The Gerontologist
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 QUICK SEARCH:   [advanced]


     


This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Buntin, M. B.
Right arrow Articles by Huskamp, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Buntin, M. B.
Right arrow Articles by Huskamp, H.
The Gerontologist 42:40-48 (2002)
© 2002 The Gerontological Society of America

What Is Known About the Economics of End-of-Life Care for Medicare Beneficiaries?

Melinda Beeuwkes Buntin, PhDa and Haiden Huskamp, PhDb

a RAND Corporation, Arlington, VA
b Harvard Medical School, Boston, MA

Correspondence: Melinda Beeuwkes Buntin, PhD, RAND Corporation, 1200 S. Hayes Street, Arlington, VA 22202-5050. E-mail: buntin{at}rand.org.


    Abstract
 TOP
 Abstract
 Overview of Medicare Payment...
 State of Science
 Research Priorities
 References
 
Purpose: This article reviews the state of science on the economics of end-of-life care for Medicare beneficiaries and outlines the research needed to fill gaps in that literature. Design and Methods: Searches of the medical, health services, and economics literature were conducted. Key topics examined were studies of spending on end-of-life care and financial, organizational, and nonfinancial barriers to high-quality end-of-life care. Results:Studies have documented poor quality of care, dissatisfaction with care, and limitations in the coverage of end-of-life care for Medicare beneficiaries. However, critical gaps in our knowledge about how to design a better end-of-life care system for Medicare beneficiaries remain. Implications: Further research on how treatment decisions at the end of life are made and prospective studies of costs, satisfaction, and outcomes are needed.

Key Words: Financing • Payment systems • Terminal illness • End-of-life care

End-of-life care is widely viewed as an area in need of improvement. Studies have documented poor quality of care, gaps in care, and patient and caregiver dissatisfaction with care received by dying patients (SUPPORT Principal Investigators [SUPPORT], 1995). Financial and organizational arrangements play a major role in determining both the setting of care at the end of life and transitions between settings. These arrangements also often inhibit the provision of high-quality care to elderly individuals at the end of life. This study provides a survey of the literature on these barriers. In addition, it highlights some of the nonfinancial barriers to high-quality care at the end of life. We also identify some of the unresolved issues in and research priorities for the economics of end-of-life care.


    Overview of Medicare Payment Systems
 TOP
 Abstract
 Overview of Medicare Payment...
 State of Science
 Research Priorities
 References
 
Medicare is the primary payer of end-of-life care services for elderly persons in the United States. Hogan and colleagues 2000Citation report that Medicare covered more than 60% of the total health care costs of decedents in the calendar year of death over the period 1992 through 1996.

Medicare's fee-for-service (FFS) payment systems for end-of-life care vary by provider type (Table 1 ). Medicare reimburses hospitals for inpatient care using the per discharge, diagnosis-related-group (DRG) system, and skilled nursing facilities (SNFs) are paid per diem rates. Hospice is reimbursed on a per diem basis, with most Medicare hospice reimbursement falling under a daily rate for routine home care. Physician services are paid for on a per visit basis. As mandated by the Balanced Budget Act of 1997, home health care providers are paid per 60-day episode of treatment, with the exception of episodes of five or fewer visits, for which providers are paid per visit.


View this table:
[in this window]
[in a new window]
 
Table 1. Principal Medicare FFS Program Provider Payment Methods Relevant to End-of-Life Care, by Provider Type

 
Medicare also contracts with managed care organizations (MCOs), which have their own internal payment systems, to provide coverage to Medicare enrollees. These MCOs are paid a monthly risk-adjusted, capitated amount; however, they are not responsible for hospice care, which is carved out of the FFS program. All of the payment systems described previously are adjusted for geographic differences in costs. The geographic adjustments largely reflect differences in wages.

Medicare's payment systems and their attendant administrative rules influence care decisions and options in three ways. First, Medicare limits coverage of some services. Enrollment of terminally ill persons in hospice care is limited by the Medicare requirement that a patient be certified as having less than 6 months to live before entering hospice and the requirement that patients electing hospice care forgo curative treatment for their illness. Patients who are terminally ill, but whose time until death is unpredictable, are ineligible for hospice services under Medicare, as are those who are unwilling to give up curative treatment for their illnesses. This often results in gaps in services. For example, psychosocial support services and outpatient prescription drugs are covered only under the hospice program and not under traditional Medicare. Long-term custodial care is not covered by Medicare, except as part of the hospice benefit.

Medicare's coverage and payment systems are generally oriented toward curative care. For example, although hospitals are paid for life-extending bypass surgeries, there is no payment for inpatient palliative care as such. Nursing home and home health care services are covered only for patients who have care needs severe enough to necessitate those levels of care and who will gain functioning as a result of such care.

Medicare's provider payment systems create financial incentives that favor certain types of end-of-life care. The type of payment, whether it is made per visit, per diem, or per discharge, influences the amount, type, and duration of services that a provider delivers. In general, prospective payment systems, such as the DRG system, create incentives to provide less care within the unit of payment, such as the discharge. FFS systems, such as the one for physicians, create incentives to provide more care. In addition, provider payment systems that do not adjust or do not adequately adjust for severity of illness can cause providers to shun very high-cost patients. For example, physicians may be relatively poorly compensated for the extra time and effort it takes to treat patients with complex problems or to counsel terminally ill patients about their options. These features of the payment system may give physicians an incentive to provide an inappropriately low level of care to dying patients. Given that many types of providers are paid on a prospective basis and receive no marginal revenue from delivering extra services, they also have an incentive to stint on extra care, including their own time in talking with patients, something Kenneth Schwartz 1995Citation calls the rationing of empathy.

The relative prices that are paid to different providers almost certainly influence the type of care that patients receive. Dying patients, therefore, may not receive the most appropriate mix of health care services. The fact that hospitals, physician groups, nursing facilities, hospices, and home health agencies are reimbursed in different ways but sometimes owned by the same entity compounds this effect by providing the owner with financial incentives to steer patients toward the type of care that would bring the highest reimbursement. Physicians are also faced with the decision of whether to refer their patients to sites of care (e.g., hospice) where they are less likely to continue seeing them regularly.


    State of Science
 TOP
 Abstract
 Overview of Medicare Payment...
 State of Science
 Research Priorities
 References
 
Spending on End-of-Life Care
A large number of studies have estimated expenditures for end-of-life care and examined differences in costs across treatment settings and subgroups of the dying.

The Proportion of Medicare Spending Attributable to Care in the Last Year of Life Has Remained Stable
In the early 1980s, as national health expenditures rose steadily, there was increased focus on the proportion of health care resources used to treat the dying (Scitovsky 1994Citation). A 1984 study showed that 28% of Medicare expenditures in 1978 was attributable to the approximately 6% of beneficiaries who died in that year (Lubitz and Prihoda 1984Citation). A number of other studies using data from the United States and other countries have found similar results regarding expenditures at the end of life (e.g., Gaumer and Stavins 1992Citation; Hogan et al. 2000Citation; Long et al. 1984Citation; Lubitz and Prihoda 1984Citation; Lubitz and Riley 1993Citation; McCall 1984Citation; Scitovsky 1988Citation; Spector and Mor 1984Citation). In the United States, the proportion of Medicare spending attributable to beneficiaries in the last year of life has remained stable at approximately 25% over the past two decades (Hogan et al. 2000Citation; Lubitz and Riley 1993Citation).

Evidence Is Mixed Regarding Differences in End-of-Life Care Spending by Age and Functional Status
Several studies—using only Medicare claims data, including Lubitz and Riley 1993Citation and Hogan and colleagues 2000Citation—have found that Medicare spending by elderly patients at the end of life decreases as age at death increases. Scitovsky 1994Citation points out that many of the services used by patients at the end of life, such as nursing home care, are not covered by Medicare and thus are not included in these calculations. She argues that, when these other expenditures are considered, end-of-life care spending for the oldest old (85 years and older) may not be lower than that for the younger old.

Two studies support Scitovsky's view that total (not just Medicare) end-of-life expenditures may not be lower for the oldest old. In a study of total expenditures for decedents age 45 and older in Manitoba, Roos, Montgomery, and Roos 1987Citation found that decedents age 85 and older had 29% higher spending in the 4 years before their deaths than decedents ages 75–84, 74% higher than decedents ages 65–74, and twice those of decedents ages 45–64. They find that the oldest old use fewer hospital services, but much more nursing home care than the other cohorts. Temkin-Greener, Meiners, Petty, and Szydlowski 1992Citation studied Medicare and Medicaid (which covers a large proportion of nursing home care for elderly persons) expenditures at the end of life for elderly decedents in Monroe County, NY, who were dually eligible for both programs. When considering Medicare spending only, they find that spending in the year of death is lowest for the oldest old. When considering spending by both programs, the oldest old had lower expenditures than the younger old in the year of death, but expenditures were not different between the two groups if expenditures in the year before the year of death were included.

Using more recent data, Hogan and colleagues 2000Citation pooled data from the 1992 through 1996 Medicare Current Beneficiary Surveys to estimate total health care spending (Medicare plus other payers) in the calendar year of death. They found that total payments were lowest for the oldest old, but note that results might be different if considering spending in the year before death as opposed to the calendar year of death.

There has been little work examining how end-of-life expenditures differ by functional status. In a study of 216 elderly patients of a large multispecialty medical practice who died between 1983 and 1984, Scitovsky 1988Citation found that mean spending for those who were unimpaired for the last year of life, those partially impaired all or part of the last year, and those totally impaired for all or part of that year was similar, but that the distribution of spending differed on the basis of impairment level. A greater proportion of spending for unimpaired individuals was devoted to hospital care whereas nursing home care represented the largest proportion of spending for the totally impaired group.

Higher End-of-Life Spending Has No Effect on Regional Mortality Outcomes
Skinner and Wennberg 1998Citation used Medicare claims data for 1992–1993 to document regional variation in the level of Medicare spending in the last 6 months of life. They found no evidence that differences in per capita spending in the last 6 months of life resulted in lower mortality rates.

Studies of Cost Savings Associated With Hospice Use Are Inconclusive Because of Methodological Issues
A number of studies have documented that decedents who use hospice care have lower average Medicare expenditures at the end of life than decedents who do not use hospice care (e.g., Banaszak-Holl and Mor 1996Citation; Gray, MacAdam, and Boldy 1987Citation; Kidder 1992Citation; Lewin-VHI, Inc., 1995; Mor and Kidder 1985Citation). Emanuel 1996Citation and Emanuel and Emanuel 1994Citation argue that most of these studies are likely to overstate any cost savings associated with hospice use because of selection bias. These studies do not randomize patients to receive either hospice or conventional care. Instead, they retrospectively compare costs for individuals who elected hospice to costs for individuals who did not. There is likely to be selection bias associated with the use of hospice: Individuals who choose hospice are less likely to have pursued aggressive medical interventions even if hospice was not an option. Thus, some of the reported differences in costs between hospice users and nonusers are likely to be attributable to unobserved differences in end-of-life care preferences for individuals in the two groups rather than to the use of hospice itself.

Using Medicare Current Beneficiary Survey data from 1992 through 1996, Hogan and colleagues 2000Citation found that total health care spending in the calendar year of death for Medicare hospice users is not significantly different from that for Medicare decedents who do not use hospice. Because of the lack of randomization, selection bias is also a potential problem for this study.

Two randomized controlled trials (Hughes et al. 1992Citation; Kane, Wales, Bernstein, Leibowitz, and Kaplan 1984Citation) have examined the effect of hospice use on cost and other outcomes. Hughes and colleagues 1992Citation report on a randomized trial of a U.S. Department of Veterans Affairs hospital-based home care program for terminally ill patients who had informal caregivers. They found that average use of inpatient care was higher for the home care program group than for the "control" group that was not assigned to the program, but average use of nursing visits was lower for the home care program group. Net per capita spending was 18% lower for the home care program group than for the control group; however, this difference was not statistically significant. Kane and colleagues 1984Citation describe a trial of a U.S. Department of Veterans Affairs hospital program that provided hospice inpatient unit and home hospice care for terminally ill cancer patients. They found no significant difference in costs or in use of hospital inpatient care between the group enrolled in the program and the "control" group not enrolled in the program.

Financial and Organizational Barriers to High-Quality End-of-Life Care
Above and beyond the amounts that are spent on decedents, the methods of reimbursement that insurers use influence the type of care and even the organization of care received by terminally ill persons. As described briefly in the payment system overview, reimbursement methods create inducements and barriers that influence the use of certain settings over others, transitions between settings, and decedents' place of death. The major sources of economic influence are described in more detail later.

The Health Care System Is Oriented Toward Acute Care
Most of the health care system is organized around providing acute care services to patients who will fully recover from their illnesses and injuries. Health insurance is designed to protect policyholders from the expenses associated with these types of illnesses. Indeed, out of a fear that they will disproportionately attract sicker than average people, many health insurers restrict care provided to people with ongoing medical problems, including people with terminal illnesses (Institute of Medicine, Committee on Care at the End of Life 1997Citation). In addition, most medical personnel are trained in institutions delivering acute care, and physicians practicing in these institutions are rewarded with greater income and prestige. Only recently have efforts begun to develop chronic-care programs—in recognition that chronic illnesses have emerged as the major causes of morbidity and mortality in the United States (Wagner et al. 2001Citation).

The acute orientation of the health care system is such that hospitals are not explicitly reimbursed for providing palliative care and thus are reportedly reluctant to deliver it. Capello, Meier, and Cassel 1998Citation point out that, nationally, a large percentage of deaths occur in hospitals and these hospitals are not specifically reimbursed for the palliative care provided there. A diagnosis code for Medicare was created by the Centers for Medicare and Medicaid Services (the agency formerly known as the Health Care Financing Administration) that would allow for the identification of patients receiving inpatient palliative care; however, no reimbursement was (or is) associated with the code, and it has been used very little (Cassel and Vladeck 1996Citation). Cassel, Ludden, and Moon 2000Citation interviewed hospital CEOs and clinicians directing hospital palliative care services to assess perceptions of barriers to high-quality palliative care in hospitals. They find that lack of financial reimbursement has resulted in an environment in which end-of-life care is not a high priority for hospitals and recommend that Medicare add a specific DRG for inpatient palliative care.

Smits, Furletti, and Vladeck 2001Citation interviewed physician experts in palliative care to identify problem areas and stories about successful improvement, and steps Medicare could take to improve the palliative care provision. They recommend implementing new standards for hospitals, nursing homes, and other providers that would require greater training of medical staffs in the principles of palliative care and greater emphasis on palliative care in quality improvement efforts. They also recommend specific reimbursement for physician visits for end-of-life care planning.

Medicare Reimbursement Systems Do Not Encourage Coordinated Care and May Result in Access Problems
Medicare reimburses separately each type of provider that a beneficiary might use at the end of life and uses a method to do so that varies from provider type to provider type. This means that providers have little financial incentive to coordinate with each other, and they often have incentives to shift patients to other types of providers. Huskamp, Beeuwkes Buntin, Wang, and Newhouse 2001Citation report that they were told of instances in which SNFs discharged dying Medicare patients to hospitals in part to avoid the additional staffing and other care costs of patients who are actively dying.

One would expect that MCOs would have the ability and incentives to coordinate care, but there is currently no evidence that this has happened. Experton and colleagues (Experton, Li, Branch, Ozminkowski, and Mellon-Lacey 1997Citation; Experton, Ozminkowski, Branch, and Li 1996Citation; Experton, Ozminkowski, Pearlman, & Thompson, 1999) tracked frail Medicare FFS, Medicare health maintenance organization (HMO), and Medicare/Medicaid beneficiaries receiving physician-prescribed home health care services. They found that whereas decedents' overall resource use did not vary by payer, those in HMOs had higher rates of preventable hospitalizations. In a 1999 report, the Education Development Center looked for exemplary models of end-of-life care in MCOs (Solomon, Romar, Sellers, & National Task Force on End-of-Life Care in Managed Care, 1999). They found some, but concluded that overall, MCOs were fearful that publicity about good end-of-life care would bring higher risk enrollees. In addition, as noted previously, Medicare HMOs are not responsible for hospice care, which may reduce coordination efforts.

Lynn, Wilkinson, Cohn, and Jones 1998Citation identified these problems as shortcomings of current FFS and managed care plans in providing end-of-life care, and have proposed a new comprehensive system of managed care for seriously ill patients nearing the end of life. This system, which they term "MediCaring," would be capitated or salaried and would offer comprehensive palliative care. Eligibility would be based on functional status rather than on prognosis.

In addition to problems with coordination of care, many providers fear that the new prospective payment systems for home health care agencies and SNFs will cause access problems for more complex patients, including those with terminal illnesses. Huskamp and colleagues 2001Citation report hospice and SNF access problems for patients with particularly high cost needs (i.e., the cost of which would greatly exceed the per diem payment).

Medicare Hospice Payment Rates Reflect Historic Patterns of Care
The Medicare hospice benefit was designed to fit the needs of a population dying mostly of cancer in the early 1980s. Since then, technological advances have changed the nature of cancer treatment, shortening hospice length of stay. Also, the range and expense of palliative interventions have increased, and hospices are caring for an increasing number of noncancer patients (U.S. Government Accounting Office, 2000). Huskamp and colleagues 2001Citation describe a number of ways in which the hospice benefit could be updated to reflect current care patterns and reimburse hospice providers more appropriately. These include rebasing hospice reimbursement rate levels to reflect the current hospice user population, care patterns, and technology, and paying more for the first and last days of a hospice stay so that providers are compensated fairly for patients with shorter lengths of stay. In addition, they suggest that policymakers consider adding an outlier policy, reimbursing for above-average travel costs, providing interim payment for patients who stabilize while on the hospice benefit, and augmenting reimbursement for respite care.

Organizational Arrangements Influence the Use of Hospice Care
The organization of a local health care market is correlated with the care delivered in that market. Wennberg and Cooper 1998Citation have extensively documented regional variation in health care use in general and the use of care by decedents. Christakis and Iwashyna 2000Citation followed a large cohort of hospice patients admitted in 1993 to examine the factors associated with duration of survival after hospice enrollment. Patients residing in markets with more hospital beds, greater hospice capacity, or a higher proportion of generalists were enrolled in hospice for more days before their death.

SUPPORT (Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments) study data have been used to analyze the associations among patient characteristics, patient preferences, health system characteristics, and place of death. Patients' preferences for at-home deaths did not explain the variation in rates of at-home versus in-hospital deaths. The supply of hospital days per beneficiary and Medicare spending on home health care in the area were positive predictors of in-hospital death. Medicare spending on hospice care and skilled nursing care was negatively associated with in-hospital death. In other words, the available supply of health care services was a better predictor of site of death than patient preferences.

There has been some controversy about the relationships between hospices and nursing homes. Petrisek and Mor 1999Citation investigated the factors associated with the proportion of nursing home residents receiving the Medicare hospice benefit. They found that for-profit, free-standing, and chain-affiliated nursing homes were more likely to have hospice patients than not-for-profit nursing homes. Nursing homes with lower percentages of Medicaid revenue, specialized units, lower levels of skilled personnel, and more competitive nursing home markets were also more likely to have hospice patients. The authors conclude that these patterns are consistent with nursing homes responding to financial incentives to increase the use of the hospice benefit. Gage and colleagues 2000Citation were commissioned to conduct a study of the hospice benefit, with a focus on the use of hospice in nursing homes, in response to investigations of financial relationships between hospices and nursing homes. They, too, found wide variation in the use of hospice in nursing homes. They also found that where hospice was used, it led to better care for nursing home residents—especially better pain control. In addition, they identified a lingering concern that the legitimate use of hospice in nursing homes in many states may lead to citations for poor quality care when patients decline. For example, hospice patients exercising their right to refuse food could be considered malnourished and neglected under some nursing home survey criteria.

Financing and Organization Influence a Patient's Place of Death
A number of polls and studies have addressed patient preferences for care at the end of life. For example, Gallup polls conducted in 1992 and 1996 found that 90% of respondents report that they would prefer being cared for in their home if they had a terminal illness and a prognosis of 6 months or less (e.g., Foreman 1996Citation; National Hospice Organization 1996Citation; Seidlitz, Duberstein, Cox, and Conwell 1995Citation). Townsend and colleagues 1990Citation, in a prospective study of a random sample of patients with cancer and a life expectancy of less than 1 year, found that 58% of patients stated a preference for dying at home during their initial interview. Fifty percent of those who would have preferred to die at home, but died in a hospital, could have died at home if there had been more supportive care available.

Wallston, Burger, Smith, and Baugher 1988Citation used data from the National Hospice Study to develop a quality-of-death measure based on patient reports of their preferences regarding their last days of life and caregiver reports of the actual outcome. They found that patients in hospice care had higher quality-of-death scores than patients in conventional care on average. However, selection bias may have influenced this result, as those choosing to enroll in hospice may do so because of stronger preferences for a "quality" death.

Mor and Hiris 1983Citation used data from the National Hospice Study to predict the site of death of hospice cancer patients on the basis of patient demographic characteristics, patients' support networks, patients' care needs, and health system factors. They found that whether the patient's hospice had inpatient beds was the dominant factor in whether or not the patient died at home, even after controlling for the sicker patients seen by the hospital-based hospices. Mor, Wachtel, and Kidder 1985Citation also compared the characteristics of patients who chose a hospital-based hospice and those who chose home hospice. They found that patients are more likely to select a hospital-based hospice if they lack a caregiver at home or if they have severe functional limitations.

These Incentives and Organizational Forces Have Different Implications for "Trajectories of Death"
Incentives and organizational forces have different implications for the "trajectories of death" identified by Lunney, Lynn, and Hogan 2001Citation. The needs, if any, of patients dying suddenly are likely met well by Medicare's coverage of intensive acute care. Terminal cancer patients may be relatively well served by the hospice benefit once they enter their final stage of decline. Even these patients, however, may desire services, such as palliative radiation therapy, that some hospices do not provide because they do not consider them strictly palliative. In addition, cancer patients generally need caregivers at home to avail themselves of hospice. Patients with organ system failure will have their medical care covered, but may need care management and coordination, patient education, and prescription drugs to enjoy a high quality of life. However, Medicare does not cover these benefits well, if at all, and qualifying for the hospice benefit is more difficult for these patients because of the uncertain trajectory of their illness (Medicare Payment Advisory Commission 1999Citation). Patients who are simply frail are probably the least well served, because Medicare does not cover their long-term needs for supportive care, and they may never qualify for hospice.

Nonfinancial Barriers to High-Quality End-of-Life Care
Although our primary aim is to outline the influences of economic and organizational factors on end-of-life care, we would be remiss in not discussing the nonfinancial barriers. This is both because nonfinancial factors have a strong effect on care and because these factors influence the financing and organization of care through the policy process and consumer demand. The nonfinancial factors that influence treatment decision making at the end of life include patient preferences for their care, prognostication accuracy, and communication between patients and physicians.

Current Methods for Making Prognoses are Imprecise
In a prospective cohort study involving interviews of 326 physicians of cancer patients admitted to 1 of 5 outpatient hospices in Chicago in 1996, Lamont and Christakis 2001Citation found that physicians tended to overestimate the prognosis of terminally ill patients. The median physician prognosis was 75 days from hospice admission, but median survival was only 26 days. Fox and colleagues 1999Citation used data from SUPPORT (with a sample of 2,067 terminally ill hospitalized patients) to examine the accuracy of currently available criteria (National Hospice Organization and SUPPORT guidelines) for identifying patients with a prognosis of 6 months or less. They concluded that neither set of criteria was effective in identifying a population with 6 months or less to live, a criterion for eligibility for the Medicare hospice benefit. If prognoses are overestimated, patients who could benefit from hospice may not be enrolled because they do not meet Medicare eligibility criteria for the benefit.

Shortcomings in Patient/Physician Communication at the End of Life Are Common
Several studies have documented shortcomings in communication between physicians and patients regarding prognosis and preferences for care (e.g., Eidinger and Schapira 1984Citation; Mackillop, Stewart, Ginsberg, & Stewart, 1988; SUPPORT, 1995; Weeks et al. 1998Citation; Wenger et al. 2001Citation). For example, Eidinger and Schapira 1984Citation found in their study that one third of the 190 patients with incurable metastatic cancer whom they studied believed they were receiving treatment to cure their condition. Using a cross-sectional survey of HIV patients to examine the prevalence of end-of-life preference discussions between practitioners and patients, Wenger and colleagues 2001Citation found that only 50% of the HIV patients surveyed had discussed some aspect of end-of-life care with their practitioners. This is a serious barrier to good care. A 1994 Louis Harris Poll found that almost 70% of a national sample preferred to make final decisions about their care if they were seriously ill. Only approximately 25% wanted their physician to make the final decisions (Institute of Medicine, Committee on Care at the End of Life 1997Citation; Medica Foundation 1994Citation).

According to Lamont and Christakis 2001Citation, when physicians and patients do communicate about prognosis, the physician often does not convey accurately her or his prognosis to the patient. Lamont and Christakis 2001Citation found that physicians often communicate to patients a more optimistic prognosis than they believe to be true. In only 37% of cases did the physician report communicating a true prognosis to the patient.

Overestimating prognosis can influence preferences for care. In a study of 917 patients with terminal cancer enrolled in SUPPORT, Weeks and coworkers 1998Citation found that patients tend to overestimate their probability of survival and that patients' survival estimates influence the treatment decisions they make. They found that patients who thought they had a high probability of surviving for at least 6 months were more than twice as likely to prefer life-extending therapy over comfort care.

Improving physician-patient communication may be difficult. SUPPORT (1995) involved a randomized controlled trial of an intervention to improve physician-patient communication and outcomes (e.g., pain control). Under the intervention, physicians received daily estimates of a patient's probability of survival, and a nurse had multiple contacts with patients, their families, and physicians to discuss preferences for care and to facilitate physician-patient communication. The intervention failed to improve communication or outcomes, showing that improvements may not come easily.

Patient and Market Characteristics Influence the Timing of Hospice Referral
Several studies have examined the patient characteristics associated with end-of-life treatment choices (e.g., Berry, Boughton, and McNamee 1994Citation; Christakis 1994Citation; Gaskin, Kong, Meropol, Yabroff, and Weaver 1998Citation; Christakis & Iwashyna, 1998, Christakis and Iwashyna 2000Citation). Christakis and Iwashyna 2000Citation found that race; age; comorbid mental illness, substance abuse, or dementia; and various market characteristics (e.g., hospice capacity, relative mix of generalist vs. specialist physicians) are associated with survival after hospice admission, whereas income and education of the patient are not associated with survival.

Use of End-of-Life Care Varies by Race and Socioeconomic Status
There are significant disparities in access to health care between racial and ethnic groups and persons of different socioeconomic status in the United States (Agency for Healthcare Research and Quality 2000Citation). Race, ethnicity, and socioeconomic status are also correlated with differential patterns in care received at the end of life. Race and socioeconomic status influence care received at the end of life, as they influence many other types of health care usage. Hogan, Lunney, Gabel, and Lynn 2001Citation recently investigated patterns in the use of care by Medicare beneficiaries in the 3 years before death. They found that African Americans used 25% less care in the 3 years before death than White persons, but 18% more in the last year of life. The primary reason for the greater use in the last year of life was greater use of inpatient care.

Gordon 1995Citation described barriers to the use of the Medicare hospice benefit by African Americans and Mexican Americans that could explain these trends. The lack of a full-time caregiver and a distrust of White care providers are cited as barriers for African Americans. People of this ethnic group are also more likely to report that they prefer to die in the hospital. Lack of familiarity with the health care system, language barriers, and circumscribed family support systems are barriers for Mexican Americans. People of this ethnic group are less likely to report that they would admit to experiencing pain or tell others they are close to death.

Differences by socioeconomic status are not limited to the United States. Catalan-Fernandez and colleagues 1991Citation presented the results of a survey of surviving relatives of cancer decedents who had been hospitalized in Majorca, Spain. Patients residing in rural areas were more likely to die at home, and patients of lower socioeconomic status were more likely to die in the hospital. In addition, they found that 31% of patients who died at home or in the hospital would have preferred to die in the other setting; however, only approximately 13% of relatives in each situation would have preferred that the death take place in a different setting. The relatives of patients dying at home had more positive perceptions of the patients' deaths.


    Research Priorities
 TOP
 Abstract
 Overview of Medicare Payment...
 State of Science
 Research Priorities
 References
 
Our review of the literature suggests two fields of inquiry for future research. First, critical gaps in our knowledge about how to design an end-of-life care system that better meets the needs of older populations must be filled. We would argue that the most critical gaps are in our understanding of how end-of-life care decisions are made and their consequences for patient satisfaction, care costs, and care outcomes. Second, researchers should study what can be done to improve the care available to dying patients under Medicare's current financing rules.

Little is known about the factors that influence treatment decision making at the end of life. For example, how do patient preferences and knowledge about end-of-life care options interact with physician attitudes and knowledge about end-of-life care options to affect decision making? Are patients' options and preferences changing as advances in medical technology allow people to live longer with chronic illnesses and with better quality of life? Although polls of the general public indicate that nearly everyone would prefer to die at home in the abstract, it would be interesting to know what dying patients and their caregivers think about these issues. Also, how are recent efforts to educate physicians and other health providers about palliative care affecting the advice and options that they provide to patients? Research is needed to address these and other related issues.

Prospective studies that are free of selection bias are needed to understand how costs, patient and family satisfaction with care, and outcomes differ across different end-of-life care treatment options. Understanding the costs and benefits of different treatment options will also help us assess the effects of changes in policy on costs and outcomes. For example, such studies could shed light on the cost to Medicare of basing eligibility for hospice-like services on functional status/severity of illness rather than on prognosis and on the potential cost of adding other Medicare end-of-life benefits. Simulation models to estimate the potential impacts of payment or benefit changes on costs and utilization patterns could be developed to assist in this task. These studies should include comparative analyses of care delivered to individuals who are dually eligible for Medicare and Medicaid, individuals who have supplemental private coverage, (i.e., Medigap or retiree coverage), and those who have Medicare coverage only. They should also examine the determinants and consequences of geographic variation in end-of-life care in more detail.

While this work is proceeding, however, research also can be conducted into how well the current system of coverage and eligibility is functioning. For example, how well do hospice per diem payments match current treatment patterns? What is the distribution of costs within the current hospice patient population, and does it differ across diagnoses or trajectories of illness? We can also monitor the extent to which current reimbursement methods create access problems or selection against high-cost patients. In particular, what are the effects of new SNF and home health agency payment systems on costs of care at the end of life, the setting where care is delivered, transitions between settings, and quality of care? The effects should be monitored and the interactions between the new payment systems evaluated to ensure that changes are not adversely affecting the care received by patients at the end of life.

Also, we need to study technological innovation in the provision of end-of-life care. Payment systems affect the development and diffusion of new technologies through the incentives they produce and their level of flexibility and generosity. For example, payment systems that are too rigid can stymie the introduction of new technologies and slow their diffusion. They may also reduce incentives to develop new technologies in the first place. At the same time that we look toward new technology, however, we need to realize that some patients do not even receive the basic low-tech care they desire. To make rational decisions about what the scale and scope of Medicare coverage of end-of-life care services should be, policymakers need to understand how their choices can affect the care available both now, and in the future, because of reimbursements' cumulative incentives for innovation.

One thing about the future does seem certain: The need to understand and improve end-of-life care will become more pressing. The baby boom generation is in the midst of caring for their aging parents and beginning to think about their own retirement and old age. The demands they will place on the health care system will be large in financial terms and in terms of expectations for rationality and responsiveness. Our current "system" of care for the dying is not likely to meet those demands.

The Forum

Book Reviews

Practice Concepts

Received for publication September 14, 2001. Accepted for publication May 7, 2002.


    References
 TOP
 Abstract
 Overview of Medicare Payment...
 State of Science
 Research Priorities
 References
 




This article has been cited by other articles:


Home page
AM J HOSP PALLIAT CAREHome page
Y. Shnoor, M. Szlaifer, A. S. Aoberman, and N. Bentur
The Cost of Home Hospice Care for Terminal Patients in Israel
American Journal of Hospice and Palliative Medicine, September 1, 2007; 24(4): 284 - 290.
[Abstract] [PDF]


Home page
GerontologistHome page
D. P. Waldrop
At the Eleventh Hour: Psychosocial Dynamics in Short Hospice Stays
Gerontologist, February 1, 2006; 46(1): 106 - 114.
[Abstract] [Full Text] [PDF]


Home page
Health Aff (Millwood)Home page
T. Gilmer, L. J. Schneiderman, H. Teetzel, J. Blustein, K. Briggs, F. Cohn, R. Cranford, D. Dugan, G. Komatsu, and E. Young
The Costs Of Nonbeneficial Treatment In The Intensive Care Setting
Health Aff., July 1, 2005; 24(4): 961 - 971.
[Abstract] [Full Text] [PDF]


Home page
Palliat MedHome page
K. I Stajduhar and B. Davies
Variations in and factors influencing family members' decisions for palliative home care
Palliative Medicine, January 1, 2005; 19(1): 21 - 32.
[Abstract] [PDF]


Home page
RadiologyHome page
R. M. Friedenberg
The End of Life
Radiology, August 1, 2003; 228(2): 309 - 312.
[Full Text] [PDF]


This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Buntin, M. B.
Right arrow Articles by Huskamp, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Buntin, M. B.
Right arrow Articles by Huskamp, H.


HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
All GSA journals Journals of Gerontology Series A: Biological Sciences and Medical Sciences Journals of Gerontology Series B: Psychological Sciences and Social Sciences