Patient-reported outcome is associated with health care costs in patients with ischaemic heart disease and arrhythmia

Abstract

Aims

Systematic use of patient-reported outcomes (PROs) have the potential to improve quality of care and reduce costs of health care services. We aimed to describe whether PROs in patients diagnosed with heart disease are directly associated with health care costs.

Methods and results

A national cross-sectional survey including PROs at discharge from a heart centre with 1-year follow-up using data from national registers. We included patients with either ischaemic heart disease (IHD), arrhythmia, heart failure (HF), or valvular heart disease (VHD). The Hospital Anxiety and Depression Scale, the heart-specific quality of life, the EuroQol five-dimensional questionnaire, and the Edmonton Symptom Assessment Scale were used. The economic analysis was based on direct costs including primary, secondary health care, and medical treatment. Patient-reported outcomes were available from 13 463 eligible patients out of 25.241 [IHD (n = 7179), arrhythmia (n = 4322), HF (n = 987), or VHD (n = 975)]. Mean annual total direct costs in all patients were €23 228 (patients with IHD: €19 479, patients with arrhythmia: €21 076, patients with HF: €34 747, patients with VDH: €48 677). Hospitalizations contributed overall to the highest part of direct costs. For patients discharged with IHD or arrhythmia, symptoms of anxiety or depression, worst heart-specific quality of life or health status, and the highest symptom burden were associated with increased economic expenditure. We found no associations in patients with HF or VHD.

Conclusion

Patient-reported outcomes at discharge from a heart centre were associated with direct health care costs in patients with IHD and arrhythmia.

Registration

ClinicalTrials.gov: NCT01926145.

Graphical Abstract

Open in new tabDownload slide

Introduction

Worldwide, heart disease is a major public health concern associated with high health care costs and considerable individual consequences for patients.^1–3 Cardiovascular diseases represent the most significant financial health care burden as well as being the leading cause of mortality and disability.^1,2 An ageing population and an increase in the number of patients with chronic heart diseases globally are expected to further increase the economic burden.¹ Thus, heart disease-related direct health care costs involving general practitioners, therapists, and pharmacological treatment are estimated to rise steadily in the future. In-hospital costs are estimated to be double from 2015 to 2035.^1–3 This situation underlines the need for alternative health care service strategies worldwide.^1,2 Proactive identification and management of services for the highest-costs patients with heart disease, who may benefit from targeted interventions to optimize health care, could be an effective way to underpin quality and reduce health costs.

Information on outcome from the patient perspective is an important endeavour in health care service planning, contributing to a more patient-centred approach.⁴ Patient-reported outcomes (PROs) are reported by the voice of patients and concern physical and mental health, symptom burden, and disease-specific health-related quality of life.^4–6 In recent years, there has been an increasing interest in patient-reported health following cardiovascular disease.⁵ Thus, PROs in patients with heart disease could have the potential to systematically incorporate the patient perspective to improve both quality and costs of health care services.⁷ Understanding the relation between PROs at discharge from a heart centre and subsequent health care costs may help to develop patient-centred high-value health care strategies for this patient group.^4,7 Studies have shown that patient-reported health status is an independent predictor of cardiovascular events, hospitalization, mortality, and costs of care in patients with heart failure (HF).⁵ However, no studies have addressed whether PROs apply to the economic burden in other diagnostic groups of heart disease.⁵ Thus, we aimed to describe whether PROs at discharge from a heart centre in patients diagnosed with ischaemic heart disease (IHD), arrhythmia, HF, or valvular heart disease (VHD) are associated with direct health care costs within the first year after hospitalization.

Materials and methods

Study population and setting

The population consisted of Danish patients from the nationwide DenHeart survey.⁸ Briefly, the DenHeart study is a cross-sectional survey including PROs at discharge (index date) from a highly specialized heart centre with 1-year follow-up data retrieved from national public registers; these registers contain information on health care service utilization and mortality.⁸ Patients hospitalized between 15 April 2013 and 15 April 2014 at a Danish heart centre were consecutively invited at discharge from the hospital to fill out a feasibility-tested questionnaire with primary validated questionnaires. Patients were recruited by nurses or research assistants and introduced to the study protocol and procedures. Patients were asked to return the questionnaire at discharge or within 3 days after discharge (index date). Patients were excluded if <18 years, had no unique Danish civil registration number and/or not able to understand Danish, or if patients were adversely ill and/or unconscious at discharge/transfer to another department.⁸

All Danish citizens are assigned a unique Danish civil registration number.⁹ We used this number to link individual-level data across the study cohort, questionnaire data and other public register data, as well as tariffs and prices on health services from the Danish Ministry of Health.¹⁰ In the present study cohort, we selected patients from the largest diagnostic groups of heart disease;⁸ a primary ICD-10 action diagnosis of IHD, arrhythmia, HF, or VHD in the Danish National Patient Register, which includes data on all hospital contacts.¹¹ Patients with an action diagnosis of congenital heart disease, infectious heart disease, heart transplant, other diagnosis, or observation for heart disease were excluded (see Supplementary material online, Table S1).

The survey was notified to the Danish Data Protection Agency (no. 2007-58-0015/30-0937) and the Danish Health Authority (FSEID-0001131) and registered at ClinicalTrials.gov (NCT01926145). Patients provided informed consent before participation. According to Danish legislation, this type of survey does not need an ethics approval.⁸ The study complies with the Declaration of Helsinki.

Patient characteristics

The Danish Civil Registration System provided data on age and sex. Socioeconomic status was defined by educational level, marital status, household income, and labour market attachment.⁹ Information on the highest completed educational level was obtained from the Educational Attainment Registry, Statistics Denmark, and divided into three levels: (i) low (elementary school); (ii) medium (upper secondary or vocational school); and high (higher education).⁸ Data on marital status (married or single) were collected from the Danish Civil Registration System.⁹ Information on household income was obtained from the Family Income Registry, Statistics Denmark¹⁰ in a 5-year period before the index date. Data on attachment to the labour market (employed, unemployed, retired, on social support, other) were collected from the Labour Force Statistics 1 year before the index date.¹⁰ Co-morbidity was assessed by diagnoses (primary and secondary) during a 10-year period before the index date in the Danish National Patient Registry.¹¹ The modified Charlson Co-morbidity Index (CCI) score (excluding points for the primary ICD-10 diagnosis at index date) was estimated¹² (see Supplementary material online, Table S2). In addition, cardiovasular co-morbidity was divided into the following categories: ventricular arrhythmia, IHD, and myocardial infarction (see Supplementary material online, Table S2). Data on baseline use of cardiovascular, anti-thrombotic, and other medical therapies were obtained from the Danish National Prescription Registry¹³ by The Anatomical Therapeutic Chemical (ATC) codes and defined as one or more redeemed prescriptions 3 months before the index date (see Supplementary material online, Table S3).

Patient-reported outcomes

Validated questionnaires with items regarding symptomatic and/or disease-specific conditions from the DenHeart survey were selected as the source of PRO data.⁸ Thus, we included the Hospital Anxiety and Depression Scale (HADS), the HeartQoL (heart-specific quality of life), the EuroQol five-dimensional questionnaire (EQ-5D-5L), and the Edmonton Symptom Assessment Scale (ESAS).⁸ Hospital Anxiety and Depression Scale is a generic measurement tool ranging from 0 to 21 points. A HADS-A score ≥8 indicates symptoms of anxiety disorder and a HADS-D score ≥8 indicates symptoms of depressive disorder within the last week.⁸ The HeartQoL is a specific questionnaire measuring heart-related quality of life during the last 4 weeks. A summarized global score ranging from 0 to 3 point is generated; a higher score indicates better heart-specific quality of life. EuroQol five-dimensional questionnaire⁸ is a generic instrument covering current health, ranging from −1 to 1 points; a higher score indicates better health status.⁸ Edmonton Symptom Assessment Scale is a generic measurement tool to assess current experiences of symptoms; the increasing score is related to higher symptom burden.⁸ To examine a pragmatic algorithm in continuous PROs without a cut-off point for clinical awareness (HeartQoL, EQ-5D-5L, and ESAS), we dichotomized these measures by the quartile in the worst end of the response score as the exposure of interest¹⁴ (see Supplementary material online, Table S4).

Health care costs

The economic burden of heart disease was estimated through the 1-year direct costs within each of the four diagnostic groups. We extracted register data on primary health care and related activity-based tariffs from the National Health Service Register;¹⁵ secondary health care and national average Diagnostic-Related Grouping tariffs from the Danish National Patient Register;¹¹ and data on prescribed medical therapies were extracted from the National Prescription Registry by retail prices as well as dispensing costs.¹³ Data on resource utilization in the primary healthcare system included activity by general practitioners, medical specialists, therapists, and others (see Supplementary material online, Table S5). Resource utilization in the secondary healthcare system was defined by inpatient admissions, bed days, and outpatient contacts. Information on medical resource utilization was obtained as redeemed cardiovascular, anti-thrombotic, or other medical therapies (see Supplementary material online, Table S3). We exacted data about dates of Defined Daily Dose for each ATC code. All cost estimates are presented in 2018 prices and presented per patient.

Other covariates

The survey also included questions about participants’ height, weight, and smoking habits. Body mass index (BMI) was calculated as (kg)/height² (m²) and overweight was defined as a BMI ≥ 25. Smoking was defined as active smoker >15 cigarettes per day (Yes/No).⁸

Statistics

Categorical data were expressed as numbers (%) and continuous data as mean ± standard deviation (SD). Fisher’s exact test, Pearson’s χ² test, and Student’s t-test were used for analysis. To study the association between PROs and direct costs of health service use, analyses were conducted using a generalized linear model. Model fit was analysed when assuming either the Gaussian distribution or the Gamma distribution for health care costs. Due to skewed cost data, the generalized linear model using the Gamma distribution and the log-scale for the analyses showed the best fit and was therefore the model of choice. Associations were expressed as the difference (%) in health care costs between the worst quartile/end and the remaining. Significance levels were reported as 95% confidence intervals (CIs). Because of clinical differences and thereby expected great variation in the use of health care and health care costs between primary action diagnostic groups, cost analyses were performed for each diagnostic group separately as well as for the whole study cohort. We performed both unadjusted analyses and analyses adjusted for age, gender, BMI, smoking, income, employment, educational level, CCI score, and death within 1 year after hospitalization. To account for the potential impact of end-of-life health care costs on results, we additionally conducted sensitivity analyses performing the adjusted analyses but excluding outliers (defined as total health care costs >200 000€) and respondents that died within the first 30 days after the study recruitment.¹⁶ We did not adjust for time preference due to the 1-year time span horizon. All analyses were conducted using STATA version 16.

Patient and public involvement

Patients and/or public were not involved in the development, design, conduct, reporting, or dissemination plans of this study.

Results

We conducted the study in a cohort of patients (n = 25 241) with IHD (n = 13 517), arrhythmia (n = 7862), HF (n = 1929), or VHD (n = 1933) as the primary diagnosis. Patient-reported outcomes were available from 13 463 eligible patients [IHD (n = 7179), arrhythmia (n = 4322), HF (n = 987), or VHD (n = 975)] with response rate 50–55% across the groups (see Supplementary material online, Figure S1). Basic characteristics of responders are presented in Table 1 (see Supplementary material online, Table S6). Table 2 shows PROs of responders. Symptoms of anxiety (HADS-A ≥8) were high for patients with IDH (32%). In patients with HF, symptoms of depressive disorder (HADS-D ≥8) were high (24%). Heart-specific quality of life was low for patients with VHD. Both health status (EQ-5D-5L) and ESAS were observed to be low in patients with HF (Table 2).

Health care costs

Mean total health care costs (direct costs) were high for patients with VHD [€48 677 (SD 29 421)] and for patients with HF [€34 747 (SD 39 810)] compared with €19 479 (SD 21 846) for patients with IHD and €21 076 (SD 21 313) for patients with arrhythmia (Table 3). The majority of direct costs was attributable to hospital admission in all diagnostic groups. Primary care costs of patients with VHD amounted to €733 (SD 508); total medical costs in patients with IHD were €1119 (SD 1088), whereas the anti-thrombotic expenditure accounted for €398 (SD 518) (Table 3) (see Supplementary material online, Tables S7 and S8).

Associations between patient-reported outcomes and total health care costs

The adjusted analyses showed that after 1 year, symptoms of anxiety (HADS-A ≥8) were associated with an increase of total health care costs with 8% for patients with IHD [0.08 (CI 0.03; 0.13)] and 9% for patients with arrhythmia [0.09 (CI 0.02; 0.15)] compared with patients with a HADS-A score <8 (Figure 1). The same consistent pattern was seen in the adjusted analyses for patients with high scores in symptoms of depressive disorder (HADS-D ≥8); IHD and arrhythmia showed a 12% [0.12 (CI 0.06; 0.19)] and 13% [0.13 (CI 0.05; 0.21)] increase in total health care costs, respectively. The worst quartile in respondents’ score in HeartQoL was similarly associated with increased total health care costs compared with patients with better scores; IHD had a 23% increase in health care costs [0.23 (CI 0.17; 0.29)] and arrhythmia showed a 15% increase [0.15 (CI 0.08; 0.22)]. The same tendencies were seen for respondents’ health status (EQ-5D-5L) and ESAS. In patients with IHD, the worst EQ-5D-5L was associated with a 20% [0.20 (CI 0.14; 0.26)] increase in total health care costs and worst ESAS showed a 17% increase [0.17 (CI 0.11; 0.22)] compared with patients in the other diagnostic groups. For arrhythmia, worst EQ-5D-5L was associated with a 16% increase in total health care costs [0.16 (CI 0.09; 0.23)] and worst ESAS with a 12% [0.12 (CI 0.05; 0.19)] increase. The analyses showed no statistically significant associations between the worst PRO scores and increased total health care costs for patients with VHD or HF (Figure 1). A sensitivity analysis did not substantially change the estimates (see Supplementary material online, Table S9).

Discussion

This study investigated whether PROs at discharge from a highly specialized heart centre of patients within the spectrum of heart disease were associated with direct health care costs after 1-year follow-up. Several novel findings in this study were observed: (i) the study showed a high cross-sectional health care consumption in patients discharged with a primary diagnosis of VHD (€48 677) and HF (€34 767) compared with patients with IHD (€19 479) and arrhythmia (€21 076); (ii) hospitalizations contributed to the highest part of direct costs and finally; (iii) our adjusted analysis indicates that symptoms of anxiety or depression as well as the experience of heart-specific quality of life, health status, or symptom burden in the worst quartile of the response scores were associated with increased total health care costs for patients discharged with IHD and arrhythmia.

In contrast to our study, a national Danish study reported that the median economic burden of first-time HF (coexisting IHD in 27% and prior myocardial infarction in 12%) in 2016 prices was €11 926 of the total annual direct costs, whereas the majority of direct costs was attributable to hospital admissions and outpatient services.¹⁷ Of notice, a systematic review including cost-of-illness studies in patients with HF, documented a large variation in costs ranging from Int$908 to Int$40 971 per patient.¹⁸ The higher annual direct costs in our HF group could be explained by methodology and the comprehensive degree of co-morbidity in our study population [coexisting HF (68%), coexisting IHD (45%), prior myocardial infarction (24%), and ≥moderate co-morbidity (30%)]. This supports the need for cost-intensive and highly specialized multidisciplinary health care services.¹⁹

A propensity-matched Danish study reported mean health care costs in 2018 prices following open-heart valve surgery of ∼€2077 in the historical control groups after 180 days follow-up (excluding pharmalogical and specialist use).²⁰ However, the study presented no data on prior diagnosis and co-morbidity could be more common in our group of patients with VHD. Additionally, due to the use of VHD diagnosis rather than surgery codes, we were not able to identify the proportion of heart valve surgery before discharge and during follow-up. A recent Canadian cost study in patients risk with severe aortic stenosis but at low surgical risk showed that index hospitalization costs for balloon-expandable trans-catheter aortic valve replacement (TAVR), self-expandable TAVR, and surgical aortic valve replacement (SAVR) were $35 812, $37 144, and $27 410, respectively.²¹ The main driver of high direct costs in our group of patients with VHD may possibly be surgery during follow-up and hospitalizations caused by high degree of disease complexity.

A national study documented that the first year after myocardial infarction, mean health care costs in 2014 prices were €12 460 and hospitalizations resulted in substantial costs.²² A review of the international literature illustrated that average costs were US $11 664 for acute myocardial infarction, $11 635 for acute ischaemic stroke, $37 611 for coronary artery bypass graft, and $13 501 for percutaneous coronary intervention. The ranges for cost estimates varied widely across countries.²³ Among the direct costs, we found that hospitalization and pharmalogical treatment constituted the largest expenditure in our group of patients with IHD. This supports that our cost estimates during 1-year after discharge could be caused by hospitalizations including invasive procedures and surgery as well as recommended expensive anti-thrombotic medical treatment.²⁴

Finally, total health care costs including societal impact covering population-based Danish patients with a first-time hospital diagnosis of atrial fibrillation were estimated to be €12 005 in 2013 prices; admission costs constituted the largest cost component.²⁵ Within our group of patients with arrhythmia, 53% had no co-morbidities, 21% had a prior diagnosis of HF, 34% were also diagnosed with IHD, and 8% also had ventricular arrhythmia, respectively. An explanation for the higher economic costs in this study could be the high cardiovascular co-morbidity in our study population and highly specialized treatment with radiofrequency ablation for atrial fibrillation, ventricular tachycardia ablation, or permanent pacemaker implantation during follow-up.²⁶ A recent study documented that catheter ablation for complex left-atrial arrhythmia including overnight hospital stay costs £350.²⁷ In addition, the DECODE registry has documented that the main driver of health care expenditures after implantable cardiovascular defibrillator/cardiac resynchronization therapy defibrillator was hospitalization related to HF.²⁸

To our knowledge, studies are scarce on PROs and their association with direct health care costs within the four largest groups of heart diseases. The EPHESUS study investigated whether the Kansas City Cardiomyopathy Questionnaire predicts costs (in hospital) in outpatients with stable HF and left ventricular dysfunction after myocardial infarction. Kansas City Cardiomyopathy Questionnaire predicts the 1-year costs in multivariable analyses.^5,29 Another study reported that symptoms of depression and anxiety (HADS) were highly prevalent in patients with stable coronary heart disease, and their long-term trajectories were associated with increased health care costs during a 3-year follow-up period.³⁰ However, differences in PRO measures and approaches to measure resource utilization and estimate direct costs are substantial compared with our study and results should thus be interpreted with caution. Our study is the first to indicate that symptoms of anxiety or depression, heart-specific quality of life, health status, or symptom burden could be PRO measures able to identify the highest-cost patients within the group of patients diagnosed with IHD or arrhythmia. Notably, we found no associations between our PRO measures and direct health care costs in the group of patients discharged with HF or VHD. An explanation could be that within diagnostic groups (like HF and VDH) with coexisting cardiovascular chronic diseases and a high burden of co-morbidity, thus particularly associated with extensive economic spending and advanced treatment during follow-up, the used PROs were non-sensitive to identify the highest-cost patients. However, the EPHESUS study indicated²⁹ that more disease- or diagnosis-specific questionnaires could potentially predict health care costs. It may be argued that more disease-specific PROs (non-generic) and their association with health care costs in all diagnostic groups should be investigated also in intervention studies. Our findings demonstrate the potential of using both generic and non-generic PROs, not only to incorporate the patient perspective but also to support the identification of highest-costs patients. with heart disease. By including PROs in addition to common clinical information, clinicians have additional tools to further optimize individualized health care.

The PROs in the present study were solely developed from a clinical perspective and may be fundamentally different from actionable measures for payers and stakeholders. We suggest that further studies shed light on how appropriate PROs can be used and developed to deliver efficient and high-value health care strategies in all subgroups of patients with heart disease.⁷

Strengths and limitations

The key strengths and limitations of the DenHeart study are presented elsewhere.⁸ The overall PRO response rate in this study may be acceptable when analysing the study population as a whole. However, individual analysis of the different diagnostic subgroups had a limited number of participants and the impact of this on the response rate for each diagnostic group was thus increased. Non-responders were older, had a higher burden of co-morbidity, and a lower socioeconomic status, which may have introduced selection bias. Complete information on resource utilization and costs based on national registers and tariffs and prices of health services contributed to the internal validity of the present study. The study did not include clinical data, which may be a confounding factor. The supplementary analyses also demonstrated higher direct costs in non-responders vs. responders; thus, we believe our associations may be underestimated rather than overestimated.

Conclusion

This study demonstrated that symptoms of anxiety or depression as well as the experience of heart-specific quality of life, health status, or symptom burden in the worst quartile of the response scores measured in patients at discharge from a heart centre were significantly associated with 1-year direct health care costs in patients with IHD or arrhythmia. We found no associations between PROs and health care costs in patients with HF or VHD.

Supplementary material

Supplementary material is available at European Journal of Cardiovascular Nursing online.

Acknowledgements

We would like to acknowledge the DenHeart research expert committee. A full roster of the DenHeart investigators can be found at http://denheart.dk/index.html. Moreover, we would like to express our gratitude to the patients who participated in the survey.

Funding

The Danish Heart Foundation (A5484).

Data availability

Due to the nature of this research, participants of this study did not agree for their data to be shared publicly, so supporting data are not available.

References

Author notes