Resource Utilization and Cost of Heart Failure associated with Reduced Ejection Fraction in Swedish Patients

Abstract

Aim:

The purpose of this study was to assess healthcare utilization and costs for heart failure patients with reduced ejection fraction (HF-REF) in Sweden.

Methods and Results:

This was a retrospective, population-based cohort study of patients diagnosed with HF-REF during a period of 18 months at 31 primary care centers in Uppsala County, Sweden. Data was obtained from computerized records from these centers, the Swedish Patient Registry, the Swedish Prescription Registry, the Cause of Death Registry, and a local echocardiography registry maintained by the Department of Physiology, Uppsala University Hospital. Main outcome measures were cardiovascular and heart-failure-related hospitalizations, outpatient visits, medication utilization, mortality (all-cause, cardiovascular, and heart-failure), and healthcare costs for HF-REF patients. During the index period, 252 heart failure patients had a left ventricular ejection fraction measurement ≤40% and were categorized as having HF-REF. More than half of the patients had ≥1 cardiovascular or heart failure-related hospitalization. On average, patients had >2 such hospitalizations annually. They also averaged ∼1 cardiovascular or heart-failure-related outpatient visit per year. All-cause mortality was high: 15.9% patients died within 1 year after the index date. The mean annual cost per patient for heart-failure-related hospitalizations was SEK 72,613 (EUR 7610). In contrast, annual prescription costs were low, on average 3% of total cost (SEK 3503, EUR 367 per patient)

Limitations:

The main limitations of this study include a short follow-up time and small sample size. Also, certain data were missing, such as echocardiograms (available for only 28% of patients), and information on patients’ New York Heart Association (NYHA) functional class, validity period for prescriptions or the units of medication prescribed, and medication dosing. Furthermore, the overall mortality could have been under-estimated, as only the primary cause of death was included in the analysis.

Conclusions:

The main burden associated with HF-REF is related to hospitalizations for heart-failure events. Effective treatment options that decrease hospitalization rates could reduce patients’ suffering and potentially offer considerable cost savings.

• Reduced ejection fraction heart failure
• Resource utilization
• Direct costs
• Health care cost

Introduction

The overall burden of heart failure on the healthcare system in Sweden is high, with an estimated prevalence (based on clinical diagnosis) of ∼1.5%1,2. While reported rates vary, reduced ejection fraction heart failure (HF-REF) likely represents close to half of these cases3,4. Prevalence of the disease increases with age, and is therefore rising with the aging of the overall Swedish population1,2. The majority of heart failure patients, particularly those with mild disease, are managed by primary care physicians5. However, as the disease becomes progressively severe, patients are more likely to be referred to a hospital. Heart failure is the most common reason for hospitalization of patients older than 65 years of age in Sweden and the fourth most common reason for hospital admission overall2,5. As many as one third of heart failure patients may require at least one hospitalization every 2 years6, and two thirds of hospitalized patients need long-term follow-up by primary care7. Specifically, HF-REF is associated with higher rates of myocardial infarction (MI) and other atheroscletoric diseases such as peripheral arterial and cerebrovascular when compared with heart failure without reduction in ejection fraction4. The combined treatment costs are substantial; the total annual cost for heart failure patients in Sweden has been calculated to be as high as 5.0–6.7 billion Swedish kronor (SEK)8. A 1999 study (expressed in 1996 prices) indicated that costs for heart failure amount to 2% of the total Swedish healthcare budget2. While hospital care comprises the major portion of this cost, primary care has been shown to account for almost one quarter of expenditures2,8.

Pharmacologic regimens including diuretics, vasodilators, inotropic agents, and beta blockers are often aimed at improving ventricular function or symptoms and patient well-being, as well as reducing hospital admissions. However, because clinical presentation of the disease is so variable, pharmacotherapeutic management may also target other aspects of heart failure such as inflammation, calcium handling, or altered expression of genes involved in cardiac myocyte metabolism or cytoarchitecture9.

International guidelines advocate echocardiography and left ventricular ejection fraction (LVEF) assessment to record cardiac dysfunction and thus diagnose heart failure4. Yet, diagnosis is confirmed by echocardiography in only ∼30% of cases in Sweden10, possibly due to limited access to this specialized test4,11. Furthermore, in Sweden, treatment of chronic heart failure in general, and particularly in primary care, often falls short of evidence-based guidelines. Medications that would reduce morbidity and mortality are both under-utilized and under-dosed10,11. In one study, while 74% of patients received treatment with an angiotensin converting enzyme (ACE) inhibitor, an angiotensin receptor blocker (ARB), or both, only 37% were on at least 50% of the recommended dose; the corresponding percentages for beta blockers were 68% and 31%, respectively10. Only 42% of patients received the recommended ACE-inhibitor/ARB plus beta blocker therapy, and only 20% received 50% of the target dose or more.

There are few publications addressing healthcare cost and resource utilization for patients with HF-REF managed by primary care in Sweden, and the existing studies are outdated (1999–2005)2,6,8,12. The aim of this study was to investigate the most recently available healthcare resource utilization and associated direct costs for these patients.

Methods

Study design

Patients with HF-REF at 31 primary care centers in Uppsala County, Sweden were included in a retrospective population-based study. This sample of physician practices corresponds to over 85% of the population in Uppsala County, and is estimated to be a representative sample reflecting the targeted heart failure population and the associated heart failure healthcare resource utilization. The Swedish healthcare system is a national, tax-financed system that provides similar heath cover to all residents5. Electronic medical records from these primary care centers were screened for the data of all patients who received an ICD-10 diagnosis of heart failure (ICD-10: 150) between July 1, 2005 and December 31, 2006 (See Appendix A for further details), and a total of 1592 patients were identified. Only patients with a LVEF assessment during the index period and, within ±3 months of the diagnosis code (n = 451) were included in the survey, and only patients with an LVEF ≤ 0.40 (categorized as HF-REF patients) are included in this report (n = 252). Patients younger than 18 years of age or whose age or gender was missing were excluded. The pre-index evaluation period included any time point before the index period but after January 1, 2005. The index evaluation period was defined as 3 months prior to, and 3 months following the index date. Patient records were then followed from the index date (the date of the first heart failure ICD-10 diagnosis code, indicating the index event) to the end of the observation period, December 31, 2007 (the post-index evaluation period; Figure 1). Data on patient demographics, diagnoses, laboratory values, and other measurements (including systolic and diastolic blood pressure, LVEF, serum creatinine, C-reactive protein, troponin, brain natriuretic peptide [BNP], N-terminal proBNP [NT-proBNP]), and outpatient clinic utilization were extracted from the computerized primary care medical records. Extraction was performed using the Pygargus Customised eXtraction Program (CXP), which has previously been used in several population-based studies in Sweden13–18. Data from primary care records were linked with the inpatient and outpatient hospital data including diagnoses, hospitalization, medication, and death based on the Swedish Patient Registry, the Swedish Prescription Registry (only available from July 2005), and the Swedish Cause of Death Registry. Data on LVEF was captured from a local echocardiography registry kept by the Department of Physiology, Uppsala University Hospital. The main outcome variables collected the number, duration, and time from the index event to subsequent cardiovascular and heart failure-related hospitalizations; the number of outpatient visits; medication utilization; all-cause, cardiovascular, and heart-failure mortality; and cost of hospitalizations, outpatient visits, and medication per patient. The official unit costs of hospitalizations, outpatient visits and per prescription costs for medications were used as the basis to calculate the costs for the respective resources per patient.

Figure 1.  Cardiovascular patients show a higher rate of hospitalization relative to heart failure patients.

Study variables and outcomes

Descriptive analyses were performed for pre-index (defined as any time prior to the date of first diagnosis) comorbidities, index patient characteristics, medication, and comorbidities, and for post-index (defined as the period from date of first diagnosis to end of study or death) medication, outpatient visits, and hospitalizations. Frequencies (numbers and proportions) and distributions (mean, median, standard deviation, and range) were compared for categorical and continuous variables, respectively. Comorbidities were assessed at pre-index and index in order to separate those which pre-dated the diagnosis of heart failure. Comorbidities identified included diabetes, cardiomegaly, bundle branch block, atrial fibrillation, mitral regurgitation, myocardial infarction (MI), coronary artery disease (CAD), peripheral vascular disease (which typically includes peripheral artery disease (PAD) and deep vein thrombosis (DVT) patients), stroke, hypertension, dependent edema, pulmonary edema, chronic obstructive pulmonary disease (COPD), and renal disease. Medication classes that were evaluated include ACE inhibitors, alpha blockers, ARBs, beta blockers, calcium channel blockers, cardiac glycosides, cardiac stimulants, diuretics (including loop-diuretics and aldosterone blockers), statins, and vasodilators. Treatment was classified either as mono-regimen therapy (prescription of one therapeutic class during the time period of interest) or multiple regimen therapy (prescriptions of more than one therapeutic class during the time period of interest). Cardiovascular hospitalization was defined as any hospitalization with a diagnosis code for unstable angina, MI, ventricular or atrial dysrhythmia, or stroke; while heart failure hospitalization was defined as any hospitalization including a diagnostic code for heart failure (reported as the primary or secondary diagnosis). Kaplan-Meier analysis was used to determine the intervals between the index event to first hospitalization, and the interval between hospitalizations, up to a maximum of five hospitalizations.

Mortality (all-cause, cardiovascular, and heart failure) was assessed during the period starting with the index date until the end of the study period. Kaplan-Meier analysis was performed to analyze the time from first admission until death.

In order to estimate the cost of resource utilization per follow-up year per patient, data collection included the number and dates of cardiovascular and heart failure related hospitalizations, number and dates of outpatient visits, number and dates of surgical and non-surgical procedures, and number and prescription dates for identified medications. Unit costs were calculated based on price lists retrieved from a study by Henriksson et al.19, the Swedish Southern healthcare region, and the Swedish Prescription Register for hospitalizations, outpatient visits, and medications, respectively. Hospitalization and outpatient unit costs were multiplied by the mean rates for hospitalizations and outpatient visits, respectively. Per prescription item costs were used to calculate per patient prescription costs. All costs were finally adjusted to the 2010 SEK value, and are presented as the total cost in SEK per patient per year. The SEK to EURO (EUR) conversions are based on the annual average conversion rate for 2010 of 1 SEK = 0.1048 EUR.

Statistical analysis

Zero-inflated negative binomial regression analysis, a variant of Poisson regression, was performed to evaluate predictors for the number of cardiovascular and heart-failure-related hospitalizations per year of follow-up. A zero-inflated negative binomial methodology was employed to account for observed over-dispersed data resulting from a high proportion of individuals with zero re-hospitalizations. Covariates that were included in the final model for both outcomes were age, gender, and inpatient v/s outpatient status during index and pre-index comorbidities. The number of index and post-index medications was also considered for inclusion in the multivariate model; however, based on Akaike Information Criteria (AIC) model fit statistic principles these variables were not included in the final model.

Results

A total of 252 patients, 78 (31%) women and 174 (69%) men, complied with the inclusion criteria. The mean age was 74.1 ± 12.0 years (Table 1). At least 1 year of follow-up was available for 83% of the patients; 17% of the patients were obese, and more than one third of patients had systolic hypertension (Table 1). Data on laboratory values were not available in the database for most of the patients. The most frequently diagnosed comorbidities for both the pre-index and index periods were hypertension, atrial fibrillation, diabetes, MI, and CAD; while the most common comorbidity in the pre-index period was hypertension (38.0%), and at index it was MI (44.0%; Table 1).

Table 1.  Baseline demographics and comorbidities.

Patient characteristics	HF-REF patients (n = 252)
	n (%)
Baseline
Gender
Male	174 (69)
Female	78 (31)
Age in years
Mean (SD)	74.1 (12)
Median (range)	77 (38–95)
Follow-up time in years
Mean (SD)	1.5 (0.7)
Median (range)	1.7 (0–2.5)
Follow-up time
Less than 1 year	42 (16.7)
1–2 years	140 (55.5)
2–3 years	70 (27.8)
Index year
2005	91 (36.1)
2006	161 (63.9)
Pre-index comorbidity
Hypertension	96 (38.0)
Atrial fibrillation	48 (19)
Diabetes	55 (21.8)
MI	51 (20.2)
CAD	47 (18.6)
PVD	23 (9.1)
COPD	15 (5.9)
Stroke	21 (8.3)
Renal disease	7 (2.7)
Mitral regurgitation	4 (2)
Dependent edema	3 (1.1)
Basal Metabolic Index
Underweight (<18.5)	9 (3.8)
Normal (18.5–24.9)	101 (41.9)
Overweight (≥25)	89 (36.9)
Obese	42 (17.4)
Index comorbidity
Hypertension	88 (34.9)
Atrial fibrillation	100 (39.6)
Diabetes	58 (23)
MI	111 (44)
CAD	42 (16.6)
PVD	17 (6.7)
COPD	18 (7.1)
Stroke	15 (5.9)
Renal disease	12 (4.7)
Mitral regurgitation	6 (2.3)
Dependent edema	6 (2.3)

HF-REF, heart failure-reduced ejection fraction; SD, standard deviation; CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; HF-REF, heart failure-reduced ejection fraction; MI, myocardial infarction; PVD, peripheral vascular disease; Index, 3 months prior to and 3 months post-HF diagnosis.

Pharmacotherapy

More than 85% of patients received at least one of the identified medications during the index period (Table 2). Beta blockers were prescribed most often (64.7% of patients, mostly selective agents), followed by diuretics (56.7%), ACE inhibitors (52.8%), and vasodilators (31.7%). Of the diuretics, loop diuretics were the most common, prescribed for 44.8% of patients. Only 12% of HF-REF patients received mono regimen therapy during the index period, vs 74% on multiple regimen therapy (i.e., more than one therapeutic drug class). The medications most frequently prescribed as monotherapy were beta blockers (44.8% of patients) and diuretics (20.7%). The most common multiple regimen therapy comprised an ACE inhibitor, beta blocker, and a diuretic that was prescribed to 10.2% of patients on multiple regimen therapy.

Table 2.  Medications at index and post-index (through study follow-up).

Index medications	HF-REF patients (n = 252)
	Index n (%)	Post-index n (%)
None	37 (14.7)	8 (3.2)
Diuretic	143 (56.7)	204 (81.9)
Thiazide	7 (2.8)	14 (5.6)
Low-ceiling nonthiazide	0 (0)	2 (0.8)
High-ceiling loop diuretic	113 (44.8)	180 (72.3)
Potassium-sparing	63 (25)	132 (53)
Potassium-sparing and low/high-ceiling	5 (2)	3 (1.2)
Diuretics in combination*	9 (3.6)	12 (4.8)
Beta blocker	163 (64.7)	215 (86.3)
Non-selective	4 (1.6)	5 (2)
Selective	159 (63.1)	212 (85.1)
Beta blockers in combination*	6 (2.4)	6 (2.4)
ACE inhibitor	133 (52.8)	191 (76.7)
Vasodilator	80 (31.7)	103 (41.4)
Calcium channel blocker	31 (12.3)	96 (38.6)
Statin	60 (23.8)	38 (15.3)
Cardiac glycoside	35 (13.9)	78 (31.3)
ARB	37 (14.7)	64 (25.7)
Alpha blocker	8 (3.2)	9 (3.6)
Top five multiple regimens (n = 186)^†
ACEI/Beta blocker/Diuretic	19 (0.102)	25 (0.115)
ACEI/Diuretic	13 (0.07)	17 (0.073)
ACEI/Beta blocker/Diuretic/Statin/Vasodilator	13 (0.07)	17 (0.073)
ACEI/Beta blocker/Statin/Vasodilator	12 (0.065)	12 (0.051)
ACEI/Beta blocker/Diuretic/Vasodilator	9 (0.048)	12 (0.051)

*Combination assessed based on codes (ATC) used for medications in prescription registry.

^†Total number of patients on multiple regimens in the index period.

ACE, angiotensin converting enzyme; ARB, angiotensin receptor blocker; HF-REF, heart failure-reduced ejection fraction.

During the post-index period almost all patients received medication (Table 3). The most commonly prescribed agents were beta blockers (86.3% of patients, with 85.1% receiving selective agents), diuretics (81.9% [loop diuretics 72.3%]), ACE inhibitors (76.7%), vasodilators (41.4%), and statins (38.6%). In contrast to the index period, the mono regimen therapy most commonly prescribed post-index was an ACE inhibitor (71.4% of those on monotherapy); however, multiple regimen treatment comprising an ACE inhibitor, beta blocker, and a diuretic remained the most common (11.5% of those on multiple regimen therapy).

Table 3.  Hospitalization and outpatient resource utilization costs per year of follow-up.

Resource	Unit costSEK (EUR)	HF-REF patients (n = 249)
		Mean number of events (Range)	Mean costs in SEK (Range)	Mean costs in EUR (Range)
Hospitalizations
Cardiovascular^†	41,088 (4,306)	2.5 (0–81.1)	102,779.1 (0–3,332,656.6)	10,771.2 (0–349,262.4)
Cardiovascular—non-heart failure^†	41,088 (4,306)	0.4 (0–13.5)	30,166.3 (0–555,442.8)	3,161.4 (0–58,210.4)
Heart failure^†	35,436 (3714)	2.0 (0–81.1)	72,612.8 (0–2,874,224.6)	7609.8 (0–301,218.7)
Outpatient clinic visits^‡
Cardiac care	3453 (362)	0.8 (0–7.6)	2814.7 (0–26,393.5)	295.0 (0–2766.0)
Gastroenterologic care	1907 (200)	0.0 (0–0.9)	16.8 (0–1718.8)	1.8 (0–180.1)
Infectious disease care	2034 (213)	0.2 (0–8.2)	365.9 (0–16,786.8)	38.3 (0–1759.2)
Internal medicine^a	3287 (344)	1.4 (0–28.1)	4552.9 (0–92,281.6)	477.1 (0–9671.1)
Neurologic care	2371 (248)	0.0 (0–1.7)	65.6 (0–3988.3)	6.9 (0–418.0)
Psychiatric care^b	2180 (228)	0.0 (0–3.2)	110.6 (0–6952.7)	11.6 (0–728.6)
Renal care	2094 (219)	0.0 (0–1.8)	58.3 (0–3697.5)	6.1 (0–387.5)
Respiratory care	2511 (263)	0.2 (0–10.0)	403.0 (0–25,088.7)	42.2 (0–2629.3)
Surgical care^c	1358 (142)	0.9 (0–11.3)	1170.3 (0–15,389.5)	122.6 (0–1612.8)
Other	1435 (150)	0.6 (0–8.6)	884.6 (0–12,310.0)	92.7 (0–1290.1)
Primary care^‡
Doctor visit	1554 (163)	1.2 (0–91.2)	1830.0 (0–141,778.1)	191.8 (0–14,858.4)
Doctor home visit	1699 (178)	0.0 (0–0.5)	6.1 (0–797.9)	0.6 (0–83.6)
Doctor telephone call	1554 (163)	0.0 (0–1.2)	45.8 (0–1932.2)	4.8 (0–202.5)
Nurse visit	1175 (123)	0.0 (0, 1.9)	31.2 (0–2254.1)	3.3 (0–236.2)
Nurse telephone call	777 (81)	0.0 (0–0.4)	1.4 (0–344.5)	0.1 (0–36.1)
Assistant nurse visit	947 (99)	0	0	0
Other visit	1108 (116)	0.1 (0–3.4)	131.6 (0–3744.9)	13.8 (0–392.5)
Medications
Cardiac glycoside	–	–	50.9 (0–460.5)	5.3 (0–48.3)
Alpha blocker	–	–	70.3 (0–6,253)	7.4 (0–655.4)
Statin	–	–	280.5 (0–8543.5)	29.4 (0–895.4)
Diuretic	–	–	594.5 (0–7581.8)	62.3 (0–794.6)
Beta blocker	–	–	862.5 (0–4159.2)	90.4 (0–435.9)
Calcium channel blocker	–	–	87.1 (0–3006.9)	9.1 (0–315.1)
ACE inhibitor			439.9 (0–5442.8)	46.1 (0–570.4)
ARB			824.5 (0–5972.4)	86.4 (0–625.9)
Vasodilator			255.0 (0–11,086.5)	26.7 (0–1161.9)

^†Hospitalization unit costs are the same for all patients and for subsequent hospitalizations.

^‡Unit costs for outpatient clinic visits include any procedure cost that could have occurred during the visit.

^aIncludes internal medicine, oncologic care, general, hematologic care, rheumatologic care, endocrine treatment, and geriatric care.

^bIncludes general, adult, and geropsychiatric care.

^cIncludes obstetric and gynecologic, surgical, plastic surgical, thoracic surgical care, hand surgical, and urologic care.

HF-REF, heart failure-reduced ejection fraction.

SD = standard deviation

Healthcare resource utilization and cost

Almost two thirds of patients had at least one cardiovascular hospitalization during the post-index evaluation period, and more than half had at least one heart failure-related hospitalization (See Appendix B for further details). On average, the patients assessed were hospitalized at least twice a year for cardiovascular (Median = 1 event; SD = 2.9) and 1.5-times per year for heart failure events (median = 1 event; SD = 2.7). For cardiovascular-related hospitalizations, the average length of stay ranged from 6 days for the first hospitalization to more than 10 days by the fifth hospitalization (See Appendix B for further details). For heart failure-related hospitalizations, the admission duration ranged from ∼7–10 days from the first to the fifth hospitalization in the post-index period. For both the cardiovascular and heart failure-related hospitalization, the median number of days between hospitalizations declined after the first hospitalization, with hospitalizations becoming progressively more rapid (See Appendix C for further details).

Outpatient visits, i.e., any clinic or primary care visit for patients with HF-REF in this study occurred at an average of 1.4 visits/year for cardiovascular disease-related reasons and at 0.9 visits/year for heart-failure-related reasons (See Appendix D for further details), while the average total number of outpatient visits per patient was 5.3/year. Surgical procedures were performed slightly more often than non-surgical procedures (1.9 vs 1.2 procedures/year), and the median total number of individual item prescriptions for evaluated medication was 18.0 per year.

The overall mortality rate over the 3-year period was 22.6% for all-cause mortality, 5.2% being cardiovascular-related, and 2.0% related to heart failure; 16% of patients died within the first year following the index date, of which 3.6% died of cardiovascular causes, and 1.6% died due to heart failure (See Appendix E for further details).

The total mean direct annual cost for HF-REF patients was SEK 118,771 (EUR 12,447). The chief component of this cost was for heart failure-related hospitalizations (Figure 2). The mean per patient cost for total cardiovascular hospitalization that included heart failure hospitalization during the first follow-up year was SEK 102,779 (EUR 10,771) (Table 3). The mean cost for non-heart-failure-related hospitalizations was SEK 30,166 (EUR 3,161), while the cost for heart-failure-related hospitalizations was SEK 72,613 (EUR 7,610). Thus, the majority (61%) of the cost for hospitalization in this patient population can be attributed directly to heart failure. The highest costs for outpatient clinic visits per patient per follow-up year were for internal medicine (SEK 4553; EUR 477). The main driver for primary care costs (SEK 1830; EUR 192) was the doctor’s visit, which occurred on average 1.2 times/year (Table 3). Regarding medication costs, the highest mean cost per item was for beta blockers (SEK 841, EUR 88; Table 3), which could be expected, as these were most commonly prescribed. Prescription costs, on the whole, were low at SEK 3503; EUR 367, or only 3% of the total annual cost per patient.

Figure 2.  The average total cost per follow-up year per HF-REF patient for each cost category—cardiovascular hospitalization cost, heart failure hospitalization cost, outpatient clinic visit cost, outpatient primary care cost, and prescription drug cost is shown in the figure.

Multivariate results suggest that patients who were treated as inpatients at index had a greater number of post-index cardiovascular-related hospitalizations per year (relative rate 1.56; p = 0.051) as well as a higher annual number of post-index heart failure-related hospitalizations (relative rate: 2.25; p = 0.010). Hospitalized patients were more often aged 65 years or older (relative rate for cardiovascular hospitalization 1.56, p = 0.034; relative rate for heart failure hospitalization 1.90; p = 0.012). The rate of hospitalizations increased significantly when there were three or more pre-index comorbidities present, compared with no comorbidities (relative rate 1.73, p = 0.023 for cardiovascular hospitalizations; relative rate 2.07, p = 0.009 for heart failure hospitalizations).

All-cause mortality for the study population was high: 15.9% of patients died within 1 year after the index date, with 3.57% of the population dying of cardiovascular-related causes and 1.6% dying of heart failure-related causes during this time. A similar mortality distribution trend was observed from 1–2 years post-index, with an overall mortality of 5.6%, and 1.59% and 0.4% mortality being recorded for cardiovascular and heart failure mortality, respectively. There were no deaths reported between 2–3 years post-index (27.8% of patients were alive as of the 2-year mark).

Discussion

The purpose of this study was to assess healthcare utilization and costs for patients with a confirmed diagnosis of HF-REF in Sweden. Overall healthcare utilization by these patients is high, with >2 hospitalizations per patient per year. The average annual total cost per patient was SEK 113,648 (EUR 11,910), which is considerably higher than the annual cost suggested in other studies. For example, a study by Agvall et al.8 of primary healthcare-based patients using prices from 2003 calculated a per-patient annual total cost of SEK 37,100, amounting to an annually adjusted total cost of SEK 40,489 (EUR 4243) in 2010 values. The patients in their study had a much lower mean number of hospital days; with all-cause hospitalization accounting for only 47% of the total cost, vs 86% of the total cost in our study. In contrast, medication costs accounted for only 3% of the total cost in our study, compared with 18% in the Agvall et al. study. Other authors have estimated the cost for medication to represent ∼11% of the total cost for heart failure, although it is possible that these populations may have a different profile compared to the study population used in this analysis2,5. Specifically, the Agvall et al. study was undertaken using data from primary care centers20, and therefore it is likely that the patient population is healthier relative to that used in our research. Additionally, the Agvall et al.20 study included a large proportion (29%) of HF patients with normal cardiac function (defined in that study as LVEF > 50%), further indication that the population may have had less severe disease with fewer comorbidities when compared to our study. The duration of hospitalization shown in our study (mean 6.7 days for first heart failure hospitalization) is in line with published data for average lengths of stay ranging from 3.5–9.5 (mean 6.5) days between hospitals5. Similar to this report, another group also calculated that the costs for institutional care, including hospitals and nursing homes, which comprises 65–75% of heart failure treatment costs2. In a study from the UK, hospitalization represented the predominant cost factor in heart failure (69%); while medication accounted for 18% of overall costs12. However, it should be noted that the definition of heart failure and study entry criteria differ among these studies.

Across of the body of literature, little is available in terms of epidemiologic data specific to REF-Heart failure. As a result, it is difficult to know if the study patient population is closely representative of the general HF population in Sweden, or worldwide; however, the treatment of patients within the Uppsala county council is in accordance with general guidelines, suggesting that the treatment patterns observed in this analysis are representative of the Swedish healthcare system.

The annual per patient hospitalization cost for heart failure in this study was SEK 72,613 (EUR 7610), representing 61% of the average total patient healthcare cost. Therapeutic interventions that prevent or delay hospitalization, whether through more effective, guideline-directed treatment and adequate management of patient comorbidities or new treatment modalities, could potentially decrease the costs of managing patients with HF-REF appreciably4,5. For example, the study results demonstrate that, despite European Society of Cardiology recommendations21, ACE inhibitors and ARBs were both used in a relatively low percentage of HF-REF patients (53% and 14%, respectively) in the index period. While these rates increase in the post-index period (ACE inhibitor use: 77%; ARB use: 26%), it is likely that hospitalizations could be prevented by improvements in medication adherence, within the index period. While not specific to Sweden, It is worth noting that the 2011 ACC/AHA/PCPI Heart Failure Performance Measure Set in the US specifies the administration of ACE inhibitors and ARBs (prescribed within a 12-month period when discharged from the hospital, or when seen in an outpatient setting) used as quality-of-care measures in HF patients22. Increased use of guideline recommended treatment would also likely add to the quality-of-life for many patients, improving symptom control. Furthermore, follow-up schedules and frequency may be optimized, leading to substantial cost savings, involving outpatient centers that employ nurses and doctors specializing in heart failure5,11,23.

Albeit retrospective, the assay method for this study to evaluate resource utilization and per patient costs by extracting data from an unselected primary care population as well as from national registries has been applied and validated in several other studies13,14. The main limitations of this study include a short follow-up time and small sample size. The short follow-up period adversely impacted extrapolated calculations based on annual incidence rates (i.e., hospitalizations/year, outpatient visits/year) increasing the maximum value for the range. Requiring a LVEF assessment within the index period (to exclude patients with preserved ejection fraction) for study inclusion severely restricted the potential follow-up period. Echocardiograms were available for only 28% of the initial potential study population with a heart failure diagnosis; this rate is in accordance with those reported elsewhere10. Information on patients’ New York Heart Association (NYHA) functional class was not accessible, and, as a result, the impact of disease severity on cost could not be assessed. Further, the results are limited by a lack of data on validity period for prescriptions or the units of medication prescribed, therefore medication possession ratio or adherence could not be calculated. Also of note, the overall mortality could have been under-estimated, as only the primary cause of death was included in the analysis. Additionally, data on devices such as pacemakers and procedures such as revascularization were missing, and therefore the results of this study are likely to under-estimate the overall cost burden. Multivariate analysis was performed with a limited number of variables as several baseline characteristics were not available. Finally, medication dosing was not available in the database, making it difficult to accurately evaluate treatment patterns and their impact on costs.

Relative to costs presented in a prior publication8, the estimated annual per patient healthcare cost for HF-REF in Sweden in this study was unexpectedly high at over SEK 113,000 (EUR 11,000). Patients in this survey have a high rate of re-hospitalization that accounts for the largest proportion of the total healthcare cost. Patients also have a high mortality, despite full access to resources in a state-based healthcare system, and the availability of current treatment options.

An improved follow-up management algorithm, possibly including better access to echocardiography in primary care with regular testing of natriuretic hormones, could potentially facilitate tailored pharmacotherapy and lead to improved management and outcomes for this population. A tailored pharmacotherapy is especially important in an elderly population with multiple co-morbidities, where there is a need to balance different goals rather than adding treatment goals for each separate disease. There is also a real need for improved treatment options, including more efficacious medication. While there are many factors at play, improved management and treatment options could potentially reduce re-hospitalization rates, decrease mortality, drive healthcare costs down, and thus alleviate the overall healthcare burden. Further research is required to determine reasons that medications which would reduce morbidity and mortality are both under-utilized and under-dosed10,11.

Transparency

Declaration of funding

This work was supported by Novartis Pharmaceuticals Corporation

Declaration of financial/other relationships

JS received compensation for his participation in the research. LS, SS, and RP are employed by Analytica International, which received funding for the research. RL is employed by Pygargus AB, which received funding for the research. RA was employed by Novartis Pharmaceuticals Corporation at the time of this research; Novartis provided funding for the research.

Acknowledgments

Thanks to Dr Dan Sandberg for providing data from the from the local echocardiography registry kept by the Department of Physiology, Uppsala University Hospital.

Thanks to Dr. Lars Benson, Luthagens Specialistläkar mottagning, Husläkargruppen in Uppsala and all public Primary Care Centers within Uppsala County Council for contributing with primary care data.

Editorial assistance was provided by Elisabeth Smith and Jacob M. Willet, MPH, both of whom received compensation from Analytica International for their work.

These data have been presented in part at the 2011 Heart Failure Congress, Gothenburg, Sweden (May 21–24).