The economic burden of HF was estimated based on a “prevalence-based approach,” which measured the costs associated with treating HF among both new and pre-existing cases of HF patients in a year [14]. A macro-costing method was used to investigate the costs of HF using the 2014 Health Insurance Review and Assessment Service-National Patients Sample (HIRA-NPS) claims data (HIRA-NPS-2014-0067). These secondary data include the claims records for the insurance-covered costs for the inpatient, outpatient, and emergency department services, and the prescription drugs of the 3 % random sample (about 1,400,000 persons) of the entire South Korean population, which consist of the NHI and Medical Aid (MA) program enrollees. This set of data was generated by using a stratified sampling with a stratification into two sets of subgroups, namely, sex (2 strata) and age (16 strata), having a total of 32 strata [15]. In South Korea, there are two tiers for the universal health security system. The NHI program is a wage-based, contributory insurance program covering about 96 % of the population, while the MA program is a government-subsidized public assistance program for poor and medically indigent individuals [16].

Adult patients with HF were defined as those aged ≥19 years who had at least one NHI or MA claim record of outpatient or inpatient services with a primary or secondary diagnosis of HF from the HIRA-NPS claims database in 2014. Due to the fact that HF in patients below 19 years old is largely attributed to congenital defects, different pathological etiologies, and lower prevalence rate than in adults, HF in patients below the age of 19 was excluded from this study [17]. Based on the literature review [7, 18‐23], the diagnosis codes for HF were identified as I11.0 (hypertensive heart disease with [congestive] heart failure), I13.0 (hypertensive heart and renal disease with [congestive] heart failure), I13.2 (hypertensive heart and renal disease with both [congestive] heart failure and renal failure), I50.x (heart failure) as listed in the International Statistical Classification of Disease and Related Health Problems 10th Revision (ICD-10 codes) for a base-case group. In an effort to reflect the patient characteristics and clinical practice in South Korea, a panel consisting of three clinicians specializing in cardiology and working in tertiary-care hospitals in South Korea was consulted. They were asked about whether the diagnosis codes of HF identified from the literature are valid for Korean patients with HF. The clinician panel suggested additional diagnosis codes (I25.5 [ischemic cardiomyopathy], I42.0 [dilated cardiomyopathy], and I42.5 [other restrictive cardiomyopathy]) to comprehensively capture patients with HF from the claims records. Thus, to minimize the over- or under-specification of patients with HF, the patient group was defined in three different ways: base-case group, narrow-definition group, and extended-definition group. The patients in the narrow-definition group have the same diagnosis codes of HF as patients in the base case group, with the former having HF as the primary diagnosis. The patients in the extended-definition group included base-case patients and those who had at least one claim record of outpatient or inpatient services with a primary or secondary diagnosis of I25.5, I42.0, and I42.5 ICD-10 codes. For each patient group, healthcare utilization associated with HF was defined as claims records for outpatient, inpatient, and emergency department services, and prescriptions with the same diagnosis codes used to define the patient group.

The economic burden of HF was estimated both from the perspective of a payer and of the society. The costs of HF from a payer’s perspective consisted only of the NHI- or MA-covered medical costs (hereafter referred to as “NHI-covered medical costs”). Societal costs included direct medical and non-medical costs, and indirect costs defined as cost of productivity loss due to morbidity and premature death. Medical costs were divided into the NHI-covered and non-NHI-covered costs. The NHI-covered costs were estimated from the HIRA-NPS claims data. Using data from the Medical Expenditure Survey provided by NHI Service, the non-NHI-covered medical costs were estimated using the ratio of the medical costs for NHI-covered services to non-NHI-covered services for patients with heart disease. [11]. Among the non-medical direct costs, the annual per-capita transportation costs were calculated as the product of the average annual number of outpatient visits and hospital admissions due to HF per patient and the average round-trip transportation costs to visit healthcare institutions, which were obtained from the 2006 Korean National Health and Nutrition Examination Survey (KNHANES). With the assumption that at least one family member takes care of the patient during the latter’s hospitalization or a helper is hired to do such, the caregiver’s cost was calculated as the product of the average annual inpatient days per patient due to HF and the average market price for the daily charge of a helper.

The indirect costs consisted of the productivity loss costs due to morbidity and mortality and were assigned for ages under 65 years old, when people are assumed to be in the labor market. The productivity loss cost due to morbidity means the opportunity costs of time lost because of hospitalization or outpatient visits, and was calculated as shown in Eq. (1) based on the human capital approach [14, 24].

i = age

j = gender

I_ij = average annual inpatient days with a diagnosis of heart failure (HF) per patient with HF by age and gender

D_ij = average daily income by age and gender

P_ij = employment rate by age and gender

O_ij = average annual number of outpatient visits with a diagnosis of HF per patient with HF by age and gender

V = average hours per outpatient visit

H_ij = average hourly wage by age and gender

The productivity loss cost due to mortality was measured based on the expected future income foregone as a result of premature death caused by HF, and was calculated as shown in Eq. (2). The age- and gender-specific number of deaths attributable to HF (ICD-10 codes of I11.0 and I50.x as a primary diagnosis) was obtained from the Annual Statistical Report of the Cause of Death by the Korean National Statistical Office (KNSO). The average annual income was derived from the age- and gender-specific average monthly income from KNSO [14].

i = age

j = gender

k = 1, 2, …, n (difference between the life expectancy and age at the time of death)

t = age at the time of death

r = annual discount rate

N_ij = number of deaths caused by HF by age and gender

Y_ij(t+k) = average annual income at the time of (t + k) by age and gender

P_ij(t+k) = employment rate at the time of (t + k) by age and gender

The estimated economic burden is presented as the annual cost per patient with HF and the total national costs of patients with HF. As the productivity loss cost due to premature death was estimated based on the total number of HF deaths in the country instead of applying the individual risk of death for each patient, it was incorporated only in estimating the total national costs. All costs were expressed in 2016 monetary value.

Sensitivity analyses were performed for the different approaches to define the patients with HF. The economic burden of HF estimated from the base-case patient group was compared with that estimated from the narrow- and extended-definition patient groups. Sensitivity analysis was also conducted for the mortality rate of HF using different data sources. The case definition of HF death used by the Annual Statistical Report of the Cause of Death by KNSO was based on the ICD-10 codes of I11.0 (hypertensive heart disease with [congestive] heart failure) and I50.x (heart failure) as a primary diagnosis, which under-identify HF cases compared to the case definition of the base-case patient group (ICD-10 codes of I11.0, I13.0, I13.2, and I50.x as a primary or secondary diagnosis) in this study. For the sensitivity analysis, the results from Roger’s cohort study of 4,537 HF patients older than 60 years were used [25, 26]. The one-year mortality rate (male, 20 %; female, 15 %) and the five-year mortality rate (male, 50 %; female, 46 %) were extracted from the study for the sensitivity analysis. The mortality estimates in Roger’s study were calculated through a review of the medical charts by a panel of physicians. Compared to the claims data, the mortality estimates in Roger’s study are more relevant for reflecting the precise and real clinical mortality rate of HF. For the mortality rate of the HF patients under 60 years, the same values from the Annual Report on the Cause of Death Statistics in South Korea were used in the base-case analysis.

To determine the impact of the disease severity of HF on the patient’s economic burden, the estimated costs across the patient subgroups were compared with the different disease severity levels. The medical costs of the HF patients younger than 65 and those older than or equal to 65 were compared. In earlier studies, it was observed that the per-capita utilization and cost of medical services to treat the same condition was significantly higher among those enrolled in the MA program in South Korea than among those enrolled in the NHI program [27, 28]. Therefore, the type of national health security program was considered a risk factor for HF patients in terms of the disease severity level and the extent of healthcare utilization. Finally, hospital admission would be a signal for a severe condition. Thus, the HF patients were divided into two groups according to experience of hospitalization, and the medical costs of the two groups were compared.

Table 1 presents the epidemiologic and healthcare utilization characteristics of patients with HF in Korea. According to the base-case analysis, a total of 475,019 adults (≥19 years old) in South Korea in 2014 were identified to have HF. The estimated prevalence rate of HF was 12.4 persons per 1,000 adults. The prevalence of HF was 9.2 times higher (47.8 vs. 5.2 per 1,000 population) in the elderly population (≥65 years old) than in the non-elderly population (19–64 years old). About two-thirds of the adult HF patients (65.1 %) in the country were 65 years or older. The highest proportion of HF patients was observed among those in the 70s (32.7 %), followed by those in the 60s (21.7 %), 80s or above (20.8 %), and 50s (16.0 %). Patients under 50 years old account for only 9.1 % of the adult HF patients. Overall, more women than men had HF throughout the age groups (57.7 % vs. 42.3 %). The prevalence of HF across genders differs, however, depending on the age. Up to the 60s, men have a higher prevalence of HF than women, while women have a higher prevalence of HF than men starting from the 70s (Fig. 1).

On average, the patients with HF had 5.01 outpatient visits, 0.24 hospital admissions, and 3.83 inpatient days for HF treatment in a year. About 12 out of 100 HF patients had hospital admissions due to HF, with 15.76 inpatient days per admission. As age increased, the probability of hospitalization increased: 14.5 % of the elderly patients aged 65 or older had hospitalization due to HF while only 7.1 % of the non-elderly patients had hospitalization due to HF. For the patients with at least one episode of HF-associated hospitalization, the average number of annual hospitalizations due to HF, including the initial hospitalization, was 2.01, and the total number of inpatient days was 31.70 days.

The average annual medical expenditure for NHI-covered services spent by individual patients to treat HF was USD 868.2 in 2014 (Table 2). From the societal perspective, the average spending of each patient in 2014 was USD 1,414.0. The total national NHI-covered medical expenditure for the treatment of HF across South Korea amounted to USD 412.4 million. While only 12 % of the patients with HF were hospitalized (Table 1), the medical expenses for inpatient services accounted for 53.4 % of the NHI-covered medical expenditure. From the societal perspective, the economic burden of HF in the country was estimated to be USD 752.8 million. Medical costs accounted for the biggest portion of the national burden (68.5 % = 54.8 + 13.7 %), followed by the caregiver’s cost (13.2 %), the productivity loss costs due to premature death (10.8 %) and morbidity (4.2 %), and transportation costs (3.4 %).

Sensitivity analyses were performed with varying approaches to define patients with HF (Table 3). When HF patients were defined in the most conservative way (using the narrow definition of patients), the total number of patients was reduced by 46.1 %, from 475,019 to 256,241 patients. Also, the total national cost was reduced by 58.7 %, from USD 752.8 to 311.1 million. On the other hand, the impact of using the extended definition of HF patients was marginal, increasing the total number of patients by only 1.1 %, from 475,019 to 498,783 patients, and increasing the total national cost by only 9.1 %, from USD 752.8 to 822.0 million. The impact of using different data sources for the mortality rate of HF was also examined through sensitivity analysis (Table 3). As both the one- and five-year mortality rates were higher than the mortality rate reported by the Annual Statistical Report on the Cause of Death, the total national costs were increased to USD 878.4 and 1,085.6 million, respectively.

The estimated medical costs substantially varied across the subgroups that were examined in this study (Table 4). The elderly patients with HF aged 65 or above spent about 1.6 times more for NHI-covered medical services than the non-elderly patients aged 19–64 years old. The annual per-capita NHI-covered medical cost for treating HF was 1.6 times higher for the patients enrolled in the MA program than for those enrolled in the NHI program. Finally, compared to the HF patients who had not experienced hospitalization associated with HF, those with at least one episode of hospitalization due to HF showed 9.7-fold higher NHI-covered medical costs for treating HF.