Prevalence quantifies the proportion of persons in a given population with a specific disease at a particular moment or over a specific period. Prevalence studies are cross-sectional and can be population-based or region-based. Population-based studies require a representative national registry and substantial investment and consequently, in practice, region-based studies are more common.

Only 2 population-based studies of the prevalence of HF have been conducted in Spain: the PRICE (Prevalencia de Insuficiencia Cardiaca en España [Heart Failure Prevalence Study in Spain]) study and the EPISERVE (Insuficiencia cardiaca en consultas ambulatorias: comorbilidades y actuaciones diagnóstico-terapéuticas por diferentes especialistas [Heart failure in outpatients: comorbidities and management by different specialists]) study. PRICE study data were drawn from 15 centers in 9 Spanish autonomous communities, selected without preestablished randomization criteria (the centers met the required characteristics and volunteered to participate). By random sampling, 2703 people older than 45 were invited to participate and 66% accepted. A diagnosis of HF was suspected in primary care (PC) using Framingham criteria and was confirmed by cardiologists if there were echocardiographic findings of significant organic or functional abnormalities. When these criteria were applied, the prevalence of HF was 6.8% and was similar in men and women. By age, the prevalence of HF was 1.3% at age 45 to 54 years, 5.5% at 55 to 64 years, 8% at 65 to 75 years, and 16.1% at >75 years.15

EPISERVE involved 507 researchers from regions all over Spain (except La Rioja), attending outpatient PC, cardiology and internal medicine clinics. Some 2534 patients were studied (5 per researcher) and HF was defined using Framingham criteria. The prevalence was 4.7%.16

Similar figures have been found by other regional studies in Spain. In Asturias, the prevalence in 2001 was 5%17; in Zaragoza, in 1994, Gallego-Catalán et al. described a prevalence of 6.3% in patients older than 65 years: 4.5% at age 65 to 74 years and 8.5% at >5 years.18 In these two regional studies, diagnosis of HF was based on Framingham criteria alone. Consequently the figures reported by PRICE, which applied more exacting criteria, would be expected to be lower than those in these regional studies. This discrepancy could be explained by the limitations of PRICE (selection bias of the participating centers and no data on 34% of the participants selected).

Other regional prevalence studies have been based on computerized registries that codify diagnoses using ICD (International Classification of Disease) criteria. The prevalence data in these registries are clearly lower: a study performed in Lérida (published in 2011) reported a prevalence of 1% in Spanish National Health Service patients older than 14 years,19 and another study performed in Madrid in the same period reported a prevalence of 0.69%.20

To contextualize these figures, a well-designed study in the United States, published in 2003, reported a total HF prevalence of 2.2%, with a significant increase from 0.7% in persons aged 45 to 54 years to 8.4% in those older than 75 years.21 In Europe, two important studies of HF diagnosis were based on the sum of clinical criteria and echocardiographic findings, as recommended in the Clinical Practice Guidelines of the European Society of Cardiology.1 One of these studies, performed in Glasgow, reported a total prevalence of 1.5%.8 The other, which was performed in Rotterdam and included the population older than 55 years (mean, 74 years), reported a prevalence of 1% at 55 to 65 years, 4% at 65 to 74 years, 9.7% at 75 to 84 years, and 17.4% at 85 years or older.14

Overall, the prevalence of HF reported in Spanish studies is higher (by approximately 2-fold) than prevalence figures described in other western countries. Although some authors believe this discrepancy could reflect a real difference and might be due to differences among the populations studied, we believe it more likely that the figures differ because of the methodological peculiarities of the various studies. As previously mentioned, participation in Spanish population-based studies has been based on practical considerations or on volunteering, which suggests a “positive” selection bias.

Other epidemiologic studies, specifically those reporting lower prevalence figures, used registries using ICD-9 or ICD-10 codes, and were essentially based on the Minimum Basic Data Set of Hospital Discharges, which has been compulsory in Spain since 1992. The validity of these registries depends on accurate coding. Several authors have reported wide interinstitutional variability and a lack of reliability in HF diagnoses recorded in administrative registries.22,23 Data retrieval is obviously limited by the quality of the data stored. Therefore, any epidemiologic study of HF based on administrative registries should validate the quality of the data by an independent audit, as well as by interobserver and intraobserver correlation studies of data coding and extraction.

We conclude this review of studies on the incidence and prevalence of HF in Spain by emphasizing their scarcity and the discrepancies between their results and those of neighboring countries, which can partly be explained by methodological failings. In the words of Dr Alonso-Pulpón, “the jury is still out” on Spanish HF studies.24 To obtain reliable information on the situation in Spain, appropriately-designed population-based studies that use current HF diagnostic criteria are needed.

Between 1980 and 1993, admissions for HF increased by 71%, and the hospitalization rate due to HF increased by 47% (from 348/100 000 in 1980 to 511/100 000 in 1993). This increase was essentially limited to patients older than 65 years, and a higher number of women were hospitalized. As a result, in 1993 there were nearly 80 000 hospitalizations for HF in Spain.7 Subsequently, Boix Martínez et al.35 showed that the number of discharges for HF rose from 25 000 to 40 000 in men and from 30 000 to 45 000 in women in only 3 years (from 1997 to 1999). Regional studies also reflected an increasing tendency for admissions for HF: in Catalonia, the absolute number of discharges for HF rose from 1735 in 1989 to 6072 in 1994—a relative increase of 250%36; whereas in Andalusia the absolute number of admissions for HF in patients older than 45 years rose from 4345 (1848 men and 2497 women) in 1990 to 10 153 (4488 men and 5665 women) in 2000—a relative increase of 230%; this increase was more pronounced in patients older than 65 years.37

Since 2003, Spain's National Institute of Statistics has gathered data on hospitalizations with a main diagnosis of HF. From 2003-2011, the number of admissions for HF in patients older than 65 years increased by 26%, at a time when the population older than 65 years grew by 13%.38 In other words, the increase in admissions for HF in the older population was 2-fold greater than the population growth.

Although data on the progressive increase in admissions for HF in Spain in the last 30 years are consistent, these figures should be interpreted with caution due to the methodological peculiarities of the studies.5,39 Admission rates for HF are obtained from hospital databases such as the Minimum Basic Data Set. Although the accuracy and quality of the coding system has improved, there remain notable limitations. Reports of patients with HF commonly contain other diagnoses—eg, the factor triggering the episode, baseline cardiomyopathy or comorbidities—that can be mistaken for the main diagnosis, depending on the criteria applied by the person responsible for coding. Moreover, each hospital has different admission criteria, and admissions for HF are spread among cardiology, internal medicine, geriatric, and emergency services, further increasing the difficulty of standardizing discharge diagnoses. Furthermore, in a complex syndrome such as HF, coding is complicated by ambiguous criteria for assigning codes. The classic ICD-9 and ICD-10 systems used in most centers have multiple codes and descriptions for HF diagnosis, and coding can be subject to interpretation. Several authors have shown notable interinstitutional variability and a lack of reliability in HF diagnoses recorded in administrative registries.22,23 All these factors obviously affect the retrieval and exploitation of Minimum Basic Data Set results.5,24

A population-based study conducted in Catalonia by Frigola et al.25 used data from the Minimum Basic Data Set to estimate the frequency of admission of patients with HF followed-up in outpatient clinics. During a 3-year follow-up, 9.5% of the patients required hospitalization for cardiovascular causes—fewer than expected.2,5 This result seems to reflect inaccurate diagnosis of HF in PC when, as in the study by Frigola et al., diagnosis is not based on objective diagnostic criteria such as those of the European Society of Cardiology.1,5,10 In contrast, 37% of admitted patients (whose HF diagnosis is more reliable) were readmitted, confirming the tendency for admissions to cluster in the initial and final stages of HF, as indicated by Desai and Stevenson32 in 2012. The independent predictors of hospitalization identified by these authors were chronic kidney disease (odds ratio [OR]=1.82), ischemic heart disease (OR=1.79), diabetes mellitus (OR=1.51), and chronic obstructive pulmonary disease (OR=1.39).25

Many of the baseline clinical characteristics of patients admitted for decompensated HF are obviously similar to those of outpatients. However, admitted patients are usually older (70% are older than 70 years) and have a greater number of comorbidities (62%) and more advanced NYHA (New York Heart Association) functional class than outpatients (NYHA III-IV in 60% of patients).40

The EAHFE (Epidemiology Acute Heart Failure Emergency) study, that included 944 patients attended in the emergency departments of 10 Spanish tertiary hospitals, investigated gender differences in the presentation and characteristics of patients with acute HF. Women were older (79.7 [9.4] vs 75.6 [10.1] years in men) and had a higher prevalence of high blood pressure (83% vs 75%), valvular heart disease (23% vs 18%) and dementia (7.4% vs 2.5%), while men had a greater prevalence of ischemic heart disease (27% vs 43%), smoking (4.4% vs 19%), chronic obstructive pulmonary disease (14% vs 29%), and chronic liver disease (1.2% vs 4.3%). Diastolic dysfunction was more frequent in women (49% vs 28%) and systolic dysfunction was more frequent in men (51% vs 72%).41

Admissions in patients with HF are usually due to decompensations. Formiga et al.42 identified infections (mostly respiratory in 29% of the patients), arrhythmias (22%), anemia (16%), and lack of treatment adherence (12%) as factors triggering decompensations.42 Of note is that most of these factors are foreseeable and can be corrected with appropriate patient follow-up, which may explain the success of outpatient HF management units in preventing admissions. The most frequent symptoms in patients hospitalized for HF were dyspnea (96%), edema (53%), chest pain (24%), and oliguria (20%).

Approximately 30% of these patients are discharged from the emergency departments rather than other services.43 About 38% of hospitalizations are in cardiology services, and 62% are in internal medicine or geriatrics. As stated earlier in relation to the outpatient management of HF, patients managed outside cardiology services are older (by a mean of 5 years), with a higher proportion of women and a greater number of associated comorbidities—particularly dementia, chronic obstructive pulmonary disease, stroke, and peripheral artery disease.44

The mean length of hospital stay for HF in the most recent studies is around 9 (5) days.25 Predictors of longer hospitalization include female sex and worse functional class at clinical presentation.45

Although episodes of decompensated HF are considered relatively benign, in-hospital mortality in patients admitted for HF is greater than mortality among those admitted for entities with a worse “reputation” such as acute coronary syndromes. Specific figures obviously depend on the characteristics of the sample studied. Hermida et al.46 analyzed patients admitted for HF to internal medicine services and found that 9.5% died during admission; in geriatric patients—who were older, had a greater number of comorbidities and worse functional class—Formiga et al.47 reported that 11% died. In the latter study, the variables independently related to an increased risk of death were creatinine level in excess of 200 μmol/l, the presence of lower limb edema, and low functional capacity.

In a study analyzing admissions for HF to all services of the Vall d’Hebron Hospital in 2002, in-hospital mortality was 6.4%, rising to 46% when total mortality from admission to 18 months postdischarge was quantified. In this study, age greater than 75 years, worse functional class, biventricular failure, and comorbidities were independent predictors of death at 18 months.40 Other authors report that low blood pressure at admission is independently associated with higher mortality and an increase in readmission compared with higher blood pressure values.48

To adequately interpret published data, it is important to distinguish between populations with HF enrolled in outpatient clinics and patient populations followed-up after hospitalization.

In series of outpatients with HF, mortality rates are lower and vary according to baseline characteristics. Hence, BADAPIC reported 6% mortality after a 13 (4) months follow-up,26 a figure lower than that of other Spanish and European studies with a longer follow-up, which typically report 20% to 30% mortality.57 This difference could be a consequence of the BADAPIC population which, as already stated, had a lower mean age (66 years) and fewer comorbidities than the other series.

Two more recent publications underline the same phenomenon: the multicenter MUSIC study reported 27% mortality in a 44-month follow-up,58 whereas, in the Badalona Hospital, Pons et al.59 reported higher mortality (37% at 36 months). In the latter study, patients had a higher mean age (69 vs 65 years), worse functional class, and a higher percentage of comorbidities (kidney failure and diabetes mellitus) and of ischemic HF.

The MUSIC study constructed a model to predict the risk of death. This model is similar to the well-known Seattle Heart Failure Risk Score and includes variables such as left atrium diameter, left ventricular ejection fraction less than 35%, intraventricular conduction abnormalities, analytic values such as hyponatremia, glomerular filtration rate, N-terminal pro-brain natriuretic peptide and troponin positivity. This model allows estimation of the risk of all-cause mortality, cardiac mortality, pump-failure death, and sudden death.

In-hospital mortality during admission for HF has already been discussed, but the prognostic implications of admission for HF go far beyond the period of hospitalization. As stated, in patients admitted to the Vall d’Hebron Hospital40 in-hospital mortality was 6.4% but at 18 months postdischarge cumulative mortality had risen to 46%.

Grigorian-Shamagian et al. studied mortality and its causes in 1360 patients hospitalized for HF, with a long mean follow-up (8 years).60 Mortality at 3.7 years was 45%. In 2005 this group, from the northwestern city of Santiago de Compostela, reported that survival at 1 year after hospitalization for HF had progressively improved over the preceding 10 years in patients with left ventricular dysfunction, whereas mortality in patients with preserved systolic function was unchanged.61 This finding seems to confirm the “real world” effectiveness of distinct treatments that reduce mortality from HF with systolic dysfunction in clinical trials in recent decades, as well as the absence of significant advances in treating diastolic HF.6

Several of the previously mentioned studies have analyzed cause of death. Pons et al.59 reported that 66% of deaths were of cardiovascular cause, essentially progression of HF (32% of total deaths), followed by sudden death, acute myocardial infarction, and other cardiovascular causes.

Grigorian-Shamagian et al.61 studied patients admitted for HF with a follow-up of up to 8 years and reported that the causes of death were decompensated HF (39%), sudden death (16%) and noncardiovascular causes (17%), myocardial infarction (15%), and vascular death (12%). Comparing these causes of death with ventricular function, these authors found no significant differences between patients with preserved function and those with depressed function, although the proportion of sudden deaths tended to be higher among patients with depressed left ventricular ejection fraction (21% vs 16%). However, they did find differences in the distribution of cause of death at 18 months follow-up postdischarge. Therefore, patients with depressed function had a cumulative risk of death from myocardial infarction of 50% in the first month after discharge, while patients with preserved function had a lower risk until 8 months after discharge. This latter group has a greater probability of death from noncardiovascular causes in the first 5 months postdischarge.

Sudden death in the context of HF remains a major challenge for clinicians. Uncertainty about the benefit of automatic defibrillator implantation is due to the low probability of sudden death reported (5.8% in Pons et al. study and 9.1% in MUSIC study) and the difficulty of accurately stratifying risk of sudden death. Hence, the indication for automatic defibrillator implantation in current guidelines extends to symptomatic patients with left ventricular ejection fraction of less than 35% if they have been under appropriate medical treatment for 3 months. The evidence is weaker for patients with nonischemic HF. Most of these patients do not receive discharges from the device and therefore research into more efficient methods of selecting those patients who could benefit from an implantable cardioverter-defibrillator remains a priority.

The high percentage of deaths from noncardiovascular causes (6.9% in Grigorian-Shamagian et al., 5.4% in MUSIC, and 9.8% in Pons et al.) reflects the high comorbidity of patients with HF who, as previously mentioned, are increasingly older and more fragile. This implies the need for more comprehensive management of individual patients and improved coordination among health care levels to allow early detection of conditions that can arise during the clinical course of patients with HF.