharge.

These findings support the notion that there is still much room for improvement in the standard of care for AHF patients. The current international guidelines for the clinical management of HF need to be further implemented in order to promote better outcomes for HF patients. Under this rationale, risk stratification of HF patients using predictors of increased risk of mortality is a vital tool for ensuring that the most appropriate healthcare plan is implemented for each patient (e.g., frequency of follow-up appointments, assignment to a telemonitoring program, among others).

Other studies have examined the contribution of multiple non-cardiac comorbidities to the mortality of patients with HF.2,14–26 In a pioneer study conducted in 2014, van Dreusen et al. analyzed non-cardiac comorbidities of patients included in the observational Heart Failure Pilot Survey (ESC-HF Pilot) and found that diabetes, chronic kidney disease, and anemia were independent predictors of all-cause mortality risk in a multivariable analysis.2 Since then, several reports, including sub-analyses from clinical studies, have confirmed this association, particularly for chronic kidney disease.19–24 Other predictors for all-cause mortality were also found, such as chronic obstructive pulmonary disease (COPD) and peripheral artery disease.25,26 We could not find any such association in our data, probably due to differences in the baseline characteristics of the population studied. In fact, compared to the pioneer study of Dreusen et al., our study population was older, with more females and fewer patients with HFrEF.2,8

In line with our findings, van Dreusen et al. did not identify sleep apnea as a predictor of mortality. However, in both studies, sleep apnea was not actively investigated; either it was mentioned on the clinical records or self-reported by patients. This may explain the contradiction with studies demonstrating that sleep apnea doubles the mortality risk of HF patients.27 There has also been an increase in the number of studies advocating the importance of sleep-related disorders in patients with cardiovascular disease.28,29 It is of note that van Dreusen et al. did not include dementia and cancer in their analysis.

In the ESC-HF Pilot study, the multivariable analysis found that predictors of one-year mortality in patients hospitalized for HF were: older age, high creatinine, low sodium, low systolic blood pressure, low EF, and other factors not included in our study.4 An Italian observational study reported similar findings.15 However, only older age had a similar result in our analysis. Likewise, we did not find a lower mortality risk in females or patients with higher LVEF, as reported elsewhere,9 despite the higher use of beta-blockers and spironolactone in patients with lower LVEF.8 These differences are probably due to the fact that our patients were all recruited after hospitalization for AHF, unlike those of the other studies, which were primarily outpatients from cardiology centers.

Emerging evidence suggests a relationship between cardiovascular disease and cancer.30,31 HF and cancer are becoming increasingly more prevalent as the population ages, and both conditions are associated with significant mortality and morbidity. Accordingly, our findings that patients admitted for HF, those >80 years, and those with active cancer are more likely to die in the next year, were expected.

Cognitive impairment has a prevalence of 28–58% among HF patients, but little is known about mortality in patients with both diseases.32 Braunstein et al. reported that Alzheimer's disease/dementia is associated with higher mortality risk.14 These observations are in line with the findings of our study since, among the population studied, mortality risk increased with age, HF, and dementia. In fact, as demonstrated by the multivariable analysis, dementia is a robust independent predictor of mortality risk. This finding calls for a dramatic change in the clinical management of dementia patients.

Impairment in renal function is associated with significant morbidity and mortality in patients with chronic HF.2,4,33 We found urea levels to be a strong predictor of one-year mortality. Indeed, blood urea nitrogen (BUN), another way of measuring urea, may increase independently of glomerular filtration rate (GFR) due to enhanced proximal and distal tubular reabsorption under HF neurohormonal activation.33,34 This is especially important in patients with decompensated HF, in whom this mechanism plays a notable role in the clinical manifestations of the disease. Considering these mechanisms, BUN may be a better prognostic indicator than estimated GFR.32 In fact, BUN has been established as a valid marker in HF mortality prediction risk models, both for in-hospital mortality and mortality during the first year after HF hospitalization.35,36 Klein et al. also concluded that high BUN levels at admission are associated with poorer survival rates in HF patients.33

Red cell distribution width is a measure of the variability in the size of circulating erythrocytes and has been linked with a higher mortality risk in both HF patients and the general population.37–39 Our results are consistent with those of previous reports, demonstrating a strong association between high levels of RDW and one-year mortality in patients with AHF, independently of other hematologic parameters, such as hemoglobin and hematocrit levels.37 Remarkably, RDW in the fourth quartile (>16.3%) at admission was found to be a very strong predictor of one-year mortality. Since the evaluation of RDW levels consists of a simple, inexpensive, and widely available test, this finding may have substantial implications for the risk stratification and clinical management of patients with AHF.

Platelet distribution width levels are a widely available platelet activation marker. Very few studies have analyzed the possible effect of PDW on mortality risk in general or specifically in HF patients. The available data suggest that the risk increases with higher values of PDW.40,41 However, we have found a contradictory result in this study, with lower PDW levels independently predicting higher one-year mortality. Thus, PDW may be an important prognostic marker in HF, but its role remains to be clarified in more extensive studies.

Our study has several limitations: (i) it is a retrospective study, (ii) it does not determine the presence and severity of chronic kidney disease, (iii) it does not determine the type, stage, or treatment for oncologic patients, and (iv) it does not examine the use of cardiac devices. Regarding this last point, it is important to stress that it was unlikely that a significant number of this study's patients had implanted such devices, considering their age and that only one-third of them had LVEF ≤40%. Moreover, it is possible that other factors with an impact on mortality have not been included in the analysis. Nevertheless, the findings of this study are sufficiently robust to prompt an improvement in the clinical management of HF patients and contribute to better health outcomes for this population.

The predictors of one-year mortality in a large cohort of patients hospitalized for HF in an Internal Medicine Department are: age ≥80 years, active cancer, dementia, high values of urea, and RDW at admission. The knowledge of such predictors allows clinicians to tailor both the treatment and follow-up of patients at higher risk of death. These data can be used to establish inexpensive and powerful lab results, such as RDW and urea, as predictors to choose not only the place where the treatment will be held (e.g., cardiac and intensive care units) and the level of monitoring during hospitalization, but also to guide patients to specialized HF clinics and to an earlier post-discharge evaluation. Further studies are needed to corroborate our findings.

Novartis Pharma supported the statistical analysis and the medical writing through an unrestricted grant.

L.T. and D.M. performed the statistical analysis, I.M. and R.L.R. prepared the manuscript, and all authors reviewed and approved the final version of the manuscript. All authors vouch for the accuracy and completeness of the data and analysis.

The authors have no conflicts of interest to declare.