Heart failure (HF) is a major and growing medical and economic problem, with high prevalence and incidence rates worldwide [1, 2]. HF is defined as a pathophysiological state in which an abnormality of cardiac function is responsible for the failure of the heart to pump blood at a rate commensurate with the requirements of the metabolizing tissues [3]. It has been estimated that 0.4–2.2% of the population in industrialized countries suffer from HF [4], with between 500,000–600,000 incident cases diagnosed each year [5]. HF affects especially the elderly, with 80% of HF-related hospitalizations and 90% of HF-related deaths occurring among patients aged 65 years or older [6]. The prevalence of HF has increased over the past decades [7]. It is expected that there will be a further rise due to a higher proportion of elderly people and better survival rates of patients with conditions such as hypertension, diabetes, etc., which trigger the development of HF [7, 8]. In addition, two thirds of the patients are readmitted to hospital within one year [9]. The mortality rate for patients with HF is high, as showed in the MAGGIC meta-analysis includes individual data on 39,372 patients with 40,2% died during a median follow-up of 2,5 years [10]. A recently published study demonstrated that the 30-day readmission rates for HF are higher than for pneumonia or acute myocardial infarction [11].

Due to the high and increasing prevalence rates, HF constitutes an enormous economic burden for the healthcare systems in industrialized countries. For example, Europa and USA spent 1–2% of their annual healthcare budget on HF [6]. The global economic burden of HF is estimated at $108 billons per annum, with $65 billons attributed to direct and $43 billons to indirect costs [12]. The US is the biggest contributor to the global HF costs and is responsible for 28.4% of total global HF spend [12]. Europa accounts for 6.83% of total global HF costs [12].

Due to the considerable cost impact of HF on healthcare systems, it is necessary to have a better understanding of the cost aspects and the specific cost drivers. In this context, Cost-of-Illness (COI) studies are an important tool to analyze the economic burden of HF and to provide information on the cost drivers to clinicians and health policy makers. On the basis of transparent and detailed cost components this is aimed at improving the planning and development of healthcare services and optimization of the allocation of healthcare expenditures and medical resources [13]. COI studies are often restricted to a certain country, deal with small patient groups or present only a part of all illness costs. Therefore, it is important to summarize the results of different COI-studies in a systematic way.

COI studies estimate the resources consumed and lost as a result of a particular disease. Results from the COI studies can improve understanding of the economic burden that a specific disease may have on society as whole, healthcare providers, and the individual patient. COI studies can also provide a fundamental basis for further economic evaluations, such as cost-effectiveness-, cost-utility- and cost-benefit analysis [14].

The aim of this study was to perform a systematic review of recently published COI studies related to HF to highlight the increasing cost impact associated with this disease and identify the major cost drivers.

Search results were transferred to EndNote, version X7, and reviewed independently by two researchers. The inclusion and exclusion criteria were adopted from the CHEC list [17] and the BMJ guidelines for authors and peer reviewers of economic submissions [18]. Although the two checklists were developed for the assessment of economic evaluations, we derived criteria that are also relevant for the evaluation of cost-of-illness studies. These criteria are listed in the appendix (Additional file 2). Furthermore they were in accordance with the check list for COI evaluation in the guide to critical evaluation of COI studies developed by Larg et al. [15].

HF can be categorized in two entities, systolic HF and diastolic HF. Systolic heart failure is defined as the entity in which the ejection fraction is reduced. In diastolic HF the ejection fraction is preserved. Owan et al. has shown that 53% of patients suffer from systolic and 47% from diastolic heart failure [19]. As our aim was to analyze all entities of HF independently of the pathophysiologic mechanism, we excluded papers dealing only with systolic/diastolic HF.

To provide a comprehensive understanding, the included studies were analyzed in terms of country, epidemiological approach (prevalent vs. incident), study period, perspective (societal, healthcare provider, etc.), main data sources and the identification of HF patients. When the study perspective or the epidemiological approach were not clearly specified in the studies, two investigators achieved consensus by discussion.

The reported costs in the included studies were transferred from the local currency in the year of the costs to the inflated values in local currency for the year 2016 [20]. This cost data was exchanged to US-dollars by using the gross domestic product purchasing power parity (PPP) [21]. This methodology can be used for COI studies in order to reach better comparability between the different currencies [22].

Study characteristics are summarized in Tables 1 and 2. Reviewed COI studies showed results for ten countries. Eight studies were conducted in Europa, and six in North America. The study population size ranged from a minimum of 115 to a maximum of 475,019. The mean age of the study population varied from 58 to 81.6 years. Three studies [23‐25] reported the mean age of different subgroups (HF-group vs. no-HF control group). Eleven articles had a prevalent and two an incident epidemiological approach. Three studies provided both types of results [25‐27]. Twelve studies adopted the perspective of a third-party payer and six articles were categorized as a prospective study.

To identify HF patients, the studies used different methods. Six articles referred to the ICD-10 (International Classification of Diseases), [27‐32], five to the ICD-9 [24, 26, 33‐35] and five [23, 25, 36‐38] didn’t specify the ICD classification.

Because HF is usually accompanied by many underlying diseases, it is often difficult to identify the correct cases of HF based on the diagnosis codes of ICD. Lee et al. [31] showed that by defining HF only as the primary diagnosis the HF cases decrease by 46% in contrast to define HF cases as primary and secondary diagnosis (using the same ICD-codes). By further analyzes he showed, that 75% of patients with HF, which was defined as a secondary diagnosis had a primary diagnosis which was related to HF, such as hypertension or angina pectoris. Thus, defining HF only as the primary diagnosis might lead to an under-specification of HF [31].

A second study of Voigt et al. [35] calculated costs for HF as the primary diagnosis (HF in isolation, HFI) and HF as one of multiple diagnosis/part of a disease milieu (HF syndrome, HFS). As a consequence, costs range between 70.8 (HFI)- 127.0 (HFS) billion dollars. This might show the great underestimation of economic burden, if defining HF only as the primary diagnosis [35].

HF shows high prevalence rates with 12.4 per 1000 persons suffering from this chronic condition [31]. The prevalence rates increase with age [29, 31, 36], with patients aged 65 or older showed 9.2-fold higher prevalence and 1.6 higher costs than 19–64 aged population [31]. Neumann et al. [29] analyzed that patients 65 or older shows 10-times higher prevalence rates than 45–65 aged patients.

Incidence rates are also high with 2.4–3.8 per 1000 persons [33]. HF results in a severe mortality rate with 1-year mortality after HF hospitalization of 24% [33]. Stafylas et al. [38] showed that the mortality rates are higher for patients, who were hospitalized for HF (24,3%) than for HF patients treated outpatient (7,7%).

The included studies reported high, similar readmission rates for HF between 42% and 44,9% [26, 28, 38].

Of the included articles, two separated the cost data by gender [28, 29] and both reported higher costs for women.

To enable better comparability, prevalence-based studies were grouped separately from the incidence-based ones.

The main economic estimates are presented in Table 4. The total annual costs per patient ranged from $868 for South Korea [31] to $25,532 for Germany [30]. Two articles [29, 35] accounted the costs not per patient, but analyzed the whole economic burden of HF per year in an aggregated way. The hospitalization costs are the greatest cost component of overall healthcare costs. Among total healthcare costs, expenditures for medication are the second largest issue [33, 37]. Due to hospitalization costs, room and board were the greatest contributor (43% of inpatient costs), following by procedures, imaging and laboratory testing [34]. Dialysis was responsible for the highest part of procedural costs, but it was needed only by a small number of patients [34].

The study of Delgado et al. [37] considered the burden of costs of informal caregiving per patient. He estimated that 59.1 to 69,8% were due to informal care costs. Additionally, he showed that the healthcare and informal care costs for HF were rapidly increasing with the number of hospital admissions, contributing to overall increases in total costs for HF. Of the included studies Dunlay et al. [34] presented the lifetime costs for HF for the longest study period.

Ory et al. [26] reported that newly diagnosed patients had significantly higher healthcare charges than the prevalent group. Furthermore, he described that – in comparison to the control group without HF- the patients with HF created four times higher total healthcare costs.

The study of Ogah et al. [32] was the only one conducted in a low/middle income country (Nigeria). 46% of the total costs were contributable to inpatient and 54% to outpatient costs [32]. Inpatient costs were lower than in high-income countries because in these countries the utilization of expensive medical equipment and surgery are higher. Ogah et al. [32] showed that 90% of direct outpatient costs are due to medication and transportation costs for monthly follow-up visits, which were mostly made through out-of-pocket payments.

Some studies estimated cost predictors, which are shown in Table 5. A higher NYHA stage [23, 25, 37, 38], kidney dysfunction [25, 38] and the comorbidity of HF and diabetes mellitus [24, 34] were considered as the most common reasons of increasing costs for HF patients. Two studies [23, 33] reported that that comorbid conditions of HF were much more associated with greater costs than prevalent HF alone. Comorbidities in HF cause ¾ of all readmissions in HF patients [33]. Dunlay et al. [34] conducted a study with an incident epidemiological approach and analyzed that diabetes mellitus caused an increase in lifetime costs of HF patients of 25%. Bogner et al. [24] also emphasized that diabetes mellitus has substantial influence on the costs of managing HF patients, extends the hospital stay and worsens the prognosis. A study analyzed that a preserved ejection fraction (> = 50%) in HF patients leads to an increasing cost impact of 24% [34]. In contrast, Liao et al. [25] showed that there was no statistically significant difference between HF patients with normal and reduced EF due to the 5-year cumulative costs.

The severity of HF is classified by NYHA (New York Heart Association) stages [1]. Of the included studies, only two [36, 37] separated the cost data by NYHA stage (Table 6). Czech et al. [36] gave very detailed information on cost data by NYHA class I-IV, estimating the total annual costs as well as the average costs of hospitalization for HF by NYHA stage. Delgado et al. [37] reported economic data for NYHA stages II-IV and estimated the total costs for HF by combining the NYHA groups III and IV. He emphasized the rising costs and the significantly increased use of health services and social services, as formal and informal care, by NYHA stage. Both studies showed that the economic burden of HF is dependent on the NYHA stage and the costs rise with advanced stages, disregarding NYHA stage I. Thus, it is important to act early and to prevent the progression of HF to more advanced and highly symptomatic forms. Czech et al. [36] reported for NYHA I high annual as well as hospitalization costs and assumed that this could be attributable to the higher proportion of special cardiology interventions in this group. Additionally, he described that NYHA IV was responsible for more than 70% of total annual costs for HF.

Two articles [25, 34] compared the cost data of patients in the first year of HF diagnosis to the previous year in order to investigate the impact of HF on the costs (Table 7). Dunlay et al. [34] reported an 318% increase in costs in the year of HF diagnosis. He also examined that the costs for HF were high at the time of initial diagnosis, decreased then, reached a stable, relative low level and increased again at the end of life. Liao et al. [25] estimated in his study that the development of HF had a greater than 200% increase in total costs compared to the year before diagnosis. This strong rise was especially generated by the inpatient cost component. After that he detected a decrease of costs, but they were still higher than in the year before HF diagnosis out to year 5, if only survivors were examined.

The main findings of our review are reported in Table 4 and are focused on the highest annual per-patient costs reported for the USA [24] and Germany [30]. Our review shows that the costs for hospital admission contribute significantly to the overall direct costs for HF (ranging from 44 to 96%). Previously published studies accounted that about two-thirds of the direct HF healthcare costs are due to hospitalization [6, 39]. The high inpatient costs are a result of high readmission rates, with 23% of HF patients readmitted to hospital stay within 6 months [6]. A further aspect of our analyses is the considerable increase in costs with advanced NYHA stage, with NYHA stage IV being the most expensive. Biermann et al. [5] estimated similar results for patients with systolic HF. An earlier review [40] estimated that patients with NYHA IV produce between 8 and 30 times higher healthcare costs than patients with NYHA II. We have shown that costs rise rapidly after a confirmed HF diagnosis (Table 7). Although HF is a chronic condition causing high lifetime costs, particularly the first year after a HF diagnosis and the end-of-life care are the most expensive ones. Two studies from the USA [41] and Canada [42] analyzed the last 180 days of life of HF patients and concluded that in the last six months of life, there is a large increase in costs and resource use. Unroe et al. [41] as well as Kaul et al. [42] analyzed that the costs during the last 180 days rose from $28,766 to $36,215 between the years 2000–2007 for the USA [41] and from $25,069 to $27,983 (Canadian dollars) between the years 2000–2006 for Canada [42].

Other studies also report a rise in costs relating to HF over time, but especially in the last two decades. Stewart et al. [39] reported a rise of direct medical costs from £716 million in 1995 to £905 million in 2000 for the UK. Liao et al. [6] presented an increase of HF-related costs for different countries (Spain, Canada, Sweden and Scotland) by 40–71%.

Our review highlighted not only the large economic burden of HF, but also the heterogeneity of the studies and lack of cost data. For better comparison of research data, future COI studies should use a standardized approach regarding methodology, in particular regarding criteria concerning the selection of HF patients and data, the inclusion of different cost drivers and the presentation of results.

Thus, further COI studies in HF should clearly state the ICD codes, which were used for identification of HF patients. Additionally, an attribution of cost data to the specific ICD code number could achieve a better comparison between COI studies. As presented in recently published papers, using different ICD codes may lead to an over- or underestimation of HF diagnosis [43, 44].

COI studies should present cost data for the whole HF group and disaggregate them consistently for sub-entities, as systolic or diastolic HF. Importantly, COI studies should emphasize clearly stating the use of the exact diagnoses (including relevant ICD codes). It should be clearly stated, if HF is defined as the primary or secondary diagnosis, as this difference influences the cost estimates. The inclusion of comorbidities would give detailed information on the cost drivers and show opportunities for decreasing costs.

In addition, the study perspective and a distinction between COI approaches should be indicated [18]. This review emphasizes that indirect costs are a significant contributor to total costs and more effort is needed to estimate these costs accurately and consistently. In addition, informal care costs are an important contributor to the COI of HF [37].

As there are many differences between healthcare systems in different countries, high-quality cost data is needed from COI studies to facilitate comparisons between countries and cost trends. Crucially, more robust data from future COI studies is needed to provide a sound basis for cost-effectiveness studies to identify the most cost-effective therapies.

Another aspect in our systematic review highlights that most of the studies are conducted in Europe and USA. But as HF is a global problem, we need more COI-studies from low-and middle-income countries as demanded in a recently published study [45].