The Basque Health Service (Osakidetza) is a public organization that provides universal healthcare free at the point of delivery to the population of the Basque Country (approximately 2.15 million). Osakidetza is divided into 13 healthcare regions each covered by an integrated healthcare organization (IHO). Each IHO manages primary and hospital-based care (with at least one referral hospital) for its catchment population, these ranging from 10,800 to 380,000.

The study population is composed of people over 40 years of age discharged after a first admission for HF (principal diagnosis: ICD-9 codes 398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, and 428.xx) from any of the hospitals of the Basque health network between January 1, 2011 and December 1, 2015. Individuals who died during hospitalization were excluded.

Baseline characteristics: These are expressed as percentages for categorical data, medians and interquartile ranges for continuous, non-normally distributed variables and mean and standard deviations (SDs) for continuous normally-distributed variables.

Prevalence and incidence of the disease: Prevalence rates were calculated for men and women and by IHO as of January 1, 2016. The numerator is all live patients ≥40 years of age with at least one admission for HF since 1997 and the denominator the Basque population ≥40 years of age after excluding individuals living in geographical areas whose main referral hospital is not part of the Basque public network of hospitals.

The annual cumulative incidence for men and women was calculated by dividing the number of first discharges after admissions for HF per year in patients ≥40 years of age divided by the Basque population ≥40 years of age at the beginning of each year. The age-standardized incidence rates were calculated for 5-year periods using the direct method and taking the total population of the Basque Country as of January 1, 2015 as a reference.

As a measure of variation between IHOs, we used the systematic component of variance (SCV), which measures variation in the deviation between the observed and expected rates, expressed as a percentage of the expected rate. The greater the SCV, the greater systematic variation (variation not likely due to chance).14

Disease burden on the health service: The total rate of unplanned admissions, rate of PAHs and length of hospital stay were calculated for each sex. Patterns in mortality by sex were explored using survival curves and hazard functions. To assess differences by calendar year, negative binomial models adjusted for age and level of comorbidity (Charlson index) were constructed for the rates of 30- and 365-day admissions and logistic regression models for mortality rates.

All statistical analyses were conducted using R version 3.3.2.

A total of 16,272 people were discharged after a first admission for HF during the study period, of whom 1163 died during hospitalization. Those who died during the index admission were older than those who survived (84.8 years; SD=8.8 vs. 79.8 years; SD=10.1). In both men and women, the probability of dying increased with age and with comorbidity status measured by the Charlson index: 5.5% if 1–2; 7.4% if 3–4; and 10.3% if 5 or greater. Overall, crude mortality during the index admission was greater among women. Those who died during admission had more hospitalizations in the previous 12 months (Table 1).

Among the 15,109 patients included in the cohort, health care provider attended a similar number of men and women (men: n=7475 patients, 49.5%; mean age=77.2±10.6 years vs women: n=7634 patients, 50.5%; mean age=82.3±8.9 years). At the index discharge, the patients of our cohort had a median of 8 different chronic conditions. The most common conditions were hypertension (12,355 patients, 81.8%), conduction disorder or cardiac dysrhythmia (10,333 patients, 68.4%) and hyperlipidemia (7199, 47.6%) (Fig. 2; Table A2). We also observed sustained increases in the mean level of comorbidity measured using the overall Charlson index (from 3±1.9 in 2011 to 3.2±2 in 2015; p<0.001) and age at the first discharge after admission for HF (79.1±10.4 in 2011 to 80.4±9.9 in 2015; p<0.001) (Table A3).

Fig. 2.

Prevalence over the study period of the most common chronic diseases by year of cohort entry. The 15 most prevalent chronic disease categories (Magnan; see Table A1).

(0.33MB).

A total of 5434 patients in the cohort (36%) had at least one hospitalization in the year prior to the index discharge. Of these 1901 (12.6%) had been admitted for a cardiovascular reason with a principal diagnosis other than HF. Moreover, 6235 (41.3%) had attended an emergency department of the Basque health network at least once and 12,514 (82.8%) had had consultations with specialists, half of these (6279, 41.6%) with a cardiologist. Regarding the use of primary care services, the 15,109 individuals in the cohort attended their health center a mean of 14.5±10.8 times for general practitioner appointments and 16±18.5 times for nursing appointments (Table 1).

As of December 31, 2015, the number of persons with the disease registered in the information system of the Basque Health System was 15,087, yielding a prevalence of 1.22% among over-40-year-olds (1.28% in men vs 1.26% in women). The prevalence in the different IHOs ranged between 0.78% and 1.33% in women and between 0.93% and 1.43% in men. The age-standardized incidence of first discharges after admission for HF ranged between 2.23 and 2.45 per 1000 women per year in women and between 3.54 and 3.8 per 1000 men per year, in the first years analyzed (Fig. 3). We did not observe any trend (either increasing or decreasing. Ptrend>0.05 in all analysis). Across the geographical area studied, there was little variation in the new discharges after admission for HF (SCV<0.1 in all the years studied, in both men and women; Table A4).

Fig. 3.

Anual incidence of heart failure among individuals over 40 years of age in the Basque Country (per 1000).

(0.08MB).

Based on the data available from hospital electronic health records, the estimated prevalence of HF among over-40-year-olds in the Basque Country is 1.22% as of December 31, 2015. The annual incidence rate has slightly increased during the study period in both women (from 2.51 to 2.9 per 1000) and men (from 2.86 to 3.13 per 1000). This increase seems to be attributable to changes in the population pyramid, as it disappears after adjusting for age.

The absolute number of first discharges after admission for HF was similar in men and women, but the mean age was approximately 5 years older in women than in men (78 vs 83 years). Most of the cohort had a high level of comorbidity, patients having a median of over eight chronic conditions. Four out of five patients had hypertension and one third of the men had chronic obstructive pulmonary disease. Overall, 42% of the patients had had contact with cardiologists in the 12 months prior to the index discharge. Among those who survived the first admission, 5% died within the following 30 days, and 25% died within the first year after discharge. Approximately 15% and 50% of the patients had at least one unplanned hospitalization within 30 and 365 days after the index discharge, respectively, 54% were for non-cardiovascular reasons (49% and 54% within 30 and 365 days, respectively) and 4 out 10 of the hospitalizations were considered to have been potentially avoidable. On the other hand, we have observed an increasing trend in the prevalence of cardiovascular risk factors in individuals diagnosed with HF.

The incidence of HF in our study, applying very restrictive criteria, was in the lower range of rates published in the literature, which vary depending on the diagnostic criteria used and the geographical area.15,16 We should note that the methodology used varies substantially between studies. Up to six methods based on signs and symptoms have been described, these yielding very different estimates of the prevalence of the disease.17

The prevalence in our study population was also in the lower range of rates published in Spain, but closer to that reported for most of the developed countries, probably due to methodological limitations of the former.5

Nevertheless, we observed some secular trends in the epidemiology of the disease. As can be inferred from other studies, the incidence of HF remains stable18 or even has decreased19 in recent years. In contrast, the prevalence is increasing undoubtedly due to population aging, but also due to increasing survival thanks to the use of successful evidence-based treatments.20

Our finding of an increase in comorbidities in these patients is consistent with results of other studies, both regarding cardiovascular diseases21 and so-called discordant comorbidities, described in the aforementioned study by Tran et al. and in the Swedish cohort of Vetrano et al.22

Finally, the rates of admission within 30 days and 1 year after discharge are similar to those described for neighboring countries, but notably lower than those found in studies in some other countries such as the USA.23

Notably, the rate of PAH reported in the latter study was higher than ours, which may point to one of the common causes of variations in clinical practice, namely, those associated with differences in the care offered.24 This finding, however, has to be interpreted with caution, as the operational definitions of PAH are substantially different in these studies.

The prevention of HF requires the prevention and control of hypertension and other modifiable risk factors for heart disease.

On the one hand, we are dealing with patients with ever more complex conditions: we observed an increase in the prevalence of hypertension, hyperlipidemia and other known risk factors, and the number of associated chronic conditions is increasing, as is the age at presentation. On the other hand, patients have been in contact with clinicians at different levels of care multiple times during the year prior to the first discharge after admission for HF, especially primary care, which means that there are numerous opportunities for the secondary prevention of the disease, for example those contacts could be better used to try to improve the low rates of compliance with guideline-directed pharmacological therapy.25 In patients with a very high cardiovascular risk, the target of maintaining systolic pressure under 120mmHg has shown to reduce the rates of mortality, cardiovascular events and hospitalizations for HF.26

In the current circumstances, with clinicians generating large quantities of data, there is a need for expert processing of the data available, in order to synthesize it and make it genuinely useful for decision making. In relation to this, the use of real world data would greatly facilitate advances toward more individualized care, through the introduction of tools to support clinical decision making.27

The cohort generated 14,559 hospitalizations within the first year after their first discharge resulting in 116,061 bed days. Though the absolute figure is notable, it is even more striking that 35% of all these hospitalizations could, in theory, have been avoided. This should make us reflect on the potential impact of interventions in this area. Measures that have been found to be effective against PAHs include strengthening continuity of care, this having been shown to reduce the rate of PAH by 5% to 8%.28 It has also been suggested that specific intervention plans, such as the current Basque population-based plan, could improve care outcomes.29

Finally, and in contrast to the dominant hypothesis that PAHs are caused by suboptimal management of the disease at the outpatient level, recent studies suggest that PAHs reflect more or less intensive use of the referral hospital, as well as of the care provided in primary care,11 or failure of prevention efforts to reduce the demand for emergency care by chronic patients.30

On the other hand, 5 out of 10 unplanned hospitalizations after the index discharge were for non-cardiovascular reasons (49% and 54% within 30 and 365 days, respectively) while HF accounted for only a third of all unplanned hospitalizations (36% and 32% within 30 and 365 days, respectively). These results are consistent with those reported in other studies and suggest the need for HF care to focus on the integrated patient health burden and better management of comorbidities in HF.31

The definition of disease in this study probably implies exclusion of most cases of less severe HF (NYHA 1–2). Although other types of data relevant to the study of HF, such as ejection fraction or brain natriuretic peptide, are electronically stored, they are not currently available in a structured format and hence were not included in our analysis. Nevertheless, the diagnostic criteria selected, together with the advanced age of patients in the cohort, make us believe that most of them will be chronic patients, in an advanced stage of the disease and with certain level of systolic and diastolic dysfunction.

On the other hand, exclusion criteria could have been a source of heterogeneity. Actually, excluding transfer on discharge could introduce bias in analysis of geographic variation. However, we think that the impact of heterogeneity on main results will be low.

Finally, we did not analyze data related to pharmacy. This could have helped to better describe our population and to detect gaps in use and dose of guideline-directed medical therapy that could partly explain these findings. In fact, according to a recently published article, in the real-world setting, the use of standard-of-care therapies, both before and after the onset of worsening HF, is low, with high unmet medical needs, suggesting that patients with worsening HF require novel treatment strategies as well as greater optimization of existing guideline-directed therapy.32 In that sense, we are working on a project that will integrate drugs prescription and dispensation trajectories into our cohort, and that could bring new insights to the real-world use of standard-of-care therapies and consequences in patients with HF.