6

Financial and Economic
Issues in End-of-Life Care

To many people, reducing expenditures at the end of life seems an easy and readily justifiable way of cutting wasteful spending and free resources to ensure universal access to health care. . . . We must stop deluding ourselves. . . .

The United States has experienced years of mounting concern about steep increases in health care spending and in the share of national resources devoted to health care. Between 1970 and 1994, personal health care spending grew from $63.8 billion to $831.7 billion (from $310.9 to $764.6 billion in 1992 dollars), and the percentage of the U.S. gross domestic product devoted to health care rose from 7.1 percent to 13.7 percent (Levit et al., 1996). Despite a recent slowdown in the rate of increase for health care spending, the stress on controlling or reducing health care costs remains intense.

Because spending on care at the end of life is so high and because much of it is financed by government programs, the cost of end-of-life care has attracted considerable attention. In the 1980s, analyses indicated that over one-fourth of Medicare expenditures in a year were accounted for by the 5 percent to 6 percent of beneficiaries in their last year of life. These figures prompted debate about whether such spending was excessive and whether advance directives, hospice care, and futility guidelines could be promoted to control costs as well as strengthen patient autonomy and improve care (Jecker and Schneiderman, 1992; Lundberg, 1992; Singer and Lowy, 1992; Fries et al., 1993; Murphy and Finucane, 1993). Analyses have cast doubt on expectations that these strategies might make a major contribution to containing health care costs (see, e.g., Emanuel and Emanuel, 1994; Teno, Lynn, Connors et al., 1997). More forceful policies, such as explicit age-based rationing, have also been discussed.

Financing of end-of-life care is also a concern for those who see incentives that in combination with cultural, legal, and other factors may impede excellent palliative care. As discussed in this chapter, financing mechanisms may encourage or discourage the use of beneficial services and may affect the mix of services, providers, and settings of care available to dying patients and their families.

This chapter examines four major questions asked about the cost of care for those who die. First, who pays for care at the end of life? Second, what is spent on this care? Third, do financing mechanisms create impediments to good care? Fourth, how might costs for care at the end of life be reduced?




WHO PAYS FOR CARE AT THE END OF LIFE?

No comprehensive national statistics document in detail the sources of payment for care at the end of life. Nonetheless, it seems clear that dying is, in considerable measure, publicly funded. Because over 70 percent of those who die each year are elderly and covered by Medicare and because thirteen percent of Medicare beneficiaries are also covered by Medicaid, those two programs undoubtedly cover a large proportion of expenses for end-of-life care. In addition to elderly people, others of those who die (e.g., some AIDS patients) are seriously enough disabled or impoverished that they qualify for Medicare or Medicaid or both. Veterans and defense health programs also pay for some end-of-life care. Moreover, some people who die without insurance or other available financial resources—for example, many homicide victims and homeless adults—die in public hospitals.

Notwithstanding Medicare's importance to its beneficiaries, the program does not cover all of their health care expenses. Data from the 1987 National Medical Care Expenditure Survey indicates that for those aged 65 or over who died in 1987, Medicare accounted for 48 percent of health expenditures during the last six months of life (52 percent for noninstitutionalized decedents and 39 percent for those in institutions) (calculated from Table 2 in Cohen, Carlson, et al., 1995). For all beneficiaries, in 1992, Medicare covered barely half (53 percent) of health care expenses with 14 percent, 10 percent, 20 percent, and 3 percent of expenses covered by Medicaid, private insurance, beneficiary out-of-pocket spending, and other sources respectively (Gornick et al., 1996).¹

For both decedents and survivors, Medicare coverage is particularly limited for long-term care, outpatient medications, and supportive services. For many, much of this care is either paid for out of pocket or provided by unpaid family members (Hing and Bloom, 1990; Pepper Commission, 1990). For example, although more than three-fourths of Medicare beneficiaries have some form of supplemental private insurance that is either self-purchased or provided by a former employer (PPRC, 1995), private insurance covers only 25 percent of prescription drug spending with almost 60 percent paid for out of pocket. Some dying patients qualify for the Medicare hospice benefit, which covers some prescription drugs and some nonmedical services (e.g., light housekeeping assistance, respite care). Community programs supported by other federal, state, local, or private sources provide some nonmedical supportive services (e.g., delivery of meals, transportation) to older people and those with disabilities.

Although over one-fourth of those who die each year are under 65, little information is available that describes how their care is financed. Some younger people who die are covered (directly or as dependents) by employer-sponsored health plans, and it seems reasonable to expect that this small group would (as in Medicare) account for a significant proportion of plan spending (IOM, 1993b). The liability of employer-sponsored plans for end-of-life care may, however, be limited when dying beneficiaries exceed maximum coverage limits or resign their jobs or when insured spouses resign or lose jobs as a result of absenteeism caused by their caregiving burdens.² Thirty-seven states provide Medicaid benefits for hospice care, which is used primarily by impoverished, nonelderly patients who do not qualify for Medicare under the disability provisions.




WHAT IS SPENT FOR CARE AT THE END OF LIFE?

Several points about spending on care at the end of life warrant emphasis. Because people who die generally have been very ill, many of these points should not be surprising. First, the small percentage of those who die each year accounts for a considerable percentage of total health care spending. A 1984 article (Lubitz and Prihoda) attracted widespread interest when it reported that the 5.9 percent of elderly Medicare beneficiaries who were in their last year of life in 1978 accounted for 27.9 percent of total Medicare spending. Other analyses suggest similar patterns through the 1980s and also dating back to 1960, before the adoption of Medicare (Lubitz and Prihoda, 1984; Scitovsky, 1984; Riley et al., 1986; Gornick et al., 1993; Lubitz and Riley, 1993). A more recent analysis looked beyond the Medicare population using data from the 1987 National Medical Care Expenditure Survey (Cohen, Carlson et al., 1995). This analysis estimated that total health care expenditures during the last six months of life for the 2.1 million people who died in 1987 (approximately 0.9 percent of the population) amounted to $44.9 billion in 1992 dollars (approximately 7.5 percent of total personal health care expenditures). For those aged 65 or over who died in 1987, this analysis found that spending during the last six months of life accounted for about 5.5 percent ($32.6 billion) of total spending.³ In this same age population, another study (Scitovsky, 1994) reported that 23 percent of survivors had no Medicare payments in 1988 compared with 3 percent of those who died.

Second, as one extends the time analyzed from the year before death to several years before, the contrast between expenditures for survivors and decedents diminishes. In 1988 average Medicare payments for decedents ($13,300) were approximately 7 times those for survivors ($1,900) (Lubitz and Riley, 1993).⁴ In 1978, the decedent-to-survivor payment ratio was 6.2 in the last year of life but dropped to 2.3 for the second-to-last year (Lubitz and Prihoda, 1984). For those who were 65 in 1974 and died in 1989, Medicare payments over this 15-year period were about twice as high for those who died as for those who survived (Gornick et al., 1993; see also Roos et al., 1987).

Third, the cause of death contributes to variations in expenditure levels and ratios for decedents and survivors both in the year of death and in the years before the final year (Riley and Lubitz, 1989; Scitovsky, 1994; Riley et al., 1995). Some conditions, such as diabetes and renal failure, show relatively long periods of higher decedent-to-survivor payment ratios. In contrast, payments for cancer decedents were notably higher than for survivors just for the year of death and the year before (Scitovsky, 1994). Among cancer patients, a recent analysis found that total Medicare payments between the time of diagnosis and the time of death were highest for those dying of bladder cancer ($57,600) and lowest for those dying of lung cancer ($29,200) (Riley et al., 1995). The difference was accounted for primarily by the shorter survival times for the latter group of patients.

Fourth, total Medicare payments per decedent drop as age at death increases; in contrast, expenditures increase with age for survivors (see Table 6.1). Some data, however, suggest that this pattern is less evident for accidents and heart attacks than for cancer, chronic obstructive pulmonary disease, and renal failure (Scitovsky, 1994). More generally, for Medicare beneficiaries who died between 1984 and 1991, higher levels of hospital charges (which cannot be equated with actual payments) were negatively associated with older age and a diagnosis of Alzheimer's disease and positively associated with being female, being on Medicaid, living in areas of higher population density, and having poor perceived health status (Culler et al., 1995). ⁵

Fifth, lower Medicare spending for older decedents appears to be partly offset by higher Medicaid spending for those who are eligible for both Medicare and Medicaid (Temkin-Greener et al., 1992; see also Scitovsky, 1988). Also, a recent paper cites yet provisional, unpublished research at the U.S. Department of Health and Human Services as indicating "that nursing home costs rise with longevity, offsetting declining Medicare costs" (Scitovsky, 1996, p. 3). The reason for this pattern is suggested by data indicating that among those aged 65 to 74, about 17 percent of those who died had spent some time in a nursing home, but for those between ages 85 and 94, the figure was 60 percent (Kemper and Murtaugh, 1991).

Sixth, although data are limited, analyses suggest that chronic and terminal illnesses have serious financial consequences for families (Bloom et al, 1985; Beck-Friis et al., 1991; Covinsky et al., 1994, 1996). These financial consequences stem in part from out-of-pocket medical expenses but also from lower patient or family income that results from absenteeism, reduced working hours, or job loss related to illness or the demands of caring for an ill family member.

Seventh and contrary to some popular thinking, the increase in overall personal health care spending is not explained by growing costs for end-of-life care. Rather, it is accounted for by population growth, general inflation in the economy, and additional medical inflation, although the contribution of each component has fluctuated widely (Levit et al., 1996). An analysis covering data for 1976, 1980, 1985, and 1988 (excluding Medicare beneficiaries enrolled in health maintenance organizations [HMOs]) revealed that the share of Medicare spending accounted for by those who died had not grown during this period (Lubitz and Riley, 1993). The authors concluded that "the same forces that have acted to increase overall Medicare expenditures have affected care for both decedents and survivors" (Lubitz and Riley, 1993, p. 1093; see Table 6.2). Moreover, another analysis has estimated that the impact on Medicare spending of increasing life expectancy is small (Lubitz et al., 1995).




DO FINANCING MECHANISMS IMPEDE
GOOD END-OF-LIFE CARE?

General Features of Insurance Plans That Might Affect End-of-Life Care

The committee searched for research on the effects of insurance on care at the end of life. It found very little. For example, the Health Insurance Experiment funded by the federal government and undertaken by Rand researchers excluded people over age 62 (i.e., those more likely to die) (Newhouse et al. 1993). What information exists is largely limited to traditional fee-for-service arrangements and may not apply to managed care, which covers a small but increasing proportion of Medicare beneficiaries.

Overall, considerable research indicates that the availability and type of health care insurance affect the use and provision of health care (see, e.g., Hadley, 1982; Lurie et al., 1984, 1986; Manning et al., 1987; Braveman et al., 1989; Hadley et al., 1991; IOM, 1993b; Newhouse et al., 1993). In part, these effects are intended and desired. For instance, health insurance eases the financial burden of illness and encourages people to get beneficial care that they might otherwise forego.⁶ Insurance, however, may also encourage the use of services of little value, thus raising costs without comparably increasing value. As described below, insurers and policymakers have devised a number of mechanisms to counter this utilization effect of insurance.

Insurers have also devised responses to another problem—biased risk selection—that may have consequences for seriously ill people. Biased risk selection arises when insurance is voluntary or when people can choose among health plans with different features. Plans that disproportionately attract the sick will cost more than those that attract healthier people.⁷ One result of this adverse selection dynamic is that insurers may refuse or restrict coverage to those with existing medical problems or may charge them more. Medicare is virtually universal for Americans aged 65 and over, which limits selection problems. Nonetheless, many beneficiaries voluntarily purchase supplemental coverage, and others voluntarily enroll in certified managed care plans; both choices may generate problems related to adverse selection. A particular concern is that competitive health plans may find it more economically attractive to compete on the basis of risk selection (i.e., attracting the well and avoiding the sick, especially the really sick) than on the basis of providing high-quality, cost-effective care.

Although insurance increases access to care, those without insurance do not necessarily go untreated, especially when they have acute or life-threatening problems (Hafner-Eaton, 1993). Emergency departments are forbidden by law from turning away critically ill people for lack of insurance or other financial resources. If an uninsured person is admitted for care and dies, the expenses incurred may be absorbed by the hospital (and, when possible, passed on indirectly to other payers) or covered by special public or charitable funds. If the person survives but is expected to remain seriously ill, efforts may be made to qualify the individual for public insurance or arrange a transfer to a public facility.

Just as being without insurance does not necessarily preclude care, participating in an insurance plan does not guarantee that the plan will pay for the specific services wanted, needed, or received by those the plan insures. Health plans vary in the services they cover, and even for covered services, other restrictions, such as requirements for prior approval of hospitalization, may apply (IOM, 1989). In addition to limiting costs, the objectives of such requirements may be to improve quality of care by reducing departures from those clinical practices associated with better outcomes, to reduce patient care for which risks exceed probable benefits, or to limit exposure to iatrogenic (care-induced) illness.

Health plan restrictions that may particularly affect people with advanced illness come in many forms. These include

• limiting the scope or level of benefits to restrict costs and encourage economical choices by consumers. Some services may be excluded altogether; for example, Medicare excludes payment for most outpatient prescription medications. In addition, plans may require patients to pay part of the cost of a service (in the form of deductibles, copayments, or coinsurance), cover a limited number of visits or days of care, or set a cap on the dollar amount of payments for selected or all services during a defined time period. Some plans (including Medicare) set no upper limit on beneficiary liability for cost-sharing.

• creating financial incentives for practitioners and providers to provide less care. Plans may establish a fixed payment per day, per case, or per capita regardless of the amount of service rendered. Some health plans are paid by employers, governments, or others on a capitated (per member per month) basis, but they may pay practitioners and providers on a fee-for-service, per case, or other basis, or they may combine partial capitation with other payments.

• requiring that patient services be authorized in advance. Many health plans insist that special personnel review hospital admissions, medical procedures, and certain other services. They may also require that a designated primary care physician authorize referrals to specialists.

• creating protocols or other tools to govern care. Plans may adopt written statements that define what services are to be provided for particular medical problems, what medications are covered (e.g., formularies), and otherwise direct care to varying degrees.

• establishing productivity standards or appointment schedules. Such schedules may limit the time physicians can spend in evaluating patients, identifying problems and concerns (such as depression or persistent pain), and discussing care options.

• covering services only if provided by designated physicians and other health care providers or applying higher cost sharing requirements if patients use physicians outside the plan network. Patients may be covered only if they use providers that have contracts with or are employed by the health plan or if they obtain explicit authorization—and pay more—for the use of out-of-plan care.

• limiting the number, type, and geographic distribution of designated providers of specialized services. Depending on the characteristics of the provider network, patients may find specialist expertise unavailable for some problems, appointments hard to get, or travel requirements difficult.

The label managed care is often applied to health plans that employ one or more of the last six strategies listed above, although some use more restrictive definitions that focus on one strategy (e.g., financial incentives) or a specific combination of strategies. This report follows the less restrictive usage. In any case, the report's major concern is not with organizational labels but with specific incentives and actions that may particuarly affect those with advanced illness. Chapter 4 has already considered some of the concerns about incentives used by managed care plans, which cover approximately 1 Medicare beneficiary in 10 (with some states having considerably higher proportions).




Financing Issues by Type of Service

Although care at the end of life is often associated with hospice services, this report has pointed out that only a small percentage of Medicare beneficiaries qualify for hospice benefits. Despite a general shift of care from hospitals to homes for both financial reasons and personal preferences, most older individuals still die in institutional settings, especially hospitals. Hospitals also play a major role in caring for incurably ill people who subsequently die elsewhere. Thus, the following discussion first considers hospitals, then nursing homes, hospice, and home care. The emphasis here is on the traditional, fee-for-service Medicare program and the possible effects of its financing provisions on care at the end of life. A later section considers Medicare managed care. To repeat a familiar theme, the effect of financing mechanisms on end-of-life care in different settings has been little examined.




Hospital Services

From the perspective of this report, major concerns about Medicare's hospital payment policies are that they may encourage premature patient discharge and discourage appropriate inpatient palliative services. Since the early 1980s, Medicare, which is administered by the Health Care Financing Administration (HCFA), has used a prospective payment scheme that pays for most hospital stays on a prospectively determined, diagnosis-related basis.⁸ In general, the scheme assumes that hospitals will get a mix of patients in each diagnosis-related category or group (DRG), some of whom will cost more than the DRG rate and some less. The differences, it is assumed, will average out over all patients. If hospitals spend less than the DRG payment, they keep the difference. They are not, in general, compensated more if they spend more unless a patient reaches the defined outlier (very high cost) category. Although other factors are also at work (e.g., pressure from private payers and employers), the structuring of the prospective payment program is generally credited with reducing hospital stays without an offsetting increase in hospital admissions.

Because Medicare provides less generous coverage for much nonhospital care, shorter hospital stays may increase the financial burden on patients and families, especially for prescription drugs. Further, because dying patients are among the sickest of Medicare patients, the physical, emotional, and financial burdens of severely shortened lengths of stay may fall heavily on this group.

One particular concern has been that prospective payment might encourage hospitals to control costs by cutting quality and limiting beneficial care, including palliative services. Early analyses suggested that Medicare beneficiaries were indeed being discharged "quicker and sicker" and in more unstable condition, but no effects on 30-day or 6-month mortality could be readily identified (ProPAC, 1989; Kahn et al., 1990; Kosecoff et al., 1990; Rubenstein et al., 1990).

Although these initial findings were reassuring, mortality is not the only issue. Premature discharge may cause physical and emotional distress for patients and families, for example, if patients are discharged more quickly than appropriate home palliative care can be organized. In addition, some dying patients may be difficult to manage outside the hospital, even with intensive palliative services. For example, some patients experience pain that can only be relieved with radiation or other therapies available in hospitals. Not only may hospitals be discouraged from treating such patients, hospice coverage for inpatient care—whatever their justification—is very limited.⁹

In response to concerns about limitations on palliative hospital care for dying patients, the Medicare program agreed to test a palliative care diagnostic code (ICD-9-CM V66.7) (Cassel, 1996). The test will help determine whether a new DRG code is needed to pay hospitals for palliative services to terminally ill patients and whether it is feasible. One question is whether it is possible to identify a homogeneous set of palliative resources—irrespective of diagnosis—so that they can be reimbursed and monitored for quality of care. The code is not intended to discourage the referral to hospice of patients who can be appropriately and feasibly managed with hospice's home and inpatient resources and coverage. Nonetheless, committee members are aware that some hospice organizations are concerned that the code could have negative effects on them.

In general, the committee urges that more attention be paid to Medicare payment policies and the hospitalization experiences of dying patients, including the experiences of those who die (and are expected to die) shortly after discharge. Relevant policies include those relating to inpatient palliative care. Among the process and outcome variables that warrant attention are trends in DNR orders and in systems for monitoring and comparing patient survival and other outcomes for hospitals and other sites of care.




Physician Services

Most physician services provided to Medicare patients (in the hospital, office, home, and nursing home or other institution) are reimbursed on a fee-for-service (FFS) basis. Since the early 1990s, these fees have been set using a resource-based relative value scale (PPRC, 1992). For beneficiaries enrolled in HMOs that receive a capitation (per person per month) payment from Medicare, different physician payment models may apply. Plans that themselves are paid on a capitated basis may pay physicians on a fully or partly capitated basis but they may also employ a discounted or other form of FFS payment that is not so different from traditional FFS. Under capitation, salary, or mixed models of physician payment, physicians may be able to earn more by seeing more patients, whereas FFS physicians can also earn more by doing more for patients.

Aside from anxieties about the general FFS incentives for overuse of surgical and other procedural services, the primary concern about Medicare's physician payment scheme is that its payment categories and payment levels may not appropriately recognize the resources needed for high-quality palliative care in different settings. The committee also heard concerns that the channeling of payments to individual physicians may make it difficult to support a coordinated interdisciplinary team or an integrated, multi-site patient record system.

For physicians treating Medicare beneficiaries who are perceived to be dying, the most relevant fees are likely to be those for "evaluation and management" services rather than for procedural services such as surgery and diagnostic tests. Codes for evaluation and management visits differ depending on several patient or other factors (e.g., whether the visit is for a new versus established patient, for an initial versus subsequent visit, and for an office or other site of care). Within the broad visit categories, different codes exist for brief, limited, extended, and other classes of visits based on the effort or practice costs involved. Although the time and effort involved in a visit with a new patient would appear to warrant a payment differential, the distinction between new and established patients could be seen as discouraging continuity of care and encouraging switching among physicians.

Visit codes do not distinguish patients with special characteristics or needs (e.g., the frail elderly), as was recommended to HCFA by the Physician Payment Review Commission, which monitors and advises on Medicare physician payment (PPRC, 1991, 1992). Thus, the fee is the same whether or not a patient has, for example, physical or cognitive impairments or special needs for education and counseling. HCFA has responded that the special needs of these patients can either be accommodated by using a higher code (for a more complex visit) or by the averaging effect across all patients and visits. However, unless these patients are separately identified and analyzed, it is not possible to determine whether such accommodation happens in practice.

PPRC has also criticized the payments for nursing home visits as too low compared to hospital visits. It argued that "there is no sound basis for assuming the work and intensity involved in evaluating and managing nursing home patients should be any less than for patients seen in the hospital or office" (1992, p. 59).

Although guidance on how to translate actual patient encounters into visit codes has been developed for physicians and carriers who administer Medicare physician payment, some ambiguity inevitably remains. Such ambiguity combined with disputes about appropriate codes and payment levels may create particular problems for physicians trying to care well for the complex physical and emotional problems presented by patients with advanced illness that is expected to prove fatal. The committee heard concerns that carriers may not accept codes that reflect the kind of careful symptom assessment and emotionally supportive care that was discussed in Chapter 3. Medicare does not explicitly dictate the scheduling of patient visits, but restrictive interpretation of visit codes may influence physicians to limit the time spent with patients in much the same way as explicit policies do.

The committee did not identify any empirical analyses of the relative effort or practice costs for different kinds of palliative care in different care settings. Such data are needed to evaluate whether the physician fee schedule adequately compensates for palliative care services. The committee urges PPRC and HCFA to consider analyses and tests of palliative care codes for physician as well as hospital care.




Nursing Home Services

Most public funding (and about half of all funding) for nursing home care is provided by Medicaid, a joint federal-state program for which nursing home patients may qualify after spending down their own financial resources. The program pays some costs for nearly 7 in 10 nursing home residents (AHCA, 1995). In 1993, elderly beneficiaries accounted for about 12 percent of the Medicaid population but 33 percent of total expenditures, much of which was accounted for by nursing home use (PPRC, 1996). Given these proportions, it seems reasonable to expect that the controversy over Medicaid's burden on state budgets will eventually shift from cuts primarily affecting younger women and their children to focus on older people and those with disabilities.

In 1987, Congress repealed legislation providing that states pay nursing homes amounts "reasonable and adequate to meet the costs incurred by efficiently and economically operated nursing facilities." Payment levels and methods vary significantly across states and these variations have "a dramatic impact on nursing home expenditures and staffing" (IOM, 1996c, p. 62).

As of 1993, only 19 states adjusted payments based on nursing home case mix (Swan et al., 1994). Lack of case-mix adjusted payment is a concern because it may discourage nursing homes from accepting sicker Medicaid patients and from providing an adequate level of care. Overall, the combination of reduced hospital length of stays, more sophisticated medical technologies, and increasing resident age and disability are contributing to a nursing home population that is more severely ill and impaired and more demanding of nursing resources than in the past. A recent IOM study of nurse staffing levels concluded that the research literature generally indicated a "strong relationship among resident characteristics, nurse staffing time requirements, and nursing costs in nursing homes" (IOM, 1996c, p. 13).

Some nursing home patients receive hospice services under the Medicare hospice benefit. Although the Office of Inspector General (OIG) of the Department of Health and Human Services (DHHS) has raised questions about duplicate billing for these patients (OIG 1997 Workplan), little appears to be known about these patients, the care provided to them, or the degree to which nursing home personnel have the appropriate expertise for meeting their needs.

The availability, quality, and cost of nursing home care for dying patients are only a subset of troubling and difficult issues involving long-term care, which also includes much home care. Another IOM committee is expected to begin examining many of these issues in the latter part of 1997. A 1996 IOM report supported federal regulations and efforts by facilities and states to improve professional nurse staffing in nursing homes and reduce waivers of federal staffing requirements (IOM, 1996c). The same report also supported the need to increase professional nurse staffing on all nursing home shifts and specifically recommended that Medicare and Medicaid payment levels be adjusted to support a required 24-hour registered nurse coverage in nursing facilities by the year 2000. The benefits of these changes would accrue to most nursing home residents.




Hospice Services

The Medicare hospice benefit was designed specifically to increase beneficiaries' access to hospice, and about 60 percent of hospice patients are covered by Medicare (Vladeck, 1996). Hospice accounts for about 1 percent of Medicare spending, and about one-tenth of one percent of Medicaid spending (NHO, 1996a).¹⁰ A study by the Minnesota Hospice Association indicated that about 77 percent of patients enrolled in hospice were covered by Medicare, 12 percent by private insurance, 4 percent by Medicaid, and 7 percent by other sources (Holst, 1996). In addition, informal care is a major resource for hospice patients. Data from the National Hospice Study suggest that about 70 percent or more of the hours of care provided for "long-stay" hospice patients was provided by informal caregivers (Mor, 1987; see also Murinen, 1986).

The Medicare hospice benefit was adopted following years of steep increases in Medicare spending. Not surprisingly, the benefit, features of which are summarized in an Addendum to this chapter, is circumscribed with an array of features intended to control program costs by:

1. limiting the number of qualifying beneficiaries

2. encouraging efficient and economical home care

3. discouraging in-patient palliative care.

Claims that hospice can reduce spending for end-of-life care are examined in a later section of this chapter.

To qualify for Medicare hospice benefits, patients must be certified as having a life expectancy of six months or less if the illness runs its natural course. This has limited the use of the hospice benefit to those, mainly cancer patients, whose dying trajectory is relatively (but, by no means, perfectly) predictable. Conversely, this provision tends to exclude many who might benefit from hospice services but who have conditions such as congestive heart failure whose course is very difficult to predict.

As discussed earlier in this report, the prognosis provision implies a degree of precision that does not exist. The committee was, thus, disturbed to learn that the DHHS Office of Inspector General was investigating long-stay hospice patients as part of its fraud and abuse initiatives.¹¹ The first audit report on a mainland hospice, released in September, 1996 claimed that the hospice erred in admitting 176 patients out of 364 patients who were either enrolled for more than 210 days or had been discharged alive over a 27-month period. The 364 cases represented about 2 percent of all patients enrolled in the hospice during that period, and the hospice overall had a median length of stay of 47 days in FY 1995 (letter from B.R. Howe to the Regional Inspector General for Audit Services, DHHS Region IV, April 29, 1996). Although hospices should not be immune from investigations of possible fraud or abuse, the committee urges regulators to exercise extreme caution in interpreting hospice stays that exceed six months as evidence of anything other than the consequence of prognostic uncertainty. To do otherwise would inappropriately penalize hospices and would threaten the trust that dying patients need to have in those who care for them. It might also discourage more timely admission to hospice of patients now referred only a few days before death, after important opportunities for physical, psychological, spiritual, and practical support have already been missed.

In addition to eligibility restrictions, the Medicare hospice payment mechanisms are also designed to limit costs. Four levels of per diem payment rates apply as does an overall cap on payments that was set for FY 1997 at $13,974 multiplied by the number of hospice enrollees. Medicare sets a cap on aggregate hospital payments at 5 percent of total payments, and the per diem rate for inpatient hospice care is only $418.93. The use of volunteers, which reflects hospice philosophy but also limits Medicare's liability for some care, is required.

At the same time that they encourage efficiency, the aggregate caps on payments for in-patient care and total payments and the per diem rates could discourage appropriate care for some patients. For example, the current per diem of $94.17 for routine home care may discourage the use of certain costly pain medications, even when less expensive drugs fail; late-night nursing visits to deal with medical or emotional crises; the appropriate application of high-technology equipment (e.g., infusion pumps); and extensive counseling services for particularly distressed patients and families (Joranson, 1994; Brown, 1996). Even more significant, the financial constraints mean that hospice care may not be suitable or feasible for patients without family or others able to provide extensive assistance with medications, hygiene, nutrition, and other care.

The last several years have seen efforts to apply capitated payment schemes to an increasingly broad range of services and providers. Hospice has not been immune, although few if any capitated contracts are in place so far, and the discussions of capitation do not appear to envision their application to Medicare beneficiaries (Hospice News Service, 1996). Capitation contracts would involve paying a hospice a small amount each month for all members (well or ill) of an HMO or other health plan and then holding the hospice financially responsible for services to terminally ill health plan members. Capitation of hospice care has been characterized as a "high-risk" mechanism because of (1) lack of data on the incidence and variability of terminal illness in non-Medicare health plan populations, (2) variability across communities, (3) lack of effective tools to detect and deflect inappropriately early referrals, and (4) lack of other information necessary to set rates and manage financial risk. The committee does not encourage this kind of capitation of hospice services.

The committee heard some concern that physicians felt financially threatened by hospice. This fear is misplaced, to the extent that it rests on the false perception that Medicare requires patients electing hospice coverage to switch to a hospice physician and leave his or her personal physician. Some individual hospices, however, apparently do encourage such a switch which may or may not be in the patient's best interest. Regardless of who the attending physician is for a hospice patient, the direct patient care services (as opposed to general oversight or consulting services) are generally covered separately by Medicare Part B rather than folded into the hospice per diem (NHO, 1996a).

Medicare beneficiaries enrolled in HMOs may elect the hospice benefit. They are not limited to using a hospice that the HMO may have contracted with for its non-Medicare enrollees, although it is not clear how well beneficiaries are informed of their options. HMOs enrolling Medicare beneficiaries are not required to provide hospice benefits themselves (although they may), and they cannot deny access to a HCFA-certified hospice for patients who wish it. When Medicare beneficiaries enrolled in HMOs elect the hospice benefit, the usual capitation payment to the HMO is reduced to a minimal level for certain covered services not required of or provided by the hospice. For HMOs paid on a reasonable cost basis, costs are covered for services not related to the terminal condition and for attending physician services provided to a hospice patient by an HMO physician.

A few HMOs operate their own Medicare-certified hospices (Della Penna, 1996), but most do not. Some HMOs and capitated physician groups have organized internal hospice-type programs to cover younger enrollees because they believe that it is good for patients and to the plan's financial advantage to do so (Baines, Gendron et al., 1996; Emma, 1996).

Although comprehensive and detailed data are scarce, hospice coverage for those not eligible for Medicare or Medicaid appears to vary enormously in terms of what services and providers are covered and with what limitations (e.g., exclusions for preexisting conditions). Hospice coverage is said to be offered by 80 percent of large- and medium-size employers (Snow, 1997).




Home Care Services

Because the Medicare hospice benefit is limited to those with a prognosis of six months or less, it is not applicable to many patients with serious illnesses but an uncertain prognosis. For these individuals, coverage for home health services can help them secure important supportive services such as medical social services and home health aide services. A major limitation of the home health benefit is that beneficiaries must be home-bound and need part-time or intermittent skilled nursing care or physical or speech therapy. Some dying patients would be able to benefit significantly from home palliative care before they become completely homebound.

As mentioned in Chapter 4, the use of Medicare-covered home care has grown rapidly in recent years. It is one of the fastest growing segments of total program spending. Between 1988 and 1994, program payments grew more than 500 percent—from $1.9 billion to $12.7 billion—and the number of certified home health agencies has grown from approximately 5,700 to 7,800 (HCFA, 1996b). The reasons for the growth in expenditures include changes in Medicare home care policies and their judicial interpretation, an aging population, a growing supply of providers, earlier discharge of patients from hospitals, and the increased feasibility of providing advanced technologies in the home.

The increase in expenditures has prompted a comprehensive assessment of the home health benefit that is examining, among other issues, quality of care, payment methods, fiscal integrity, and effective program management (Clauser, 1994; Vladeck and Miller, 1994). One part of the initiative is demonstration projects to test a prospective per-episode payment system to encourage efficiency and cost control. Preliminary results suggest difficulties in establishing episode lengths given the skewing of episodes.

In general, Medicare coverage for home care services has provoked both support and skepticism as well as some evidence of fraud and abuse (Kane et al., 1991; IOM, 1994b; GAO, 1995a). The skepticism has largely to do with claims that home care can reduce total health care costs by substituting for hospital or nursing home care (Benjamin, 1993). One of the largest studies of home health care concluded that "home care was used mostly by those not at risk for entering a nursing home, costs increased,. . . and benefits . . . are few and fleeting" (Weissert, 1991, p. 68). Another study suggested a "complementarity" in home health and nursing home use. Home health services "tended to be used by persons with serious health problems whose disability appeared to be more a consequence of illness and whose chronicity may have been tied to the duration of the health problem . . . [whereas skilled nursing facility] use seemed to be concentrated among those with serious functional disability of potentially longer standing" (Manton et al., 1994).

Some have questioned whether any savings that do exist from averted hospital or nursing care are offset by a shift to paid home care away from unpaid care by family members (Pepper Commission, 1990). Analyses indicate that Medicare home care users are more likely than nonusers to be disabled, living alone, poor, and receiving Medicaid (Clauser, 1994). The Medicare benefit does not require a deductible, copayment, or previous hospitalization, although this may change given the concern about recent spending increases. In contrast to hospices, about half of home health agencies are for-profit, and these agencies tend to have more aide visits, fewer therapy services, longer episodes of care, and more visits per episode than other categories of agencies (ProPAC, 1996).




Medications

One of the long recognized limitations of the Medicare benefit is that it does not cover most outpatient prescription medications. The major exception is if a patient is enrolled in hospice and is prescribed drugs for symptom management. Some (usually expensive) supplemental policies cover outpatient prescription drugs. Medicaid programs generally cover outpatient medications, but a number limit the availability of medications by placing limits on the number of refills, the number of prescriptions covered in a month, or the quantity of medication (Soumerai et al., 1987; Joranson, 1994). These restrictions can interfere with effective, continuous management of pain and other problems.

Medicare's disparity in inpatient and outpatient coverage for medications may encourage the use of more expensive interventions. For example, pain relief through oral medications or transdermal patches is not covered for outpatients, but use of an infusion pump to provide pain relief at home can be covered under some circumstances (Joranson, 1994). The latter is not only more expensive but is somewhat more likely to produce certain kinds of complications. A patient without the resources to cover oral morphine and without assistance from family members might even be admitted to a hospital for intravenous or subcutaneous medications, which are covered in that setting. Ferrell (1993) points out that having morphine so administered might cost $4,000 compared with $100 for an oral morphine solution.

One study of elderly patients treated for selected chronic illnesses suggested that "limiting reimbursement for effective drugs puts frail, low-income, elderly patients at increased risk of institutionalization in nursing homes and may increase Medicaid costs" (Soumerai et al., 1991, p. 1,072). No studies appear to have specifically examined the effects of limited outpatient prescription drug policies on patients at the end of life, but it seems reasonable to expect that these policies also put this group in jeopardy as well.




WHAT ABOUT PROPOSALS TO REDUCE THE
COST OF END-OF-LIFE CARE?

The discussion to this point has focused primarily on patient concerns as they are affected by efforts to limit Medicare costs. The concern with Medicare's precarious financial position is likely to make the cost of care for the dying one issue—among many—for those attempting to protect the solvency of the program.

As noted in the introduction to this chapter, various strategies for reducing the cost of end-of-life care have been put forth. Five are briefly considered below: broader use of hospice care; expanded use of advance directives; consumer choice strategies; futility guidelines; and explicit rationing of care. A suggestion not discussed below is that a closer physician-patient relationship would reduce costs associated with fragmented care, inattention to patient circumstances and concerns, litigation arising from mistrust, and similar problems (Scitovsky, 1996). The committee did not make formal recommendations on any of these options, but it does encourage continued investigation and testing of innovative strategies to discourage the use of nonbeneficial or marginally beneficial care without jeopardizing people's health and well-being.




Hospice Care

In addition to its benefits for patients and families, hospice has been proposed as a way to reduce costs, particularly for Medicare. The basic argument is that hospice saves money by decreasing the amount of expensive hospital care provided to terminally ill patients. After reviewing the literature, the committee concluded that the impact of hospice on overall health care costs remains uncertain.

Proponents of the cost savings position cite two studies—one published in 1985, which showed that expenditures for hospice patients in the last month of life are almost two-thirds lower than patients not in hospice (Mor and Kidder, 1985), and the second published in 1995, which reported that hospice saved Medicare $1.52 in Part A and B expenditures for every dollar spent (NHO, 1995). The more skeptical (see, e.g., Kidder, 1992; Scitovsky, 1994; Emanuel, 1996) point to the one randomized trial of hospice (Kane et al., 1984) showing no cost savings. Skeptics also criticize the nonrandomized studies for probable selection bias (e.g., patients who choose hospice may be more willing than others to forego aggressive life-prolonging interventions), measurement problems (e.g., reliance on charges rather than costs, exclusion of costs to families), and time period evaluated (e.g., in some studies, only one month prior to death).

One recent review of the literature suggests that for those patients for whom hospice care is applicable, cost savings of 25 percent to 40 percent may be generated in the last month of life (Emanuel, 1996). Savings in the last six months of life may drop to 10 percent to 17 percent and 0 percent to 10 percent in the last year. Because not all terminally ill patients are candidates for hospice care for the reasons described in earlier chapters, such savings cannot be readily extrapolated to all patients.

This committee believes that hospice programs should be promoted primarily on noneconomic grounds related to their emphasis on medical and nonmedical care goals, attention to families and others close to their patient, provision for nonmedical supportive services (e.g., respite care, bereavement counseling), and similar features. Nonetheless, given the accelerating financial pressures on Medicare, it is reasonable to consider continued research to identify the circumstances under which hospice care can reduce the cost of end-of-life care. One area for such research involves the application of the new guidelines for qualifying patients with selected noncancer diagnoses, which were discussed in Chapter 4.




Advance Directives

Advance directives specifically and advance care planning more generally are not health care financing mechanisms per se, and the arguments in their favor have primarily to do with patient and family control over end-of-life decisionmaking and with thoughtful consideration of peoples' goals at the end of life (see Chapters 3 and 7). Nonetheless, the potential of advance directives to reduce costs has also attracted attention. The idea is that written directives would eliminate the uncertainty or the dedication to rescue that drives clinicians treating comatose or otherwise mentally incompetent patients to provide ineffectual life-prolonging interventions. In addition, advance care planning might help some people to prepare for decisions that they may face while conscious and mentally competent (e.g., when to elect hospice care, when to accept or refuse another round of chemotherapy) and to be less susceptible to pressure from clinicians or family to agree to unwanted interventions when a crisis arises.

Some research suggests that patients who sign advance directives generate lower costs for their final hospitalizations than patients without such directives (Weeks et al., 1994). Other research, however, indicates the patient and family preferences are often not effectively communicated to or understood by clinicians (SUPPORT Principal Investigators, 1995; see also Chapter 3).

A sense of the outer limits of cost savings comes from an estimate that if every patient who died in 1988 "executed an advance directive, chose hospice care, and refused aggressive, in-hospital interventions at the end of life . . . the total savings in health care expenditures would have been $18.1 billion . . . or about 3.3 percent of all health care spending" (Emanuel and Emanuel, 1994, p. 542). The savings to the Medicare program would have amounted to about 6 percent of program spending in 1988. Clearly, it is unreasonable to contemplate universal acceptance and implementation of these measures. It is reasonable, however, to expect that some unwanted care and avoidable costs are incurred when care is provided contrary to the oral or written directive of a patient or family. Even if savings were only 10 percent of the high estimate cited above, the amounts, while small in the context of overall spending and spending increases, would not be meaningless.

The savings estimates for advance directives are, in general, vulnerable to the same criticisms that have been raised about research on hospice cost savings (e.g., that those willing to sign directives are not representative of the larger population) (Emanuel, 1996). In particular, skepticism about projections of significant cost savings rests on doubts about (1) the willingness of people to sign such directives, (2) their readiness to forego life-extending care when death is a real, not an abstract, prospect, and (3) the ability of health care systems to implement directives (Schneiderman et al., 1992; Teno et al., 1993; GAO, 1995b; SUPPORT Principal Investigators, 1995; Levinsky, 1996; Scitovsky, 1996; Lynn, Harrell et al., 1997; see also Chapters 3 and 7).

Although there is no evidence to suggest that significant cost savings would soon result, some steps could be taken to increase the potential for advance care planning to avoid some services—and expenditures—that most would agree are unwanted, ineffective, and even damaging to patient well-being. For patients who have completed advance directives (or taken similar steps), better procedures are needed to assure that information on patient wishes is readily available and considered at critical decision points, for example, the transfer of a seriously debilitated, very elderly patient from a nursing home to a hospital. The Oregon initiative described in Chapter 3 provides examples of such procedures.

Some have argued that managed care organizations are particularly well suited to encourage use of advance directives (Fade and Kaplan, 1995). A number of managed care organizations have undertaken systematic efforts to encourage greater use of advance directives (Baines, Barnhart et al., 1996; Christensen, 1996; Hammes and Rooney, 1996). Some caution is, however, warranted about promoting advance directives as a cost-containment measure in this context, and managed care organizations have particular reason to be hesitant about appearing too intent on encouraging people to sign directives limiting certain forms of care. For example, Sulmasy (1995), in discussing concerns about conflict of interest, suggested that advance directives (particularly if assisted suicide were to become legal) could be a "chillingly effective way to control the cost of managed care" (p. 245). He also notes that some have suggested lower insurance premiums for those who have advance directives (see Washington Post, May 2, 1993, C3).

Overall, continued efforts to encourage various forms of advance care planning and goal setting make sense at the margin, notwithstanding the social and cultural obstacles. One caveat: should people suspect that they were being pressured to sign advance directives and that such directives might be used inappropriately to limit care, then the result might be both higher costs and more use of advanced technologies that patients or families would forego in an environment of greater trust. Also, patients might try to "game" or manipulate a system that conditioned coverage on signing advance directives by later revoking the directives (assuming that such directives would continue to be revocable). As Callahan has pointed out, "advance directives were designed to assure the private good, not the public . . . to give patients a safeguard against being overpowered by overzealous physicians or institutions" (Callahan, 1996).




Consumer Choice Strategies

A market-oriented strategy to reduce health care spending proposes that those who wish aggressive but marginally beneficial treatments (or, in some versions, aggressive beneficial care) should be allowed that option but only if they absorb some of the additional costs that such a choice might entail (Eddy, 1991a; Havighurst, 1992, 1995). People could contract for a different "intensity" or "quality" or "extensiveness" of coverage as provided in different health plan options. The differences might be set forth in alternative clinical protocols for specific health care problems. Thus, one person might choose a health benefits package that covered only palliative services for persons with prostate cancer of a certain stage and prognosis, whereas another person might choose (and pay more for) a package that also would pay for curative or life-extending care under the same circumstances.

It may be unrealistic to expect people—particularly those not immediately concerned about life-threatening illness—to analyze and understand how health plans differ in their protocols for caring for myriad different illnesses and combinations of medical problems (IOM, 1993b). Nonetheless, with the support of clinicians and others, people might be helped to understand a choice—similar to that involved in electing hospice care—between an option that allowed for aggressive, curative or life-extending care for conditions with a poor prognosis and an option that provided for palliative services appropriate to the person's disease stage and prognosis.




Medicare Managed Care

Managed care strategies have the potential to reduce costs for end-of-life care in much the same way that they appear to reduce costs for healthier populations, in particular, by reducing the use of inpatient hospital services. For example, patients admitted to the hospital but determined to be likely to die within a few days may be discharged to die elsewhere if they do not meet the criteria for acute inpatient care. The overall impact of such strategies on the cost and quality of care would vary depending on the regard given to the wishes of patients approaching death and the alternative care available (e.g., hospice, nursing home, home health care). Alternative care will offset some of the savings from avoided hospitalization unless it also prevents rehospitalization or is provided on an unpaid basis.

As discussed in Chapter 4, the proportion of Medicare beneficiaries enrolled in HMOs is small, although much higher in a few states and growing nationwide. Although research examining patients who die is very limited, the results are consistent with other research that raises concerns about hospital discharge and home care policies and practices as they affect the frail elderly, the chronically ill, and other vulnerable groups.

The point is not that managed care organizations cannot provide good care for terminally ill patients but that financial incentives may tempt some to avoid enrolling the ill people or to behave in ways that compromise the quality of care for those with life-threatening illnesses. Speaking generally of the chronically ill, Jones has argued that market dynamics argue "for investing less, at the margin, in improvements or plan features that increase value to the chronically ill" (Jones, 1996b, p. 4). He identifies strategies that plans might use to discourage enrollment by the chronically ill. These range from using advertising images that suggest that the plan is for healthy not ill people to "keeping the numbers and availability of specific types of health professionals, clinic facilities, and other resources that attract the chronically ill to a minimum, thus ensuring long waiting times" (Jones, 1996b, p. 4). Governments, accrediting bodies, and other standard-setting organizations can attempt to devise standards that minimize these strategies. Unless, however, the financial incentives for capitated plans to avoid the chronically ill are changed, avoidance strategies can be expected to continue.

What can be done to change incentives? One approach is to devise a scheme for paying health plans that adjusts capitation payments to compensate them properly for the risk they assume in enrolling sicker people but that also avoids overpaying for healthy enrollees. Despite some success in improving methods for risk-adjusting payments, no method yet appears equal to the challenge (Newhouse et al., 1989; IOM, 1993b; Newhouse, 1994; Ellis et al., 1996; Gruenberg et al., 1996; Jones, 1996b). As far as the committee is aware, no special attention has been paid to the question of adjustments for those who die, and methods usually either use full-year participants only or annualize adjustments. Any explicit adjustment in rates to account for deaths could raise policy and political questions if it appeared to pay plans more if their members died. Current proposals to cut the Medicare capitation payment to HMOs across-the-board rather than to risk-adjust payments would, if anything, intensify the incentives to avoid people with serious health problems.

Another approach relies on performance measurement and reporting schemes that would identify, publicize, and discourage poor performance and that also would identify apparently good performance that was mainly a function of having a relatively healthy group of enrollees. Issues in measuring the quality of care have been discussed in Chapter 5. Given the difficulties that existing schemes are having in devising feasible and informative measures for a modest array of relatively common health problems, collecting sufficiently sensitive process and outcomes measures for the most vulnerable patients appears to be a distant although desirable prospect. Moreover, the strategy would not be successful unless the negative effects of being identified as a poor performer were strong enough to overcome financial disincentives to being identified as a good plan for chronically ill people (and thereby disproportionately attracting them).

A third approach to encourage health plan interest in sicker people is to devise special payment strategies for people with serious chronic illnesses or for patients with life-threatening illnesses who do not fit the hospice model (Jones, 1996b; Lynn, 1996b). Jones has discussed how Medicare, Medicaid, and very large employers might develop competitive care arrangements for people with specific chronic illnesses that would balance cost, quality, and access objectives in relation to the special needs of these patients (Jones, 1996b). He argues that Medicare could take the lead in experimenting with such options. For terminally or seriously ill Medicare patients who do not fit the current hospice model, Lynn has suggested the creation of coverage that would favor lower-technology supportive care provided by multidisciplinary teams (Lynn, 1996b). The thrust of this proposal is, however, not to save money but to make supportive care more available without increasing costs. Nonetheless, a test of such an option (or two or three variants that might be expected to have different cost effects) might indicate the extent to which such a strategy might affect overall costs and the extent to which it was accepted by patients, families, and practitioners. The generalizability of findings from such a test would need careful attention.

Of some interest in this regard, Medicare is considering tests of special capitation payment arrangements for patients with end-stage renal disease (Farley et al., 1996). Researchers assessing strategies for establishing such payments conclude that more work remains to be done on several design and technical issues before such arrangements are feasible. They also note that enrollee protection policies (e.g., quality assurance programs, grievance and appeal procedures) are particularly important for chronically ill Medicare beneficiaries who "may be especially vulnerable to reductions in access or quality resulting from cost cutting actions by plans" (Farley et al., 1996, p. 141).




Futility Guidelines

Futility has been defined in a number of ways (Veatch and Spicer, 1992; Halevy and Brody, 1996). Miles, for example, has suggested that it be viewed as "a medical conclusion that a therapy is of no value to a patient and should not be prescribed" (Miles, 1992, p. 310). Such conclusions might best be applied through professionally developed guidelines that focused on therapies that "a vast majority of people would not accept for themselves" (Miles, 1992, p. 313). Another view, which also argues for professional guidelines and for qualitative definitions of futility, suggests that futility might be viewed in probabilistic terms as a point on a continuum where the probability of benefit is very low, perhaps 1 in 100 (Schneiderman, 1994). Others have suggested community-based guidelines, such as those developed in Denver and Sacramento (Murphy and Barbour, 1994; see also Appendix C).

Expectations that "futility" guidelines would produce large cost savings should be restrained. Based on their analyses of data on a large group of seriously ill hospitalized patients, Teno and colleagues found that guidelines limiting treatment for patients with only a 1 percent or less chance of surviving 2 months would have saved about 13 percent of hospital charges for these patients (Teno, Murphy et al., 1994). They emphasized, however, that most of these savings came from 12 patients, 6 of whom were under age 51 and probably would not be very likely targets for aggressive care-limiting guidelines. They also found that 90 percent of all patients with this poor prognosis died within a week and that about a third died after life-sustaining treatments were withdrawn. The authors interpreted this as indicating that "some limits on aggressive treatment are already in place" (p. 1206). In a smaller study of pediatric intensive care unit (PICU) patients and a definition of futility that was deliberately broad for analytical purposes, Sachdeva et al. (1996) found a very small number of futile care days in the PICU and thus a small potential for savings based on limiting care.




Rationing

Rationing is another term with varied meanings (see, e.g., Aaron and Schwartz, 1984; Mechanic, 1985; Brook and Lohr, 1986; Reagan, 1988; Grumet, 1989; Morreim, 1989; Englehardt, 1991; Hadorn and Brook, 1991; Wiener, 1991; Hall, 1994). One traditional view sees rationing as the fair allocation of a specific limited resource (e.g., food during famine, organs for transplant). A more expansive definition includes limiting consumption of a potentially beneficial good or service by various explicit or implicit means (e.g., ability to pay, inconvenience, queuing, limits on technological innovation). Both these concepts of rationing exclude strategies for eliminating clearly inappropriate or harmful care and, instead, focus on care that may have some but not "enough" benefit to warrant coverage.

Determinations about what constitutes enough benefit or what constitutes essential—not just beneficial—care are partly subjective and partly scientific questions (Brook and Lohr, 1986; Eddy, 1996). One problem is that the evidence base for such determinations is very limited. Although it will continue to expand, evidence will still be unlikely to provide a firm foundation for treatment decisions about many specific combinations of patient circumstances (e.g., an 80-year-old man with mild dementia and congestive heart failure) (see, e.g., IOM, 1992). This is not an argument against evidence-based decisionmaking, only a caution that expectations should probably be modest.

Rationing of potentially beneficial services already exists—at least implicitly—in forms such as patient cost sharing, administrative mechanisms that limit access to services or providers, and various provider payment mechanisms. Although implicit rationing mechanisms have prompted varying amounts of criticism, more controversial are various options or proposals for the explicit rationing of potentially beneficial services (see, e.g., Brook and Lohr, 1986; Veatch, 1986; Morreim, 1989; Aaron and Schwartz, 1990; Daniels, 1994; Hall, 1994; Callahan, 1995). These proposals are more complex than such traditional devices as the exclusion from insurance plans of certain classes of services (e.g., long-term care) or conditions (e.g., preexisting medical problems). Such exclusions, which have been subject to continuing legislative attack, have been justified less on grounds of resource allocation than on grounds that the dynamics of a voluntary insurance market demanded them (IOM, 1993b). The discussion below focuses on proposals to ration on the basis of age or community consensus.




Age-Based Rationing

Age is one possible basis for rationing health services (see, generally, Callahan, 1987; Smeeding, 1987; Daniels, 1988; Binstock and Post, 1991; Winslow and Walters, 1992; Jecker and Schneiderman, 1994; Joint International Research Group, 1994; Callahan, 1996). For example, if you were over a specified age, open-heart surgery and similar life-prolonging interventions would not be offered or covered. (In the United States, it is probable that such interventions would still be provided for people financially able and willing to pay for them on their own.) A softer but more complex variant would be to set different limits for different age groups, perhaps covering life-prolonging treatments for those under age 85 who were conscious and had good cognitive functioning.

A less explicit approach to age-based rationing—along the lines suggested in Box 6.1—is also possible. Such an approach might be most extensively employed with the very old and also with others having a very grave prognosis.

Savings from age-based rationing have been questioned, particularly if restrictions were largely limited to the very old. Some researchers have pointed to evidence suggesting that the older old may already get less intensive care than younger people with equivalent health or functional status (Scitovsky, 1988; Farrow et al., 1996; Hamel et al., 1996). Scitovsky has concluded that "high-cost medical services may already be allocated to older people in their last year of life in a more rational manner than is generally assumed, with their age and functional status taken into account" (Scitovsky, 1988, p. 656). That is, the care actually provided is not "highly technical" or "center[ed] around often inappropriate cure-oriented services provided in hospitals" (Temkin-Greener, 1992, p. 699).

Without much more detailed data and analysis, it is not possible to determine the basis of apparent age-conditioned treatment intensity. It could be associated with overtreatment of younger persons or undertreatment of older persons or both.

The committee believes that public debate over age-based rationing has not really begun and may not be joined directly until the baby boom generation gets nearer the age—about the mid-fifties—when medical problems and costs begin to rise significantly. Such a debate could involve a constructive or destructive consideration of issues of justice, biological uncertainty, and potential benefits and burdens of treatments at different points in the life span.




Rationing Based on Community Consensus

A variant on rationing strategies combines evidence and clinical judgment with a public process for debating and weighing priorities. The best known example of such an approach, the so-called Oregon plan, is worth special attention because it has actually moved from the realm of theory to action and because it explicitly recognized the value of palliative care. The plan was envisioned as an "ongoing process to achieve consensus on our policy goals and the principles that will guide our health policy and as a framework in which resource allocation and reallocation can take place" (Gibson, 1993, p. 57; see also Fox and Leichter, 1991). Those involved in crafting the plan anticipated a near universal access to health insurance within the state, which would mute concerns about economic inequities.

A central element of the plan was a ranked list of health services on the basis of clinical effectiveness and social value. Services would be covered in rank order as far as revenues and judgments about budget priorities allowed. The ranking effort involved both experts and the community (e.g., through surveys and town meetings), clinical and economic data, and evidence and values (e.g., a subjective review of the ranked list by officials).

The priority-setting scheme explicitly dealt with several kinds of services involving life-threatening conditions and terminal illness. It considered essential coverage to include acute, potentially fatal conditions for which treatment could be expected to prevent death and offer full recovery (e.g., appendectomy for acute appendicitis) or to prevent death without a clear prospect of full recovery (e.g., treatment for strokes and serious burns). Essential services also included care for chronic incurable conditions for which treatment could be expected to improve life span and quality of life (e.g, treatment for diabetes and comfort care such as pain management and hospice care) (Cotton, 1992). The next highest priority of the Oregon plan included important services involving mostly nonfatal conditions. The lowest-ranking group included "fatal or nonfatal conditions of which treatment causes minimal or no improvement in quality of life, such as severe brain injury, aggressive treatment for extremely premature infants, and end stages of cancer" (Cotton, 1992).

Criticisms of the Oregon plan have involved a mix of technical, political, financial, and ethical arguments (Brown, 1991; Daniels, 1991; Eddy, 1991b; Grannemann, 1991; Cotton, 1992; Hadorn, 1992; Menzel, 1992; Azevedo, 1993; Hansson et al., 1994; Kaplan, 1994; Firshein, 1996). One criticism of this approach has been that it does not allow for differences among patients in the same diagnosis/treatment category. That is, even with over 600 categories, some patients fitting a high ranking diagnosis/treatment category may have characteristics that make them unlikely to benefit from a usually beneficial treatment whereas some patients in a low ranking category might have characteristics that give them a good chance of benefiting from a usually less effective treatment. Thus, even if evidence exists about differences in benefit based on differences in patient characteristics, the ranking scheme would not accommodate it. In addition, the analytical strategy has been criticized for potentially discriminating against those with disabilities, and concern about such discrimination was a major factor in the federal government's initial cool reaction to Oregon's request for a Medicaid waiver for the project (Menzel, 1992; Campbell, 1993).

Instead of operating on a statewide or community basis, a rationing scheme could, in principle, operate on a health plan basis with representatives of health plan members setting priorities (Medica Foundation, 1994, 1996). Such a strategy could involve hundreds of individual priority-setting exercises for different health plans. In addition to raising some of the same concerns as other rationing schemes, such an approach appears vulnerable to the same charges of overcomplexity and limits on consumer understanding that were raised for the consumer choices strategies described above.




CONCLUSION

Care at the end of life appears to be, in large measure, financed by public programs, especially Medicare. Although explicit comparisons are not available, trend data on Medicare expenditures suggest that care at the end of life has not become disproportionately more expensive and technologically intensive than other care in recent decades, although it is clear that a considerable part of Medicare spending is accounted for by those who die. Data also indicate that Medicare spending at the end of life drops with increasing age of decedents, a drop that is partially offset by increased Medicaid spending for less technologically intensive nursing home care. Except for hospice, few studies have examined the impact of Medicare or other financing practices on care at the end of life. The committee found reason to be concerned about the possible effect of several features of both the traditional Medicare program and its managed care alternatives.

The committee believes it is important to know more about the patterns and the impact on dying patients and their families of early hospital discharge, including the impact on total costs, extent of readmissions, smoothness of transition to home care (with or without hospice benefits), management of symptoms, quality of life, and family burden. The information gained from the test of the Medicare palliative care code should be helpful in evaluating the nature of hospital palliative care and its cost compared with alternatives.

Several areas of physician payment also warrant investigation. They include the fit between different patient evaluation and management codes and the time and effort required to provide palliative services in different settings. The adequacy of payment levels for nursing home and home health visits is of particular concern to the committee. The committee is also concerned that Medicare's fee-for-service program has financial incentives that do not encourage continuity of care, interdisciplinary care, or supportive care for a broad range of dying patients.

Medicare's hospice benefit might seem to overcome these difficulties, and indeed, its per diem payment, coordination with Part B physician services and with capitated managed care, and required interdisciplinary care team are illuminating and auspicious components. Hospice programs, however, cannot regularly take on care of persons without homes, family caregivers, or predictable fatal illnesses. Likewise, persons who still should receive costly drugs or supports, such as respirators or intensive care, often cannot be enrolled in hospice, and such treatments are not likely to be made more available within the hospice program.

Proposals to test new ways of financing care for those with serious chronic illness warrant serious attention. Not only is this desirable as a way of responding to the problems faced by people with less predictable prognoses but it also responds to concerns about the incentives of Medicare's current method of capitating HMOs. Seriously chronically ill persons cost much more than the Medicare capitated payment, which discourages plans from attracting such patients by developing a reputation for excellence in their care. The financing mechanism acts counter to quality. Without the testing of alternative financing strategies by the HCFA, state Medicaid programs, and health plans, it is not clear how problems with both managed care and FFS financing can be resolved.




ADDENDUM

Medicare coverage and payment provisions are complicated by the division of the program into two parts involving different services, payment methods, and beneficiary cost sharing provisions (HCFA, 1995). Medicare Part A covers mostly inpatient hospital care with limited coverage for nursing home, home health agency, and hospice care. Part B covers physician services, limited services by other professionals (e.g., chiropractors and dentists), hospital outpatient services, freestanding ambulatory surgery, physical therapy, durable medical equipment, and some other services or supplies. Beneficiaries share in the cost of their care, with the specifics varying by type of service and other conditions. For example, for 1997, the Part A hospital deductible was $760. Medicare does not cover outpatient prescription or nonprescription drugs, long-term (nonacute) care, and most routine preventive services; it covers a very limited amount of behavioral health care.

About 13 percent of Medicare beneficiaries—the dually eligible—are also covered by Medicaid. Payments of intermediate care and nursing facility care account for about one-third of all Medicaid spending. Medicaid covers required patient cost sharing for Medicare covered services for those who meet means testing criteria.




Major Features of Medicare Hospice Benefit

Eligibility

General conditions: A person must be entitled to Medicare Part A benefits and certified to have a terminal illness (defined by statute as having a life expectancy of six months or less if the illness runs its normal course).

Election and waiver statement: A person must sign a statement electing the hospice benefit and waiving the right to standard Medicare benefits for curative treatments for (or related to) the terminal condition.

Election periods: A person starts with an election period for hospice care of 90 days. Subsequent elections may be for 90, then 30 days, and finally for an indefinite period.

Revoking the benefit: If a person revokes his or her election, benefits for the election period in question are forfeited. Benefits may be reelected for the next election category, except that a person who is in the indefinite election category forfeits any further coverage for hospice care.




Coverage

General: Services are covered if reasonable and necessary for palliation or management of the terminal illness and related medical problems.

Inpatient care: Short-term inpatient care is covered if it is necessary to manage symptoms or provide respite for home caregivers.

Home care: A variety of medical and nonmedical services are available to patients at home (and sometimes in nursing homes or other institutional settings—including nursing visits, physical therapy, physician visits, medical social services, services from home health aides, counseling services (including spiritual and bereavement counseling for family members), homemaker services, outpatient drugs, and medical appliances. Continuous home care is covered if necessary to keep a patient at home during a medical crisis.




Physician Payment

Employee attending physician: These physicians are paid under the hospice daily rate for technical (e.g., lab) and administrative services and under Medicare Part B for professional services.

Outside attending physician: These physicians are paid under Medicare Part B for professional and administrative services and as part of the hospice daily rate for technical services.

Consulting physicians: These physicians are paid under Medicare Part B for professional services and under the daily rate for technical services.




Payment to Hospice

General conditions: Payment is made on a prospective per diem basis except for physician services, which are paid for on a fee-for-service basis.

Levels: Different per diem rates are set for routine home care ($94.17 in FY 1997), continuous home care ($549.65 for 24-hour care or $22.90 per hour for a minimum of 8 hours), inpatient respite care ($97.41), and general inpatient care for palliative services ($418.93).

Cap on inpatient care: For a hospice, payments are reduced if total inpatient care days exceed 20 percent of the total number of hospice care days for all Medicare patients.

Overall cap: For a hospice, total payments are limited to an amount equal to the number of Medicare patients multiplied by a statutory cap amount ($13,974 in FY 1997). (Physician services covered by Medicare Part B are not included in the cap.)



---

NOTES

¹Coverage percentages range from zero for outpatient prescription drugs and 6 percent for nursing home care to 87 percent for inpatient hospital care. Disabled beneficiaries have more of their expenses covered by Medicaid and "other sources" (25 percent and 11 percent) than do aged beneficiaries (12 percent and 2 percent). For those who qualify for federal-state Medicaid, the program is an important source of payment for nursing home care. It pays about 47 percent of all nursing home bills, compared to Medicare's 8 percent (Levit et al., 1996). Patients and families cover most of the rest of the cost of nursing home care (42 percent for aged beneficiaries but 14 percent for disabled beneficiaries) (Gornick et al., 1996).

²Federal law provides that those who leave jobs under certain circumstances may be able to extend their coverage, at their own expense, for 18 or, less often, 36 months (IOM, 1993b). Few aggregate data are available on the cause of departures from employer health plans, but most are probably related to job changes, changes in marital status, retirement, switches to a spouse's health plan, and similar factors. Data on employer coverage or payments for those who die are virtually nonexistent. This is not surprising in part because death is a relatively rare event in the employer-covered population and in part because linking information on decedents to health plan coverage—or disenrollment—is expensive, although advances in computing power and the development of massive national health databases could make it easier in the future.

³This analysis indicated that 15.9 percent of Medicare spending for elderly beneficiaries in 1987 was accounted for by care in the last six months of life for those who died (Cohen, Carlson et al., 1995). The authors of this study point out that the equivalent (six months) figure for the 1988 (last year of life) data reported by Lubitz and Riley (1993) is about 21 percent (because about three-quarters of Medicare costs in the last year of life are incurred in the last six months). After noting that sampling error for their 1987 data could put the figure as high as 19.4 percent, the authors attribute their lower figure to differences in study years and estimation methods.

⁴Very high-cost care does not appear to be common for Medicare beneficiaries who die. About 5 percent of decedents generated Medicare payments of $40,000 or more in 1988, although less than 1 percent of surviving beneficiaries generated such high payments (Scitovsky, 1994).

⁵Also of interest is Scitovksy's analysis (1988) of sample data on decedents (mostly white and middle class) who received care from physicians at a large, multispecialty group practice. It indicated that total expenses (including expenses for care provided outside the group) were about 20 percent lower for those aged 80 or over compared to younger decedents. Expenses for hospital care were substantially lower, but payments for nursing home and home health were substantially higher for older decedents. For those aged 65 or over, total expenses did not vary by impairment status, but those who were totally impaired had substantially lower expenses for hospital care and higher expenses for nursing home care. High nursing home expenses were especially characteristic of those aged 80 and over who were totally impaired.

⁶Historically, the earliest sickness insurance arrangements were not designed to cover medical care expenses but to help offset wage losses and funeral expenses (see IOM, 1993b, Chapter 2, for a brief history of the evolution of health insurance). As medical care and hospital services became more effective and expensive, attempts began to be made to insure hospital care. Attempts to cover inpatient physician services followed.

⁷In most countries, the difficulties of private and voluntary health insurance have led to some kind of public insurance mandate. In the United States, several attempts since the early 1900s to establish national health insurance have been turned back. After World War II, a variety of factors encouraged a significant growth in the scope of private health insurance. Although perhaps half of those over age 65 had some form of insurance by the early 1960s, the difficulties faced by many older people in securing private health insurance eventually led to the adoption of Medicare.

⁸In addition to taking diagnosis into account, the assignment of a DRG to a particular patient may reflect other characteristics of a patient or a patient's care, such as age, operating room procedures, secondary diagnoses, or—rarely—costs so high that the case qualifies as an outlier. Hospital payments are also affected by their case mix and other factors.

⁹In addition, the American Academy of Hospice Physicians (Byock, 1996, p. 6) testified to the committee that "current diagnosis and treatment coding often leads to a charade in which tests are ordered and treatments begun to secure financial coverage for basic care to continue in the patient's preferred setting. This usually occurs in the hospital where the DRG for the patient's diagnosis has been exceeded, but it also applies to maintenance of skilled nursing visits at home for patients who are debilitated but not in acute crisis." As was true for many other concerns, the committee did not locate empirical data to support or contradict this claim.

¹⁰Cancer is the most frequent diagnosis, accounting for 60 to 80 percent of all hospice patients (Christakis and Escarce, 1996; Strahan, 1996). Another 10 to 13 percent have heart-related diagnoses. The National Hospice Organization estimates that 1 out of 3 patients with AIDS or cancer is cared for in hospice, a much higher proportion than for heart disease. Some Alzheimer's patients receive hospice care, but predicting the length of the terminal period of this disease is difficult (Volicer, 1996).

¹¹In 1995 the OIG began, under the name Operation Restore Trust, a series of investigations of fraud and abuse in home health agencies, nursing homes, durable medical equipment suppliers, and hospice organizations. The inclusion of hospice came after audits of two Puerto Rican hospices reported 77 percent of one hospice's patients were ineligible because they were not terminally ill (NHO Newsline, July 15, 1995).