Kaul P, McAlister FA, Ezekowitz JA, Bakal JA, Curtis LH, Quan H, Knudtson ML, Armstrong PW. Resource Use in the Last 6 Months of Life Among Patients With Heart Failure in Canada. Arch Intern Med. 2011;171(3):211-217. doi:10.1001/archinternmed.2010.365
Copyright 2011 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2011
Heart failure (HF) is a debilitating and chronic condition associated with significant morbidity and mortality. More than 500 000 Canadians live with HF, and approximately 50 000 new patients are diagnosed each year.1 Often considered a disease of elderly individuals, the prevalence of HF increases notably with age. In 2003, 0.2% of Canadians younger than 50 years reported having HF; the corresponding figure for those 80 years or older was 6.7%.2 Rates of HF are expected to increase exponentially in coming years as a result of several factors, including the aging of the Canadian population, improved survival rates after myocardial infarction, and the development of new therapies for more effective treatment of those with HF.3- 5 Quality of life for patients with HF is frequently poor, and long-term prognosis includes a 5-year mortality rate of 69% among hospitalized patients in the community.6- 8
A comprehensive assessment of the economic burden of HF on the Canadian health care system has yet to be conducted. However, in 2004 to 2005, HF was the fourth most expensive medical condition in terms of hospitalization costs.9 An important factor in the escalating costs of HF may be the high cost of end-of-life (EOL) care.10- 13 Despite the high mortality rates associated with the disease, neither location of death nor EOL costs among patients with HF in Canada have been examined. Therefore, we examined trends in resource use and costs during the last 6 months of life among elderly patients with HF who died between 2000 and 2006 in the province of Alberta, Canada.
Our HF cohort consists of all residents of Alberta 20 years or older with a hospitalization, outpatient visit (including emergency department visit), or physician claim with a diagnosis of HF between April 1, 1999, and March 31, 2007 (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM ] code 428 or ICD, 10th Revision [ICD -10] code I50 [after April 1, 2002]). The following data were available for patients in the cohort: (1) hospital admissions for the period April 1, 1992, through March 31, 2007; (2) outpatient visits (including emergency department visits) for the period April 1, 1997, through March 31, 2007; (3) physician claims for the period April 1, 1993, and March 31, 2007; (4) pharmaceutical claims for the period April 1, 1993, through March 31, 2007, for patients 65 years or older; and (5) vital status of patients as of December 31, 2006. The health ethics board of the University of Alberta, Edmonton, approved this study.
Using a previously published definition to identify the study population,14 we included patients who met 1 of the following criteria: (1) had at least 1 hospitalization with a diagnosis of HF in any diagnosis field; or (2) had at least 3 outpatient or physician claims with a diagnosis of HF on different days in a consecutive 20-month period. We extended this definition to include patients with at least 1 emergency department visit with a diagnosis of HF.
Patients who died between January 1, 2000, and December 31, 2006, and who were 65 years or older at the time of death were included in the study. We defined the date of HF incidence as the date of (1) the earliest inpatient diagnosis, (2) the earliest emergency department diagnosis, or (3) the third outpatient or physician claim diagnosis, whichever occurred first. We calculated the duration of HF as the difference between the date of death and the date of HF incidence. To keep the patients being compared across the years of the study homogeneous and relatively at the same stage of disease, we excluded patients with HF duration greater than 8 years.
We considered a comorbid condition to be present at the time of death if it was coded in any inpatient, outpatient, or physician claim in the preceding 365 days. We used previously established definitions based on ICD-9-CM and ICD-10 codes (as of April 1, 2002) to identify comorbid conditions.15,16
We examined resource use related to acute care hospitalizations, outpatient clinic visits, emergency department visits, physician services, and pharmaceutical costs in the 180 days before death. For patients with at least 1 hospitalization, we calculated the mean number of days in the hospital and the mean number of days in an intensive care unit (ICU) or cardiac care unit (CCU). Duration of ICU or CCU stays was measured in days for 2000 through 2001 and in hours for 2002 through 2006. We converted the latter to days to facilitate year-by-year comparisons. As of April 1, 2002, data on whether a patient was discharged from an acute care hospital to a continuing care facility, home care, or other facility (eg, hospice, disability service) were available. We calculated the percentage of patients who used these services at any time during the last 6 months of life.
Each inpatient stay was assigned a Canadian Institute for Health Information case mix group, which groups patients according to the most responsible diagnosis, operating room procedures, surgical or medical hierarchy, and diagnoses. A complexity overlay, which incorporates the presence of comorbid conditions, and an age group overlay are applied to enhance homogeneity in terms of resources used. Each resulting combination of case mix group, complexity level, and age overlay is assigned a dollar value based on an established “bottom-up” activity-based costing system developed by the provincial health ministry, Alberta Health and Wellness.17 Cost weights were available only for 2001 through 2006; therefore, we assigned the 2001 values to hospitalizations that occurred in 2000.
We calculated the total number of patients using emergency department and outpatient services and the mean number of visits per patient in the last 6 months of life. Similar to hospital costs, these visits were assigned costs using the provincial ambulatory care case mix group. We used physician claims to calculate total physician costs and the mean number of physician office visits. The cost of in-hospital drugs is included in the case mix group cost weights. In Alberta, all prescription medications for elderly patients are covered, and we used pharmaceutical claims to calculate the mean total outpatient drug costs for the last 6 months of life for each patient. We used the Consumer Price Index to adjust all costs to 2006 Canadian dollars.18 We calculated total costs to the health care system at the EOL among patients with HF as the sum of costs related to all hospitalizations (acute care and other hospitalizations, including mental health and rehabilitation), emergency department visits, outpatient visits, physician services, and drug costs. In addition, we examined rates of specific diagnostic and therapeutic procedures, including stress tests, cardiac catheterizations, coronary artery bypass graft surgery, pacemakers, implantable cardioverter-defibrillators, and hemodialysis.
We examined the percentage of patients who died in hospital overall and in specific age groups. Among patients who died in hospital, we examined the length of stay for the last hospitalization and whether it included a stay in an ICU. We calculated 6-month costs by location of death (ie, whether the patient died in an acute care hospital).
We present categorical variables as frequencies and continuous variables as means (SDs). We used Cochran-Mantel-Haenszel tests to examine temporal trends. We used generalized linear models with log links and Poisson distributions to examine unadjusted and adjusted relationships between covariates and total costs in the last 6 months of life. Covariates included age at death, sex, comorbid conditions, duration of HF, and year of death. We used SAS statistical software (version 9.2; SAS Institute Inc, Cary, North Carolina) for all analyses.
Among 89 792 adults 20 years or older with HF, 47 970 died between January 1, 2000, and December 31, 2006. Of these, 37 103 (77.3%) were 65 years or older at the time of death. Patients with HF durations greater than 8 years (n = 3959) were excluded (their baseline characteristics and trends in EOL costs are provided in the eTable). Table 1 shows the characteristics of the final study population of 33 144 patients. There was a slight increase in age at death and the mean number of days that patients lived with HF during the study period. Rates of dementia, diabetes mellitus, hypertension, and renal disease increased significantly during the study period. In contrast, rates of ischemic heart disease, cerebrovascular disease, and peripheral vascular disease significantly declined. Over 40% of the patients had 4 chronic conditions, in addition to HF, at the time of death.
As shown in Table 2, the percentage of patients with at least 1 hospitalization in the last 6 months of life decreased from 84.0% in 2000 to 76.2% in 2006 (P < .001). Over 80% of hospitalizations in the last 6 months of life were for reasons other than HF. Among those hospitalized, the mean length of stay in the last 6 months of life remained similar in all years. In general, ICU days accounted for less than one-tenth of the total number of days in hospital. Mean hospitalization costs per patient increased by approximately $4000, from $21 995 in 2002 to $26 186 in 2006.
In contrast to the reduction in the use of inpatient services during the study period, patients who died in later years of the study were substantially more likely to receive outpatient care in the last 6 months of life than those who died earlier in the study (from 52.8% in 2000 to 69.8% in 2006; P < .001). Of those receiving outpatient care, the mean number of visits to outpatient clinics increased from 6.4 in 2000 to 7.7 in 2006 (P < .001). On the one hand, the mean number of visits to physician's offices (including family physicians and general practitioners) during the last 6 months of life also increased from 6.9 to 8.8 (P < .001). On the other hand, the percentage of patients using emergency department services and the mean number of visits per patient decreased slightly. There was a marginal increase in mean emergency department costs per patient. The percentage of patients discharged to a continuing care facility in the last 6 months of life was approximately 16%. The percentage of patients discharged to home care increased from 15% to 18% (P < .001). The percentage of patients who underwent diagnostic tests and invasive procedures in the last 6 months of life remained low throughout the study period. Only the use of hemodialysis increased significantly, from 4.8% in 2000 to 7.0% in 2006 (P < .001).
Table 3 summarizes the mean cost per patient for each type of service. There was an increase in all costs in the final 6 months of life. Although the largest relative increases were in outpatient costs (75%), physician costs (49%), and drug costs (42%), hospitalization costs accounted for most costs in the last 6 months of life.
The percentage of patients who died in acute care hospitals decreased from 60.4% to 54.0% (Table 4). The likelihood of dying in hospital decreased with increasing age. In 2000, 71% of patients aged 65 to 69 years died in hospital, compared with 49% of patients 90 years or older. This pattern remained similar in 2006: 68% of patients aged 65 to 69 years and 43% of patients 90 years or older died in an acute care hospital. Costs in the last 6 months of life among patients who died in hospital were more than double those for patients who did not. The hospitalization during which death occurred was a major driver of resource use, as evidenced by the substantial lengths of stay (approximately 20 days). However, the proportion of patients with an ICU stay during the last hospitalization decreased from 15.7% to 12.1% (P < .001).
Table 5 shows multivariate predictors of total costs to the health care system. There was an inverse relationship between age and EOL costs. With the exception of dementia, all comorbid conditions were associated with significant increases in costs, the highest of which was renal disease. Six-month costs were highest among patients who died within 1 year of being diagnosed, compared with patients whose duration of HF was longer.
To our knowledge, this study is the first to examine EOL costs at a population level among patients with HF in Canada, a country with a single-payer health care system and universal access. We found that EOL was associated with substantial health care resource use among patients with HF. In 2006, the mean cost per patient with HF in the last 6 months of life was $27 983, more than 5 times the per-capita health care expenditure in Alberta ($4995) in the same year.19
Hospitalizations accounted for over 70% of costs in the last 6 months of life. Although the percentage of patients with at least 1 hospitalization decreased during the study period, the mean length of stay remained high. One possible explanation for the high rates of hospitalization at the EOL may be the limited availability of alternative venues of care. In countries with adequate hospice facilities, such as the United States, the use of these resources by Medicare beneficiaries with HF has increased, and the use of these resources prevents hospital admissions.20- 22 In Alberta, where elderly individuals comprise approximately 10% of the population of 3.3 million,23 fewer than 15 000 beds in continuing care facilities and fewer than 5000 beds in supported living facilities are available through the provincial health care department. The inadequacy of these alternative levels of care may explain why less than 20% of the patients in this study were discharged to a continuing care facility in the last 6 months of life.
Access to long-term care facilities and home health care resources may also have affected the percentage of patients who died in an acute care hospital, which remained well over 50% throughout the study period. Costs in the last 6 months of life among patients who died in hospital were more than double those for patients who did not. This difference in costs is likely attributable to the long lengths of stay during which the deaths occurred. The lower costs of out-of-hospital deaths may be confounded by sudden deaths; however, resuscitated patients with cardiac arrests brought to an ICU may also account for some of the in-hospital deaths. The substantial impact of location of death on costs can be illustrated as follows: reducing the number of hospital deaths by 10% in 2006 would have saved the health care system approximately $11 million (486 patients multiplied by mean cost savings of $22 374 per patient).
Location of death may have influenced more than costs, it may have also contributed to quality of life. Heyland et al24 reported low levels of patient satisfaction with EOL care in hospitals in Canada, especially among patients with HF. Although location of death was not listed among the most important issues at the EOL, 44% of patients hospitalized at tertiary care centers for advanced HF expressed a preference to die at home, whereas 25% had no preference regarding location of death.25
An issue repeatedly identified as being of critical importance at the EOL is effective communication by the physician regarding quantity and quality of life.25,26 A major barrier to the initiation of discussions about palliative care in HF is the difficulty of establishing the trajectory of the disease.27 However, several prognostic models in HF have been developed that may assist in decisions concerning the treatment and management of the disease.7,28,29 Canadian Cardiovascular Society guidelines26 recommend that patients with HF be approached early in the disease process and that decisions regarding resuscitation and the use of palliative care be reviewed on a regular basis.
Our study shows that although the percentage of patients hospitalized is decreasing and more patients are dying outside of hospitals, costs of care in the last 6 months of life have continued to increase. It is important to put these cost increases in perspective, however. For example, the average hospital cost (in 2006 Canadian dollars) for patients with renal failure undergoing hemodialysis in Alberta increased from $15 102 in 2001 to 2002 to $19 803 in 2005 to 2006: a cost increase of 31%.17 In contrast, the increase in EOL costs in patients with HF was less than 20% during this time period, which may suggest that although substantial, EOL care may be becoming more cost-efficient over time. This theory is supported by the observed trade-off between reductions in inpatient care and increases in outpatient and physician office visits in our study. New and expensive health technologies, such as implantable cardioverter-defibrillators, have been cited as culprits for the increase in health care costs. However, we found the rates of these procedures to be low.
The changing clinical profile of patients with HF is also likely to account for some of the increase in costs at the EOL. Patients who died in 2006 were older and had higher rates of comorbid conditions (particularly hypertension and renal disease) than those who died in 2000. As shown by the multivariate analysis, the presence of comorbid conditions, especially renal failure, was associated with significantly higher costs.
Our study has some limitations. Although the study included costs of inpatient, outpatient, emergency department, and physician services and outpatient drugs in a single-payer system with universal access, costs associated with continuing care and home care were not included. Therefore, estimates of total costs during the last 6 months of life are likely to be underestimates of the true costs to the health care system. We also did not account for indirect costs to family members who provide home care for dying relatives. Finally, we used administrative data to identify patients' clinical characteristics. Although these data are population based, they lack clinical detail and are dependent on the completeness and accuracy of coding. To ensure the validity of the data, we used a previously established algorithm for examining records in the year prior to death to identify comorbid conditions.30
In conclusion, resource use in the last 6 months of life among patients with HF is changing, with a reduction in acute care hospitalizations, in-hospital deaths, and an increase in the use of outpatient services. While EOL costs are substantial and seem to be increasing, the rate of increase seems more modest than that among other high-cost patients, such as those with renal failure receiving hemodialysis. Although it is commonly assumed that inappropriate use of expensive technologies like biventricular pacemakers and implantable cardioverter defibrillators are driving EOL costs, their use in this population is low. Increasing the availability of alternative venues of care, such as long-term care and home care, may be effective in further reducing hospitalizations and containing costs.
Correspondence: Padma Kaul, PhD, Department of Medicine, University of Alberta, 331 Environmental Engineering Bldg, 112th Street and 87th Avenue, Edmonton, AB T6G 2M8, Canada (email@example.com).
Accepted for Publication: April 26, 2010.
Published Online: October 11, 2010. doi:10.1001/archinternmed.2010.365. This article was corrected on October 22, 2010.
Author Contributions: Dr Kaul had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Kaul, McAlister, Bakal, Curtis, and Armstrong. Acquisition of data: Kaul, Ezekowitz, Bakal, Quan, and Knudtson. Analysis and interpretation of data: Kaul, McAlister, Ezekowitz, Bakal, Curtis, Quan, and Armstrong. Drafting of the manuscript: Kaul and Armstrong. Critical revision of the manuscript for important intellectual content: McAlister, Ezekowitz, Bakal, Curtis, Quan, and Knudtson. Statistical analysis: Kaul, Bakal, and Quan. Obtained funding: Kaul and Armstrong. Administrative, technical, and material support: Kaul, Knudtson, and Armstrong. Study supervision: Kaul and Armstrong.
Financial Disclosure: Dr Curtis has been a consultant for Pfizer Inc and has received research grants or contracts from Allergan Inc, Eli Lilly & Co, GlaxoSmithKline, Johnson & Johnson Inc, Medtronic Inc, Merck Co Inc, Novartis, OSI Eyetech, Pfizer Inc, and Sanofi-Aventis US LLC.
Funding/Support: This study was funded by a Canadian Institutes of Health Research (CIHR) operating grant. Drs Kaul and Ezekowitz are supported by CIHR New Investigator Awards and by Population Health Investigator Awards from Alberta Innovates–Health Solutions. Dr McAlister is supported by a Senior Health Scholar Award from Alberta Innovates–Health Solutions and the Patient Health Management Chair at the University of Alberta.
Role of the Sponsors: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
Disclaimer: Although this study is based in part on data provided by Alberta Health and Wellness, the interpretation and conclusions contained herein are those of the researchers and do not necessarily represent the views of the Government of Alberta.
Additional Contributions: The Alberta Cardiac Access Collaborative provided access to the data. Damon M. Seils, MA, Duke University, provided assistance with manuscript preparation. Mr Seils did not receive compensation for his assistance apart from his employment at one of the institutions where the study was conducted.