Skilled Nursing Facilities: Medicare Payments Need to Better Account for
Nontherapy Ancillary Cost Variation (Letter Report, 09/30/1999,
GAO/HEHS-99-185).

Pursuant to a congressional request, GAO provided information the
Medicare payments for skilled nursing facilities' (SNF) services under
the new prospective payment system (PPS), focusing on: (1) whether the
SNF payment rates incorporate the costs of nontherapy ancillary
services; and (2) analyzing the PPS design and nontherapy ancillary cost
variation to assess whether payments are distributed appropriately.

GAO noted that: (1) SNF PPS rates were calculated using the full
historical costs of nontherapy ancillary services, updated for
inflation; (2) costs associated with unnecessary care and improperly
billed services may have boosted these historical costs above what was
warranted, resulting in generous PPS payment rates; (3) however, the
Balanced Budget Act of 1997 explicitly reduced payments by not
accounting for total cost increases, raising concerns about whether the
adjustment process adequately accounts for cost increases that occurred
between the base-year and the first PPS payment year; (4) although the
case-mix adjustments to payments for each patient under PPS is intended
to account for changes in costs due to shifts in the mix of treatments,
evidence indicates that for some types of patients, these adjustments
may not be adequate; (5) a full audit of SNF base-year and current costs
and medical reviews of service provision would be needed to establish
the actual relationship between the costs of medically appropriate care
and payments; (6) nontherapy ancillary costs were not used to develop
the payment adjusters that raise or lower the average payment to account
for resource need differences across patients; (7) as a result, per diem
payments may not be adequate for types of patients who are likely to
incur high nontherapy ancillary costs or may be excessive for those
groups of patients with low expected nontherapy ancillary costs; (8) in
1995, nontherapy ancillary service costs comprised 16 percent of total
daily SNF costs, indicating that failure to adequately account for
nontherapy ancillary cost variation could result in substantial under-
or overpayments; (9) this potential misallocation could contribute to
beneficiary access problems if certain patients are identified prior to
SNF admission as requiring nontherapy ancillary costs higher than the
PPS rate; (10) the Health Care Financing Administration is investigating
possible refinements to PPS that could address these problems; and (11)
in the meantime, increasing SNF payments will not improve the allocation
of the payments but will only increase program outlays and possible
overpayments to certain facilities.

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

 REPORTNUM:  HEHS-99-185
     TITLE:  Skilled Nursing Facilities: Medicare Payments Need to
	     Better Account for Nontherapy Ancillary Cost Variation
      DATE:  09/30/1999
   SUBJECT:  Patient care services
	     Skilled nursing facilities
	     Prices and pricing
	     Health care costs
	     Health insurance cost control
	     Medical services rates
IDENTIFIER:  Medicare Prospective Payment System
	     Medicare Program
	     Medicare Skilled Nursing Facilities Program

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** 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.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************

Cover
================================================================ COVER

Report to the Chairman, Subcommittee on Health, Committee on Ways and
Means, House of Representatives

September 1999

SKILLED NURSING FACILITIES -
MEDICARE PAYMENTS NEED TO BETTER
ACCOUNT FOR NONTHERAPY ANCILLARY
COST VARIATION

GAO/HEHS-99-185

SNF Nontherapy Ancillary Costs

(101791)

Abbreviations
=============================================================== ABBREV

  BBA - Balanced Budget Act of 1997
  HCFA - Health Care Financing Administration
  HHS - Department of Health and Human Services
  MDS - Minimum Data Set
  OIG - Office of the Inspector General
  PPS - prospective payment system
  RUG - Resource Utilization Group
  SNF - skilled nursing facility

Letter
=============================================================== LETTER

B-283595

September 30, 1999

The Honorable William M.  Thomas
Chairman
Subcommittee on Health
Committee on Ways and Means
House of Representatives

Dear Mr.  Chairman: 

Over the last decade, Medicare payments for skilled nursing facility
(SNF) services have increased dramatically, with spending rising on
average over 23 percent per year between 1990 and 1996.  To curb this
growth, the Balanced Budget Act of 1997 (BBA) replaced Medicare's
existing cost-based payment methodology with a prospective payment
system (PPS).\1 PPS payments for SNFs--which provides facilities an
all-inclusive daily payment, adjusted for the complexity and expected
care needs of each patient--began being phased in on July 1, 1998.\2

Concerns have been raised about whether the rates under the new
payment system account for disparate patient costs, particularly high
or low nontherapy ancillary service costs, which include drugs,
laboratory tests, radiology procedures, respiratory therapy, medical
supplies, intravenous therapy, and other nonroutine services.  These
concerns have prompted legislative proposals to raise SNF PPS
payments for all or some types of patients.  In this context, you
asked us to (1) assess whether the SNF payment rates incorporate the
costs of nontherapy ancillary services and (2) analyze the PPS design
and nontherapy ancillary cost variation to assess whether payments
are distributed appropriately. 

To complete this study, we reviewed the provisions of BBA and the
Health Care Financing Administration's (HCFA) interim rule and final
rule on the prospective payment system and consolidated billing for
SNFs, which took effect on July 1, 1998, to determine the extent to
which nontherapy ancillary cost variation was accounted for in the
payment rates.  We also analyzed provider cost reports from fiscal
year 1995 (the most recent available data) to estimate the average
costs per day, the components of daily costs, and the variations in
costs across Medicare-certified SNFs.  We conducted our work between
December 1998 and August 1999 in accordance with generally accepted
government auditing standards.  (For a detailed discussion of our
scope and methodology, see app.  I.)

--------------------
\1 P.L.  105-33, section 4432(a). 

\2 There is a 3-year transition to the new payment system during
which payments are a blend of facility-specific and national average
per diem rates.  In the first year, payments are 75-percent
facility-specific; 50-percent facility-specific in the second year;
and 25-percent facility-specific in the third.  The facility-specific
portion is based on each facility's updated 1995 costs.  SNFs are
being phased in according to the start of their fiscal year. 

   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

SNF PPS rates were calculated using the full historical costs of
nontherapy ancillary services, updated for inflation.  Costs
associated with unnecessary care and improperly billed services may
have boosted these historical costs above what was warranted,
resulting in generous PPS payment rates.  However, BBA explicitly
reduced payments by not accounting for total cost increases, raising
concerns about whether the adjustment process adequately accounts for
cost increases that occurred between the base-year and the first PPS
payment year.  Although the case-mix adjustments to payments for each
patient under PPS are intended to account for changes in costs due to
shifts in the mix of treatments, evidence indicates that for some
types of patients, these adjustments may not be adequate.  A full
audit of SNF base-year and current costs and medical reviews of
service provision would be needed to establish the actual
relationship between the current costs of medically appropriate care
and payments. 

Nontherapy ancillary costs were not used to develop the payment
adjusters that raise or lower the average payment to account for
resource need differences across patients.  As a result, per diem
payments may not be adequate for types of patients who are likely to
incur high nontherapy ancillary costs or may be excessive for those
groups of patients with low expected nontherapy ancillary costs.  In
1995, nontherapy ancillary service costs comprised 16 percent of
total daily SNF costs, indicating that failure to adequately account
for nontherapy ancillary cost variation could result in substantial
under- or overpayments.  This potential misallocation could
contribute to beneficiary access problems if certain patients are
identified prior to SNF admission as requiring nontherapy ancillary
costs higher than the PPS rate. 

HCFA is investigating possible refinements to PPS that could address
these problems.  In the meantime, increasing SNF payments will not
improve the allocation of the payments but will only increase program
outlays and possible overpayments to certain facilities. 

   BACKGROUND
------------------------------------------------------------ Letter :2

Medicare covers up to 100 days of care in a SNF for beneficiaries who
need skilled nursing or rehabilitative care on a daily basis
following a hospital stay of at least 3 days.  Medicare pays for
routine services, such as room and board, skilled nursing care,
social services, and supplies and equipment.  It also pays for
ancillary services, including physical, occupational, respiratory,
and speech therapies; laboratory services; radiology procedures; and
drugs. 

      COST-BASED REIMBURSEMENT HAD
      FEW INCENTIVES TO CONSTRAIN
      COSTS
---------------------------------------------------------- Letter :2.1

Prior to the implementation of PPS, Medicare paid SNFs on a
reasonable cost basis.  Routine nursing and room and board costs were
paid up to specified limits, with higher limits applied to
hospital-based SNFs than to freestanding ones.  New providers were
exempt from the cost limits for up to their first 4 years of
operation.  In addition, providers that demonstrated higher than
average costs as a consequence of atypical patients or patterns of
care could be granted exceptions to the routine cost limits. 

Unlike payments for routine costs, payments for ancillary (therapy
and nontherapy) costs were not subject to limits.  Services had to
meet medical necessity criteria, but there was little Medicare review
of their use.  As a result, facilities had few incentives to
constrain costs or to restrict ancillary service provision to only
necessary services--increases in ancillary service costs increased
payments.  In fact, payments for ancillary services increased 17 to
20 percent annually between 1992 and 1995, compared with 5 to 7
percent for routine services. 

Despite the growth in Medicare expenditures, funding for program
safeguards decreased by 50 percent between 1989 and 1995.  Limited
auditing of cost reports and medical review of claims raised concerns
that ancillary cost growth was not entirely due to increases in the
service needs of Medicare beneficiaries. 

      PPS IMPLEMENTED TO CONTROL
      SPENDING
---------------------------------------------------------- Letter :2.2

To curb the rise in Medicare SNF spending, BBA required HCFA to
implement a PPS for SNFs.  HCFA designed an all-inclusive per diem
payment approach to replace the cost-based reimbursement methodology
(see app.  II).  The per diem payment, which is adjusted for
differences in the resource needs of patients and for geographic
differences in labor costs, covers all routine, ancillary, and
capital costs incurred in treating a SNF patient. 

The per diem rate has three components--one for nursing (nursing
care, social services, and nontherapy ancillary services), one for
therapies (physical, occupational, and speech), and a non-case-mix
services component (for example, capital, maintenance, dietary)--that
are totaled to determine the overall payment.  The nursing and
therapy components are adjusted upward for patients who are expected
to be more resource-intensive--and thus more costly to care for--or
downward for patients who are expected to be less resource intensive
than average.  The non-case-mix component covers costs that are
assumed to be uniform across all patients and, therefore, is not
adjusted. 

The adjustments of nursing and therapy payments are based on a
case-mix classification system--Resource Utilization Group, version
III, or RUG-III--developed by HCFA contractors.  The system comprises
44 distinct patient groups distinguished by patient clinical
condition, functional status, and expected use of certain types of
services.  Each case-mix group has a corresponding "nursing relative
weight" that reflects the costliness of providing services to
patients in that group relative to the average costliness of patients
across all groups.  Of the 44 RUG-III groups, 14 describe patients
who require substantial therapy services and have an associated
"therapy relative weight." The remaining case-mix groups are assumed
to require a minimal amount of therapy services and are paid a fixed
non-case-mix therapy payment.  The payment for each day of care for a
patient is the sum of three partsthe nursing component (the product
of the nursing base rate and the nursing relative weight for the
appropriate RUG-III group), the therapy component (the product of the
therapy base rate and the appropriate relative weight or a flat
amount, depending on the RUG-III category), and the non-case-mix
amount (see fig.  1).  (App.  II contains a more complete discussion
of the payment amount calculation.)

   Figure 1:  Overview of SNF PPS
   Rate Calculation

   (See figure in printed
   edition.)

   AVERAGE SNF PAYMENTS INCLUDE
   HISTORICAL NONTHERAPY ANCILLARY
   COSTS
------------------------------------------------------------ Letter :3

HCFA used 1995-reported SNF costs, including those for nontherapy
ancillary services, as the basis for the 1999 base payments under
PPS.  Given the lack of incentives under the prior payment approach
to control ancillary costs, the 1995 costs may be higher than
warranted due to inefficient service provision, the costs of
unnecessary care, and improper billings.  On the other hand, some
contend that the method for updating the 1995 costs to 1999 levels
underestimated appropriate cost increases over that period.  Without
a systematic review of SNF costs, service provision, and payments, it
is not possible to determine the appropriateness of the resulting
1999 base rates. 

The base PPS payment amounts include (1) the per diem routine,
ancillary, and capital costs reported by SNFs in fiscal year 1995 and
(2) an estimate of the per diem average amount paid during that year
for ancillary services furnished to SNF patients by external
providers (such as outside laboratories.) The 1995 nontherapy
ancillary costs were thus fully included in the calculation of the
base rates.  Total costs were updated for inflation between the 1995
base year and 1999 by the SNF market-basket index minus 1 percent, as
required by BBA.\3

There is evidence that base year spending was higher than it should
have been due to unwarranted growth in ancillary expenditures and
unnecessary costs or inappropriate billing for services, which was
undetected because of minimal program oversight.  We have reported
that it is likely that the base year costs include too many services
and that the costs per service were inappropriately high.\4 Likewise,
in its review of the SNF PPS, the Department of Health and Human
Services' (HHS) Office of the Inspector General (OIG) noted that the
rate-setting process did not adequately exclude costs for medically
unnecessary care or the amount of improper SNF payments.\5 Due to
these factors, the level of overpayments is not known. 

The adequacy of the method of updating the 1995 costs to 1999 costs
has been called into question.  Some contend that actual SNF costs
rose faster than the inflation adjuster because SNFs were treating
more complex patients and providing more intensive treatments.  Some
of this increase, however, will be accounted for by the case-mix
adjustment to the payments.  To the extent that higher costs are due
to a different mix of patients than is measured by the case-mix
adjustment method, the national portion of the payments will be
higher.\6 Reflecting congressional concerns about excessive cost
increases due to inefficient or inappropriate service provision under
cost-based payments, BBA explicitly reduces SNF per diem payments by
requiring the use of an inflation adjuster that is less than the
expected increase in SNF costs, as measured by the market-basket
index. 

Additional information is required to determine the adequacy and
appropriateness of payments.  Thoroughly audited Medicare cost
reports, patient assessment data, and beneficiary claims are needed
to establish the appropriateness of facility costs, and medical
reviews of services provided to Medicare beneficiaries would
determine if any unnecessary care had been provided.  Together, this
information would provide a clearer picture of what Medicare should
be paying for services and could be used to identify and assess the
appropriateness of any cost growth that remains unaccounted for by
the inflation adjuster or the case-mix adjustment to payments. 

--------------------
\3 The market-basket index measures the annual change in the prices
of goods and services providers use in producing health care
services. 

\4 Balanced Budget Act:  Any Proposed Fee-for-Service Payment
Modifications Need Thorough Evaluation (GAO/T-HEHS-99-139, June 10,
1999). 

\5 OIG, Review of the Health Care Financing Administration's
Development of a Prospective Payment System for Skilled Nursing
Facilities, A-14-98-00350 (Washington, D.C.:  HHS). 

\6 During the transition, the facility portion of the payment is not
case-mix adjusted because it already includes facility-specific
costs. 

   PPS CASE-MIX ADJUSTMENTS MAY
   NOT APPROPRIATELY DISTRIBUTE
   SNF PAYMENTS
------------------------------------------------------------ Letter :4

The SNF case-mix-adjustment system does not directly account for the
variation in nontherapy ancillary costs across patients because only
variations in nursing time were used to establish the relative
weights for the case-mix groups.  As a result, SNF payments may not
vary consistently with the expected variation in patient costs.  This
could disadvantage those facilities that treat many patients with
high nontherapy ancillary costs and may create access problems for
patients who are identified as having high nontherapy ancillary needs
prior to admission.  The dollars at stake are substantial.  In 1995,
Medicare nontherapy ancillary costs accounted for 16 percent ($45) of
the daily costs of care.  To the extent that payments do not
adequately reflect nontherapy ancillary costs, some SNFs could
receive substantial overpayments relative to the expected costs of
their mix of patients, while others could be underpaid. 

In order to assess the adequacy of the payments for nontherapy
ancillary costs across the different case-mix categories, average
patient-level costs and average payments would need to be compared. 
These data are not yet available.\7

However, facility-level data indicate that there is a ninefold
variation in average nontherapy ancillary costs per day.  By
comparison, the relative weights used to adjust payments for these
costs only allow payments to vary by about two and a half
times--suggesting that PPS could be overpaying some facilities and
underpaying others.  Further, by comparing the range in potential
payments for nontherapy ancillary costs to facilities' costs, we
found that two-thirds of the SNFs had reported costs either below or
above that range.  This also indicates that there could be
substantial over- or underpayments under PPS to certain facilities. 

--------------------
\7 Until data are available, many elements required to classify
patients into the 44 RUG-III groupssuch as the frequency and
duration of therapies, the number of physician visits or order
changes, and activities of daily livingcannot be reproduced using
existing claims information. 

      CLASSIFICATION SYSTEM'S
      RELATIVE WEIGHTS NOT LIKELY
      TO ADEQUATELY ACCOUNT FOR
      PATIENT COST VARIATION
---------------------------------------------------------- Letter :4.1

The RUG-III classification system groups similar types of patients
based on their expected level of resource use.\8 To adjust payments,
each group is assigned two weights:  one based on the average cost of
providing nursing services to the patients in the group relative to
overall patient averages and the other based on relative therapy
costs.  Although total nontherapy ancillary costs were included in
the base nursing rate, these costs were not considered in the
calculation of the nursing relative weights.  Rather, nursing time
was used to develop the relative weights.  If nontherapy ancillary
costs are correlated with nursing time, the nursing weights will
appropriately distribute payments according to patients' nontherapy
ancillary resource needs.  If this is not the case, the payments for
some groups of patients will be too high and for others, too low. 

According to HCFA, it incorporated nontherapy ancillary costs into
the nursing base rate because its analysis showed that patients in
the RUG-III categories with high nursing relative weights tend to
have high nontherapy ancillary charges.  However, this does not
necessarily mean that weights based only on nursing time are adequate
to distribute payments for nontherapy ancillary services.  At the
time the classification system was developed, nontherapy ancillary
costs did not comprise a substantial share of SNF costs\9 --they now
do, averaging approximately 16 percent of SNF per diem costs in
1995.\10 Thus, if the relative weights do not adequately account for
these costs, the total per diem payment may not be appropriate.  For
example, some patients requiring relatively limited nursing time
might have costly nontherapy ancillary needs, such as the
administration of expensive drugs.  Without other service needs that
would place these patients in higher weighted groups, they would get
assigned to case-mix groups with lower relative weights that may not
fully reflect their high nontherapy ancillary costs.  If nursing
homes identify these patients and choose not to admit them, the
patients may need to stay in a hospital longer to receive the care
they need.  Our work and work conducted by HHS' OIG found that some
patients who require extensive services are more difficult to
place.\11 Conversely, payments may be too high for patients with
relatively low nontherapy ancillary use.  For example, an unstable
patient may require significant amounts of nursing time for
monitoring but may not be receiving treatments involving many
nontherapy ancillary resources. 

HCFA has acknowledged concerns about whether the case-mix adjustment
method appropriately accounts for nontherapy ancillary cost variation
and is sponsoring research to determine if the accuracy of the rates
could be improved by refining the RUG-III system to explicitly
incorporate nontherapy ancillary services.  HCFA also is
investigating whether relative weights based on ancillary charges,
rather than the current weights based on nursing time, would be more
appropriate for adjusting the nontherapy ancillary component of the
payment amount.  It anticipates completing these research projects by
January 1, 2000.  Any payment system refinements resulting from these
projects would be implemented starting October 1, 2000, before the
transition to the full PPS is complete. 

--------------------
\8 The need for certain nontherapy ancillary services, such as
chemotherapy, radiation therapy, and parenteral feeding, are used to
classify patients into some case-mix groups.  The costs of these
nontherapy ancillary services, however, are not used in calculating
the relative weights for these case-mix groups. 

\9 For example, pharmacy costs, the largest, was 5 percent of nursing
costs.  See Brant E.  Fries, Don P.  Schneider, and others, Refining
a Case-Mix Measure for Nursing Homes:  Resource Utilization Groups
(RUG-III), Medical Care, Vol.  32, No.  7 (1994), pp.  668-85. 

\10 PPS provides incentives for SNFs to lower their provision of
nontherapy ancillary services and to negotiate lower prices paid for
them.  As a result, the share of total costs that are attributable to
nontherapy services may have declined since 1995. 

\11 OIG, Office of Evaluation and Inspections, Early Effects of the
Prospective Payment System on Access to Skilled Nursing Facilities,
OEI-02-99-00400 (Washington, D.C.:  HHS, Aug.  1999). 

      NONTHERAPY ANCILLARY COST
      VARIATION WIDER THAN RANGE
      IN PPS PAYMENTS
---------------------------------------------------------- Letter :4.2

Measuring the effect of omitting nontherapy ancillary costs in
computing the RUG-III relative weights on patients and facilities
requires data on patient characteristics not currently available.\12
However, our analysis of facility-level information revealed that
two-thirds of facilities have average nontherapy ancillary costs that
are outside of the range of potential PPS payments.  This means that
many facilities could be over- or underpaid. 

According to our analysis, nontherapy ancillary costs averaged about
$45 per day in 1995 (see table 1).\13 Although for the majority of
SNFs, these costs averaged below $40, the most expensive providers of
these services (the top 10 percent) had daily costs of $95 or more,
while the least expensive providers (the bottom 10 percent) had costs
below $11.  (See app.  III for a more complete presentation of
facility cost variation.) Thus, facilities with the highest
nontherapy ancillary costs were nine times more expensive than the
bottom 10 percent of facilities.  Patient-level costs could vary
considerably more than these facility averages. 

                          Table 1
          
            Average Daily SNF Reported Costs for
                  Medicare Patients, 1995

                                Average per-    Percent of
Cost category                      day costs         total
------------------------------  ------------  ------------
Nontherapy ancillary services            $45           16%
Therapy services\a                        78            28
Routine\b                                153            56
==========================================================
Total                                   $276          100%
----------------------------------------------------------
\a Therapy costs include speech, occupational, and physical therapy. 

\b Routine costs include room and board, nursing, and other costs. 

Source:  GAO analysis of 1995 SNF Medicare cost reports. 

By contrast, PPS payments for nontherapy ancillary services will
range from about $35 to almost $80 per day, depending on the RUG-III
category of the patient.\14 A comparison of reported costs to the
possible range in payments indicates that two-thirds of the SNFs had
average daily nontherapy ancillary costs either below or above the
range of potential payments established in PPS (see table 2).  This
may be an underestimate of the proportion of patient days that would
be under- or overpaid because each facility treats patients across a
range of the RUG-III categories.  Therefore, these facility averages
may mask the extreme payment and cost variation across patient days. 

                          Table 2
          
            Facility-Level Nontherapy Ancillary
          Reported Costs Compared to Estimated PPS
                    Payment Range, 1995

                                   Number of
Nontherapy ancillary costs        facilities       Percent
------------------------------  ------------  ------------
Costs less than estimated              5,291           53%
 payment range
Costs within payment range             3,185            32
Costs above estimated payment          1,539            15
 range
==========================================================
Total                                 10,015          100%
----------------------------------------------------------
Source:  GAO analysis of 1995 SNF Medicare cost reports. 

--------------------
\12 In developing the payment rates and the relative weights, HCFA
did not have a national sample of patient-level data to classify
patients into the RUG-III groupings.  Instead, it used available
claims data and decision rules to group patients into 10 broad
categories, using a model known as the MedPAR analog. 

\13 App.  III shows the average costs and the distribution of costs
for the major nontherapy ancillary services. 

\14 Nontherapy ancillary costs account for approximately 43 percent
of the nursing base rate.  Therefore, the nontherapy ancillary
portion of the nursing rate is $47 for urban facilities (43 percent
of $109.48) and $45 for rural facilities (43 percent of $104.88.) The
range in the weights for urban facilities is ($47 x .75) to ($47 x
1.7); the range for rural facilities is ($45 x .75) to ($45 x 1.7). 
We examined this variation only for the 26 RUG-III patient groups
that account for most Medicare-covered stays. 

   CONCLUSIONS
------------------------------------------------------------ Letter :5

Total Medicare payments for all SNFs are likely to be adequate, if
not generous, to cover the costs of nontherapy ancillary services. 
However, the PPS case-mix adjustment method may not appropriately
account for the variation in the nontherapy ancillary costs and thus
may not correctly raise or lower payments across the patient groups
to reflect expected differences in nontherapy ancillary needs. 
Therefore, Medicare payments for certain patient groups may be too
high or too low, relative to the average.  Any assessment of the
adequacy of total Medicare payments to any SNF, however, would need
to consider total Medicare costs and payments over the entire year. 

HCFA is aware of the concern about this issue.  It has commissioned
research to assess the extent of any payment distributional problem
and evaluate the possibility of refining the RUG-III classification
system and weights to explicitly account for nontherapy ancillary
cost variation.  These refinements will become even more important as
the 3-year transition to fully prospective rates proceeds. 

In the meantime, increasing SNF payments for all or some RUG-III
groups will not address the allocation problem.  It would simply add
costs to the program and increase overpayments without improving the
distribution of payments across patient categories and SNFs.  Rather,
as a first step, the extent of any maldistribution of SNF payments
across case-mix groups needs to be assessed.  If any distributional
problems are identified, the RUG-III relative weights would have to
be recalculated to better target payments to the case-mix groups that
contain patients with high expected nontherapy ancillary needs. 

   AGENCY COMMENTS
------------------------------------------------------------ Letter :6

In written comments on a draft of this report, HCFA shared GAO's
concerns for PPS' potential effects on medically complex patients
under the SNF PPS.  HCFA noted that it is expediting research that
will allow it to refine the payment system for nontherapy ancillary
services and affirmed its commitment to assessing potential changes
that could affect quality of care and access to skilled nursing care
for Medicare beneficiaries. 

HCFA also provided technical comments, which we incorporated where
appropriate.  Among these, HCFA stressed two important advantages of
a PPS.  First, under PPS, SNFs receive an all-inclusive per diem
payment, which is fungible among the various services provided to SNF
patients.  SNFs do not receive separate payments for nontherapy
ancillary services.  Second, because of the all-inclusive nature of
the payment, SNFs are encouraged to provide services in an efficient
manner.  Providers may choose to provide fewer nontherapy ancillary
services and to negotiate lower prices paid for them.  Because our
sample was based on 1995 costs, any reductions in ancillary pricing
or utilization will not be reflected in these data.  HCFA's letter is
reprinted as appendix IV. 

---------------------------------------------------------- Letter :6.1

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

If you or your staff have any questions, please call me or Laura
Dummit, Associate Director, at (202) 512-7114.  Other major
contributors include Carol Carter, Jennifer DuLac, Daniel Lee, and
Dana Kelley. 

Sincerely yours,

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

SCOPE AND METHODOLOGY
=========================================================== Appendix I

To determine how HCFA incorporated nontherapy ancillary costs into
the SNF PPS, we reviewed (1) the provisions of the Balanced Budget
Act of 1997 that mandated the new SNF PPS; (2) the SNF PPS interim
rule, which took effect on July 1, 1998; (3) the SNF PPS final rule,
which took effect September 30, 1999; and (4) associated research
concerning SNF payment policies.  We also discussed HCFA's
implementation of the SNF PPS with officials at its Division of
Inpatient Post Acute Care. 

To determine the variation in SNF costs, we analyzed the 1995 SNF
Minimum Data Set (MDS), which contains cost, financial, and other
statistical information for Medicare-certified SNFs from the Medicare
cost report.  We used fiscal year 1995 data because they were the
most complete data available at the time of our analysis.  Based on
input from HCFA officials, we calculated per diem ancillary (therapy
and nontherapy), routine, and total costs for each facility.\15 To
control for regional wage differences, we adjusted costs for wage
differences across geographic areas according to the methodology
prescribed in the regulations.\16 Finally, based on input from HCFA
officials, examinations of the regulations, and our own
determinations, we excluded SNFs that met any of the following
conditions:  (1) cost report periods less then 10 months or greater
than 13 months, (2) low or no Medicare utilization, (3) extremely
high or low routine or ancillary costs,\17 or (4) no identifiable
wage index.  These conditions reduced the analytic file from 12,276
to 10,015 facilities. 

Due to data limitations, we could not examine SNF costs by case-mix
group.  The RUG-III classification system uses variables that were
not in the 1995 cost report or claims files.  Therefore, we focused
our analysis on average per diem costs at the facility level. 
Although this limited our ability to examine the impact of the
payment system under the new provisions, comparisons were adequate to
establish a potential problem with the distribution of payments under
PPS. 

--------------------
\15 Ancillary costs are costs for specialized services that are
directly identifiable to individual patients.  Therapy-ancillary
costs include speech, occupational, and physical therapy costs. 
Nontherapy ancillary costs are all other ancillary cost categories,
including drugs, medical supplies, labs, and X rays.  Routine costs
include regular room, dietary, nursing, and other services for which
a separate charge is not made.  All costs are after the allocation of
overhead expenses. 

\16 64 Fed.  Reg.  41, 643-41, 683 (1999) (to be codified at 52
C.F.R.  409, 411, 413, 489). 

\17 We excluded SNFs with no ancillary or routine costs and excluded
SNFs whose routine or ancillary costs were within the top or bottom
0.25 percent for each group of hospital-based and freestanding SNFs. 
We chose this approach over HCFA's typical approach of excluding
values equal to the mean plus three standard deviations or minus
three standard deviations since it would have eliminated many of the
high-cost providers without eliminating any of the extremely low-cost
providers.  Because many of the high-cost providers were hospital
based, HCFA's approach would have eliminated a disproportionate
number of hospital-based SNFs. 

MEDICARE'S PROSPECTIVE PAYMENT
SYSTEM RATE CALCULATION
========================================================== Appendix II

Under PPS, SNFs are paid for their Medicare patients on a per diem
basis.  Each patient is grouped into 1 of 44 RUG-III categories based
on their clinical condition, functional status, and expected use of
certain services (see table II.1).  A base payment is adjusted for
each RUG-III category to account for the nursing and therapy costs
associated with treating the average patient in that group. 

                         Table II.1
          
                       RUG-III Groups

                                                    Number
                                                        of
Service     Clinical condition/need                 groups
----------  --------------------------------------  ------
Rehabilita  Patients who require rehabilitation in      14
 tion        one of five groups based on the
             number of therapy minutes per week:
             --Ultra: 720 or more therapy minutes
             per week
             --Very high: 500 to 719 therapy
             minutes per week
             --High: 325 to 499 therapy minutes
             per week
             --Medium: 150 to 324 therapy minutes
             per week
             --Low: 45 to 149 therapy minutes per
             week
Extensive   Patients who require intravenous             3
 services    feeding or medications, suctioning,
             tracheostomy care, or are on a
             ventilator/respirator
Special     Patients with cerebral palsy;                3
 care        quadraplegia; multiple sclerosis;
             pressure ulcers; fever with vomiting,
             weight loss, or dehydration; tube
             feeding and aphrasia; or receiving
             radiation therapy.
Clinically  Patients with burns, coma, septicemia,       6
 complex     pneumonia, internal bleeding,
             chemotherapy, wounds, kidney failure,
             urinary tract infections, oxygen, or
             transfusions
Impaired    Patients with poor cognitive                 4
 cognition   performance
Behavior    Patients with behavior symptoms such         4
 problems    as wandering, hallucinations, or
             physical or verbal abuse of others
             (unless other condition would place
             patient in other category)
Reduced     No special clinical conditions; RUG         10
 physical    groups based solely on patient
 function    ability to perform activities of
             daily living
----------------------------------------------------------
Each payment has three components to cover different types of costs: 
nursing, therapy, and other services.\18 The nursing component is
calculated by multiplying the nursing weight assigned to each RUG-III
category by the nursing base rate ($109.48 for urban facilities in
1998).  The nursing weight reflects nursing, social services, and
nontherapy ancillary resources necessary for providing care to the
average patient within the associated RUG-III category.  For the 26
RUG-III groups that will cover the majority of Medicare patients, the
nursing weight ranges from 0.75 to 1.7.\19

The costs of nontherapy ancillary services are included in the
nursing component of the payment amount.  Consequently, the nursing
weights determine the payment range for nontherapy ancillary
services.  The nursing payment component, which covers nursing,
social service, and nontherapy ancillary costs, ranges from $82.11 to
$186.12 for urban facilities, depending on the RUG-III category, and
$78.66 to $178.30 for rural facilities. 

The therapy component consists of either a therapy case-mix amount or
a therapy non-case-mix amount, depending on the RUG-III category and
the amount of therapy resources required.  For high-therapy-use
groups, the therapy case-mix amount is calculated by multiplying the
therapy weight by the therapy base amount ($82.67 for urban
facilities in 1998).  The therapy weight reflects resources necessary
to provide physical therapy, speech therapy, or occupational therapy
to the average patient within the associated RUG-III group.  Patients
who require minimal therapy services receive the therapy non-case-mix
amount.  This fixed amount reflects costs incurred to provide lower
levels of therapy services. 

The non-case-mix component is a fixed amount assigned to all RUG-III
groups.  This amount covers administrative, overhead, and other
general patient care costs. 

Table II.2 shows the PPS rate calculations for urban SNFs for the
upper 26 case-mix groups. 

                                        Table II.2
                         
                         PPS Rate Calculations for Urban SNFs for
                               the Upper 26 Case-Mix Groups

                                                                         Non-
                          Nursing                                       case-
                        component\a          Therapy component\b         mix      Total
                     ------------------  ----------------------------  --------  --------
                               Weight x            Weight x      Non-
RUG-III category     Relative      base  Relative      base     case-            (A+B+C+D
(code)                 weight  rate (A)    weight  rate (B)   mix (C)       (D)         )
-------------------  --------  --------  --------  --------  --------  --------  --------
Rehabilitation
-----------------------------------------------------------------------------------------
Ultra C (RUC)            1.30   $142.32      2.25   $186.01              $55.88   $384.21
Ultra B (RUB)            0.95    104.01      2.25    186.01               55.88    345.90
Ultra A (RUA)            0.78     85.39      2.25    186.01               55.88    327.28
Very high C (RVC)        1.13    123.71      1.41    116.56               55.88    296.15
Very high B (RVB)        1.04    113.86      1.41    116.56               55.88    286.30
Very high A (RVA)        0.81     88.68      1.41    116.56               55.88    261.12
High C (RHC)             1.26    137.94      0.94     77.71               55.88    271.53
High B (RHB)             1.06    116.05      0.94     77.71               55.88    249.64
High A (RHA)             0.87     95.25      0.94     77.71               55.88    228.84
Medium C (RMC)           1.35    147.80      0.77     63.66               55.88    267.34
Medium B (RMB)           1.09    119.33      0.77     63.66               55.88    238.87
Medium A (RMA)           0.96    105.10      0.77     63.66               55.88    224.64
Low B (RLB)              1.11    121.52      0.43     35.55               55.88    212.95
Low A (RLA)              0.80     87.58      0.43     35.55               55.88    179.01

Extensive services
-----------------------------------------------------------------------------------------
Level 3 (SE3)            1.70    186.12                        $10.91     55.88    252.91
Level 2 (SE2)            1.39    152.18                         10.91     55.88    218.97
Level 1 (SE1)            1.17    128.09                         10.91     55.88    194.88

Special care
-----------------------------------------------------------------------------------------
Level C (SSC)            1.13    123.71                         10.91     55.88    190.50
Level B (SSB)            1.05    114.95                         10.91     55.88    181.74
Level A (SSA)            1.01    110.57                         10.91     55.88    177.36

Clinically complex
-----------------------------------------------------------------------------------------
ADL high, with           1.12    122.62                         10.91     55.88    189.41
 depression (CC2)
ADL high, without        0.99    108.39                         10.91     55.88    175.18
 depression (CC1)
ADL medium, with         0.91     99.63                         10.91     55.88    166.42
 depression (CB2)
ADL medium, without      0.84     91.96                         10.91     55.88    158.75
 depression (CB1)
ADL low, with            0.83     90.87                         10.91     55.88    157.66
 depression (CA2)
ADL low, without         0.75     82.11                         10.91     55.88    148.90
 depression (CA1)
-----------------------------------------------------------------------------------------
\a The urban SNF base rate for nursing is $109.48; the rural SNF
nursing base rate is $104.88. 

\b The urban SNF base rate for therapy is $82.67; the rural SNF base
rate for therapy is $95.51. 

--------------------
\18 The labor-related portion of the rate is adjusted by the hospital
wage index to reflect the wage level in each SNF's market area. 

\19 Patients classified into 1 of the upper 26 RUG-III categories are
deemed to be eligible for Medicare coverage.  Patients classified
into 1 of the lower 18 RUG-III categories are reviewed on a
case-by-case basis to determine Medicare eligibility. 

TYPES OF AND VARIATION IN
NONTHERAPY ANCILLARY SERVICE
COSTS, 1995
========================================================= Appendix III

Drug and medical supply costs accounted for the highest shares of
nontherapy ancillary service spending in 1995 (see table III.1). 
Drugs were the most commonly provided service and were also the most
expensive service on average.  Virtually all (99 percent) of the SNFs
in 1995 had reported drug costs, averaging almost $20 per day, and
making up over half (58 percent) of all nontherapy ancillary costs. 
Ten percent of facilities had drug costs of $37 per day or more. 
Medical supplies were the next most common service (supplied in 90
percent of the SNFs) and made up about 18 percent of total nontherapy
ancillary costs, or $9 per day.  Again, the top 10 percent of
facilities had costs well above that, at $22 or more per day. 

                        Table III.1
          
              Distribution of Daily Nontherapy
            Ancillary Costs by Cost Center, 1995

                       Range of costs
                ----------------------------
                                                Percent of
Nontherapy          10th                90th          SNFs
ancillary       percenti            percenti     reporting
service               le      Mean        le         costs
--------------  --------  --------  --------  ------------
Drugs                 $7       $20       $37           99%
Medical                0         9        22            90
 supplies
Oxygen therapy         0         8        26            54
Labs                   0         2        10            26
All other cost         0         2         7            42
 centers\a
Intravenous            0         2         4            22
 therapy
Radiology              0         1         5            32
----------------------------------------------------------
\a Electrocardiology, dental, and other nontherapy ancillary cost
centers. 

Source:  GAO analysis of 1995 SNF Medicare cost reports. 

Facility-level nontherapy ancillary costs ranged from less than $11
at the 10th percentile to $95 or greater at the 90th percentile (see
table III.2).  Although most of the facilities in our sample had
average daily nontherapy ancillary costs below $40, 8 percent of
facilities had costs that exceeded $101 per day.  The type of
institution may explain some of the variation in daily nontherapy
ancillary costs, as seen in table III.3.  For 67 percent of
freestanding SNFs, these costs were $40 or less.  However, only 25
percent of hospital-based SNFs had daily nontherapy ancillary costs
of $40 or less. 

                        Table III.2
          
          Per Diem Nontherapy Ancillary Costs, By
                      Percentile, 1995

Percentile                                      Daily cost
--------------------------------------  ------------------
1st                                                     $2
5th                                                      7
10th                                                    10
25th                                                    18
50th (median)                                           33
75th                                                    60
90th                                                    95
95th                                                   123
99th                                                   181
----------------------------------------------------------
Source:  GAO Analysis of 1995 SNF Medicare cost reports. 

                              Table III.3
                
                Daily Nontherapy Ancillary Costs, Range
                        Across Facilities, 1995

                                Number of facilities,
                                       by type
                                ----------------------
                                                                Cumula
                                                                  tive
                                                        Percen  percen
                                        Hospit            t of    t of
Average daily nontherapy        Freest     al-          facili  facili
ancillary cost                  anding   based   Total    ties    ties
------------------------------  ------  ------  ------  ------  ------
$0-10                            1,012     104   1,116     11%     11%
11-20                            1,888     123   2,011      20      31
21-30                            1,558      94   1,652      16      48
31-40                            1,147     105   1,252      13      60
41-50                              781     100     881       9      69
51-60                              548     110     658       7      76
61-70                              399     150     549       5      81
71-80                              290     140     430       4      85
81-90                              203     141     344       3      89
91-100                             148     128     276       3      92
101+                               349     497     846       8     100
======================================================================
Total                            8,323   1,692  10,015    100%    100%
----------------------------------------------------------------------

(See figure in printed edition.)Appendix III
COMMENTS FROM THE HEALTH CARE
FINANCING ADMINISTRATION
========================================================= Appendix III

(See figure in printed edition.)

*** End of document. ***