Medicare: More Beneficiaries Use Hospice but for Fewer Days of Care
(Letter Report, 09/18/2000, GAO/HEHS-00-182).

Nearly 360,000 Medicare beneficiaries enrolled in a hospice program in
1998, more than twice the number that chose hospice care in 1992.
Although cancer patients account for more than half of Medicare hospice
patients, growth in use has been particularly strong among individuals
with other diagnoses, such as heart disease, lung disease, stroke, and
Alzheimer's disease. Many are doing so closer to the time of death; the
average period of hospice care fell from 74 days in 1992 to 59 days in
1998. Greater federal scrutiny of compliance with program eligibility
requirements may have contributed to a decline in beneficiaries' average
number of days of hospice care. Between 1992 and 1999, the number of
hospices participating in Medicare rose 82 percent, with large providers
and those in the for-profit sector accounting for a greater proportion
of the services delivered. At the same time, hospice industry officials
report cost pressures from declining patient enrollment and increased
use of more expensive forms of palliative care, such as radiation and
chemotherapy, for pain relief.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  HEHS-00-182
     TITLE:  Medicare: More Beneficiaries Use Hospice but for Fewer
	     Days of Care
      DATE:  09/18/2000
   SUBJECT:  Patient care services
	     Health care programs
	     Home health care services
	     Beneficiaries
	     Health care cost control
	     Medical services rates
IDENTIFIER:  Medicare Health Maintenance Organizations Program
	     Medicare Program

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO/HEHS-00-182

Appendix I: Methodology for Developing Information on Medicare Hospice Use
and Hospice Providers

32

Appendix II: Patterns and Trends in Hospice Supply and Use

35

Appendix III: Comments From the Health Care Financing
Administration

49

Table 1: Medicare Hospice Use Rates in Selected States and the
District of Columbia, 1998 10

Table 2: New Hospice Enrollees by Primary Diagnosis, 1992 and
1998 11

Table 3: Hospice Use Rates for Selected Diagnoses, 1992 and 1997 12

Table 4: Hospice Benefit Use by Weekly Intervals, 1992 and 1998 14

Table 5: Growth in Medicare Hospice Programs by Provider Characteristics,
1992 and 1999 27

Table 6: Growth in Number and Size of Hospices in States With Large Medicare
Populations, 1992 and 1998 28

Table 7: Number of Hospice Providers by Control, Affiliation, Size,
and Location, 1992-99 35

Table 8: Percentage of Hospice Providers by Control, Affiliation, Size,
and Location, 1992-99 36

Table 9: Average Number of Days Beneficiaries Used Hospice by
Provider Control, Affiliation, Size, and Location, 1992-98 37

Table 10: Use of Medicare Hospice Benefits by Cancer and Noncancer Patients,
1992-98 38

Table 11: Number of Hospice Users by Selected Common
Diagnoses, 1992-98 39

Table 12: Percentage of Hospice Users by Selected Common
Diagnoses, 1992-98 40

Table 13: Average Number of Days Beneficiaries Used Hospice by
Selected Common Diagnoses, 1992-98 41

Table 14: Hospice Use and Supply by State, 1992 and 1998 42

Table 15: Percentage Change in Hospice Use and Supply by State,
1992-98 44

Table 16: Change in Percentage of Providers, Users, and Days of
Care by Provider Type, 1992-98 46

Table 17: Variation in Hospice Use Among Medicare Decedents by
State, 1998 47

Figure 1: Number of Medicare Hospice Benefit Users, 1992-98 8

Figure 2: Decline in Average and Median Hospice Lengths of
Service, 1992-98 13

Figure 3: Average Days of Use for Cancer and Noncancer Patients,
1992-98 15

Figure 4: Decline in Average and Median Days of Hospice Care for
Cancer and Noncancer Patients, 1992-98 21

Figure 5: Growth in the Number of Medicare Hospices and New
Entrants, 1992-99 25

BBA Balanced Budget Act of 1997

CDC Centers for Disease Control and Prevention

CON certificate of need

HCFA Health Care Financing Administration

HHA home health agency

HHS Department of Health and Human Services

HMO health maintenance organization

IOM Institute of Medicine

MSA metropolitan statistical area

OIG Office of Inspector General

SNF skilled nursing facility

Health, Education,
and Human Services Division

B-284476

September 18, 2000

The Honorable Charles E. Grassley
Chairman, Special Committee on Aging
United States Senate

The Honorable John B. Breaux
Ranking Minority Member, Special Committee on Aging
United States Senate

The Honorable Ron Wyden
United States Senate

The Medicare hospice benefit was designed to serve terminally ill persons
who choose not to continue medical treatment to cure or extend life but
instead to receive services focused on comfort and pain relief as well as
emotional and spiritual support. Although hospice services are available in
inpatient settings, most services are provided in the patient's home or
residence. The hospice benefit is available to beneficiaries with a
prognosis that they have 6 months or less to live, but hospice users may
receive services for longer than 6 months, provided that they are
periodically recertified as eligible.

Some patient advocates, hospice providers, and others contend that certain
Medicare beneficiaries for whom hospice care is appropriate may have
difficulty in gaining access to hospice care or receiving services in a
timely manner. However, officials of the Health Care Financing
Administration (HCFA) and others assert that the benefit is basically
working as intended and meeting the needs of those who choose to use it. You
asked us to examine (1) the patterns and trends in hospice use by Medicare
beneficiaries, (2) factors that affect the use of the hospice benefit, and
(3) the availability of hospice providers to serve the needs of Medicare
beneficiaries.

To address these issues, we analyzed HCFA claims and enrollment and provider
data for a 9-year period ending in December 1999.1 (A detailed description
of our methodology is in app. I.) In addition, we interviewed officials of
hospice programs representing various affiliations, enrollment sizes, and
locations; officials responsible for processing Medicare hospice claims;
HCFA officials; patient advocacy groups; experts in palliative medicine; and
trade association representatives. We also reviewed relevant research as
well as Medicare program regulations, manuals, and guidance. We performed
this review from November 1999 through August 2000 in accordance with
generally accepted government auditing standards.

The number of Medicare beneficiaries choosing hospice services has increased
substantially. In 1998, nearly 360,000 Medicare beneficiaries enrolled in a
hospice program, more than twice the number who elected hospice in 1992. Of
Medicare beneficiaries who died in 1998, about one in five used the hospice
benefit, but use varies considerably across the states. Although cancer
patients account for more than half of Medicare hospice patients, growth in
use has been particularly strong among individuals with other common
diagnoses such as heart disease, lung disease, stroke, and Alzheimer's
disease. Although more beneficiaries are choosing hospice, many are doing so
closer to the time of death. The average period of hospice use declined from
74 days in 1992 to 59 days in 1998. Half of Medicare hospice users now
receive care for 19 or fewer days, and care for 1 week or less is common.

Many factors influence the use of the Medicare hospice benefit. Decisions
about whether and when to use hospice depend on physician preferences and
practices, patient choice and circumstances, and public and professional
awareness of the benefit. Along with these factors, increases in federal
scrutiny of compliance with program eligibility requirements may have
contributed to a decline in the average number of days of hospice care that
beneficiaries use.

The growth in the number of Medicare hospice providers in both urban and
rural areas and in almost every state suggests that hospice services are
more widely available to program beneficiaries than in the past. Between
1992 and 1999, the number of hospices participating in Medicare increased 82
percent, with large providers and those in the for-profit sector accounting
for a greater proportion of the services delivered. At the same time,
hospice industry officials report cost pressures from declining patient
enrollment periods and increased use of more expensive forms of palliative
care, such as radiation and chemotherapy for pain relief. Because reliable
data on provider costs are not available, however, the effect of these
reported cost pressures on the overall financial condition of hospice
providers is uncertain. As required by the Balanced Budget Act of 1997
(BBA), HCFA began collecting information in 1999 from hospice providers
about their costs to allow a reevaluation of the Medicare hospice payment
rate.

The Medicare hospice benefit, authorized in 1982 under part A of the
Medicare program, covers medical and palliative care services for terminally
ill beneficiaries. A Medicare-certified hospice provides physician services,
nursing care, physical and occupational therapy, home health aide services,
medical supplies and equipment, and short-term inpatient hospital care for
pain control and symptom management. In addition, the hospice benefit
provides coverage for several services not generally available under the
regular fee-for-service Medicare benefit. These include outpatient
prescription drugs for treating pain and other symptoms of the terminal
illness, homemaker services, short-term inpatient respite care, and
bereavement counseling for the patient's family.2

Patients may receive services from freestanding hospice providers or from a
hospice program based in a home health agency, hospital, or skilled nursing
facility. For each day a beneficiary is enrolled, the hospice provider is
paid an all-inclusive, prospectively determined rate, depending on the level
of hospice care provided (routine home care, continuous home care, inpatient
respite, or general inpatient care).3 Initial payment rates were based on
cost data reported by 26 hospice programs that participated in Medicare's
hospice demonstration project from 1980 to 1982.4 Since 1993, these rates
have been updated by an annual statutory adjustment factor

tied to inflation in the hospital market basket (a measure of the cost of
goods and services purchased by hospitals nationwide).5

Eligibility for hospice services requires that the beneficiary's physician
and the hospice medical director (or other physician affiliated with the
hospice) certify that the individual's prognosis is for a life expectancy of
6 months or less, if the terminal illness runs its normal course.
Beneficiaries who elect hospice must waive all other Medicare coverage of
care related to their terminal illness, although they retain coverage for
services unrelated to their terminal illness. A beneficiary can cancel his
or her election of hospice benefits at any time and return to regular
Medicare, and beneficiaries are free to reselect hospice coverage at a later
date. While there are currently no limits on the number of days an
individual can receive hospice care, a beneficiary's prognosis must be
reaffirmed at 90 days, at 180 days, and every 60 days thereafter.6

The hospice eligibility requirement that a beneficiary be certified as
having a prognosis of 6 months or less has been an ongoing concern expressed
by advocates and providers. The requirement has been challenged as difficult
to implement and a deterrent to hospice referrals, especially for
beneficiaries with noncancer diagnoses. Research suggests that it can be
difficult for physicians to accurately predict whether a patient is likely
to die within 6 months.7 It is particularly difficult to estimate life
expectancy for persons with noncancer diagnoses because the course of their
disease

is likely to be erratic.8 For example, patients with heart disease are more
likely to die suddenly than persons with cancer, who commonly have a period
of steady decline before death. Similarly, very elderly people in frail
health or with certain chronic illnesses may experience long periods of
declining health punctuated by several medical crises--any one of which can
be fatal. In such cases, physicians may find it difficult to justify a
hospice referral for beneficiaries who appear to be relatively stable, and,
as a result, the physicians may delay initiation of hospice services until a
medical crisis occurs shortly before death.

Used Have Declined

From 1992 to 1998, the number of Medicare beneficiaries enrolling in hospice
more than doubled, with growth in all population subgroups and in all
states. Growth was particularly rapid among beneficiaries with diagnoses
other than cancer. At the same time, many beneficiaries had shorter stays.
On average, the days of hospice service used per beneficiary declined by
about one-fifth during the 7-year period and beneficiaries with diagnoses
other than cancer experienced the sharpest reductions.

Our analysis of Medicare claims data indicates substantial growth in hospice
use. The number of beneficiaries electing hospice care increased 2 times
from 1992 to 1998, from about 143,000 to nearly 360,000 persons annually.
(See fig. 1.) Across most demographic groups, the use of hospice services
has grown at a relatively consistent rate. Thus, hospice users today are
similar to users in 1992; the distribution of enrollees by race has not
changed (89 percent are white), and the proportion of enrollees who are
women has climbed only slightly (from 50 to 54 percent). However, the use of
hospice services grew more rapidly among beneficiaries aged 80 and older
than it did among younger beneficiaries. This age group now makes up 47
percent of Medicare hospice enrollees, up from 35 percent in 1992.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

Overall, 19 percent of Medicare beneficiaries who died in 1998 received
hospice services, compared with 8 percent in 1992.9 However, this measure
understates the proportion of Medicare beneficiaries who choose hospice care
among those for whom the benefit was intended. According to a former
president of the National Hospice Organization, "when the number of deaths
nationwide is adjusted to reflect only those that are likely to be
appropriate for hospice care, the percentage of dying patients cared for in
hospice care is probably about 40 percent."10

Some groups of beneficiaries are more likely to choose hospice services than
others. For example, 20 percent of white Medicare beneficiaries who died in
1998 elected hospice services, compared with 15 percent of black
beneficiaries who died that year. Similarly, the use of hospice services is
more common among beneficiaries who are enrolled in Medicare health
maintenance organizations (HMO) at the end of life than among those in
fee-for-service plans. Of the beneficiaries who died in 1998, 27 percent of
those enrolled in an HMO elected hospice, compared with 18 percent of
fee-for-service beneficiaries.11 (See app. II for detailed information about
hospice use rates among decedents.)

In addition, the proportion of Medicare decedents who used the hospice
benefit varies widely by state. For example, in 1998, the number of hospice
users as a share of Medicare decedents was more than four times higher in
Arizona than in Maine. Table 1 shows states with the highest and lowest
rates of hospice use in 1998.

 State            Hospice users as a percent
                  of Medicare decedents
 U.S. total       19%
 Highest use
 Arizona          36
 Colorado         30
 Florida          29
 Oregon           26
 New Mexico       25
 Nevada           24
 Lowest use
 Alaska           6
 Maine            8
 South Dakota     11
 Tennessee        11
 Wyoming          11
 Washington, D.C. 12

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and Medicare decedent data from the denominator file, an
abbreviated version of HCFA's Medicare enrollment database.

Noncancer Diagnoses

Although people who die from cancer are more likely to choose hospice
services than are those who die from other conditions, the use of hospice
services by beneficiaries with noncancer diagnoses has increased rapidly.
From 1992 to 1998, hospice enrollment by beneficiaries with cancer increased
91 percent, while enrollment by beneficiaries with all other conditions
increased 338 percent. The most dramatic growth in use was among individuals
with other terminal conditions, such as heart disease, lung disease, stroke,
or Alzheimer's disease. About 43 percent of beneficiaries who elected
hospice in 1998 had noncancer diagnoses, compared with about 24 percent in
1992. Table 2 shows the distribution of new hospice enrollees by primary
diagnosis.

                                      1992                 1998
 Primary diagnosisa
                                      Number   Percent     Number  Percent
 All cancer                           108,232  75.6        206,190 57.4
 Lung                                 29,966   20.9        57,841  16.1
 Prostate                             10,052   7.0         15,494  4.3
 Breast                               7,602    5.3         13,093  3.6
 Colon                                6,697    4.7         13,278  3.7
 Pancreatic                           6,359    4.4         12,116  3.4
 Other                                47,556   33.2        94,368  26.3
 All noncancer                        34,878   24.4        152,759 42.6
 Congestive heart failure             6,141    4.3         24,248  6.8
 Chronic obstructive pulmonary disease4,112    2.9         15,765  4.4
 Stroke                               2,140    1.5         13,282  3.7
 Alzheimer's disease                  1,591    1.1         11,836  3.3
 "Ill-defined conditions"             888      0.6         7,599   2.1
 Other                                20,006   14.0        80,029  22.3
 Total                                143,110  100.0       358,949 100.0

aPatients entering hospice may have more than one medical condition or
diagnosis. Data include only the first, or principal, diagnosis listed for
each patient.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

For many of the leading causes of death, the proportion of elderly decedents
who use the hospice benefit has increased. In 1997, about half of the people
aged 65 and older who died from cancer had used hospice services, compared
with about one-fourth in 1992.12 This pattern generally held for breast
cancer, lung cancer, and prostate cancer. However, the use of hospice
services is even more common among persons with other types of cancer;
roughly 75 percent of people age 65 and older who died from brain or liver
cancer in 1997 used hospice services before death. The proportion of elderly
decedents who used hospice services also increased among beneficiaries who
died from other causes. Table 3 shows the change in hospice use rates from
1992 to 1997 for common hospice diagnoses.

                   Hospice users as a percent of decedents
 Primary diagnosis aged 65 and older
                   1992                1997
 Cancer            26%                 47%
 Breast            26                  48
 Colon             16                  31
 Lung              27                  47
 Pancreatic        30                  53
 Prostate          29                  50
 Noncancera        2                   9

aThe denominator used for calculating noncancer use rates includes people
who died unexpectedly (for example, from a first heart attack or injuries
sustained during an automobile accident) and thus are not candidates for
hospice care.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and Centers for Disease Control and Prevention (CDC) mortality
data from the WONDER/PC Data File. CDC's "Compressed Mortality" provides
data through 1997 only.

Although more Medicare beneficiaries are receiving hospice services, on
average, they are receiving fewer days of care than did beneficiaries in the
past.13 From 1992 to 1998, average length of stay declined 20 percent (from
74 to 59 days), while median length of stay declined 27 percent (from 26 to
19 days).14 (See fig. 2.)

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

The overall decline in average length of service appears to have been driven
by both (1) a reduction in the proportion of beneficiaries with very long
hospice stays and (2) an increase in the share of users with very short
stays.15 (See table 4.) From 1992 to 1998, the share of hospice enrollees
with more than 6 months of service use declined from 9.3 to 7.3 percent.
Over the same period, the proportion of beneficiaries who used hospice for a
very brief period before death rose sharply. In 1998, 28 percent of all
beneficiaries using hospice care did so for 1 week or less.

 Weeks of hospice care     Total hospice users    Total hospice days useda
 used                      1992      1998         1992         1998

 1 or less                 20.9%     28.0%        1.2%         1.8%
 2                         14.5      15.2         2.1          2.7
 3                         9.8       9.1          2.4          2.7
 4                         7.2       6.4          2.4          2.7
 5 to 8                    17.2      14.6         9.4          10.0
 9 to 12                   8.6       7.7          8.0          9.0
 13 to 16                  5.6       5.3          7.3          8.7
 17 to 26                  7.0       6.4          13.6         15.6
 More than 26              9.3       7.3          53.6         46.8

aIncludes all days of hospice care each beneficiary used.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

The decline in the average number of hospice days used has been especially
dramatic among beneficiaries with a primary diagnosis other than cancer.
While these beneficiaries historically had many more days of care than
cancer patients, the average number of days used declined 38 percent between
1992 and 1998. In comparison, average days used by beneficiaries diagnosed
with cancer declined by 14 percent. As a result, differences in length of
stay across diagnosis categories have narrowed considerably. In 1998, cancer
patients used an average of 54 days while noncancer patients used 68 days,
on average. Figure 3 compares the decline in the average number of hospice
days used for beneficiaries with cancer and noncancer diagnoses.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

At the state level, average length of service declined in 42 of 50 states
and the District of Columbia from 1992 to 1998, and variation in average
length of service across states lessened considerably. (State data appear in
app. II.) In 1992, 27 states had average service periods within 10 days of
the 74-day national average. By 1998, 36 states had average service periods
within 10 days of the 59-day average.

Several factors influence beneficiary choice about whether and when to use
hospice care. These include physician preferences and referral practices,
individual patient choice and circumstances, and general awareness of the
benefit among the public and professional communities. In addition, recent
federal oversight of compliance with patient eligibility requirements may
have affected certain beneficiaries' use of the hospice benefit.

Affect Hospice Use

Although Medicare beneficiaries and their families make the decision about
whether and when to initiate hospice services, physician willingness to
discuss options for end-of-life care is important to the decision. However,
the research literature indicates that not all physicians are comfortable
discussing end-of-life care, and some may hesitate to suggest hospice care
for other reasons, such as concerns about relinquishing control of their
patients' care. Even when the issue has been broached, some beneficiaries
choose instead to continue curative or life-extending treatments. Patient
advocacy groups, several medical societies, and others have called for
greater public and professional awareness of options for care of the dying,
which has led to a range of educational efforts designed to increase
awareness of hospice and its benefits.

Physician Preferences and Practices

The research literature indicates that because patients and their families
rely heavily on physician recommendations for treatment, including
recommendations for end-of-life care, physicians are an influential factor
in patient entry into hospice. Physicians initiate most referrals to
hospice, and they may continue to care for their patients after enrollment
as part of the hospice team. However, research has shown that many
physicians are poorly trained in the care of the dying and are often
uncomfortable discussing options for end-of-life care or the cessation of
curative treatment.16 A recent review of 50 top-selling textbooks from
several medical specialties found that most provided inadequate information
on end-of-life issues, with oncology textbooks among those particularly
likely

to provide no information about key aspects of end-of-life care.17 Physician
referral to hospice may be limited by other factors, as well. For example,
experts in the area of palliative care, as well as the research literature,
suggest that some physicians may not be aware that they can continue to
provide services after a beneficiary has entered hospice and may delay
referral out of concern about losing control of the patient's care.18

Patient Choice and Circumstances

The use of hospice services by Medicare beneficiaries requires not just
awareness of the benefit and a physician's certification of prognosis but
also acceptance that death is the outcome of their illness and the choice by
beneficiaries to give up a portion of their standard Medicare benefits in
order to receive hospice care. Once enrolled, no other services related to a
patient's terminal condition are covered under Medicare. HCFA officials and
others also noted that improvements in cancer care and the addition of new
treatment options may be prompting some beneficiaries to pursue new curative
options until very shortly before death, thus contributing to the trend of
shorter hospice stays. Other beneficiaries may favor continuing aggressive,
life-extending treatments up until the time of death and not enter hospice
at all. According to HCFA officials, it may be that some terminal patients
do not want hospice care, and that should be their right.

Research suggests that beneficiaries who do not consider hospice care may be
unwilling to confront the terminal nature of their illness, may not know
that the alternative exists, or may misunderstand the services available
through hospice care. The Institute of Medicine (IOM) noted that patients
are influenced by the general unwillingness to accept limits of all types,
including those of aging and death.19 A Gallup poll in 1996 found that
although a majority of people expressed interest in hospice care, most also
said that they would still seek curative care.

In some cases, a beneficiary's circumstances may complicate hospice
enrollment. Hospice is designed to allow the beneficiary to remain at home
during his or her last few weeks of life, where family and friends are
expected to deliver most of the routine day-to-day care. Hospice staff offer
more specialized care and respite care to give family members a break when
they need one. Thus, some hospice programs limit participation to
beneficiaries who have a caregiver at home. Others permit such beneficiaries
to enter the hospice program with the understanding that transfer to a
nursing home will be required when their needs for assistance reach a
certain stage.20

Public and Professional Awareness

Public and professional awareness also influences the use of the Medicare
hospice benefit. The need for greater public and professional understanding
of options for end-of-life care, including hospice, has been highlighted in
several recent congressional hearings and in other public forums.21 In
addition, several medical societies, patient advocacy groups, and the
hospice industry have undertaken a variety of efforts to educate their
members and the public about end-of-life care options. For example, the
American Medical Association and the Robert Wood Johnson Foundation are
developing a core curriculum for educating physicians in end-of-life care.
The Medicare Rights Center, a consumer advocacy and education organization,
is conducting a national campaign to increase awareness of the Medicare
hospice benefit among health professionals. Also, the National Hospice and
Palliative Care Organization has published a variety of materials on public
education and outreach strategies for its members.

Beneficiary Use of Services

Industry and patient advocacy groups contend that recent federal scrutiny of
provider compliance with program eligibility requirements has
inappropriately limited access to hospice for certain beneficiaries. While
federal scrutiny may have contributed somewhat to the existing trend toward
shorter hospice enrollment periods, continued growth in the number of
beneficiaries receiving hospice services makes it difficult to identify the
extent to which federal scrutiny may have deterred access. Furthermore, the
use of hospice services has increased most rapidly among beneficiaries with
diagnoses other than cancer--those for whom arriving at a 6-month prognosis
may be more difficult.

In 1995 and 1996, the Department of Health and Human Services' (HHS) Office
of the Inspector General (OIG) investigated the eligibility status of
Medicare beneficiaries receiving hospice services as part of a larger
investigation of fraud and abuse in Medicare. Specifically, OIG reviewed the
admission decisions made for hospice patients with very long stays at 12
hospices in four states; it found that many of these patients did not meet
eligibility criteria upon admission to hospice. OIG followed this effort
with other reviews of beneficiary eligibility, encompassing a larger sample
of hospices, and found that the vast majority of Medicare beneficiaries
receiving hospice services were eligible for such services.22

Patient advocacy groups and the hospice industry assert that this federal
scrutiny of compliance with the 6-month eligibility rule has had a chilling
effect on entry into hospice for noncancer beneficiaries, for whom it may be
more difficult to establish a 6-month prognosis with confidence. They
contend that hospice providers are more cautious about admitting
beneficiaries with noncancer diagnoses as a result, leading to delays in
hospice entry for persons wishing to use the benefit. Although the
percentage increase in beneficiaries electing hospice slowed somewhat from
1995 through 1998 compared with the prior period, it is difficult to know
what portion of this slower growth is attributable to the effect of federal
scrutiny and what portion is attributable to other factors, such as the
larger base of beneficiaries already using hospice.

The OIG scrutiny of beneficiary eligibility may have contributed to later
hospice entry for some beneficiaries, to the extent that hospice providers
responded to the oversight with greater caution about beneficiary
eligibility. However, the trend toward fewer days of hospice use began
before the period of federal scrutiny. As shown in figure 4, the average
length of service for both cancer and noncancer hospice users peaked by
1994, before scrutiny of the hospice benefit increased. Furthermore,
physician groups we spoke with did not cite caution among hospice providers
about beneficiary eligibility as a primary barrier to the initiation of
hospice services for their patients. According to the American Society of
Clinical Oncology, barriers to timely hospice care for cancer patients
include the attitudes of physicians and patients toward death and reluctance
to talk about death until the very end of life.

Noncancer Patients, 1992-98

Note: Operation Restore Trust (ORT) was a joint initiative among HCFA, OIG,
and the Administration on Aging designed to identify vulnerabilities in the
Medicare program.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

While the OIG reviews were under way, the National Hospice Organization
developed guidelines to assist physicians and hospices in determining a
6-month prognosis for patients with selected noncancer diagnoses. These
included amyotrophic lateral sclerosis (ALS), dementia, human
immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS),
heart disease, pulmonary disease, liver disease, stroke and coma, and kidney
disease. In order to enhance accuracy and uniformity in the claims review
process, HCFA distributed these guidelines to the intermediaries that
process hospice claims for Medicare to use in assessing compliance with
benefit requirements.23 The intermediaries have since adopted these
guidelines as formal local medical review policies.24

Concerns have been raised that using these guidelines as a standardized
basis for determining Medicare hospice eligibility limits access to hospice,
particularly for patients with noncancer diagnoses. Industry representatives
assert that the guidelines require further development to improve prognostic
confidence and accuracy before they would be appropriate as formal medical
review policies.

However, intermediaries point out that while medical review policies specify
clinical criteria for establishing a patient's 6-month prognosis, they allow
for variation in individual cases. For example, one intermediary's medical
review policy for heart disease notes that "some patients may not meet the
criteria, yet still be appropriate for hospice care, because of other
comorbidities or rapid decline." According to Medicare program guidance to
all hospices, the fact that a hospice patient lives beyond 6 months does
not, by itself, constitute grounds for a determination that the patient was
never eligible for hospice care or that Medicare does not cover services
provided to the patient. Typically, if a question is raised as to whether a
patient is terminally ill, the intermediary asks the hospice to furnish

information necessary to affirm the patient's prognosis. The rates of
medical reviews of claims began increasing in 1995, at HCFA's direction.25
Four of the five intermediaries reported that, by 1999, review rates ranged
from 0.8 to 4.2 percent of all hospice claims processed.26

Nearly all the hospice providers we spoke with said they consult their
intermediaries' medical review policies as part of the admission screening
process. Asked about the effect of the review policies on admitting
patients, some hospices reported that using these criteria has decreased the
likelihood of admitting patients with noncancer conditions, while others
said that the review criteria have increased the likelihood of admissions or
have had no effect at all.

Concerns

Sustained growth in the number of hospice providers participating in
Medicare and in their distribution throughout the country suggests that
hospice services are now more widely available to program beneficiaries.
While all sectors of the hospice industry have grown over the past decade,
recent growth has been particularly strong in the for-profit sector and
among large hospice programs. At the same time, hospice industry officials
report growing cost pressures from shorter patient stays and changes in the
practice of palliative care. Because data on provider costs are not
available, however, it is not clear how these cost factors affect providers
and beneficiaries.

Although the overall rate of growth has slowed somewhat in the past few
years, new hospice providers continue to enter the Medicare program every
year. As shown in figure 5, the number of Medicare-certified hospice
providers nationwide grew by 82 percent from 1,208 in 1992 to 2,196 in

1999.27 Each year during the period, additional hospice programs became
certified for Medicare, although the number of new entrants declined from
274 in 1994 to 46 in 1999, and the number of hospices leaving Medicare
exceeded the number of new entrants in 1999. (Many of those leaving were
based in home health agencies (HHA) that may have closed because of changes
in HHA payments enacted in BBA.) The increased number reflects not only new
hospices but also growing participation in Medicare. (See app. II for more
detail on changes in hospice supply and distribution.) In 1989, we estimated
that only about 35 percent of the approximately 1,700 hospice providers
nationwide participated in Medicare. By 1998, the National Hospice and
Palliative Care Organization estimated that 80 percent of hospices were
certified to serve Medicare patients.

Source: GAO analysis of annual Medicare Provider of Service Files.

Over this period, all types of hospice providers grew, in rural and urban
areas and in almost every state. From 1992 to 1999, the rate of growth was
greatest among for-profit providers and those in rural areas. Also, large
providers accounted for an increasing share of the services delivered. (See
table 5.) The number of for-profit providers increased nearly fourfold, and
the number of large hospice programs (those serving 500 or more Medicare
patients per year) more than tripled over the period.28 In addition, the
number of rural providers increased by 116 percent while the number of
urban-based providers increased 64 percent. Even with high growth in these
sectors of the industry, the majority of hospices are small programs (with
fewer than 100 Medicare patients per year), organized as not-for-profit, and
located in urban areas.

                Number of hospices
 Characteristic                     Percent change 1992-99
                  1992      1999
 All hospices   1,208     2,196     82%
 Affiliation
 Freestanding   466       877       88
 Hospital-based 327       553       69
 HHA-based      403       730       81
 Control
 For-profit     151       593       293
 Not-for-profit 957       1,365     43
 Government     63        146       132
 Other          36        75        108
 Sizea
 Small          795       1,244     56
 Medium         370       816       121
 Large          43        136       216
 Locationb
 Urban          823       1,350     64
 Rural          384       829       116

Note: Detailed information was not available for all providers.

aWe categorized hospices small if they served fewer than 100 Medicare
beneficiaries during the year, medium if they served 100 to 499 patients per
year, and large if they served 500 or more beneficiaries per year.

bWe categorized a hospice urban if it was located within a metropolitan
statistical area (MSA) and rural if it was located outside an MSA.

Source: GAO analysis of annual Medicare Provider of Service Files.

Even with certificate-of-need (CON) requirements that apply to hospice
providers in 14 states, the number and size of Medicare hospice providers
increased in almost every state from 1992 to 1998.29 Among states with large
Medicare enrollment and no CON requirements, the most dramatic growth was in
Texas, where the supply of hospice providers relative to the size of the
Medicare population nearly doubled. Among the CON states with large Medicare
populations, providers increased the number of patients served, while growth
in the number of providers was constrained. For example, in Florida and New
York the number of hospices per million beneficiaries remained virtually
unchanged; however, the number of beneficiaries each hospice served grew 66
and 105 percent, respectively. (See table 6.)

                           1992                           1998

     State         Providers      Hospice         Providers       Hospice
                  per million    users per       per million     users per
                beneficiaries    provider      beneficiaries     provider
 U.S. total    35               118           60               157
 California    25               152           48               189
 Florida (CON) 15               579           14               961
 Illinois      30               155           54               183
 Michigan      39               133           54               219
 New Jersey    29               141           37               210
 New York
 (CON)         19               158           20               325
 North
 Carolina      62               80            63               131
 (CON)
 Ohio          36               129           53               212
 Pennsylvania  30               85            56               155
 Texas         35               153           69               162

Note: These 10 states combined accounted for 54 percent of all Medicare
enrollees in 1998.

Source: GAO analysis of annual Medicare Provider of Service Files.

Providers Is Uncertain

Even as the hospice industry has grown, changes in the use of the hospice
benefit and the delivery of hospice care have raised concerns about cost
among providers. Most significantly, declines in the average enrollment
period have resulted in fewer days over which providers can spread the fixed
costs associated with a patient's stay in hospice. In addition, providers
report that changes in the practice of palliative medicine have made the use
of higher-cost services more common. However, because reliable data on
provider costs are not available, it is not clear how these factors may
effect hospices' financial status or their ability to serve Medicare
beneficiaries.

Industry representatives point out several areas of change that they contend
are adversely affecting the financial condition of providers. Specifically,

� Under Medicare's per diem payment system for hospice care, hospices have
traditionally offset the higher-cost days that occur at admission and during
the period immediately preceding death with lower-cost days of less
intensive care.30 For example, costs for admitting and assessing a new
patient, establishing a care plan, and delivering medical equipment are
incurred during the first few days of enrollment and do not vary with the
patient's period of service. As enrollment periods have declined, hospices
have had fewer days over which they can spread the higher costs associated
with the start and end of a patient's stay.

� As more patients enter hospice later in the course of their terminal
illness, they enter with higher levels of impairment and in need of more
intensive services. In addition, the shift in the mix of patients by
diagnosis may have increased the average service needs for the overall
hospice population. According to the most recent National Home and Hospice
Care Survey, hospice patients with noncancer diagnoses are somewhat more
likely than those with cancer to be functionally impaired and thus may
require more services on a regular basis from hospice agencies. 31

� Physicians and patients are calling on hospice programs to provide a
broader array of palliative services than in the past. Costly treatments
such as chemotherapy and radiation--traditionally used for curative
purposes--are increasingly used in the hospice setting to manage pain and
other symptoms. Furthermore, some new palliative care treatment options,
such as the transdermal administration of narcotic pain medication, may
offer better symptom control for some patients but often at greater expense.

To the extent that hospice providers believe that Medicare payments do not
adequately cover their costs, they may have an incentive to limit their
acceptance of patients who need more intensive services or limit the types

and amount of services they make available.32 Providers may also respond by
choosing not to admit patients who are expected to be more expensive.
However, hospice officials we interviewed reported being able to enroll most
patients who were referred. With the exception of patients lacking
sufficient informal caregiver support, the potential cost of care and
payment rates were not generally cited as factors limiting the admission of
eligible patients.

Data to assess how declining patient stays and changes in palliative care
practices affect overall provider costs are not currently available. While
certain more expensive services may be provided more frequently, the share
of total costs that these services currently represent is unknown.33
Furthermore, we do not know the extent to which providing more expensive
medications or treatments to hospice patients may reduce the need for other
services such as nursing visits. In response to BBA requirements, HCFA has
begun collecting hospice cost data to use in evaluating the adequacy of
current levels of Medicare reimbursement. Officials anticipate that audited
hospice cost data will be available beginning in late 2001.

Trends in the use of the Medicare hospice benefit during the 1990s indicate
that beneficiaries with all types of terminal illnesses are making use of
hospice services in greater numbers every year. In particular, the types of
patients selecting hospice have expanded broadly--from mostly beneficiaries
with cancer to a nearly even split among those with cancer and those with
other terminal conditions. In spite of these trends in use, and the
widespread availability of hospice providers, patient advocates and the
industry are concerned about the trend toward using fewer days of hospice
care. Because many factors influence the use of hospice care, and potential
demand is difficult to determine, the extent to which the Medicare hospice
benefit may be underutilized is not clear.

We provided a draft of this report to HCFA for review. In its comments, HCFA
discussed the importance of the hospice benefit to the Medicare program and
efforts to ensure that beneficiaries, physicians, and hospice providers
understand the benefits' coverage and eligibility criteria. Furthermore,
HCFA stated that it does not believe the underutilization concerns of
hospice advocates and the industry should be discounted. It noted that
enrollment in hospice may not be an option for beneficiaries who lack family
support at home or that it may be delayed for patients who wish to continue
curative care treatments. HCFA's comments appear in appendix III. The agency
made technical comments that we incorporated where appropriate.

As we agreed with your office, unless you publicly announce the report's
contents earlier, we plan no further distribution of it until 30 days after
the date of this letter. We will then send copies to the Honorable Donna
Shalala, Secretary of HHS; the Honorable Min DeParle, Administrator of HCFA;
and others who are interested. We will also make copies available to others
on request.

If you or your staff have any questions, please call me at 202-512-7119 or
Rosamond Katz, Assistant Director, at 202-512-7148. Other major contributors
were Eric Anderson, Jenny Grover, Wayne Turowski, and Ann White.
Janet Heinrich
Associate Director, Health Financing and
Public Health Issues

Methodology for Developing Information on Medicare Hospice Use and Hospice
Providers

Our study is an analysis of national hospice enrollment, use patterns, and
industry developments from 1992 through 1998. We examined Medicare
beneficiary claims data for hospice services to determine hospice use rates
for different groups of beneficiaries. We also gathered descriptive
information about the hospices that provided the services.

We used the Medicare Hospice Standard Analytic File of the Health Care
Financing Administration (HCFA) to identify beneficiaries who enrolled in
hospice during the study period and to determine their pattern of hospice
use. To conduct an analysis of hospice enrollment by year, we assigned
beneficiaries to the year of their first hospice claim. We excluded
beneficiaries from our analysis if total payment for a beneficiary was less
than $75 or was $1 million or more, if a beneficiary at the time of a first
hospice claim was younger than 20 or older than 110, or if a beneficiary's
residence was not in one of the 50 states or the District of Columbia.

Our analysis of beneficiary use includes information on

� age at the time of entry into hospice (younger than 65, 65 to 74, 75 to
84, 85 and older),

� gender,

� race and ethnicity (white, black, Hispanic, and other),

� state of residence,

� enrollment in managed care or fee-for-service Medicare (based on status in
the month of death, from the HCFA denominator file), and

� primary diagnosis (three-digit International Classification of Disease
code).

Analysis of the beneficiary claims data showed that 98 percent of
beneficiaries had only one hospice provider, and 97 percent had only one
diagnosis code. Therefore, we conducted all further analysis on the basis of
the provider and diagnosis listed in the first hospice claim for each
beneficiary. We calculated the period of enrollment by summing the number of
days covered by each claim, even if they covered discontinuous periods of
service, and excluded duplicate claims.34 Because records of hospice use are
not complete for beneficiaries who entered hospice during the later years of
our study period, we adjusted the claimed days of hospice service for 1996
to 1998 to better account for beneficiaries with very long stays.35 Our
adjustment factor was calculated from 1992-95 data on the proportion of
total beneficiary claim days accounted for within the first 2 calendar years
of hospice use.

We also described hospice use rates among different groups of Medicare
decedents. To calculate the rate of hospice use, we identified the number of
Medicare decedents each calendar year who had used hospice before death. We
used the HCFA Denominator File to identify all Medicare decedents belonging
to each demographic group in our analysis.36 Because the HCFA Denominator
File does not contain information about beneficiary diagnosis, we used
Centers for Disease Control and Prevention (CDC) mortality data to determine
the number of deaths among people aged 65 and older.37

We used annual Medicare Provider of Service Files to identify hospice
characteristics. These files contain data on provider certification and
status, such as facility and service characteristics, provider type, and
location. We included hospices that received total Medicare payments of $75
or more during our study period. The provider identification number from the
first hospice claim for each beneficiary was matched with the Provider of
Service file data available for that provider. We characterized providers by

� type of control (profit, not-for-profit, or government),

� affiliation (freestanding, hospital-based, home health agency, and skilled
nursing facility),

� state,

� urban or rural location, and

� number of Medicare beneficiaries receiving services from each hospice each
year (small defined as fewer than 100 beneficiaries, medium defined as 100
to 499 beneficiaries, and large defined as 500 or more beneficiaries).

Patterns and Trends in Hospice Supply and Use

                                                                   Percent
 Category         1992  1993  1994  1995  1996  1997   1998  1999  change
                                                                   1992-99
 Total providers  1,208 1,433 1,667 1,910 2,129 2,238  2,281 2,196 82%
 New entrants     193   240   274   266   243   174    103   46    -76
 Departures and
 closures         6     7     32    26    34    60     118   77    1,183
 Control
 For-profit       151   220   303   410   531   599    623   593   293
 Not-for-profit   957   1,088 1,201 1,313 1,384 1,418  1,425 1,365 43
 Government       63    83    107   120   137   140    144   146   132
 Other and
 unknown          37    42    56    67    77    81     89    92    149
 Affiliation
 Freestanding     466   548   615   699   774   836    877   877   88
 Hospital-based   327   371   420   476   526   557    570   553   69
 HHA-based        403   502   615   713   804   814    802   730   81
 Skilled nursing
 facility (SNF)   11    11    14    19    20    22     21    19    73
 based
 Sizea
 Small            795   923   1,041 1,204 1,327 1,350  1,313 1,244 56
 Medium           370   458   551   620   697   768    829   816   121
 Large            43    52    75    86    105   120    139   136   216
 Locationb
 Urban            823   954   1,073 1,231 1,357 1,421  1,423 1,350 64
 Rural            384   478   591   676   767   808    847   829   116

Note: Detailed information was not available for all providers.

aSmall = fewer than 100 patients a year. Medium = 100-499 patients a year.
Large = 500 or more patients a year.

bUrban = metropolitan statistical area. Rural = nonmetropolitan statistical
area.

Source: GAO analysis of annual Medicare Provider of Service Files.

 Category            1992   1993   1994   1995  1996   1997   1998   1999
 Certified providers 1,208  1,433  1,667  1,910 2,129  2,238  2,281  2,196
 Control
 For-profit          13%    15%    18%    21%   25%    27%    27%    27%
 Not-for-profit      79     76     72     69    65     63     62     62
 Government          5      6      6      6     6      6      6      7
 Other and unknown   3      3      3      3     3      3      3      4
 Affiliation
 Freestanding        39%    38%    37%    37%   36%    37%    38%    40%
 Hospital-based      27     26     25     25    25     25     25     25
 HHA-based           33     35     37     37    38     36     35     33
 SNF-based           1      1      1      1     1      1      1      1
 Sizea
 Small               66%    64%    62%    63%   62%    60%    58%    57%
 Medium              31     32     33     32    33     34     36     37
 Large               4      4      5      5     5      5      6      6
 Locationb
 Urban               68%    67%    64%    64%   64%    63%    62%    61%
 Rural               32     33     35     35    36     36     37     38

aSmall = fewer than 100 patients a year. Medium = 100-499 patients a year.
Large = 500 or more patients a year.

bUrban = metropolitan statistical area. Rural = nonmetropolitan statistical
area.

Source: GAO analysis of annual Medicare Provider of Service Files.

                                                           Percent change
 Category         1992  1993  1994  1995 1996  1997  1998
                                                           1992-98
 All U.S hospices 73.9  75.7  74.7  67.6 64.8  61.4  59.3  -20%
 Control
 For-profit       83.1  87.1  87.7  79.8 77.8  74.6  71.4  -14
 Not-for-profit   72.1  73.3  71.8  64.4 61.0  57.2  55.4  -23
 Government       75.8  71.8  70.6  65.9 65.4  61.6  58.4  -23
 Affiliation
 Freestanding     81.7  83.6  81.0  73.4 70.6  66.9  64.5  -21
 Hospital-based   69.9  71.1  70.5  63.1 61.3  57.6  54.0  -23
 HHA-based        64.4  66.8  67.9  61.5 57.7  54.5  53.7  -17
 SNF-based        53.4  55.4  61.2  55.9 47.3  50.8  51.7  -3
 Sizea
 Small            75.1  76.2  74.4  70.4 69.2  65.9  66.6  -11
 Medium           73.2  75.3  74.9  67.5 64.2  61.3  59.5  -19
 Large            74.0  76.0  74.6  66.2 63.5  59.4  56.3  -24
 Locationb
 Urban            72.1  74.3  73.7  66.6 63.7  60.4  58.0  -20
 Rural            86.8  85.2  81.2  73.9 71.8  67.6  66.9  -23

aSmall = fewer than 100 patients a year. Medium = 100-499 patients a year.
Large = 500 or more patients a year.

bUrban = metropolitan statistical area. Rural = nonmetropolitan statistical
area.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and provider data from annual Medicare Provider of Service
Files.

                                                                                       Percent
 Category                                                                              change
           1992       1993       1994       1995      1996       1997       1998
                                                                                       1992-98
 Medicare
 enrollees 34,852,914 35,557,195 36,169,499 36,758,10137,268,560 37,633,245 37,998,035 9%
 Medicare
 decedents 1,633,088  1,718,881  1,733,054  1,770,360 1,791,588  1,806,655  1,833,053  12
 Hospice
 users     143,110    180,097    223,044    258,481   294,931    325,324    358,949    151
 Cancer
 patients  108,232    128,678    151,701    169,080   182,038    195,114    206,190    91
 Noncancer
 patients  34,878     51,419     71,343     89,401    112,893    130,210    152,759    338
 Average
 days used 73.9       75.7       74.7       67.6      64.8       61.4       59.3       -20
 Cancer    62.5       63.2       63.6       60.5      58.4       55.8       53.9       -14
 Noncancer 109.2      106.9      98.4       80.9      76.3       70.6       67.8       -38
 Median
 days used 26         26         26         24        23         21         19         -27
 Cancer    26         27         27         26        25         23         23         -12
 Noncancer 25         26         25         21        18         16         15         -40

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and provider data from annual Medicare Provider of Service
Files.

                                                                     Percent
 Primary                                                             change
 diagnosis    1992    1993    1994    1995   1996    1997    1998
                                                                     1992-98
 All patients 143,110 180,097 223,044 258,481294,931 325,324 358,949 151%
 Cancer       108,232 128,678 151,701 169,080185,038 195,114 206,190 91
 Breast       7,602   8,761   10,476  11,626 12,296  12,661  13,093  72
 Colon        6,697   8,402   9,426   10,722 11,762  12,401  13,278  98
 Lung         29,966  35,528  41,694  46,747 51,772  54,940  57,841  93
 Pancreatic   6,359   7,597   8,860   9,771  10,592  11,371  12,116  91
 Prostate     10,052  11,808  13,577  14,368 15,441  15,249  15,494  54
 Other        47,556  56,582  67,668  75,846 83,173  88,492  94,368  98
 Noncancer    34,878  51,419  71,343  89,401 112,893 130,210 152,759 338
 Alzheimer's  1,591   2,549   4,060   6,050  7,839   9,608   11,836  644
 Stroke       2,140   3,528   5,426   6,972  8,973   10,910  13,282  521
 Chronic
 obstructive
 pulmonary    4,112   6,407   8,506   9,975  11,911  14,041  15,765  283
 disease
 Congestive
 heart failure6,141   9,118   12,432  15,172 18,293  20,661  24,248  295
 "Ill-defined"
 conditions   888     991     1,306   1,651  2,593   4,845   7,599   756
 Other        20,006  28,826  39,613  49,581 63,284  70,145  80,029  300

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

 Primary
 diagnosis       1992     1993    1994    1995     1996    1997    1998
 All patients    143,110  180,097 223,044 258,481  294,931 325,324 358,949
 Cancer          75.6%    71.4%   68.0%   65.4%    62.7%   60.0%   57.4%
 Breast          5.3      4.9     4.7     4.5      4.2     3.9     3.6
 Colon           4.7      4.7     4.2     4.1      4.0     3.8     3.7
 Lung            20.9     19.7    16.7    18.1     17.6    16.9    16.1
 Pancreatic      4.4      4.2     4.0     3.8      3.6     3.5     3.4
 Prostate        7.0      6.6     6.1     5.6      5.2     4.7     4.3
 Other           33.2     31.4    30.3    29.3     28.2    27.2    26.3
 Noncancer       24.4%    28.6%   32.0%   34.6%    38.3%   40.0%   42.6%
 Alzheimer's     1.1      1.4     1.8     2.3      2.7     3.0     3.3
 Stroke          1.5      2.0     2.4     2.7      3.0     3.4     3.7
 Chronic
 obstructive
 pulmonary       2.9      3.6     3.8     3.9      4.0     4.3     4.4
 disease
 Congestive
 heart failure   4.3      5.1     5.6     5.9      6.2     6.4     6.8
 "Ill-defined"
 conditions      0.6      0.6     0.6     0.6      0.9     1.5     2.1
 Other           14.0     15.9    17.8    19.2     21.5    21.4    22.3

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

 Primary diagnosis            1992   1993  1994   1995   1996   1997  1998
 All patients                 73.9   75.7  74.7   67.6   64.8   61.4  59.3
 Cancer                       62.5   63.2  63.6   60.5   58.4   55.8  53.9
 Breast                       85.1   88.0  86.3   80.4   78.1   73.9  70.0
 Colon                        62.8   64.6  67.3   63.8   62.1   59.8  58.6
 Lung                         56.7   57.1  58.0   56.2   54.6   52.5  51.0
 Pancreatic                   51.8   52.6  52.8   54.0   50.0   48.5  45.9
 Prostate                     80.6   79.7  81.5   76.9   70.8   67.4  64.5
 Noncancer                    109.2  106.9 98.4   80.9   76.3   70.6  67.8
 Alzheimer's                  173.7  180.9 156.3  125.1  113.4  100.9 99.7
 Stroke                       89.3   79.8  77.9   58.5   54.2   49.1  46.3
 Chronic obstructive
 pulmonary disease            139.8  132.2 119.3  102.4  97.9   89.2  85.5
 Congestive heart failure     110.7  108.2 96.7   78.8   75.3   69.1  70.9
 "Ill-defined" conditions     124.8  132.4 108.6  95.6   96.9   86.4  77.9

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File.

                      1992                               1998

 State                    Average Average                    AverageAverage
         Users  Providers  daily  days of   Users  Providers  daily days of
                          census   care                      census  care
 U.S.   143,110 1,208     24      74       358,949 2,281     23     59
 Ala.   2,162   29        15      74       7,112   66        23     85
 Alaska 0       0         0       0        67      3         2      53
 Ariz.  3,611   16        32      53       11,098  40        37     53
 Ark.   1,138   27        11      95       3,621   59        12     81
 Calif. 13,251  87        28      68       34,293  181       25     53
 Colo.  2,332   26        15      61       6,179   38        20     48
 Conn.  1,565   9         21      46       3,951   29        15     44
 Del.   568     3         36      72       1,049   5         35     63
 D.C.   238     2         18      60       472     3         16     44
 Fla.a  21,407  37        120     76       38,422  40        148    63
 Ga.    2,276   28        14      65       9,363   62        26     68
 Hawaiia394     7         8       55       1,074   7         17     43
 Idaho  615     13        11      87       1,282   30        7      62
 Ill.   7,276   47        34      78       16,147  88        24     55
 Ind.a  1,009   17        11      69       6,066   61        16     62
 Iowa   1,601   28        14      88       4,560   59        12     60
 Kans.  1,025   17        17      97       2,562   32        14     65
 Kya    3,276   27        36      107      5,367   29        35     76
 La.    1,462   21        12      63       4,065   38        15     57
 Mainea 310     11        5       61       687     15        8      64
 Md.a   1,906   22        14      58       5,387   30        22     47
 Mass.  3,800   33        22      69       7,074   44        20     49
 Mich.  6,670   50        26      70       16,186  74        30     56
 Minn.  2,222   24        17      68       5,449   63        14     62
 Miss.  611     12        12      86       2,925   39        17     94
 Mo.    3,119   33        22      87       8,849   76        17     60
 Mont.  402     9         10      78       1,008   17        10     65
 Nebr.a 519     9         12      74       2,042   27        12     63
 Nev.   794     4         43      78       2,589   9         37     50
 N.H.   351     9         6       61       1,219   22        8      59
 N.J.   4,652   33        25      66       9,225   44        27     50
 N. Mex.919     12        18      83       2,361   30        13     66
 N.Y.a  7,743   49        27      62       17,540  54        44     52
 N.C.a  4,696   59        22      100      9,022   69        24     71
 N. Dak.343     8         9       78       983     14        9      52
 Ohio   7,492   58        23      68       19,073  90        27     50
 Okla.  939     14        14      72       5,946   56        23     88
 Oreg.  2,483   25        19      68       6,304   42        19     49
 Pa.    5,200   61        14      60       18,286  118       21     52
 R.I.a  646     7         19      73       1,478   8         22     50
 S.C.   1,406   16        20      83       4,050   34        22     73
 S. Dak.248     6         9       78       691     13        8      54
 Tenn.a 1,522   23        13      74       4,924   60        13     61
 Tex.   10,427  68        39      93       24,517  151       26     69
 Utah   168     4         5       48       1,747   17        14     50
 Vt.a   308     8         6       56       561     9         7      46
 Va.    1,349   10        30      83       5,994   46        11     58
 Wash.  2,631   20        26      72       6,845   29        32     53
 W. Va.a1,078   12        18      69       2,294   21        18     70
 Wis.   2,862   41        13      70       6,231   54        17     58
 Wyo.   30      3         1       54       295     12        4      65

aState has a certificate-of-need requirement for hospice.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and provider data from annual Medicare Provider of Service
Files.

  State  Users  Providers     Average       Average
                           daily census  days of care
 U.S.    151%   89%        -3%           -20%
 Ala.    229    128        55            15
 Alaska  a      a          a             a
 Ariz.   207    150        17            3
 Ark.    218    119        10            -14
 Calif.  159    108        -11           -22
 Colo.   165    46         33            -21
 Conn.   152    222        -27           -1
 Del.    85     67         -5            -10
 D.C.    98     50         -14           -20
 Fla.b   79     8          23            -17
 Ga.     311    121        83            6
 Hawaiib 173    0          126           -13
 Idaho   108    131        -40           -27
 Ill.    122    87         -28           -31
 Ind.b   501    259        42            -9
 Iowa    185    111        -16           -33
 Kans.   150    88         -17           -38
 Ky.b    64     7          -3            -31
 La.     178    81         25            -12
 Maineb  122    36         66            7
 Md.b    177    36         58            -19
 Mass.   86     33         -5            -28
 Mich.   143    48         18            -21
 Minn.   145    163        -21           -9
 Miss.   379    225        51            13
 Mo.     184    130        -26           -31
 Mont.   151    89         1             -18
 Nebr.b  293    200        0             -15
 Nev.    226    125        -15           -36
 N.H.    247    144        30            -2
 N.J.    98     33         8             -23
 N. Mex. 157    150        -27           -23
 N.Y.b   127    10         66            -15
 N.C.b   92     17         9             -29
 N. Dak. 187    75         6             -32
 Ohio    155    55         14            -25
 Okla.   533    300        67            20
 Oreg.   154    68         1             -30
 Pa.     252    93         49            -13
 R.I.b   129    14         16            -34
 S.C.    188    113        10            -12
 S. Dak. 179    117        -13           -30
 Tenn.b  224    161        1             -16
 Tex.    135    122        -32           -26
 Utah    940    325        161           11
 Va.     344    360        -37           -29
 Vt.b    82     13         19            -21
 Wash.b  160    45         22            -26
 W. Va.b 113    75         3             -3
 Wis.b   118    32         26            -17
 Wyo.    883    300        241           38

aAlaska had no Medicare-certified hospice providers in 1992.

bState has a certificate-of-need requirement for hospice.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and provider data from annual Medicare Provider of Service
Files.

               1992                  1994                  1996                  1998
 Category
               Providers Users Days  Providers Users Days  Providers Users Days  Providers Users Days
 Control
 For-profit    13%       16%   18%   18%       19%   23%   25%       24%   28%   27%       26%   30%
 Not-for-profit79        81    79    72        77    74    65        73    69    62        71    67
 Government    5         2     2     6         2     2     6         2     2     6         2     2
 Affiliation
 Freestanding  39%       50%   55%   37%       49%   53%   36%       50%   54%   38%       52%   55%
 Hospital-based27        20    19    25        19    18    25        19    18    25        19    17
 HHA-based     33        28    25    37        30    27    38        30    27    35        29    27
 SNF-based     1         2     2     1         1     1     1         1     1     1         1     1
 Sizea
 Small         66%       23%   23%   62%       18%   18%   62%       17%   18%   58%       15%   17%
 Medium        31        49    49    33        51    51    33        50    50    36        49    50
 Large         4         28    28    5         31    31    5         33    32    6         36    34
 Locationb
 Urban         68%       88%   86%   64%       87%   86%   64%       86%   85%   62%       86%   84%
 Rural         32        12    14    35        13    14    36        14    15    37        14    16

Note: Percentages indicate percentage shares.

aSmall = fewer than 100 patients a year. Medium = 100-499 patients a year.
Large = 500 or more patients a year.

bUrban = metropolitan statistical area. Rural = nonmetropolitan statistical
area.

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and provider data from annual Medicare Provider of Service
Files.

   State    Hospice users as percent of Medicare decedents
 Ariz.      36%
 Colo.      30
 Fla.       29
 Oreg.      26
 N. Mex.    25
 Nev.       24
 Mich.      23
 Ohio       22
 Tex.       22
 Del.       21
 Okla.      21
 Calif.     20
 Ga.        20
 Ill.       20
 Wash.      20
 Ala.       19
 Iowa       19
 Utah       19
 Idaho      18
 Kans.      18
 Ky.        18
 Minn.      18
 Mo.        18
 R.I.       18
 Hawaii     17
 Md.        17
 Pa.        17
 Ark.       16
 Conn.      16
 Nebr.      16
 N.H.       16
 N.J.       16
 N.C.       16
 N. Dak.    16
 S.C.       16
 Wis.       16
 Mass.      15
 Mont.      15
 N.Y.       15
 Ind.       14
 Va.        14
 W. Va.     14
 La.        13
 Miss.      13
 Vt.        13
 D.C.       12
 S. Dak.    11
 Tenn.      11
 Wyo.       11
 Maine      8
 Alaska     6
 U.S. total 19

Source: GAO analysis of claims data from the Medicare Hospice Standard
Analytic File and Medicare decedent data from the denominator file, an
abbreviated version of HCFA's Medicare enrollment database.

Comments From the Health Care Financing Administration

(201009)

Table 1: Medicare Hospice Use Rates in Selected States and the
District of Columbia, 1998 10

Table 2: New Hospice Enrollees by Primary Diagnosis, 1992 and
1998 11

Table 3: Hospice Use Rates for Selected Diagnoses, 1992 and 1997 12

Table 4: Hospice Benefit Use by Weekly Intervals, 1992 and 1998 14

Table 5: Growth in Medicare Hospice Programs by Provider Characteristics,
1992 and 1999 27

Table 6: Growth in Number and Size of Hospices in States With Large Medicare
Populations, 1992 and 1998 28

Table 7: Number of Hospice Providers by Control, Affiliation, Size,
and Location, 1992-99 35

Table 8: Percentage of Hospice Providers by Control, Affiliation, Size,
and Location, 1992-99 36

Table 9: Average Number of Days Beneficiaries Used Hospice by
Provider Control, Affiliation, Size, and Location, 1992-98 37

Table 10: Use of Medicare Hospice Benefits by Cancer and Noncancer Patients,
1992-98 38

Table 11: Number of Hospice Users by Selected Common
Diagnoses, 1992-98 39

Table 12: Percentage of Hospice Users by Selected Common
Diagnoses, 1992-98 40

Table 13: Average Number of Days Beneficiaries Used Hospice by
Selected Common Diagnoses, 1992-98 41

Table 14: Hospice Use and Supply by State, 1992 and 1998 42

Table 15: Percentage Change in Hospice Use and Supply by State,
1992-98 44

Table 16: Change in Percentage of Providers, Users, and Days of
Care by Provider Type, 1992-98 46

Table 17: Variation in Hospice Use Among Medicare Decedents by
State, 1998 47

Figure 1: Number of Medicare Hospice Benefit Users, 1992-98 8

Figure 2: Decline in Average and Median Hospice Lengths of
Service, 1992-98 13

Figure 3: Average Days of Use for Cancer and Noncancer Patients,
1992-98 15

Figure 4: Decline in Average and Median Days of Hospice Care for
Cancer and Noncancer Patients, 1992-98 21

Figure 5: Growth in the Number of Medicare Hospices and New
Entrants, 1992-99 25

1. Data were not available before 1991.

2. Outpatient medications and respite care are subject to limited
coinsurance by beneficiaries. Beneficiaries pay 5 percent of the cost of
drugs or $5 per prescription, whichever is less, and they pay 5 percent of
hospice payments for respite care but no more than the Medicare hospital
deductible, which is $776 for calendar year 2000.

3. For routine home care, the hospice is paid a daily rate for each day a
beneficiary is enrolled, regardless of whether or not services are provided
each day. Routine home care is the default payment rate when other levels of
care are not provided. In 1998, 97 percent of hospice care days were paid at
the routine home care rate.

4. Cost data included the kinds of services furnished, their costs, and the
frequency of their provision.

5. The daily base payment rates for fiscal year 2000 are $98.96 for routine
home care, $577.59 for continuous home care, $102.37 for inpatient respite
care, and $440.22 for inpatient hospital care. These base rates are adjusted
by the hospital wage index to reflect differences in wage levels in
different areas of the country.

6. Before 1990, Medicare payments to hospices were limited to a 210-day
lifetime maximum per patient. From 1990 to 1997, coverage was limited to a
lifetime maximum of four benefit periods but an unlimited number of days.

7. A recent study found that physicians are inaccurate in their predictions
of how long their terminally ill patients have to live. See N. Christakis
and E. Lamont, "Extent and Determinants of Error in Doctors' Prognoses in
Terminally Ill Patients: Prospective Cohort Study," British Medical Journal,
Vol. 320 (Feb. 19, 2000), pp. 469-73.

8. Researchers have found that for seriously ill hospitalized patients with
advanced chronic obstructive pulmonary disease, congestive heart failure, or
end-stage liver disease, it was not possible to accurately identify in
advance those who would die within 6 months. For example, the most inclusive
criteria excluded 58 percent of patients who died within 6 months of
discharge, while 70 percent of patients identified as likely to die within 6
months survived longer. In contrast, the most restrictive criteria
eliminated 99 percent of patients who died within 6 months but 53 percent of
patients identified as likely to die within 6 months survived longer. See E.
Fox and others, "Evaluation of Prognostic Criteria for Determining Hospice
Eligibility in Patients With Advanced Lung, Heart, or Liver Disease,"
Journal of the American Medical Association, Vol. 282, No. 17 (Nov. 4,
1999), pp. 1638-45.

9. A recent study found that about 18 percent of Medicare beneficiaries who
died in 1996 used hospice services. See Department of Health and Human
Services, Assistant Secretary for Planning and Evaluation, Office of
Disability, Aging and Long-Term Care Policy, Medicare's Hospice Benefit: Use
and Expenditures, 1996 Cohort (Mar. 2000).

10. John J. Mahoney, "The Medicare Hospice Benefit--15 years of Success,"
Journal of Palliative Medicine, Vol. 1, No. 2 (1998), pp. 139-46.

11. We classified Medicare beneficiaries as enrolled in an HMO or
fee-for-service care at the end of life on the basis of their status in the
month they died. When an HMO enrollee elects hospice, the monthly HMO
payment is reduced, and the HMO then bills Medicare on a fee-for-service
basis for any covered medical services unrelated to the terminal illness.

12. A recent study found that 51 percent of all 1998 Medicare cancer
decedents used hospice care. See Medicare Payment Advisory Commission,
Medicare Beneficiaries Costs and Use of Care in the Last Year of Life (May
2000).

13. Not all beneficiaries who elect the hospice benefit die while in hospice
care. Each year, between 10 and 14 percent of beneficiaries using hospice
deselected the benefit. Slightly more than half of these beneficiaries die
within 30 days of the time they stop receiving hospice services.
Beneficiaries who received hospice services for longer than 6 months were
twice as likely as those enrolled for less than 6 months (25 percent versus
12 percent) to leave hospice care alive. Similarly, beneficiaries with
noncancer diagnoses were more likely than those with cancer to disenroll
from a hospice program, although the difference was not as great--about 15
and 11 percent, respectively.

14. The small proportion of beneficiaries with very long periods of
enrollment skews the average length of hospice service. Annually, more than
80 percent of new hospice entrants complete their hospice use in the year in
which they enroll, and 97 percent complete their hospice use by the end of
the year following their initial enrollment. However, some of the remaining
beneficiaries receive services for many more years. Because records of
hospice use are not complete for beneficiaries who entered hospice during
the later years of our study period, we adjusted the claimed days of hospice
service for 1996 to 1998 to better account for beneficiaries with very long
stays. See app. I for more information on our methodology.

15. The distribution of patients between cancer and noncancer diagnoses is
the same for very long (more than 6 months) and very short (1 week or less)
periods of hospice use, 47 percent and 53 percent, respectively.

16. A recent study found that physicians commonly make errors in their
predictions of life expectancy, most often predicting that patients have
longer to live than they really do. The authors concluded that this may
contribute to late referral to hospice care and noted that physicians are
uncomfortable making predictions of expected life span, in part because
neither medical training nor published literature treats prognostication as
important. See Christakis and Lamont, pp. 469-73.

17. M. W. Rabow and others, "End-of-Life Care Content in 50 Textbooks From
Multiple Specialties," Journal of the American Medical Association, Vol.
283, No. 6 (Feb. 9, 2000), pp. 771-78.

18. Visits by the patient's attending physician are not covered by the
hospice payment and continue to be paid by Medicare part B coverage.

19. IOM, Approaching Death: Improving Care at the End of Life (Washington,
D.C.: National Academy Press, 1997).

20. For a detailed discussion of hospice use by nursing home residents, see
Department of Health and Human Services, Assistant Secretary for Planning
and Evaluation, Office of Disability, Aging and Long-Term Care Policy, Use
of Medicare's Hospice Benefit by Nursing Facility Residents (Mar. 2000).

21. U.S. Senate Special Committee on Aging, "The End of Life: Improving
Care, Easing Pain and Helping Families" (July 17, 2000); U.S. House of
Representatives, Committee on Government Reform, "Improving Care at the End
of Life With Complementary Medicine" (Oct. 19, 1999); U.S. Senate, Committee
on Health, Education, Labor, and Pensions, "Pain Management and Improving
End-of-Life Care" (Oct. 13, 1999).

22. OIG reported that 81 percent of beneficiaries in its sample were
eligible for hospice care, while 7 percent were ineligible. OIG was not able
to determine eligibility for 12 percent of beneficiaries. A larger
proportion of hospice beneficiaries in nursing homes were found ineligible.
See HHS, Medicare Hospice Beneficiaries: Services and Eligibility, pub.
OEI-04-93-00270 (Apr. 1998). For a discussion of the OIG report, see HHS,
Assistant Secretary for Planning and Evaluation, Office of Disability, Aging
and Long-Term Care Policy, Important Questions for Hospice in the Next
Century (Mar. 2000).

23. Intermediaries contract with HCFA for paying providers for services
provided to Medicare beneficiaries. They review all hospice claims for
accuracy and completeness before payment and review a sample of claims to
confirm that beneficiaries were eligible for the hospice services provided.

24. Local medical review policies are medical criteria, specific to a
service or diagnosis, that may assist in determining compliance with program
eligibility requirements. More recently, some intermediaries have adopted
policies for adult failure to thrive, and policies for specific cancers are
also under development.

25. Claims are selected for medical review on the basis of a variety of
factors, including beneficiary length of stay, beneficiary diagnosis, and
provider use of hospice continuous home care or inpatient care. Before 1995,
intermediaries reviewed a very small proportion of hospice claims.

26. One fiscal intermediary did not provide consistent data on rates of
medical review.

27. The total number of Medicare hospice providers peaked at 2,281 in 1998.
In 1998 and 1999, the 195 hospice program closures exceeded the 149 new
program entrants for the first time. A disproportionate number of hospice
closures were among those based in home health agencies (HHA). Although HHA
hospices represent approximately one-third of all hospices, they accounted
for 43 percent of those that closed over the 2-year period. As we reported
in Medicare Home Health Agencies: Closures Continue, With Little Evidence
Beneficiary Access Is Impaired (GAO/HEHS-99-120, May 26, 1999), 14 percent
of HHAs closed between October 1997 and January 1999.

28. For-profit hospices provided about 30 percent of the hospice days
beneficiaries used in 1998, up from 18 percent in 1992. Large hospices
(those serving 500 or more patients in a year) provided 38 percent of
hospice days in 1998 compared with 28 percent in 1992.

29. In 14 states, a hospice may be required to obtain a CON before becoming
eligible to provide services. CON programs are run by state planning or
regulatory agencies and are designed to limit the numbers of providers of a
given type, based on established formulas or criteria for need.

30. Hospice representatives we interviewed reported that the hours of
nursing, social work, and administrative time the typical patient requires
are nearly twice as high during the first and last weeks of a patient's care
as they are during the intervening weeks.

31. National Center for Health Statistics, The National Home and Hospice
Care Survey: 1996.

32. IOM has expressed concern that Medicare's current system of paying a
fixed amount per day may discourage the provision of care for some patients,
such as the use of costly pain medications or equipment or the provision of
late-night nursing visits or extensive counseling services. See IOM,
Approaching Death: Improving Care at the End of Life.

33. Initial Medicare hospice rates were based on data from HCFA's 1980-82
hospice demonstration project, allocated over an average beneficiary
enrollment period of 70 days. The rates took into consideration the kinds of
services furnished, their cost, and the frequency of their provision at that
time.

34. Our analysis showed that 95 percent of all beneficiaries during the
period of our analysis received hospice services during one continuous time
period.

35. The small proportion of beneficiaries with very long periods of
enrollment skews the average length of hospice service. Although 97 to 98
percent of all those electing hospice complete their hospice use by the end
of the year following their initial enrollment, our data for 1992 and 1993
show that some of the remaining 2 to 3 percent of beneficiaries may receive
services for as many as 6 or 7 years.

36. The HCFA Denominator File includes information about all Medicare
beneficiaries with entitlement (eligible for Medicare part A or part B) in a
given year. We excluded decedents who were not eligible for Medicare part A
benefits as well as those who lived outside the 50 states and the District
of Columbia.

37. Data collected from the dataset Compressed Mortality. This file provides
mortality data through 1997 only.
*** End of document. ***