Medicare Home Health Care: Prospective Payment System Will Need
Refinement as Data Become Available (Letter Report, 04/07/2000,
GAO/HEHS-00-9).

Pursuant to a legislative requirement, GAO reviewed the Health Care
Financing Administration's (HCFA) research on a home health prospective
payment system (PPS), focusing on: (1) the objectives, findings, and
costs of the research and demonstration projects HCFA has funded that
were related to the design of the PPS; and (2) how these projects
contributed to the proposed PPS design and which design decisions were
based on incomplete information.

GAO noted that: (1) HCFA has sponsored a number of research and
demonstration projects on payment design and home health care users and
service delivery since 1987 at a cost of almost $27 million; (2) despite
these important efforts, key features of a PPS were not evaluated in
these projects, which limits the ability to evaluate the effects of
certain payment policies on home health care service delivery and
spending; (3) HCFA's major home health agencies (HHA) payment
demonstration project provided evidence that HHAs would reduce their
costs of providing home health visits when paid under a PPS model that
tightly limited both their profits and their losses; (4) furthermore,
the demonstration did not develop a case-mix adjustment method to alter
payments for expected differences in resource use across groups of
patients; (5) however, an ongoing research project has constructed an
initial case-mix adjustment method for the PPS and will continue to
refine this method as more data become available; (6) other
HCFA-sponsored research projects have documented the variation in home
health care service delivery; (7) these projects have demonstrated that
methods for quality measurement and monitoring are not well developed,
which will impair the ability to evaluate the effect of payment changes;
(8) although HCFA's research and demonstration projects have proven
useful in designing the PPS, information gaps remain; (9) these gaps
mean that the PPS could cause unintended consequences for some
beneficiaries, some HHAs, or the level of Medicare spending; (10)
concerns remain about whether the case-mix adjustment method will
adequately group patients with like resource needs and then
appropriately adjust payments for beneficiaries in each group; (11)
furthermore, how a patient is classified and how much the agency is paid
are very dependent on whether, and how much, therapy services are
provided; (12) without adequate design features, Medicare could overpay
for unneeded services or underpay for required care, resulting in
beneficiaries facing access problems or receiving poor quality of care;
(13) although the change from cost-based payments to prospective
payments is intended to help Medicare control its spending, how costs
and service provision will change under the new system is unknown; and
(14) therefore, HCFA will need to have sufficient resources to monitor
service delivery across types of beneficiaries and across HHAs so that
inadequate or medically inappropriate care can be identified.

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

 REPORTNUM:  HEHS-00-9
     TITLE:  Medicare Home Health Care: Prospective Payment System Will
	     Need Refinement as Data Become Available
      DATE:  04/07/2000
   SUBJECT:  Home health care services
	     Health care cost control
	     Health resources utilization
	     Claims processing
	     Health insurance
	     Medical information systems
	     Health care programs
	     Internal controls
IDENTIFIER:  Medicare Program
	     Medicare Prospective Payment System
	     HHS Home Health Care Services and Training 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-00-9

Appendix I: Scope and Methodology

32

Appendix II: HCFA Projects to Develop a Prospective Payment
System for Home Health Agencies

33

Appendix III: Comments From the Health Care Financing
Administration

37

Table 1: HCFA-Funded HHA Research Spending, 1988-99, by Project Category 16

BBA Balanced Budget Act of 1997

HCFA Health Care Financing Administration

HHA home health agency

HHRG home health resource group

HHS Department of Health and Human Services

IPS interim payment system

MedPAC Medicare Payment Advisory Commission

OASIS Outcome and Assessment Information Set

PPS prospective payment system

SCIC significant change in condition

Health, Education. and
Human Services Division

B-282264

April 7, 2000

Congressional Committees

Medicare spending for home health care rose from $3.7 billion in 1990 to
$17.8 billion in 1997, making it one of the fastest growing components of
the program. Since then, spending has moderated. In 1998 Medicare outlays
were $17.3 billion, and expenditures in 1999 are projected to be around $15
billion. The historic rise in expenditures for home health care primarily
was due to more beneficiaries receiving services and more visits provided
per user. To control spending, the Congress enacted fundamental payment
reforms in the Balanced Budget Act of 1997 (BBA). Notably, the act required
the Secretary of Health and Human Services to develop a prospective payment
system (PPS) to replace cost-based payments for home health agencies (HHA).
The BBA outlined the general terms for the HHA PPS, specifying that it pay a
fixed, predetermined amount for a unit of service, adjusted for patient
characteristics that affect the costs of providing care.

The Health Care Financing Administration (HCFA), within the Department of
Health and Human Services (HHS), has been conducting research on a home
health PPS for some time. This research is especially important because
designing an effective payment system is difficult, given Medicare's broad
definition of who qualifies for home health coverage and the lack of
standards for what constitutes appropriate care. The challenges lie in
defining the service unit that will be used for payment purposes and
developing the case-mix adjustment method to vary payments for differences
in patient needs. These and other system design decisions, as well as their
implementation, will determine the extent to which the PPS rewards HHAs that
deliver care efficiently, protects beneficiaries from inadequate care, and
ensures that the Medicare program is only paying for medically necessary
care.

Concerned about HCFA's ability to develop the PPS on time, the Congress, in
the Omnibus Consolidated and Emergency Supplemental Appropriations Act of
1999 (P.L. 105-277), delayed the PPS implementation date by a year to
October 1, 2000. The legislation also mandated three reports on the PPS for
HHAs. First, the Secretary, HHS, was required to submit a report on HCFA's
research relevant to the PPS for HHAs and the schedule for implementing the
PPS. Second, the Medicare Payment Advisory Commission (MedPAC) was required
to analyze the Secretary's report and make recommendations regarding its
contents. Finally, we were asked to review and report to the Congress on the
expenditures for HCFA's research on the home health PPS. (Addressees are
listed at the end of this letter.)

After reviewing the other mandated reports and in consultation with your
staffs, we agreed to (1) document the objectives, findings, and costs of the
research and demonstration projects HCFA has funded that were related to the
design of the PPS and (2) assess how these projects contributed to the
proposed PPS design and determine which design decisions were based on
incomplete information. To address these objectives, we examined materials
and reports for projects listed in HCFA's Active Projects Reports, the
Secretary's report to the Congress, and MedPAC's letter to the Congress. In
addition, we analyzed the proposed design for the PPS1 and discussed the
proposal with HCFA officials and their primary contractor, Abt Associates,
Inc. (App. I contains a complete description of our methodology.) We
performed our work between December 1998 and February 2000 in accordance
with generally accepted government auditing standards.

Recognizing the need for home health payment reform, HCFA has sponsored a
number of research and demonstration projects on payment design and home
health care users and service delivery since 1987 at a cost of almost $27
million. Despite these important efforts, key features of a PPS were not
evaluated in these projects, which limits the ability to evaluate the
effects of certain payment policies on home health care service delivery and
spending. HCFA's major HHA payment demonstration project provided evidence
that HHAs would reduce their costs of providing home health visits when paid
under a PPS model that tightly limited both their profits and their losses.
The demonstration did not examine alternative levels of payments.
Furthermore, the demonstration did not develop a case-mix adjustment method
to alter payments for expected differences in resource use across groups of
patients. However, an ongoing research project has constructed an initial
case-mix adjustment method for the PPS and will continue to refine this
method as more data become available. Other HCFA-sponsored research projects
have documented the variation in home health care service delivery. These
projects have demonstrated that methods for quality measurement and
monitoring are not well developed, which will impair the ability to evaluate
the effect of payment changes.

Although HCFA's research and demonstration projects have proven useful in
designing the PPS, information gaps remain. These gaps, coupled with
substantial variation in the way home health care services are delivered and
the lack of standards for what constitutes appropriate care, mean that the
PPS could cause unintended consequences for some beneficiaries, some HHAs,
or the level of Medicare spending. For example, the proposed unit of
payment, a 60-day episode, is likely to be too long for many beneficiaries
and could result in unnecessary expenditures if payments are not adequately
adjusted for patient needs. Also, the level of payments per episode will be
based on national average costs, which could result in sharp revenue
increases for some agencies and large declines for others. Concerns remain
about whether the case-mix adjustment method will adequately group patients
with like resource needs and then appropriately adjust payments for
beneficiaries in each group. Furthermore, how a patient is classified and
how much the agency is paid are very dependent on whether, and how much,
therapy services2 are provided--something that is directly controlled by
HHAs. Without adequate design features, Medicare could overpay for unneeded
services or underpay for required care, resulting in beneficiaries facing
access problems or receiving poor quality of care.

Although the change from cost-based payments to prospective payments is
intended to help Medicare control its spending, how costs and service
provision will change under the new system is unknown. Therefore, HCFA will
need to have sufficient resources to monitor service delivery across types
of beneficiaries and across HHAs so that inadequate or medically
inappropriate care can be identified. We are recommending that HCFA devote
the resources necessary to perform these monitoring activities. And, as more
information becomes available about what services are delivered within an
episode and how long visits last, efforts should be pursued to develop
criteria for service adequacy and appropriateness and to identify the
outcomes for beneficiaries. In the interim, a risk-sharing arrangement, in
which aggregate Medicare payments are adjusted at year-end to reflect a
provider's actual costs, would mitigate any unintended consequences of the
payment change. Limiting an HHA's losses or gains would help protect the
industry, the Medicare program, and beneficiaries from possible negative
effects of the PPS until more is known about how best to design the PPS and
the most appropriate home health treatment patterns. We are also
recommending that as new data are available and experience is gained with
the PPS, HCFA should study practice patterns and provider responses to the
PPS and make any needed modifications to the PPS design and implementation.

Medicare's home health care benefit enables certain beneficiaries with
post-acute-care needs (such as recovery from joint replacement) and chronic
conditions (such as congestive heart failure) to receive care in their homes
rather than in other settings. To qualify for home health care, a
beneficiary must be confined to his or her residence ("homebound");3 require
intermittent skilled nursing, physical therapy, or speech therapy; be under
the care of a physician; and have the services furnished under a plan of
care prescribed and periodically reviewed by a physician. If these
conditions are met, Medicare will pay for part-time or intermittent skilled
nursing; physical, occupational, and speech therapy; medical social
services; and home health aide visits.4 Beneficiaries are not liable for any
coinsurance or deductibles for these services and may receive an unlimited
number of visits, provided the coverage criteria are met.

Coverage

Between 1990 and 1997, Medicare home health payments rose an average of 25.2
percent annually. This increase was due primarily to a steady rise through
1997 of the proportion of beneficiaries receiving home health care and in
the number of visits per person served. The number of home health users per
1,000 beneficiaries increased from 57 to 109, and the average number of
visits per user doubled from 36 to 73.5 This growth can be attributed to
many factors, including changes in patient demographics and in the delivery
of health care services, particularly following the introduction of the
inpatient hospital PPS in 1983, which encouraged hospitals to discharge
beneficiaries to alternative settings. Most notably, the relaxation of
coverage guidelines as a result of legislative changes and a landmark court
case resulted in more services being provided.

Originally, home health care coverage distinguished between services
provided under Medicare part A (hospital insurance) and part B (supplemental
medical insurance). Under part A, up to 100 visits a year were provided to a
beneficiary following a hospital or nursing home stay, with no beneficiary
cost-sharing. A beneficiary could also receive up to 100 visits under part
B, with no institutional stay required, but had to pay 20 percent
coinsurance. In 1980, legislation removed the institutional stay requirement
under part A, eliminated the coinsurance requirement under part B, and
lifted the visit limits under parts A and B. It also changed the program
financing, with part A covering all home health care received by
beneficiaries unless the beneficiary only had part B coverage, in which case
part B would pay for the services.

Following these changes, home health care spending nearly doubled from 1980
to 1985, which prompted a series of additional regulatory actions by HCFA to
tighten the benefit and coverage. This, in turn, triggered a class action
suit in 1987 (Duggan v. Bowen) in which a coalition of beneficiaries and
providers charged that Medicare's interpretation of the statutory phrase
"part-time or intermittent" with regard to covered visits was too narrow,
leading to the denial of care for eligible beneficiaries. Under a settlement
agreement in 1989, the coverage guidelines for home health care were
broadened, allowing more beneficiaries to qualify for more visits. As a
result of these expansions, the benefit was transformed from one that
primarily covered patients receiving short-term care following an acute
event to one that covers chronic and long-term care patients as well.

At the same time this growth occurred, program controls were essentially
nonexistent. Few claims were subject to medical review and most were paid
without question. Previous work conducted by GAO and by the HHS Office of
Inspector General has documented that some of the care provided was not
medically necessary or lacked supporting documentation.6

The majority of home health users receive few visits, but a small and
growing proportion make extensive use of the benefit. According to MedPAC,
51 percent of recipients received fewer than 30 visits and accounted for 9
percent of all home health visits in 1996. By contrast, 15 percent of users
had 150 visits or more and accounted for 59 percent of all Medicare home
health visits that year. Approximately one-third of the beneficiaries in
this latter group received over 300 visits.7

Home health users also differ in the mix of services they receive. A small
proportion of Medicare users appears to need long-term care and gets a
significant amount of aide services,8 as opposed to skilled care, from the
program. In 1996, about 56 percent of the visits for beneficiaries who had
100 visits or more were for home health aide services. By contrast, only 6
percent of all visits to short-term users--beneficiaries who received nine
or fewer visits--were for aide services; skilled nursing care comprised over
75 percent of their total visits.

There also is marked variation in home health use across geographic areas
and types of agencies. For example, Medicare home health users in Maryland
received an average of 37 visits in 1997, with an average payment per user
of $3,088. In that same year, home health users in Louisiana received an
average of 161 visits per user, with an average Medicare payment per user of
$9,278. This wide variation in usage has been evident even after controlling
for patient diagnosis. Patterns of care also differed across agency types.9
Proprietary agencies tended to deliver more visits per beneficiary than
other types of agencies and to provide more aide visits. For example, in
1993, beneficiaries who received care from proprietary agencies were given
an average of 69 home health aide visits, compared with 43 and 48 visits
from voluntary and government agencies, respectively.10 Such variation could
be due to a variety of factors, including provider responses to financial
incentives, differences in patient needs, regional practice patterns,
states' varying Medicaid coverage and eligibility policies, and the use of
home health care to substitute for services in other clinical settings.

The variation in service provision may also reflect the lack of standards
for what constitutes necessary or appropriate home health care. As a result,
it is not clear when home health care is warranted or when services should
be stopped. Many home health users have chronic and multiple needs, so the
beginning and end of care for a particular problem may overlap with care for
another. Furthermore, even the most basic unit of service--the visit--is not
well defined. Only recently have HHAs been required to record the time
involved in a visit, and services and procedures provided are still not
documented on the payment record.

Before the BBA, agencies were paid on the basis of their costs, up to
preestablished per-visit limits equal to 112 percent of the national average
cost for each type of visit. Although there was a separate payment limit for
each type of visit (skilled nursing; physical, occupational, or speech
therapy; medical social service; or home health aide), the limits were
applied in the aggregate for the agency. That is, costs above the limit for
one type of visit would still be paid if costs were sufficiently below the
limit for other types of visits. There were no incentives to control the
volume of services delivered; as a result, agencies could enhance their
revenues by providing beneficiaries with more services.

The BBA changed the home health payment method beginning October 1, 1997,
with the implementation of the interim payment system (IPS), a temporary
measure to bring Medicare spending under control until the PPS is
implemented. The per-visit limits are generally lower under the IPS, and
agencies are subject to a Medicare revenue cap that is based on an aggregate
per-beneficiary amount.11 Generally, the per-beneficiary amount reflects
each agency's historical average payments for treating a Medicare
beneficiary and the regional or national average amount.12 An agency's
revenue cap is the product of its per-beneficiary amount and the number of
patients it serves.

The IPS is designed to limit the Medicare per-visit payment and to give
agencies the financial incentives to limit the number of services each
provides to beneficiaries. It does not, however, limit the services an
individual beneficiary can receive, nor does it restrain the number of
beneficiaries an HHA can serve. Because the revenue cap applies to the
average payment for a beneficiary for a year of services at an HHA, the
services and costs associated with any particular beneficiary can vary. To
ensure that Medicare payments will cover its costs, an HHA will need to keep
the average cost of its visits below the per-visit limits and keep its
average cost per Medicare beneficiary below its per-beneficiary amount. It
can do this by delivering visits more efficiently, changing the mix or
reducing the number of visits provided to each user, increasing the
proportion of lower-cost patients it treats, or some combination of these
strategies.

Adapting to the IPS involves greater challenges for some agencies than
others. HHAs that have provided more visits per beneficiary or that have
higher per-visit costs than the average will have to change the way they
deliver care. Agencies that treat few beneficiaries, agencies with few
low-cost patients, or those with costly treatment patterns may find the
adjustments more extensive and difficult. For agencies that have changed
treatment patterns or their mix of patients since 1994, the limits, which
are based on historical agency-specific costs, may not reflect current
service provision. New agencies, particularly those located in high-cost
regions, may be disadvantaged compared with established agencies because
their aggregate revenue caps are based on the national median
per-beneficiary amount.

The IPS may not adequately protect beneficiaries from underservice or the
Medicare program from paying more than is warranted in the longer term. The
revenue caps are based on each HHA's historical costs, which is intended to
account for the differences in the mix of patients across agencies. An
agency's historical patient mix, however, may not reflect the costs of the
agency's current patients. Furthermore, the IPS does not mitigate the
substantial cost differences across HHAs that may not have any relationship
to patient needs.13 In addition, a system of cost-based payments subject to
limits does not incorporate incentives for providers to reduce costs below
the limits, thus it may not curb Medicare spending on lower-cost patients.

Controls

Under a PPS, payments are established in advance of service delivery.
Payments may vary with patient characteristics or other factors that affect
costs. The payment is divorced from an individual provider's actual cost of
delivering care. Providers that on average deliver care for less than the
payment amount profit; conversely, providers lose if their average service
costs are higher than the payment.

HHA efforts to control the cost of service delivery can result in
unacceptable reductions in the quantity or quality of care. Features of the
PPS--including the unit of payment, the level of payments, and the method of
adjusting payments for differences in patients' needs--should therefore be
designed to minimize inappropriate provider responses to the financial
incentives of the payment system. Risk-sharing arrangements between the
payer and provider that limit providers' profits and losses may be added to
the PPS to temper incentives to either over- or underserve. Other strategies
outside the payment mechanism, such as medical review and examination of
utilization patterns, could be used to monitor provider responses to the
payment system and to design modifications to the PPS to help ensure the
delivery of medically appropriate and necessary services.

The unit of payment defines the bundle of services covered by the
payment--for example, it could be a home health visit or all of the visits
during a course of treatment (an episode of care) that may span many days.
Selecting between smaller and larger units involves trade-offs between
control over the volume of services provided and beneficiary protections
against underservice. With individual services as the unit of payment,
providers have an incentive to deliver more services to increase their
revenues, and, as a result, concerns about underservice are minimized.
Defining an episode as the unit may encourage providers to focus on being
more efficient by changing the intensity and mix of services delivered to
patients during each episode, but it may also create incentives to stint on
care.

The level of payment for each unit can affect access and the adequacy of
services as well as overall program spending. The level is usually based on
a historical average cost of a unit, either nationally or regionally, or by
provider. Because information to identify the costs of efficiently
delivering only appropriate services is generally not available, an average
amount is assumed to be adequate to ensure that sufficient numbers of
providers will continue to supply services. More generous payments would
tend to ensure that more providers continue to serve Medicare beneficiaries
if access proves to be a problem. When the payment level equals a national
or regional average, providers with higher-than-average costs will need to
lower their costs by, for example, shortening the length of visits, changing
the mix of visits within an episode, or reducing the number of visits in an
episode. Such changes may improve efficiency but also could reduce the
quality of care. However, the lack of standards for home health care means
that distinguishing between added efficiency and reduced quality will often
be difficult.

Under a PPS, adjusting payments to reflect expected resource needs for
individual patients is critical to maintaining appropriate access and
sufficient service provision while ensuring that payments and program
expenditures are not unnecessarily high. The proposed PPS design
incorporates a case-mix adjustment method to accomplish these objectives. A
case-mix adjustment method has two parts, both of which must be reasonably
precise to result in appropriate payment adjustments. First, categories are
developed to group patients with comparable levels of expected resource use.
Patients are usually assigned a category on the basis of clinical,
functional, and other characteristics that are predictive of service costs.
It is important that these categories not be subject to manipulation, such
that providers could inappropriately assign patients to a particular group
simply to boost payments.

Second, a relative weight is associated with each patient category. Each
relative weight reflects the average costliness of the patients in that
case-mix group compared with the costliness of all patients. To determine
the payment for a given patient, therefore, requires first assigning the
patient to the appropriate case-mix group and then multiplying the relative
weight for that group by the average payment amount. Thus, how well the
case-mix groups categorize patients with similar resource needs and how well
the relative weight adjusts the payment will determine the fairness and
adequacy of the payments under the PPS.

A PPS may be combined with a risk-sharing arrangement that limits the losses
and gains a provider can experience over a period of time. Risk-sharing
involves considering the provider's actual cost of delivering services in
determining the final payment.14 Although risk-sharing reduces incentives to
eliminate inefficient or inappropriate service use, it still may be
appropriate because of the protections it affords beneficiaries against
underservice and to the program against excessive payments, particularly
when other mechanisms do not provide these safeguards.

HCFA published its proposed PPS design in a proposed rule in the Federal
Register on October 28, 1999. The basic unit of payment would be a 60-day
episode of care. An episode would begin with a beneficiary's first visit and
end on the 60th day after that visit. For beneficiaries requiring additional
care, a second episode would begin on day 61 and end on day 120. There would
be no limits on the number of episodes for any beneficiary. The basic 60-day
episode payment would cover all of a beneficiary's home health care during
that period, regardless of the actual days of care or visits provided.

The basic 60-day episode payment would incorporate two adjustments--a
case-mix adjustment based on a clinical classification system and a wage
adjustment to reflect the variation in labor costs across geographic areas.
The case-mix adjustment method would assign patients to one of 80 home
health resource groups (HHRG) on the basis of patient assessment data from
the Outcome and Assessment Information Set (OASIS)15 and the projected
number of therapy visits in the patient's plan of care. The wage adjustment
would be based on the geographic area in which the beneficiary received
services and would be applied to the labor portion of the episode payment.

In cases in which the basic episode of care is interrupted, payments would
be prorated.16 Interruptions occur when a beneficiary elects to transfer to
another HHA, when a beneficiary is discharged when treatment goals are
attained but then returns to the same HHA, or when a significant change in
the beneficiary's condition (SCIC) results in a new OASIS assessment. For
beneficiary-elected transfers or a discharge and return to the same agency
during a 60-day episode, the payment would be adjusted to reflect the actual
length of time the beneficiary remained under the agency's care before the
intervening event. In the case of an SCIC, payments would have two prorated
components: one for the portion of care before the SCIC and one for the care
provided after a second OASIS assessment triggered a new HHRG.

The level of aggregate agency payments was established in the BBA. PPS rates
are to be set so that Medicare expenditures are equivalent to what would
have been spent under the IPS, with those limits reduced by 15 percent.17
The base episode payments are to be set at the national average cost of
providing the average number and mix of services within an episode.

HCFA's PPS design incorporates two additional features to adjust payments
for beneficiaries with unusual costs. For beneficiaries with exceptionally
high costs, an HHA would receive an "outlier" payment. The outlier payment
is not intended to cover the full cost of the case, but it raises the
payment above the standard episode amount. For beneficiaries with extremely
low service use within an episode (four or fewer visits), HHAs would receive
a low-utilization payment adjustment. Instead of the per-episode payment,
payments would be on a per-visit basis, without a case-mix adjustment, but
adjusted for geographic differences in labor costs.

New HHA data reporting requirements, while not available in time to affect
the initial PPS design, may prove useful for subsequent refinements. As part
of an assessment requirement, agencies must report OASIS and therapy service
provision data on all patients, which will be used to assign a patient to an
HHRG. In addition, the BBA required agencies to report the duration of
visits in 15-minute increments. The OASIS, therapy, and visit time
information will help in understanding what services are provided and what
resources are needed during visits and episodes of care and in developing
patient outcome indicators.

HCFA has sponsored a number of research and demonstration projects, at a
total cost of $27 million over 12 years, to better understand the nature of
home health care, the characteristics of its users, and how these factors
should be reflected in the payment system (see table 1). (For more detail on
funded projects, see app. II.) The Home Health Agency Prospective Payment
Demonstration involved six different projects in two phases at a cost of
$14.9 million. In both demonstration phases, Medicare tested limited
features of a PPS, paying agencies prospective rates for visits in phase I
and episodes of care in phase II and adjusting payments to limit the profits
and losses of an agency. Phase II yielded valuable information on the effect
of episode-based payments on costs and service use. The tested models did
not, however, yield information about the appropriate level of payments.
Furthermore, a case-mix adjustment method was not tested in either phase.
Eight other HHA payment projects were funded at a cost of $12.1 million. An
ongoing project developed the case-mix adjustment method that will be used
in the PPS and will continue to refine this method as more data become
available. Other projects investigated determinants of and variation in home
health utilization and expenditure growth to address questions about quality
of care and appropriate service delivery.

Table 1: HCFA-Funded HHA Research Spending, 1988-99, by Project Category

 Project category                                            Expenditure
 Home Health Agency Prospective Payment Demonstration, phase
 I                                                           $6.5 million

 Home Health Agency Prospective Payment Demonstration, phase
 II
                                                             8.3 million
 Case-mix research                                           3.9 million
 Related home health research                                8.2 million
 Total                                                       $26.9 million

Source: GAO analysis of data from Active Projects Report: Research and
Demonstrations in Health Care Financing, eds. 1988-98.

The Omnibus Budget Reconciliation Act of 1987 directed HHS to conduct a
demonstration project on prospective payment methods for HHAs. The
demonstration tested only selected PPS features. In the two phases of the
demonstration, alternative units of payment were evaluated; however, other
key PPS design features were not. The PPS models tested in the demonstration
were shaped, in part, by the need to minimize the financial risk placed on
HHAs in order to secure their participation.

Phase I, conducted from 1990 to 1995, tested the effect of per-visit
prospective payments on agency costs and patterns of care. HHAs in the test
group were paid a fixed, prospective rate per visit based on each agency's
historical costs. Payments were not subject to a case-mix adjustment method
to account for patient cost differences that were due to variation in
resource needs. Retrospective payment adjustments were made so that gains
and losses for each test agency were limited to 5 percent of
Medicare-allowable costs. The evaluation of this demonstration concluded
that prospective per-visit payments did not affect agency costs or the
provision of care when the HHA was at little financial risk.

Phase II of the demonstration is testing the effect of prospective
episode-based payments on HHA costs and service delivery. Phase II started
in 1995 and will continue until the PPS is implemented. Test agencies are
paid a single prospective rate for the first 120 days of services to a
patient and a prospective per-visit rate for any additional visits. As in
phase I, payment rates are based on each agency's historical costs.
Therefore, HHAs with higher historical costs continue to receive higher
payments. The rates may be adjusted at year-end if the cost patterns of the
agency's current patient mix do not match its historical costs. Agency
losses are shouldered primarily by HCFA and agency profits over 5 percent of
Medicare-allowable costs are shared with Medicare.18 Preliminary evaluation
of phase II suggests that HHAs respond to an episode-based PPS by
controlling their per-episode costs by reducing the number of visits
provided.19 This is notable, given the demonstration's risk-sharing
arrangement, which limited an agency's financial losses and gains and thus
reduced the incentives to change the way care was delivered. That is, HHAs
controlled their costs even though their financial losses or gains were
restricted. The evaluation will continue to monitor the demonstration's
effect on patient outcomes and access and on the use of other health
services.

Phase II of the demonstration showed that expanding the unit of payment for
home health care from a visit to a 120-day episode can change the way HHAs
provide services, resulting in lower per-episode spending. The risk-sharing
arrangements incorporated into the demonstration mitigated any large swings
in payment to HHAs. Phase II did not continue to pay for long-term patients
on a per-episode basis--instead it switched to per-visit payments after 120
days. Neither phase of the demonstration adjusted payments for differences
in expected beneficiary resource use, so they did not yield a case-mix
adjustment method for use in a PPS. Finally, neither phase tested
alternative levels of payment for services across agencies, as payments were
based on agency-specific historical payments.

HCFA also funded research to measure and monitor the quality of care
provided by HHAs participating in the demonstration. Both phase I and phase
II of the demonstration evaluated the effect of prospective payment on the
quality of care. Results indicate that a PPS may not negatively affect the
measured patient outcomes and that HHAs can reduce costs for home health
services without compromising quality or outcomes. Furthermore, the effects
of the payment approach tested in the demonstration on the quality of care
may be understated because outcome measures to evaluate care, especially for
chronically ill patients, are not fully developed.

HCFA has been researching a case-mix adjustment method for home health
services for some time, although the demonstration did not test one. An
early research project to develop patient categories was completed in 1991.
It entailed a significant primary data collection phase because available
claims data contained little of the patient-level information necessary to
adequately group patients to reflect their expected resource needs. The
study concluded that service-dependent descriptors such as nursing
treatments were more predictive of patient resource use than measures like
functional status or medical diagnosis. Service-dependent patient
descriptors, however, are vulnerable to provider manipulation (that is,
providers can add services to increase payments) and therefore are not ideal
for use in a payment system. Furthermore, the project's data were collected
before the issuance of revised coverage guidelines in 1989, so they do not
reflect current treatment patterns for many patients.

In 1996, HCFA funded a major study to develop a case-mix adjustment method
to be used in the HHA PPS.20 The resulting preliminary method will be based
on 20 patient descriptors from the OASIS assessment instrument and a measure
of patients' therapy use during the home health episode. These data
elements, measuring clinical severity, functional status, and therapy
service use, are used to assign a patient to one of 80 HHRGs. Each HHRG is
assigned a relative weight that reflects the costliness of the beneficiaries
in that category relative to the other case-mix groups. The case-mix
adjustment method will continue to be refined as the research project is
completed.

Care

Six other HCFA-funded research projects, at a total cost of $8.2 million,
were initiated to contribute to the design of the PPS or to understand its
potential effect on service delivery. The majority of these studies
addressed the quality of home health care, which could be affected by PPS
payment incentives to reduce costs. HCFA funded efforts to develop home
health care quality measures and monitoring methods. The projects did not,
however, establish who should receive home health care or any appropriate
standards of care. Given the continued lack of agreement, the quality
measures and monitoring methods may not be adequate to assess changes in
home health care quality associated with new payment policies.

Other studies were designed to explain factors contributing to variation in
service mix and costs. Research on the regional variation in costs found
that only a portion of these differences in service use could be explained
by patient characteristics. Some of the differences appeared to be due to
variations in practice patterns unrelated to patient needs, indicating that
policies to reduce service use would not necessarily constrain needed care.
Another study examined the growth in home health expenditures. It found that
spending increases were fueled by rising utilization, not higher costs per
visit. This suggests that an episode-based payment method, rather than a
per-visit approach, would be more promising in controlling Medicare home
health outlays.

Although HCFA has used information from its research efforts to shape the
design of the PPS, the PPS is based on incomplete knowledge and will likely
require modifications as more data become available. Given the wide
variation in patients and service use, the 60-day episode unit of payment
will not reflect the service patterns of many beneficiaries. Because the
episode length does not closely match many beneficiaries' service needs, the
PPS will rely heavily on the case-mix adjustment to calibrate payments to
different types of patients. In addition, payments based on national average
costs and service provision will redistribute payments across providers and
may result in over- or underpayments. Furthermore, because of widely varying
service patterns, it has been difficult to develop a case-mix adjustment
method that adequately describes resource use, particularly for long-term
users. The proposed method relies heavily on therapy treatments for its
classification of patients into payment groups, which may contribute to
inappropriate care if providers change therapy regimens for financial
reasons. Therefore, although the PPS is intended to affect provider behavior
and thereby moderate Medicare spending, the limited research evidence on the
appropriate payment design, the broad range of beneficiaries, and the lack
of standards for care make it likely that the HHA PPS will have unintended
consequences for beneficiary access, quality of care, and Medicare spending.

HCFA has proposed a 60-day episode of care as the unit of payment in the
PPS, rather than the per-visit payment or the 120-day episode tested in the
demonstration. HCFA believes that the 60-day episode has several advantages
because it conforms to Medicare physician certification requirements for
home health care and the OASIS reassessment schedule. In addition, the
majority of beneficiaries historically have received services for fewer than
60 days and thus one episode payment would cover their care.

While a 60-day episode is logical for administrative reasons, it may not be
appropriate for all beneficiaries, and the case-mix adjuster may not be
robust enough to adequately calibrate payments under these circumstances.
Although for the majority of beneficiaries care is completed within 60 days,
care for many patients is completed in a much shorter period. Medicare's
home health benefit is broadly defined, and different groups of home health
users have unique care needs. For example, prior research has demonstrated
that the health status and patterns of care of long-term users of home
health care, as described by functional limitations, differ substantially
from those of short-term users. It is possible that a single payment unit,
combined with the limitations of the case-mix adjuster, may not reflect the
variation in these distinct groups of beneficiaries.

The effect of HCFA's designation of 60-day episodes as the unit of payment
attempts to balance competing concerns about level of service to Medicare
beneficiaries and Medicare outlays. The proposed unit of payment will impose
more discipline on HHAs to lower total costs of caring for short-term users
than a shorter unit such as a visit. It also may not be as easy for
providers to increase the number of units of service to augment revenues.
The 60-day episode payment, however, creates incentives to lower the
intensity or cost of services in the episodes--by shortening visit lengths
or by reducing the number of visits provided within the episode. This
anticipated response to the episode payment may not negatively affect
patient care, given the questionable appropriateness of previous
utilization. However, because there is a lack of accepted standards of home
health care, it will be difficult to assess the effect of any changes in
service patterns. Furthermore, if HHAs merely delay services in order to
extend care over multiple episodes to increase provider payments, Medicare
expenditures will rise inappropriately.

HCFA has anticipated that the payments for patients who receive few visits
could be too high under a 60-day episode amount. Thus, as one refinement, it
has developed a specific payment policy for low-cost patients--the
low-utilization payment adjustment. This policy would result in lower
payments for these beneficiaries than would be made under a case-mix
adjusted episode amount and is intended to partially counter the incentive
to generate additional low-cost episodes. Agencies, however, will have
strong financial incentives to ensure that all beneficiaries receive enough
services to qualify for the full episode payment. If agencies do increase
the number of episodes provided, Medicare expenditures will increase.

Payments

HCFA used the most recent available cost data (fiscal year 1997) to develop
the HHA payment rates. The rates reflect the national average number of
visits in an episode and the national average cost per visit, by type of
visit. Because the prospective rates reflect national average service use
and visit costs, the PPS will result in considerable redistribution of
payments across providers.

The proposed rule includes HCFA's estimates of the effect of the PPS on
HHAs. Payments to rural, freestanding, for-profit HHAs are expected to fall
17 percent when the PPS is implemented, but payments are expected to
increase 46 percent to rural, freestanding, government HHAs. Payments to
urban for-profit agencies would decrease 18 percent compared with a 20
percent increase to urban nonprofit agencies. The appropriateness of these
estimated changes in payment and resulting provider responses are not known.
Higher revenues may lead some HHAs to provide more services, while others
may not change their patterns of care, thus leaving little justification for
the higher Medicare payments. Conversely, lower payments may encourage some
providers to increase their efficiency or to provide fewer services. Because
there are no established treatment guidelines or outcome measures for home
health care, it will be difficult to evaluate the effect of any service
changes on quality of care.

Differences in Patient Needs

A case-mix adjustment method for home health care must be robust to
adequately predict resource use for the wide range of types of patients who
receive care. Without a good measurement system, payments will be too high
for some types of patients and too low for others, creating provider
incentives to seek or to avoid certain types of patients for financial
reasons. The proposed case-mix adjustment method, based on HCFA-funded
research, predicts patient resource use about as well as methods used in
PPSs for other services, but it is heavily dependent on therapy services
delivered during the HHA episode for its ability to do so. Therefore,
providers have incentives to manipulate therapy treatments to maximize their
Medicare revenues.21 This could be a particular problem in home health care
because the patterns of care are so variable and the standards of care are
so ill-defined.

Given the diversity of patients receiving care, the large variation in
practices, and the lack of standards for appropriate care, the case-mix
adjustment method may not closely track resource use, resulting in large
over- and underpayments for certain types of patients. In particular, HCFA's
analysis indicates that resource use of long-term users may not be
adequately accounted for by the case-mix adjustment method. This is because
the variable measuring therapy use is the major predictor of costs, yet
these patients generally use fewer of these services. As a result, the PPS
payments for long-term patients in subsequent episodes may be consistently
too high or too low, which could distort provider incentives regarding
beneficiary treatment.

Anticipating the occasional exceptionally costly patient, HCFA has proposed
outlier payments to cushion the losses an HHA would incur on a particular
beneficiary. Moreover, an outlier policy should counter the incentive to
avoid certain types of patients that can be identified prior to treatment.
Outlier payments, however, may not adequately protect beneficiary access or
address HHA concerns about losses if the basic case-mix adjustment method,
which affects payments for all beneficiaries, is not robust.

Substantial variation in patient service needs and costs of care, combined
with the 60-day episode unit of payment and the potential limitations of the
case-mix adjustment method, could result in inappropriately high or low
payments for particular beneficiaries or certain HHAs. Risk-sharing
arrangements that limit HHA losses or gains from Medicare could be
incorporated into the PPS design to adjust aggregate payments to account for
actual agency costs. Such arrangements could moderate the effects of
inadequate payments, the incentives to manipulate services to maximize
profits, and the uncertainties associated with payment rates that are based
on averages when so little is known about appropriate patterns of home
health care. On the other hand, risk-sharing dampens the incentive of the
PPS to provide care more efficiently.

A risk-sharing arrangement that limits the amount an HHA can lose or gain
would involve a year-end settlement that compares an HHA's actual
Medicare-allowed costs with its total Medicare payments. Payments above the
costs would be constrained to a specific percentage, as would agency losses.
Both phases of the demonstration included this form of risk-sharing, but
HCFA's proposed PPS does not.

Although the PPS is intended to provide incentives to HHAs to deliver care
more efficiently by allowing them to earn profits while risking losses on
treating Medicare beneficiaries, extreme gains or losses could have
unintended consequences. For example, the possibility of large gains might
encourage providers to underserve beneficiaries because the HHA could retain
all payments in excess of costs. Conversely, unlimited losses could
undermine the quality of care and could eventually lead to reduced access
for Medicare beneficiaries. These potential problems are compounded by the
lack of standards for appropriate home health care that preclude effective
monitoring of these HHA behaviors.

Over the past 12 years, HCFA has sponsored several research projects related
to the use of home health care and payment policy for HHAs. These projects
have provided valuable insights into the appropriate design of the PPS. Yet,
questions about key PPS design components and their effect on service
delivery and costs remain. The research offered little to explain the
variation in service costs and patterns of care that is not tied to therapy
service use, which is valuable information to have to help evaluate the
effects of the PPS on beneficiary access and quality of care.

The proposed HHA PPS would create strong financial incentives to providers
to change the way they deliver care that could compromise quality of care
and could result in unintended increases in Medicare home health spending.
It is likely that extensive monitoring of home health service delivery will
be required to ensure that HHAs do not respond to these financial incentives
either by inappropriately reducing care within an episode or by providing
care that is not medically necessary in order to gain payments for
additional episodes. Such monitoring is complicated by the current lack of
accepted standards for home health care against which changes may be
measured.

Uncertainties about the appropriate specification of key design features and
provider responses to the PPS suggest moderating the effect of a largely
untested PPS. Until data are available to refine the PPS to ensure
appropriate beneficiary access and payment levels, a risk-sharing approach
could moderate unintended changes. Although risk-sharing may dampen the PPS
incentive to provide care more efficiently, we believe such a trade-off is
appropriate to protect beneficiaries, HHAs, and the Medicare program.

Finally, key PPS features may need to be modified as experience is gained
with the system and more data become available. While such revisions are
common when major changes are made to payment methods, the current gaps in
information mean that HCFA should be prepared to develop and implement
substantial improvements. OASIS, therapy service, and visit length data
should help define what services beneficiaries receive for specific clinical
conditions. These data should be analyzed to determine whether the unit of
payment, the level of payments, and the case-mix adjustment method need
refinement. These data could also be used to assess the relationship between
service use and appropriate patient outcomes.

In order to minimize unintended consequences on beneficiaries, HHAs, and
Medicare, and to narrow information gaps in the PPS design, the
Administrator of HCFA should take the following actions:

ï¿½ Ensure that adequate resources are devoted to utilization monitoring and
medical review to ensure that Medicare does not make inappropriate payments
for home health services and that quality of care is not compromised.

ï¿½ Incorporate a risk-sharing arrangement into the PPS design, consistent
with methods tested in the demonstration, until available analyses indicate
that it is no longer needed to protect beneficiaries, HHAs, or the Medicare
program.

ï¿½ Modify the PPS design, as appropriate, on the basis of continued study of
the variations in service use and patient needs and the effects of the
change in payment method on service use.

In commenting on a draft of this report, HCFA agreed overall with our
exposition of Medicare's home health care benefit; the research and findings
funded by HCFA; the difficulties inherent in changing Medicare's payment
method; and that careful monitoring of HHA, beneficiary, and program
experience will be needed under the PPS. The agency agreed with two of our
recommendations but raised concerns about our recommendation to incorporate
a risk-sharing arrangement into the payment system. It also commented about
some of our concerns regarding the overall design of the PPS. HCFA also made
one technical comment, which we have incorporated.

HCFA agreed with our first recommendation to ensure that adequate resources
are devoted to utilization monitoring and medical review. In its comments,
HCFA outlined its planned efforts to (1) ensure that patient classification
and billing data are accurate and payments are appropriate and (2) provide
quick feedback on beneficiary outcomes and impacts for use in future PPS
refinements. Its activities include using OASIS data to check the accuracy
of data reporting and payments and ensure that HHA services properly address
the identified needs of beneficiaries; conducting medical reviews of claims;
and creating an aggressive surveillance system to assess the impact of the
payment changes. It plans to compile claims information on a real-time basis
to improve its ability to identify significant changes in provider behavior.
Furthermore, it will target its investigative efforts on areas identified as
potential vulnerabilities to ensure that the payment system is being
implemented correctly and that agencies are responding to it appropriately.
We support HCFA's efforts to ensure data and payment accuracy; nevertheless,
we urge HCFA to devote sufficient resources to review efforts to detect
unnecessary episodes and flag underservice within episodes on an ongoing
basis as well as at the start of the PPS.

HCFA raised two major concerns with our second recommendation to incorporate
risk sharing into the PPS design. First, it believes that such a policy is
not needed, given the adjustments included in the PPS that, in combination
with its monitoring activities, it believes will be sufficient to protect
agencies, beneficiaries, and the program. These adjustments include a
case-mix measurement system to calibrate payments on the basis of patient
needs, unlimited episode payments to account for long-term patients, outlier
payments for extraordinarily high-cost episodes, and significant change in
condition (SCIC) policies that vary payments when a beneficiary's condition
changes substantially during the episode. HCFA's second concern was that
implementing a risk-sharing arrangement would be operationally difficult and
could threaten meeting its deadline for the PPS. HCFA stated that a
risk-sharing arrangement would be more costly for HHAs and HFCA to
administer because it would require comparisons of payments and
provider-specific costs and require auditing of HHA costs to determine
allowable costs. Such an arrangement would also make it more difficult to
estimate payment levels to achieve budget neutrality.

While we agree with HCFA that the four payment adjustments included in the
proposal are all important to calibrate payments for individual episodes, we
believe that, given the incentives under a PPS and the historically
substantial variation in utilization of the benefit, they are insufficient
by themselves. These episode-level adjustments will help ensure that
payments for certain beneficiaries are not too extreme, but they will not be
sufficient to ensure that agencies with treatment patterns that are very
different from the average are protected from extraordinary losses or do not
gain inappropriately from extreme profits. This moderation to agency-level
losses or gains is needed until HCFA better understands geographic and
agency differences in treatment and agrees on appropriate HHA service so
that PPS payments can be calibrated appropriately and underservice can be
avoided. A risk-sharing mechanism would temper the incentives of the
proposed PPS, protect beneficiaries from underservice, and shield HHAs from
large losses that high-cost cases may engender if not adequately addressed
by the outlier policy. Coupled with the inadequacies of the proposed
case-mix adjustment method, we believe this moderation is critical to
protect beneficiaries from inadequate care and safeguard the Medicare
program from paying for services that were not needed or were not provided.
We are sympathetic to HCFA's concerns that a risk-sharing method adds
complexity to the payment calculations, and we do not believe that the
implementation of the PPS should be delayed in order to incorporate one.
However, we believe that the magnitude of potential overpayments to some
HHAs and underpayments to others warrants this added complexity.
Furthermore, we believe that the Medicare program already has experience in
administering each component of a risk-sharing approach. For example, for
many providers, Medicare has estimated current-year costs to adjust current
payments. In addition, HCFA has used payment methods that blend prospective
rates with provider-specific costs, and it has implemented complex budget
neutrality requirements in other payment policies. Though HCFA may need to
proceed with its plans so that it can expedite replacing the IPS, we believe
it should consider incorporating a risk-sharing arrangement in the future.

HCFA agreed with our third recommendation to modify the PPS design, as
appropriate, on the basis of experience under the PPS and continued research
on the variation in service use and patient needs. HCFA will continue to
refine its case-mix measurement system and will evaluate whether the
15-minute billing data could be used in this refinement. Likewise,
examination of data on users with multiple episodes may suggest different
episode lengths or case-mix groupings for the population of long-term users.

HCFA also discussed some of our general concerns expressed in the report:

ï¿½ We noted that the proposed case-mix adjustment method is heavily dependent
on the level of therapy services provided, which can be manipulated by HHAs
to boost payments. In its discussion of the case-mix adjustment method, HCFA
notes that therapy services are an important component of home health care
and that these home health patients are likely to fall into two
groups--those who need significant amounts of therapy and those who do not.
We agree that distinguishing among patients on the basis of their service
need is appropriate. However, this case-mix adjustment method distinguishes
between those who use therapy services and those who do not. We caution
against relying on therapy service use to designate home health case-mix
categories because of the financial incentives this creates for
inappropriate service provision. We agree that the payment system needs to
appropriately account for therapy services; therefore, we urge HCFA to
refine the case-mix adjustment method so that it reflects patient needs, not
service provision. We also agree that it will be important to monitor and
assess how well the case-mix adjustment method accounts for the costs of
long-stay patients. These beneficiaries make up a substantial share of all
users, yet because they use fewer therapy services than other patients and
their experience was not used in developing the case-mix adjustment method,
refinements may be needed.

ï¿½ We also raised concerns that using 60 days as the length of the episode
may not be appropriate for patients with shorter lengths of stay. HCFA noted
that many such stays are not predictable; that the 60-day length is
appealing from an administrative perspective; and that a shorter period
would undermine the notion of an "episode," while a longer period would
result in overpayments for many patients. We agree that beneficiary needs
for services are not always predictable from the outset and that longer
episode lengths would be likely to result in considerable overpayments for
many cases. However, because the majority of home health episodes last 60
days or less, and a considerable share of stays are under 30 days, we
believe that HCFA needs to evaluate current service patterns and make
refinements to the episode length if necessary.

ï¿½ We are also concerned that agencies will have an incentive to provide
enough visits to qualify for the full episode payment rather than the
low-utilization payment adjustment. HCFA agrees that the low-utilization
payment adjustment policy is likely to increase service provision and
proposes a behavioral offset, not discussed in the proposed rule, that would
decrease payments for all cases to account for added spending resulting from
an increased number of episodes. HCFA also indicates that this will be a
payment area that will be closely monitored. We concur that this is a
vulnerability in the PPS and that this situation warrants monitoring, but we
are unsure whether reducing the average payment through a behavioral offset
adjustment is appropriate.

ï¿½ Finally, HCFA said that the report did not adequately discuss that
historical utilization is not necessarily an appropriate basis for setting
PPS rates because of services that were not medically necessary or lacked
supporting documentation. We agree and have noted this in the text.
Moreover, because historical utilization was used as the basis for the PPS,
we believe this is further support for incorporating risk sharing into its
design.

HCFA's comments appear in their entirety as appendix III.

We are sending copies of this report to the Honorable Nancy-Ann Min DeParle,
the Administrator of HCFA; interested congressional committees; and other
interested parties. We will also make copies available to others upon
request.

If you have any questions about this report, please call me or Laura Dummit,
Associate Director, at (202) 512-7114. Major contributors to this report
included Carol Carter, Jean Chung, and James E. Mathews.

William J. Scanlon
Director, Health Financing and
Public Health Issues

List of Addressees

The Honorable William V. Roth, Jr.
Chairman
The Honorable Daniel Patrick Moynihan
Ranking Minority Member
Committee on Finance
United States Senate

The Honorable Arlen Specter
Chairman
The Honorable Tom Harkin
Ranking Minority Member
Subcommittee on Labor, Health and
Human Services, and Education
Committee on Appropriations
United States Senate

The Honorable Tom Bliley
Chairman
The Honorable John D. Dingell
Ranking Minority Member
Committee on Commerce
House of Representatives

The Honorable Bill Archer
Chairman
The Honorable Charles B. Rangel
Ranking Minority Member
Committee on Ways and Means
House of Representatives

The Honorable John Edward Porter
Chairman
The Honorable David R. Obey
Ranking Minority Member
Subcommittee on Labor, Health and
Human Services, Education and
Related Agencies
Committee on Appropriations
House of Representatives

Scope and Methodology

To develop the information for this study, we examined contracts awarded as
described in HCFA's Active Projects Reports from 1987 through 1998.22 We
included research projects that were conducted as part of the Home Health
Agency Prospective Payment Demonstration and projects that addressed issues
related to home health payment, such as quality assurance, case-mix
measurement and adjustment, and the relationship between service volume and
patient outcome. We excluded home health research projects that did not have
direct implications for a payment system (see app. II). We summarized the
research and demonstration findings and tabulated the costs of these
projects but did not evaluate the scope, methodology, or findings of the
projects.

Our analysis of HCFA's proposed PPS was based on the proposed rule23 and
briefings and discussions with HCFA officials. We also examined the project
reports prepared by Abt Associates, Inc., on the case-mix measurement system
and followed up on several methodological issues.

HCFA Projects to Develop a Prospective Payment System for Home Health
Agencies

Continued

           Project            Expenditures             Description
 Home Health Agency
 Prospective Payment
 Demonstration
                                              This project implemented and
                                              monitored the design of phase
                                              I of the demonstration. The
                                              demonstration tested
                                              prospectively set payments
                                              per visit by type of
                                              discipline (that is, skilled
                                              nursing, home health aide,
                                              physical therapy,
                                              occupational therapy, speech
 Implementation of the Home                   therapy, and medical social
 Health Agency Prospective                    services). HHAs that agreed
 Payment Demonstration.                       to participate were randomly
                                              assigned to either the
 Project No.: 500-90-0024                     per-visit prospective payment
                              $1,608,319a     group or the control group,
 Contractor: Abt Associates,                  which was paid according to
 Inc.                                         existing Medicare payment
                                              rules. HHAs paid on a
 Period: 1990-95                              per-visit basis shared the
                                              financial risks and rewards
                                              with Medicare--agencies were
                                              reimbursed for any losses
                                              greater than 5 percent of
                                              their Medicare-allowable
                                              costs and gave back any
                                              profits greater than 5
                                              percent of that amount.
                                              Forty-seven agencies
                                              participated in the
                                              demonstration for 3 years.

 Evaluation of the Home                       This project evaluated the
 Health Prospective Payment                   results of phase I of the
 Demonstration                                demonstration. The findings
                                              indicated that a per-visit
 Project No.: 500-90-0047     $3,406,668a     PPS had no significant effect
                                              on quality of care, selection
 Contractor: Mathematica                      and retention of patients,
 Policy Research, Inc.                        cost per visit, visit volume,
                                              use of non-Medicare services,
 Period: 1990-95                              and use and reimbursement of
                                              Medicare-covered services.
                                              This project reviewed the
                                              quality of care provided by
 Quality Review for Phase I                   HHAs participating in phase I
 of the Home Health Agency                    of the demonstration. Nurses
 Prospective Payment                          reviewed patient records for
 Demonstration                                a sample of Medicare
                                              beneficiaries. They found
 Project No.: 500-91-0096                     that quality of care was
                              $1,499,085a     unaffected by per-visit
 Contractor: New England                      prospective payment. Patient
                                              access to care was also
 Research Institute, Inc.                     unaffected by the payment
                                              method. Patients treated by
 Period: 1991-94                              agencies paid on a per-visit
                                              basis were generally similar
                                              to patients treated by
                                              control agencies.
                                              This project, implementing
                                              and monitoring phase II of
                                              the demonstration, tests a
                                              per-episode prospective
                                              payment approach. Agencies
                                              are paid a prospective amount
                                              for a patient's first 120
 Phase II Implementation of                   days of care and a per-visit
 the Home Health Agency                       amount for subsequent care
 Prospective Payment                          during the episode. Payment
 Demonstration                                rates are based on each
                                              agency's costs in a base
 Project No.: 500-95-0011     $1,811,184b     year. Loss protection and
                                              profit-sharing provisions
 Contractor: Abt Associates,                  mitigate the financial risks
 Inc.                                         for participating agencies.
                                              Ninety-one agencies from 5
 Period: 1995-2000                            states--CA, FL, IL, MA, and
                                              TX--were randomly assigned to
                                              either the prospective
                                              payment or the control group.
                                              At the participating
                                              agencies' request, phase II
                                              has been extended until the
                                              HHA PPS is implemented in
                                              October 2000.
                                              This program evaluation
                                              addresses two key issues
                                              relating to a per-episode
                                              payment system: program
                                              impact and HHA decisions and
                                              operations. The evaluation
                                              estimates the effect of the
                                              demonstration PPS on cost,
                                              service use, access, and
                                              quality. Evaluation results
                                              to date are based on
 Evaluation of Phase II of                    first-year data from 51,000
 the Home Health Agency                       home health episodes from 85
 Prospective Payment                          demonstration agencies. On
 Demonstration                                average, the cost per episode
                                              was lower in PPS HHAs by
 Project No.: 500-94-0062     $3,732,642b     $419, or 13 percent; however,
                                              the cost per visit was higher
 Contractor: Mathematica                      than in the control HHAs. The
 Policy Research, Inc.                        number of visits in the
                                              120-day period was 17 percent
 Period: 1994-99                              lower for patients in PPS
                                              agencies, and there was
                                              little change in the
                                              proportion of visits across
                                              health disciplines. The
                                              average length of episodes
                                              under the PPS decreased by 15
                                              percent. The evidence
                                              suggests that there were no
                                              increases in the use of other
                                              health care services and that
                                              patient outcomes were not
                                              compromised.
                                              This project provides for the
                                              development and
 Quality Assurance for Phase                  implementation of a quality
 II of the Home Health Agency                 review mechanism for use by
 Prospective Payment                          HHAs participating in phase
 Demonstration                                II of the demonstration. All
                                              participating agencies are
 Project No.: 500-95-0028                     required to collect patient
                                              status data at prescribed
 Contractor: Center for       $2,799,265b     intervals. Data on patient
                                              outcomes are provided to
 Health Policy Research                       individual agencies that, in
                                              turn, may use this
 Period: 1995-2000                            information to make
                                              adjustments to treatment
                                              protocols. In addition, the
                                              project will compare patient
                                              outcomes for PPS and control
                                              agencies.
 Subtotal                     $14,857,163
 Related research projects
                                              This project developed a
                                              method for classifying
                                              patients to predict resource
                                              requirements and measure
                                              treatment outcomes of
                                              Medicare patients in HHAs.
 Develop and Demonstrate a                    Data on 73 variables were
 Method for Classifying Home                  collected from the medical
 Health Patients to Predict                   records of about 9,000
 Resource Requirements and to                 recently discharged Medicare
 Measure Outcomes                             home health patients. The
                                              analysis indicated that home
 Project No.: 17-C-98983/3    $967,836a       health care was primarily
                                              provided to a white,
 Contractor: Georgetown                       suburban, young-elderly
                                              population. Less advantaged
 University School of Nursing                 and higher-risk patients were
                                              more likely to receive care
 Period: 1987-91                              in nursing homes. Findings on
                                              resource use suggested that
                                              nursing diagnoses and nursing
                                              interventions were better
                                              predictors of home health
                                              utilization than were
                                              functional status or medical
                                              diagnosis.
                                              This project synthesized the
                                              research literature on
                                              prospective payment and
 Analysis of Home Health                      examined outlier cases and
 Costs and Service                            possible volume adjustments
 Utilization Issues                           using Medicare claims.
                                              Findings suggested that a
 Project No.: 99-C-99169/5                    120-day home health episode
                              $189,607c       was appropriate for the
 Contractor: University of                    demonstration. The study also
 Minnesota Research Center                    found significant regional
                                              variation in the average
 Period: 1991-92                              length of episodes. This
                                              study helped shape phase II
                                              of the Home Health Agency
                                              Prospective Payment
                                              Demonstration.
                                              This project developed and
                                              tested outcome-based measures
 Development of Outcome-Based                 or indicators of quality for
 Quality Measures for Home                    Medicare home health
 Health Services                              services. Outcomes were
                                              developed according to
 Project No.: 500-88-0054                     different types of patient
                              $2,699,298a     care needs defined by patient
 Contractor: Center for                       condition taxonomy termed
                                              Quality Indicator Groups.
 Health Policy Research                       Using longitudinal data
                                              collected on about 3,000
 Period: 1988-94                              patients, the investigators
                                              tested the reliability,
                                              validity, and utility of each
                                              outcome measure.
                                              This study addressed regional
                                              differences in the
                                              utilization of home health
                                              services. The authors
                                              examined factors such as
                                              patient characteristics,
                                              supply of home health
 Regional Variation in Home                   agencies and staff, and
 Health Episode Length and                    availability of alternatives
 Number of Visits per Episode                 to home health care across
                                              high-use regions and low-use
 Project No.: 500-89-0047/38  $168,600c       regions. In the highest-use
                                              regions, HHAs served very
 Contractor: Lewin/VHI, Inc.                  frail patients who may not
                                              have access to alternative
 Period: 1993-94                              sources of care and few
                                              resources to purchase those
                                              available. By contrast,
                                              agencies in low-use regions
                                              served much less frail and
                                              less chronically ill patients
                                              who may have better access to
                                              alternative sources of care.
                                              The objective of this project
                                              was to develop and consider
                                              options for controlling home
                                              health expenditures. The
                                              first phase of the project
                                              used secondary data to
                                              examine the composition of
                                              the Medicare home health
 Sources of Medicare Home                     expenditures from 1985 to
 Health Expenditure Growth:                   1989 and 1989 to 1991 (that
 Implications for Control                     is, growth in number of
 Options                                      people served, visits per
                                              person, mix of visits, and
 Project No.: 17-C-90107/1                    visit charges; and
                              $385,764a
                                              attributing growth to types
 Contractor: Brandeis                         of agencies by auspice and
 University/Heller Graduate                   scale). The second phase
 School                                       examined data from the
                                              Regional Home Health
 Period: 1992-95                              Intermediary database to
                                              measure variation in types of
                                              patients served. Results
                                              indicated that the rise in
                                              home health expenditures was
                                              primarily driven by an
                                              increase in the number of
                                              home health visits received
                                              per patient.
                                              This demonstration is to
                                              develop outcome-oriented
                                              quality assurance measures
                                              and promote continuous
                                              quality improvement in HHAs.
                                              It is designed to serve two
                                              purposes: to increase HCFA's
 Design and Implementation of                 capacity to assess the
 Medicare Home Health Quality                 quality of Medicare home
 Assurance Demonstration                      health care services and to
                                              improve HHAs'outcomes. The
 Project No.: 500-94-0054                     quality assurance approach
                              $3,234,881b     would complement existing
 Contractor: Center for                       home health certification and
                                              review programs and could be
 Health Policy Research                       used with current survey and
                                              certification approaches. The
 Period: 1994-99                              study's conceptual framework
                                              is based on home health
                                              outcome measures developed
                                              under the HCFA-funded study
                                              entitled "Development of
                                              Outcome-Based Quality
                                              Measures in Home Health
                                              Services."
                                              This study examines the
                                              relationship between the
                                              volume of home health
 Maximizing the Cost                          services provided by HHAs and
 Effectiveness of Home Health                 patient outcomes. The study
 Care: The Influence of                       will determine whether upper-
 Service Volume and                           and lower-volume thresholds
 Integration With Other Care                  exist that can be used to
 Settings on Patient Outcomes                 define the range of services
                                              most beneficial to patients.
 Project No.: 17-C-90435/8    $1,496,245b     In addition, the study will
                                              test whether it is possible
 Contractor: Center for                       to improve patient care and
                                              control costs by
 Health Policy Research                       strengthening the role of the
                                              physician and better
 Period: 1994-99                              integrating home health care
                                              with other services during an
                                              episode of care. Interim
                                              findings have not been
                                              released.
                                              This project will develop a
                                              case-mix classification
                                              system for the national home
                                              health PPS. The resulting
                                              case-mix system will be based
                                              on serial 60-day episodes.
 Case-Mix Adjustment for a                    The Outcome and Assessment
 National Home Health                         Information Set (OASIS),
 Prospective Payment System                   which was developed for
                                              outcome-based quality
 Project No.: 500-96-0003/02                  assurance and improvement for
                              $2,966,524b
                                              Medicare HHAs, will be
 Contractor: Abt Associates,                  examined to determine whether
 Inc.                                         the data elements can be used
                                              to construct a case-mix
 Period: 1996-99                              classification system.
                                              Significant features of this
                                              project include its
                                              measurement of resource use
                                              and emphasis on easily
                                              understandable patient
                                              groupings.
 Subtotal                     $12,108,755
 Total                        $26,965,918

aAmount represents actual expenditures.

bAmount represents amount allocated for an ongoing project.

cNumber represents amount allocated for project. HCFA could not provide
actual expenditures for this project.

Sources: Active Projects Report: Research and Demonstrations in Health Care
Financing (1987-98); Report to the Congress from the Secretary of Health and
Human Services (Washington, D.C.: Jan. 4, 1999); and discussions with HCFA
officials.

Comments From the Health Care Financing Administration

(101792)

Table 1: HCFA-Funded HHA Research Spending, 1988-99, by Project Category 16
  

1. "Medicare Program: Prospective Payment System for Home Health Agencies,"
proposed rule, Federal Register, Vol. 64, No. 208 (Oct. 28, 1999).

2. Therapy services include physical, speech, and occupational therapies.

3. A beneficiary is homebound when he or she has a condition that results in
a normal inability to leave home except with considerable and taxing effort,
and absences from home are infrequent or of relatively short duration or are
attributable to receiving medical treatment.

4. "Part-time or intermittent" means skilled nursing and home health aide
services furnished any number of days per week as long as they are furnished
fewer than 8 hours each day and for 28 or fewer hours each week.

5. These data on service use include Medicare fee-for-service beneficiaries
only.

6. A 1999 study found that 19 percent of the services in four states were
improper or highly questionable and did not meet Medicare reimbursement
requirements. HHS Office of Inspector General, Review of Medicare Home
Health Services in California, Illinois, New York and Texas, A-04-99-01194
(Washington, D.C.: HHS, Nov. 1999).

7. MedPAC, Report to the Congress: Context for a Changing Medicare Program
(Washington, D.C.: MedPAC, June 1998).

8. For Medicare coverage, aide services include personal care (such as help
with dressing and bathing), simple wound dressing changes and assistance
with medications that do not require the skills of a licensed nurse, routine
exercises, and routine care of orthotic and prosthetic devices.

9. Agencies may be not-for-profit (or voluntary, including visiting nurse
associations), for-profit (or proprietary), or government-owned.

10. Medicare: Home Health Utilization Expands While Program Controls
Deteriorate ( GAO/HEHS-96-16, Mar. 27, 1996).

11. Originally, the IPS per-visit limit was based on 105 percent of the
national median per-visit cost. Section 5101(b) of the Omnibus Consolidated
and Emergency Supplemental Appropriations Act, 1999 (P.L. 105-277) revised
the IPS and increased the per-visit limit to 106 percent of the national
median per-visit cost.

12. For an agency that had been in operation for a full year before October
1, 1994, the per-beneficiary amount is calculated as 98 percent of a blend
of 75 percent of the agency's own fiscal year 1994 average per-beneficiary
payment and 25 percent of the comparable regional average. The
per-beneficiary amount for new agencies--those that had not participated in
Medicare for a full year by October 1994--equals 100 percent of the 1994
national median per beneficiary payment. The Omnibus Consolidated and
Emergency Supplemental Appropriations Act, 1999 (P.L. 105-277, sec. 5105
(a)) made several changes to the revenue cap. For HHAs with per-beneficiary
amounts less than the national median, limits were increased by one-third of
the difference between their amount and the national median. The cap for new
HHAs (as defined by the BBA) was increased from 98 to 100 percent of the
national median. Further, HHAs that opened after October 1, 1998, have
per-beneficiary limits equal to 75 percent of the national median, reduced
by 2 percent. The following year, the Medicare, Medicaid, and SCHIP Balanced
Budget Refinement Act of 1999 (P.L. 106-113, sec. 302) increased the
per-beneficiary limits by removing the 2 percent reduction for all
providers, thus increasing Medicare payments.

13. The IPS is intended to account for historical differences in treatment
patterns and patient mix through the use of each HHA's own historical costs.
This may result in inefficient or inappropriate service use being
incorporated in the payments. In addition, the IPS' use of the national
median for new agencies means that there is no adjustment for differences in
these agencies' own treatment patterns or patient mix.

14. One such method would compare a provider's total Medicare payments with
its actual allowable Medicare costs. For example, if payments exceed actual
costs by 10 percent or more, these profits would be shared with Medicare.
But if costs exceeded payments by 10 percent or more, a portion of these
losses would be shouldered by Medicare.

15. OASIS contains patient-level data on medical condition; demographic
characteristics; supportive assistance; sensory, neurological, emotional,
and behavioral status; activities of daily living; and instrumental
activities of daily living. All agencies were required to begin using OASIS
effective July 19, 1999.

16. The proration calculations assume that costs are evenly distributed
throughout an episode.

17. The Medicare, Medicaid and SCHIP Refinement Act of 1999 delays the 15
percent reduction in the payments required under the PPS by the BBA until 12
months after implementation of the PPS. It also requires the Secretary of
Health and Human Services to report to the Congress within 6 months of
implementation of the PPS on the need for the 15 percent or other reduction.

18. HCFA reimburses each test agency for 99 percent of its losses in the
first demonstration year, 98 percent of its losses in the second year, and
97 percent in the third and subsequent demonstration years, subject to total
payments being no greater than Medicare-allowable costs. HCFA retains 25
percent of an agency's profit, if it is between 5 percent and 15 percent of
the HHA's total allowable costs, and 100 percent thereafter.

19. First-year results from phase II showed a 17 percent reduction in the
number of visits provided by study agencies, compared with agencies in the
control group.

20. This study is scheduled to be completed by April 2000.

21. Balanced Budget Act: Implementation of Key Medicare Mandates Must Evolve
to Fulfill Congressional Objectives (GAO/T-HEHS-98-214, July 16, 1998 ) and
Medicare Post-Acute Care: Better Information Needed Before Modifying BBA
Reforms (GAO/T-HEHS-99-192, Sept. 15, 1999 ).

22. Active Projects Report: Research and Demonstrations in Health Care
Financing is produced annually and summarizes active intramural and
extramural projects.

23. "Medicare Program: Prospective Payment System for Home Health Agencies,"
Proposed Rule, Federal Register, Vol. 64, No. 208 (Oct. 28, 1999).
*** End of document. ***