All consecutive ambulatory patients referred to a structured HF clinic of a university hospital from 1 August 2001 to 31 December 2012, regardless of etiology, were included in an outpatient setting. The criteria for clinical practice referral to the HF unit have been reported elsewhere.20,21 Briefly, the criteria were HF with at least 1 hospitalization and/or reduced left ventricular ejection fraction (LVEF) < 40%. Most patients were referred from cardiology and internal medicine departments, and a few were from the emergency room/short-stay unit or other hospital departments. Less than 10% of patients were admitted to the HF unit for asymptomatic reduced LVEF after acute myocardial infarction. Two patients were excluded due to valvular surgical repair between baseline and 6 months.

All patients were seen regularly during follow-up visits at the HF clinic according to their clinical needs. Follow-up visits included a minimum of 1 visit from a nurse every 3 months and 1 visit from a physician (cardiologist, internist, or family physician) every 6 months, as well as optional visits from specialists in geriatrics, psychiatry, and rehabilitation.20,21 During the baseline visit, patients provided written consent for analytical samples and the use of their clinical data for research purposes.

The study was performed in compliance with the law protecting personal data in accordance with the international guidelines on clinical investigation of the World Medical Association's Declaration of Helsinki.

The main outcome was death from all causes. The number and causes of death during follow-up were obtained from clinical records at the HF unit, other hospital departments, other hospital records, or by contacting the patient's relatives. Data were verified using the databases of the Catalan and Spanish health systems. Five patients were lost during follow-up and were adequately censored in the survival analysis.

Hemoglobin was determined at first visit and at 6 months. Anemia was defined according to World Health Organization (WHO) criteria (hemoglobin < 13g/dL for men and < 12g/dL for women). Patients were classified relative to their hemoglobin values as nonanemic (both measurements normal), transiently anemic (anemic at the first visit but not at 6 months), newly anemic (nonanemic initially but anemic at 6 months), or persistently anemic (anemic in both measurements). Resolution of anemia was defined as normalization of hemoglobin levels (≥ 13g/dL for men and ≥ 12g/dL for women).

Categorical variables are expressed as frequencies and percentages. Continuous variables are expressed as mean±standard deviation or as median [interquartile range] for cases with skewed distribution. Normal distribution was assessed with normal Q-Q plots. Statistical differences between groups were assessed using the chi-square test for categorical variables, Student t test for continuous variables with normal distribution, or the Mann-Whitney U test for nonnormal distributions. Univariate Cox proportional hazards regression analysis was performed using all-cause mortality as the dependent variable and the anemia prespecified subgroup as the independent variable. Proportional assumptions needed to use Cox proportional hazard regression models were tested for all variables. To fulfil the assumption of linearity, the logarithmic function of the covariable HF duration was used. A multivariable Cox proportional hazards model was also created with the same dependent and independent variables and different covariables were included in the model due to their significance in the univariate analysis or because they were considered clinically significant: age, sex, New York Heart Association (NYHA) functional class, duration of HF, etiology of HF, LVEF, diabetes mellitus, hypertension, atrial fibrillation, renal failure, chronic obstructive pulmonary disease, peripheral vasculopathy, heart rate, systolic blood pressure, and treatments (backward stepwise). Statistical analyses were performed using SPSS 15 (SPSS Inc.; Chicago, Ilinois, United States). A 2-sided P < .05 was considered statistically significant.

At baseline, anemia was present in 550 patients (46.9%). One quarter of anemic patients (14.5% of total patients) had normalized hemoglobin levels at 6 months (transient anemia), and a similar proportion of patients had developed new anemia (12.5% of the cohort) at 6 months, while 32.4% of patients were persistently anemic (Figure 1). Patients with any kind of anemia (n = 697, 59.4%) were older (P < .001), had a lower glomerular filtration rate (Chronic Kidney Disease Epidemiology Collaboration formula) (P < .001), were in a worse functional class (P < .001), and had a lower duration of HF (Table 1). No significant differences were found between groups regarding the etiology of HF or associated comorbidities. On the other hand, the use of an angiotensin-converting enzyme inhibitor (P < .001) and beta-blockers (P = .03) was lower among patients with anemia during follow-up (Table 1). Table 2 shows the differences in demographic and clinical characteristics among anemic patients based on the prespecified categories of anemia. Remarkably, patients with persistent anemia were older and had a worse functional class, better LVEF, and lower estimated glomerular filtration rate.

Figure 1.

Prevalence of anemia categories. Anemia was present in 59% of patients and was persistent in most of them.

(0.04MB).

Patients with new-onset anemia differed from those with transient anemia only in that they were older and received more furosemide, mineralcorticoid receptor antagonists, and digoxin, the latter 3 probably reflecting worse clinical status despite similar NYHA functional class.

A total of 494 (42.1%) patients died during a mean follow-up of 3.7±2.8 years after the 6-month visit. Causes of death were worsening HF in 147 patients (29.8%), sudden death in 58 patients (11.7%), acute myocardial infarction in 36 patients (7.3%), stroke in 16 (3.2%), cardiovascular procedure in 8 (1.6%), other cardiovascular causes in 28 patients (5.7%), noncardiovascular causes in 168 patients (34.0%), and unknown cause in 33 patients (6.7%). Mortality was significantly higher in the presence of anemia (any type) (hazard ratio [HR] = 2.08; 95% confidence interval [95%CI], 1.71-2.53; P < .001). In a sensitivity analysis, the presence of anemia at baseline based on the WHO criteria showed an independent association with mortality in the comprehensive multivariable analysis: HR = 1.39; 95%CI, 1.15-1.67; P = .001.

When anemia strata were taken into account, the worst prognosis was for persistent anemia (HR = 2.59; 95%CI, 2.09-3.20; P < .001), followed by new-onset anemia (HR = 1.75; 95%CI, 1.31-2.34; P < .001) and transient anemia (HR = 1.42; 95%CI, 1.06-1.91; P = .02) (Figure 2). On comprehensive multivariable Cox regression analysis, including as covariates age, sex, NYHA functional class, HF duration, HF etiology, LVEF, diabetes mellitus, hypertension, atrial fibrillation, renal failure, chronic obstructive pulmonary disease, peripheral vasculopathy, heart rate, blood systolic pressure and treatments during follow-up, the type of anemia was an independent predictor of all-cause mortality (P < .001) (Table 3).

Figure 2.

Nonadjusted survival curves according to anemia strata. The worst prognosis was observed for persistent anemia (HR = 2.06; 95%CI, 2.10-3.21; P < .001), followed by new-onset anemia (HR = 1.76; 95%CI, 1.32-2.35; P < .001) and transient anemia (HR = 1.42; 95%CI, 1.06-1.91; P = .02). 95%CI, 95% confidence interval; HR, hazard ratio.

(0.12MB).

Anemia was studied in two thirds of the patients. Fifty percent of the patients with baseline anemia had an iron deficiency component (defined as ferritin < 100μg/dL or < 300μg/dL with transferrin saturation index < 20%) while the rest had chronic disease or renal failure-associated anemia. Only 199 (17%) patients of the total cohort (one third of anemic patients) received any treatment to resolve the anemia or to restore hemoglobin levels during the first 6 months between the 2 hemoglobin samples, and iron supplementation was the most common treatment (82% of the treated patients). Erythropoietin was used in 6.5%, and transfusion was needed in 4.5% of anemic patients. Of interest, iron supplementation and the use of erythropoietin were not independently related to survival, but blood transfusion doubled the risk of death in the multivariable analysis (Table 3).

We used only 2 measures of hemoglobin 6 months apart without more information about changes in hemoglobin levels during the remaining follow-up. Furthermore, we used a fixed cutoff value for anemia definition (even with sex-specific cutoffs). As in all published studies based on 2 clinical or analytical determinations, our analyses were performed in “completers,” that is, patients with both baseline and 6-month hemoglobin data available for analysis. Therefore, those patients that died before the 6 month visit were not included in the study, which could have introduced an unavoidable bias in the analysis performed. It is not possible to predict the association of changes in hemoglobin with prognosis in “noncompleters”.

Although our study cohort was formed by an unselected population of HF patients treated in a specific multidisciplinary HF unit of a tertiary hospital, most of them were referred from the cardiology ward. Patients were predominantly male, and the most common cause of HF was ischemic heart disease. Therefore, the results of this study must be interpreted carefully when considering the overall population.