Humanistic burden and economic impact of heart failure – a systematic review of the literature

Systematic literature review

A systematic search was performed using MEDLINE^® and MEDLINE In-Process, Embase, and the Cochrane Library via Ovid between January 2002 and May 2017 for the humanistic burden SR, and between January 2007 and May 2017 for the economic impact SR. Supplementary searches included reviews of congress abstracts between 2015 and 2017 (or the most recent 2 years available) for the following meetings: International Society for Pharmacoeconomics and Outcomes Research US and European congresses, American Heart Association Scientific Sessions, European Society of Cardiology (ESC) Congress, World Congress of Cardiology and Cardiovascular Health, American College of Cardiology Annual Scientific Session, and ESC Heart Failure. The search strings used to identify evidence are listed in Table 1. This review was not prospectively registered with the National Institute for Health Research International Prospective Register of Systematic Reviews (PROSPERO), but the protocol is available on request from the corresponding author.

Citation screening and full text review

Abstracts and titles identified were screened by an independent reviewer to determine whether they met the PICOS (patient, interventions, comparisons, outcomes, and study design) eligibility criteria (Table 2), in accordance with 2009 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines²¹. All publications that met the entry criteria for the review were obtained as full articles and reassessed against the review criteria. Owing to the large number of citations meeting the predefined inclusion criteria, a decision was made to restrict data extraction from eligible studies. For both SRs, data were extracted only from full publications of observational studies. Data were not extracted if the study population included fewer than 200 patients for the humanistic burden SR, and fewer than 100 patients for the economic impact SR. The restriction criteria used are summarized Table 2.

Data extraction

For each publication, information was extracted into a data-extraction table. For each study, NYHA classification, age, and comorbidities including CKD and DM were recorded.

Search results

In the initial searches, 11 622 papers were identified, of which 2186 were removed as duplicates, and 9436 papers were included for electronic screening. Electronic screening identified 8166 papers that did not meet the inclusion criteria. In total, 124 papers were identified for inclusion following full paper review: 54 papers reporting HRQoL and 71 papers reporting costs and resource use (including one reference that reported data for both outcomes). A PRISMA flow diagram is shown in Figure 1. The studies identified were grouped according to the key data presented, as shown in Figure 2.

Figure 1. PRISMA flow diagram, showing the flow of studies identified through the systematic review process.

HRQoL, health-related quality of life. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2. Summary of evidence identified in this systematic review, showing the number of studies identified, grouped according to the key data presented.

HF, heart failure. HRQoL, health-related quality of life. ^aSome studies reported data that fell into more than one category.

HRQoL in patients with and without HF

A small number of studies assessed HRQoL in patients with HF compared with individuals without HF, or patients with other chronic diseases. In a study conducted in the Netherlands, patients with HF (NYHA class II–IV) reported worse scores across all domains of the 36-item Short-Form Health Survey (SF-36) than a population of age- and sex-matched community control individuals (Figure 3)²². A Danish study also found that patients who had been discharged from hospital reported lower 12-item Short-Form Health Survey (SF-12) physical component summary (PCS; mean±standard deviation [SD]: 42.5±11.1 vs 47.7±13.4, p<0.001) and mental component summary (MCS; mean±SD: 48.2±10.9 vs 51.5±14.7, p<0.001) scores than a reference group representative of the national population²³. A further cohort survey, conducted in the UK, compared patients with HF with those who had asthma, chronic obstructive pulmonary disease (COPD), DM, epilepsy, or stroke. Patients with HF reported lower baseline 5-dimension EuroQol questionnaire (EQ-5D) utility and visual analogue scale (VAS) scores than patients with asthma, DM, epilepsy, and stroke, and similar scores to patients with COPD²⁴.

Figure 3. SF-36 scores in patients with heart failure compared with the general population²².

This figure provides a graphical representation of the component scores for the SF-36 HRQoL questionnaire for patients with HF (stage II–IV) compared with community-dwelling elderly people who do not have HF. BP, bodily pain. GHP, general health perception. HF, heart failure. HRQoL, health-related quality of life. MH, mental health. PF, physical functioning. RL-M, role-limiting (mental). RL-P, role-limiting (physical). SF, social functioning. SF-36, 36-item Short-Form Health Survey. V, vitality.

HRQoL by HF severity

Studies from Sweden²⁵, Australia²⁶, and the USA^27,28 assessed HRQoL by NYHA HF class and reported a greater humanistic burden with more severe disease (Figure 4). Based on the evidence identified, there was a reduction in HRQoL between NYHA HF classes I and II^25,27,28 and classes II and III^25–28, and between classes III and IV^25–28, assessed using the Minnesota Living with Heart Failure Questionnaire (MLHFQ) and the Kansas City Cardiomyopathy Questionnaire (KCCQ), both of which are cardiac-specific instruments. One study also reported decreasing SF-36 PCS and MCS scores with increasing NYHA HF class²⁵. For SF-12 scores, there was a greater reduction between NYHA classes I and II, and smaller reductions between more severe NYHA classes (Figure 4). A consistent decrease in utilities between classes II and III (–0.042) and classes III and IV (–0.041), measured using the 5-level EQ-5D (EQ-5D-5L) was also reported by a single study alongside a decline in MLHFQ scores (Figure 4)²⁶.

Figure 4.

Humanistic burden with increasing New York Heart Association class measured using: (a) MLHFQ^25,27, (b) KCCQ^26,28, (c) SF-36²⁵, and (d) EQ-5D-5L²⁶.

This figure provides a graphical representation of HRQoL scores by NYHA class as measured by the MLHFQ, KCCQ, SF-36, or EQ-5D-5L questionnaires. Better health-related quality of life is represented by higher KCCQ, SF-36, and EQ-5D-5L scores, and lower MLHFQ scores. EQ-5D-5L, 5-level, 5-dimension EuroQol questionnaire. HFpEF, heart failure with preserved ejection fraction. HFrEF, heart failure with reduced ejection fraction. HRQoL, health-related quality of life. KCCQ, Kansas City Cardiomyopathy Questionnaire. MCS, mental component summary. MLHFQ, Minnesota Living with Heart Failure Questionnaire. NYHA, New York Heart Association. PCS, physical component summary. SF-36, 36-item Short-Form Health Survey.

Two further studies, both conducted in the USA, reported a negative correlation between HRQoL and increasing HF severity. A study of patients with HF living in a rural location reported a negative association between NYHA HF class and MLHFQ score, but did not report results by separate NYHA classes²⁹. Another study investigated HRQoL in a subgroup of patients with HF as a comorbidity, from a population with CKD in a subanalysis of the Hemodialysis Study, a randomized controlled trial of patients undergoing haemodialysis. Reductions in both mental and physical HRQoL were reported with increasing HF severity (categories ranging from ‘absent’ to ‘severe’). Scores in the role-limiting (physical), bodily pain, general health, vitality, social functioning, and mental health domains, as well as overall PCS scores, decreased with increasing HF severity³⁰.

Few longitudinal studies reporting HRQoL in HF were identified. Of the studies that were purely observational, two reported a decrease in HRQoL over time, whereas the remaining three reported an improvement in HRQoL across the study period. A UK cohort study investigating HRQoL in patients with HF relative to those with other chronic diseases reported generic and disease-specific HRQoL scores at baseline and 1 year. Patients with HF were the only group to report a reduction in HRQoL over the follow-up period, with significantly lower EQ-5D VAS scores indicating worse HRQoL at 1 year²⁴. A US study reported HRQoL over a follow-up of up to 6.6 years, adjusting for the effect of time on disease deterioration³⁰. PCS scores decreased over time, along with scores for six of the eight SF-36 domains; MCS scores and scores for the role-limiting (mental) domain did not decrease over time³⁰.

A study in the Netherlands, however, reported that patients with a higher level of education experienced a statistically significant improvement in the role-limiting (emotional) domain of the SF-36 across an 18-month longitudinal study (p=0.015). No significant difference was found in the remaining SF-36 domains³¹. Patients in a Canadian study in whom HF was diagnosed following an emergency department (ED) visit reported increases in physical HRQoL scores over a 12-month period (SF-12 PCS score: 1.6–3.4; KCCQ-physical limitation score: 5.6–9.4)³². An analysis modelled health trajectories over 12 months defined by patients’ KCCQ scores over time using data from the Patient-Centered Disease Management for Heart Failure trial. Patients with a poor or moderate health status trajectory experienced improvements in KCCQ scores between baseline and 3 months followed by a stabilization in scores up to 12 months, whereas those who had a trajectory of ‘marked improvement’ reported an increase in HRQoL over the whole 12-month period³³.

In a number of the longitudinal studies identified, healthcare professionals provided support to patients over the follow-up period, which was considered likely to impact HRQoL. These studies are summarized in Extended data: Supplemental Table 1³⁴. Studies in Norway³⁵ and in Spain³⁶ reported a statistically significant improvement in MLHFQ scores between baseline and 6 or 12 months, respectively, for patients monitored over the study period. In the Spanish study, patients continued to experience an increase in HRQoL over 5 years³⁶. An improvement in MLHFQ scores over 12 months following enrolment into an HF clinic was also reported by patients in a Canadian study³⁷.

HRQoL in patients with HF by ejection fraction status

Few studies comparing HRQoL in patients with HFpEF and those with HFrEF were identified. In a study conducted in Germany, patients with HFpEF reported better physical functioning, measured using the SF-36, than patients with HFrEF³⁸. A study conducted in the USA reported that patients with HFpEF experienced no greater impact on HRQoL than those with HFrEF as disease severity increased. However, patients with HFpEF, with KCCQ scores in the lowest quartile, had a higher risk of mortality or hospitalization as well as lower rates of event-free survival than those who had scores in the highest quartile²⁷.

HRQoL in patients with HF and comorbidities

Comorbidities in patients with HF were typically associated with reductions in HRQoL (Extended data: Supplemental Table 1³⁴). In studies conducted in Spain and the USA, respectively, patients with HF and iron deficiency³⁹ or pain⁴⁰ reported higher overall and physical summary MLHFQ scores, representative of worse HRQoL, than those without these comorbidities. However, a further study in the USA found that although patients with anaemia reported worse baseline KCCQ scores than those without, deterioration in HRQoL over a 3-month period was similar in both populations²⁸. Patients in a study in the Netherlands who experienced difficulties with sexual activity reported worse HRQoL than those without difficulties⁴¹. In a US study, excessive daytime sleepiness and cognitive impairment were associated with comparatively lower HRQoL⁴². In a second US study, patients who had comparatively lower haemoglobin levels at baseline reported worse KCCQ and MLHFQ scores than those with higher haemoglobin levels⁴³.

Patients in Australia, the UK, Ireland, the USA, and Norway who had ischaemic heart disease with ischaemic HF, angina, or a myocardial infarction as a comorbidity reported PCS scores that were lower than expected for the general population (more than one SD below the standardized mean of the general US population), whereas MCS scores were within one SD of the mean⁴⁴.

HRQoL and mental health in patients with HF

Poor mental health was generally associated with worse HRQoL in patients with HF. Comparative studies were identified that focused on depression, suicidal ideation, panic disorder, and distress (Extended data: Supplemental Table 1³⁴).

Patients with HF and depression had worse HRQoL than those without depression in studies from Germany, Italy, and the USA. Statistically significant reductions in scores across all SF-36 domains and the KCCQ summary domains were found with increasing severity of depression in two studies in Germany^45,46. A further German study reported an association between lower SF-36 physical functioning (PF) and bodily pain scores as severity of depression increased⁴⁷. Furthermore, in a study in the USA, PF was negatively correlated with depression⁴⁸. Worse health status, as defined by KCCQ score, was associated with a comparatively higher risk of depression at baseline in another US study³³; and a German study indicated that patients with type D personality, predictive of depression, reported worse HRQoL (MLHFQ scores) than those with a different personality type⁴⁹.

Suicidal ideation and ideas of self-harm were reported to be associated with reduced HRQoL⁵⁰. Patients with panic disorder⁵¹ and those who experienced distress at baseline and/or follow-up⁵² reported worse HRQoL than those who did not. Another study reported an association between patients’ end-of-life preferences and their physical HRQoL (SF-36 PCS scores)⁵³.

Patients’ perceived health status was reported to have a moderate impact on MLHFQ score in a Canadian study⁵⁴, and social functioning, measured using the KCCQ, was found to be a predictor of perceived health status, particularly in male patients with HF in the USA⁵⁵. High levels of self-care were associated with better MLHFQ or KCCQ scores in studies in Italy⁵⁶ and the USA⁵⁷, respectively, as well as lower hospitalization rates in the Italian study.

Hospitalization and HRQoL in patients with HF

Patients hospitalized for HF experienced comparatively better HRQoL once discharged from hospital (Extended data: Supplemental Table 1³⁴). A study in the USA and Canada found that EQ-5D scores were higher at the time of hospital discharge than when patients were admitted to hospital (baseline), and remained higher than baseline 30 days after discharge from hospital⁵⁸. Another US study showed that MLHFQ scores improved significantly during the first month post-discharge (p<0.001)⁵⁹. Furthermore, patients in a study in the USA who did not experience readmission to hospital within 30 days of discharge experienced better HRQoL (KCCQ scores) than those who did⁶⁰. Fear of hospitalization had a negative impact on patients’ MLHFQ scores in an Italian study⁶¹, and unplanned hospitalization was identified as a predictor of better SF-12 PCS scores in a patient population who had previously been hospitalized for HF in a study in Denmark²³.

Impact of disease management on HRQoL in patients with HF

Owing to the chronic nature of HF, the implementation of appropriate disease management is an important part of treatment, and evidence identified in this SR indicates that it is associated with HRQoL improvements (Extended data: Supplemental Table 1³⁴). Patients in the UK⁶² and Germany⁶³ with HF and either atrial fibrillation or sinus rhythm reported an improvement in HRQoL following cardiac resynchronization therapy and treatment with β-blockers, respectively. The use of left ventricular assist devices (LVADs) as bridging or destination therapy was also associated with an improvement in HRQoL in a US study⁶⁴, and patients with an LVAD had better HRQoL than patients assessed for, or awaiting, heart transplant in a UK study⁶⁵.

HRQoL as a predictive factor in patients with HF

Several studies were identified that reported HRQoL as a predictive factor both of mortality and of outcomes in patients with HF (Extended data: Supplemental Table 1³⁴). Scores in the SF-36 and SF-12 general health and PF domains were predictors of mortality in studies conducted in the Netherlands⁶⁶ and the USA⁶⁷, and were associated with an increased risk of hospitalization and ED visits in another study in the USA⁶⁸. Two further studies, one in Italy⁶⁹ and the other multinational⁷⁰, found that KCCQ score was a predictor of mortality, and a study in the USA reported a negative association between MLHFQ score and cardiac event-free survival⁷¹. Physical and depressive symptoms, in addition to spiritual well-being and comorbidity count, were negatively correlated with HF-specific health status (partially determined by KCCQ score) in patients in the USA⁷². An Australian study examined the contribution of MLHFQ and EQ-5D scores to composite scoring systems in HF trials⁷³.

Impact of sociodemographic status on HRQoL in patients with HF

In several studies, patient ethnicity and sex were shown to impact HRQoL (Extended data: Supplemental Table 1³⁴). A study in the USA found that Hispanic patients reported better HRQoL than those of white or black ethnicity⁷⁴. In two Canadian studies, women reported worse physical and overall HRQoL, respectively, than men at baseline^32,75, and worse physical HRQoL at 12 months⁷⁵. Sociodemographic factors including marital status, education, income, and employment were correlated with worse HRQoL in studies in the Netherlands³¹ and in Canada, France, and the USA⁷⁶.

Total costs for HF

The overall costs associated with the treatment for HF varied widely across studies. This is likely to result from disparities in study settings and populations, which meant that the patients included differed in terms of disease severity, clinical history, and comorbidities.

Five studies conducted in Europe reported total annual healthcare costs associated with HF. The lowest annual per-patient costs reported were €3150 in a German study⁷⁷, followed by €5700 in a Swedish study⁷⁸, and €6571 in a study conducted in Spain⁷⁹. Two Italian studies, each of which included only patients who had experienced an HF-related hospitalization, reported total annual costs of €11 100⁸⁰ and €11 864⁸¹, respectively.

The lowest total annual costs reported in the USA were for a subpopulation of individuals enrolled in Medicare; patients with HF who were not classed as having a low income or dual eligibility for Medicare and Medicaid (non-low-income/dually eligible [LI/DE] cohort) incurred per-patient costs of US$13 897 annually. This value was US$17 840 for patients with HF who were classed as LI/DE⁸². The highest annual US costs identified were from a study comparing patients with HF who died within 1 year and those who survived. Surviving patients incurred total mean per-patient costs of US$36 426 per annum⁸³. A further four US studies reported annual per-patient costs in the range of US$16 912–29 456^84–87. Two additional US studies recorded total costs for patients in the period before they died as a result of HF; these data are not included in the overall range, but are presented in Extended data: Supplemental Table 2^88–90. In addition to evidence from Europe and the USA, total annual costs associated with HF were reported as C$27 809 in a Canadian study⁹¹ and ¥28 974 in a study from China⁹².

Rather than reporting annual total costs for HF, two studies presented data collected over different time periods (Extended data: Supplemental Table 2⁹⁰). A UK study reported the total costs incurred in the year before an HF event and in the 36 months after the event⁹³, and a US study reported total Medicare payments over 30 days for patients discharged from hospital, grouped based on dyspnoea severity⁹⁴.

Costs and resource use for patients with and without HF

Eight studies, all from the USA, compared costs and resource use for patients with HF with those for individuals without HF, or with other chronic diseases. Total annual costs were significantly higher for patients with HF than for those without HF (p<0.001): 4.6 times higher (US$27 152 vs US$5952) during 2010–2011, and 4.3 times higher during 2002–2011 (US$23 854 vs US$5511)⁸⁵. A breakdown of constituent costs in this study is shown in Figure 5. Another study compared allowed monthly Medicare costs and resource use for patients with HF and all individuals who made Medicare claims (fee-for-service population). Per-patient allowed monthly costs for HF were 3.2 times higher than those incurred by the average fee-for-service patient (US$3395 vs US$1045)⁹⁵. Patients with HF also had comparatively higher rates of inpatient admission, 30-day readmission, and use of skilled nursing facilities, meaning that higher costs were incurred for all of these resources, compared with the Medicare fee-for-service population.

Figure 5. Healthcare costs for patients with and without heart failure in the USA⁸⁵.

This figure provides a graphical representation of annual healthcare expenditure with constituent costs for individuals with or without HF between 2002 and 2011. HF, heart failure.

Additional evidence indicated that HF is associated with increased resource use. A study of patients who were admitted to hospital for trauma and subsequently discharged found that the presence of HF was associated with a 71% increased risk of 30-day hospital readmission (24% vs 14%; p<0.0001)⁹⁶. An examination of the resource use associated with hospital discharge for patients with HF, compared with all patients discharged, showed that costs for cardiology, supplies and devices, coronary care, and operating room services were 1.3–3.3 times higher for patients with HF⁹⁷. Although few studies focused on indirect costs, one study examining care requirements was identified. In patients aged over 50 years, individuals with HF were significantly more likely than those without HF to make use of either formal (paid) care (HF: 9.1%; no HF: 1.5%; p<0.001) or informal care (HF: 33%; no HF: 8.6%; p<0.001), defined as care provided by a family member or unpaid volunteer⁹⁸. Patients with HF required significantly more informal care per week than those without HF (32.1 hours vs 25.1 hours; p=0.002).

Three studies compared the costs associated with HF and those for other chronic diseases. One study showed that similar total annual costs were associated with HF and obesity (US$1642 and US$1908, respectively); however, these conditions were considerably more costly than hypertension, which was associated with costs of US$431 per annum⁹⁹. A second study reported that total annual costs were nearly twofold higher for HF than for DM, partly owing to higher inpatient and outpatient costs⁸⁷. Finally, a study reporting disease-specific costs showed that HF incurred higher annual medical and pharmacy costs than asthma, coronary artery disease, COPD, DM, hyperlipidaemia, and hypertension⁸².

Costs for HF as a comorbidity

Three studies showed that HF is associated with additional costs in patients with DM. A UK study estimated that annual inpatient care costs for a typical 60-year-old man with type 2 DM are more than six times higher in the presence of HF than if HF does not occur (£3191 for HF vs £459 for no HF)¹⁰⁰. The occurrence of HF in a patient with DM was also estimated to result in 84% higher non-inpatient costs. The cost of preventable hospitalization for HF in patients with DM was reported as US$7949 in a US study¹⁰¹, and a second US study found that acute care for an HF episode incurred costs of US$23 758 per event-year, plus US$1904 in ongoing management costs¹⁰². The impact of HF in obese patients was also examined in one study. Patients with obesity but no HF incurred annual total costs of US$1908; however, this rose to US$5276 in patients with HF as well as obesity⁹⁹.

Two US studies reported the cost impact of comorbidities in patients with HF (Extended data: Supplemental Table 2⁹⁰). Shaya et al. estimated the annual cost savings that could be made via a 20% reduction in a range of comorbidities, the most costly of which were cardiovascular disease and DM¹⁰³. Smith et al. estimated the contribution of various comorbidities to the overall cost of care for patients with HF, finding that DM incurred the greatest costs of the comorbidities examined in the study¹⁰⁴.

Costs by severity of HF

Only one study, conducted in Japan, reported costs by HF severity¹⁰⁵. Hospitalization costs were modelled across NYHA classes: compared with NYHA class II, classes III and IV were associated with an additional US$490 and US$640, respectively (p<0.001).

Costs by HF ejection fraction status

Two Swedish studies collected data on the total annual direct costs associated with HFrEF¹⁰⁶ and HFpEF¹⁰⁷. These studies, however, were not conducted simultaneously or designed to compare costs by ejection fraction, so limited inference should be drawn from across-study comparisons. HFrEF was associated with slightly higher annual costs than HFpEF (€12 447 vs €11 344), as a result of higher costs associated with inpatient care, hospitalizations, and medication. Annual outpatient clinic costs were approximately 43% higher for HFpEF than for HFrEF (€1561 vs €1094).

Two US studies compared resource use between patients with HFrEF and those with HFpEF. One study found no significant difference between the two groups in terms of 30-day hospital readmission rate. However, patients with HFrEF had a significantly greater length of stay (LoS) in hospital than patients with HFpEF (HFrEF: 10.9 days; HFpEF: 8.5 days; p=0.027)⁹⁶. In a second study, there were no significant differences between groups in terms of adjusted 30-day readmission rates, annual hospitalization rate, LoS, or pharmaceutical dispenses. Although the absolute differences between cohorts were small, HFpEF was associated with significantly more outpatient visits (HFpEF: 21.5; HFrEF: 20.1; p<0.002) and ED visits (HFpEF: 3.24; HFrEF: 2.94; p<0.002) annually than HFrEF, as well as a significantly higher rate of readmission within 1 year (HFpEF: 58%; HFrEF: 55%; p=0.010)¹⁰⁸.

Costs for inpatient care and hospitalization in patients with HF

Inpatient care and hospitalization were identified as major cost drivers in HF and were reported as the single largest contributor to costs in multiple studies across different geographies. There was relatively wide variation across studies in the percentage of costs contributed by inpatient care, which can be accounted for by differences between populations and disparities in the types of costs included in each study.

In total, five European studies presented inpatient care or hospitalization as a percentage of total or direct costs: one each from Germany⁷⁷ and Italy⁸¹ and three from Sweden^78,106,107. Inpatient care or hospitalization contributed 69–87% of total costs in these studies (Extended data: Supplemental Table 3⁹⁰). Ten studies reported relevant data from North America, of which eight were conducted in the USA^{83,85–88,95,109,110}, one took place in Canada⁹¹, and one conducted analyses using data from a clinical trial with study centres in the USA, Canada, and France¹¹¹. The percentage of total costs contributed by inpatient or hospitalization costs across these studies varied between 47%⁸⁵ and approximately 87%^106,109 (Extended data: Supplemental Table 3⁹⁰). Three studies included only patients who died from HF; however, the contribution of inpatient care to total costs in these studies fell within the overall range for all studies^83,88,111. An additional US study reported the proportion of total claims for HF accounted for by inpatient claims (60.5%) and outpatient claims (39.5%)⁹⁹. One additional study, conducted in China, reported that inpatient care contributed 66% of total costs (Extended data: Supplemental Table 3⁹⁰)⁹².

A German study estimated the future costs of inpatient care for HF, predicting that the overall budget impact in Germany would be €1.80 billion in 2025, up from €1.27 billion in 2009¹¹². This predicted increase arose from a 14% increase in the cost allowance per patient with HF in hospital and a 23% increase in the number of patients with HF in hospital.

Several other studies reported inpatient or hospitalization costs (Extended data: Supplemental Table 2⁹⁰). Seven studies, two from the UK^100,113 and five from the USA^19,114–117, collected data on overall inpatient/hospitalization costs associated with HF; however, these were not reported as a percentage of total costs. Additionally, five US studies reported the costs associated with a single hospitalization^118–122, and one reported the cost of an ‘acute HF hospital episode’ as US$10 775¹²³.

Hospitalization and readmission rates for patients with HF

Patients’ risk of hospitalization for HF or readmission within 1 year is likely to be influenced by factors such as disease severity, and consequently a wide range of rates were reported across the studies identified. There was a larger evidence base for all-cause 30-day readmission rates; these studies indicated that between 15% and 30% of patients hospitalized for HF are likely to be readmitted within a month after discharge from hospital.

Seven studies^{79,80,82,87,109,111,124} reported 1-year hospitalization rates (Extended data: Supplemental Table 4⁹⁰). An Italian study found that HF-related and all-cause hospitalization rates were considerably lower in patients who had never been hospitalized for HF than in those who had (HF-related: 0.4% vs 22%; all-cause: 15% vs 59%)⁸⁰. All-cause hospitalization was 30.8% in a Spanish study⁷⁹. In the USA, HF-related hospitalization rates ranged from 6%⁸² to 22%¹⁰⁹, and all-cause hospitalization rates ranged from 33% to 57%¹⁰⁹. In a multinational study, 38% of patients experienced all-cause hospitalization¹¹¹. A study in China reported relatively high rates of HF-related hospitalization within 1 year, at 34.8%¹²⁴. A further 13 studies reported data on hospitalizations (Extended data: Supplemental Table 2⁹⁰)^{78,95,106,107,113,115,116,121,122,125–128}, but did not present data as a percentage of patients experiencing hospitalization, did not report rates over 1 year, or included only patients who died of HF.

Studies that reported 1-year hospital readmission rates, of which five were from the USA^{86,108,118,129,130}, and one each from Italy⁸¹, Australia¹³¹, and China⁹², are summarized in Extended data: Supplemental Table 5⁹⁰. HF-related readmission rates ranged from 13.8% in a US study¹³⁰ to 46.1% in the Italian study⁸¹, and all-cause readmission rates ranged from 55% in the USA¹⁰⁸ to 73% in Australia¹³¹.

Hospital readmission rates at 30 days were reported in studies from the USA, Australia, and China (Extended data: Supplemental Table 6⁹⁰). In the USA, HF-related rates of readmission were 6.4–12.5%^{94,120,121,129,132,133}, and all-cause rates were 17.1–30.4%^{86,94–96,108,114,119–121,129,130–138}. An additional US study reported all-cause 30-day readmission rates after hospital discharge following implantation of an LVAD as 27.6%¹³⁹. Similar rates were reported in Australia (HF-related: 11%; all-cause: 27%)¹³¹ and slightly lower rates reported in China (HF-related: 4.1%; all-cause: 16.2%)⁹². In addition, a study from Spain reported readmission rates in the 90 or 180 days after discharge from hospital (Extended data: Supplemental Table 2⁹⁰)¹⁴⁰.

Two US studies focused on preventable readmissions. Gunadi et al. reported results from a care programme intended to reduce readmission rates, showing that the decrease in variable cost for each avoided hospital readmission was US$5652¹⁴¹. Chen et al. examined the rate of preventable hospital readmissions based on patients’ location and found that those in remote rural areas had a 27% lower risk of 30-day preventable readmission than patients living in urban areas¹⁴².

Other costs and resource use

A small number of the studies identified reported data that did not fall into any of our key areas of interest (Extended data: Supplemental Table 2⁹⁰). A Spanish study reported prescription drug costs¹⁴³, a study from the Netherlands reported care home and nursing home costs for patients at the end of life¹⁴⁴, and three North American studies reported hospital LoS⁵⁸, nursing contacts¹⁴⁵, and the association between heart rate and medical costs in patients with HF¹⁴⁶, respectively.

Key findings – HRQoL

The evidence identified in this SR illustrates that HF is associated with a substantial humanistic burden. For the purposes of this review, we consider humanistic burden to represent the impact of HF on the individual affected, as indicated by patient-reported outcomes. Patients with HF experience worse HRQoL than both individuals without HF and patients with other chronic diseases, and the burden of HF is greater for patients with more severe disease than for those at earlier stages. Poor mental health, comorbidities, and hospitalization are likely to be associated with worse HRQoL in patients with HF, whereas appropriate disease intervention can help to bring about improvements in HRQoL. This is supported by evidence that HRQoL decreases over time in the absence of disease management, but for patients who were provided support, HRQoL increased over time. Taken together with evidence indicating the positive impact of disease management and self-care, and structured support, this indicates possible ways of improving HRQoL for patients with HF.

Key findings – costs and resource use

The majority of the economic studies identified in this SR were conducted in the USA, with a smaller evidence base in Europe and relatively few studies from the rest of the world. HF is associated with considerable healthcare costs, and patients with HF incur higher costs and greater resource use, including inpatient, outpatient, and informal care, than individuals without HF or with other chronic diseases. The occurrence of HF as a comorbidity is also associated with extra costs. There is evidence to indicate that costs rise with increasing severity of HF, although this is based on only one study. The results of studies reporting costs and resource use by ejection fraction were mixed and did not provide sufficient evidence to conclude whether HFrEF or HFpEF incurs higher costs. As the diagnosis of HF has previously hinged on EF, with therapy development focused on reduced EF, limited evidence may be available on the treatment of HFpEF ¹⁴⁷. Patients with HF are at a relatively high risk of hospitalization or readmission; accordingly, inpatient care and hospitalizations were identified as key cost drivers.

Several studies suggested that improved disease management, with the aim of reducing the number of inpatient cases and reducing patients’ risk of hospital readmission, could be a way to limit the economic impact of HF. Specific strategies discussed included use of case-management programmes, identification of risk factors for readmission, and optimization of medication, for example by improving adherence.

Evidence gaps

This SR identified a broad range of studies; consequently, this review has focused on key areas of interest that best illustrate the impact of HF. Some studies have been presented in less detail, either because they did not discuss these key themes or because the data reported were not readily comparable with the rest of the evidence base, owing to differences in study population, setting, or time frame.

Although a large number of studies were identified, some gaps in the evidence base were apparent. In studies comparing patients with HF and individuals with other chronic conditions, disease severity in the individual patient populations was not specified, nor was the impact of treatment interventions over the period of the study. This is particularly relevant when comparing different diseases, for which the effect of treatment upon HRQoL may vary considerably. Further studies on HRQoL in patients with HF, particularly those assessing the incremental effects of the disease over time or with increasing severity, would be valuable as part of an overall approach to identify means of improving patients’ well-being.

A relatively small number of studies that assessed the economic impact of HF included detailed clinical information. The majority of studies in this SR, particularly those from the USA, which was the country with the largest evidence base, identified patients with HF in administrative claims databases using International Classification of Diseases diagnosis codes. This meant that factors such as comorbidities, HF severity by NYHA classification, and disease history were not known for all patient populations. These factors can have a large impact on disease outcomes and associated costs; therefore, some of the variation between the costs reported in different studies is likely to be attributable to factors that are not recorded. Economic studies in which patients’ medical records are available might therefore be valuable to obtain more granular data on cost drivers in HF. In addition, there is little published information on the indirect costs associated with HF, with very few studies reporting data on informal care for patients with HF; thus, the wider societal impact of the disease is not immediately apparent in the literature.

Underlying data

All data underlying the results are available as part of the article (included under extended data), and no additional source data are required to support our results.

Extended data

figshare: Supplemental content 1 – Supplemental Table 1, http://doi.org/10.6084/m9.figshare.8099915³⁴.

figshare: Supplemental content 2 – Supplemental Tables 2–6, http://doi.org/10.6084/m9.figshare.8099969⁹⁰.

The systematic review protocol is available on request by contacting the corresponding author.

Reporting guidelines

PRISMA checklist: figshare: Supplemental Content 3 – PRISMA checklist, http://doi.org/10.6084/m9.figshare.8100020¹⁴⁹.