Medicare Payment Issues Affecting Inpatient Rehabilitation Facilities (IRFs)

CRS Report for Congress
Medicare Payment Issues Affecting
Inpatient Rehabilitation Facilities (IRFs)
Updated January 24, 2006
Sibyl Tilson
Specialist in Social Legislation
Domestic Social Policy Division

Congressional Research Service ˜ The Library of Congress

Medicare Payment Issues Affecting
Inpatient Rehabilitation Facilities (IRFs)
Medicare spending on post-acute care, either those services provided in a facility
after an acute hospitalization or home health services provided to eligible
beneficiaries in the community, has elicited increasing attention as program spending
on these services has grown. Beneficiaries can receive post-acute care in multiple
settings, elevating the importance of identifying the most appropriate, cost-effective
setting to provide necessary care. Recent implementation of prospective payment
systems for the different settings has amplified concerns that post-acute providers are
making decisions about beneficiaries’ rehabilitative care in response to financial
incentives rather than deciding on the basis of which setting is the most appropriate
for the care needs of the patient.
Inpatient rehabilitation facilities (IRFs) are one post-acute provider participating
in Medicare. IRFs, either freestanding hospitals or distinct units of hospitals, are
exempt from Medicare’s payment system used to pay acute care hospitals. The
majority of IRFs participating in Medicare are distinct parts units of acute care
hospitals. Medicare is the largest single payer for IRF services. Starting in 2002,
Medicare began implementing a prospective payment system specifically for IRFs
(IRF-PPS). Much of this report describes the IRF payment system and concludes
with an FY2006 payment calculation.
Recent administrative actions by the Centers for Medicare and Medicaid
Services (CMS) to enforce the newly constituted “75% rule” have been causing a
certain amount of consternation among the provider community. The 75% rule
specifies criteria, including qualifying medical conditions and compliance thresholds
(the percentage of patients treated that have those conditions), that a facility must
meet in order to be paid as an IRF and not as a lower-paid general hospital. Pending
local coverage determinations (LCDs), the medical review policies established by
Medicare contractors regarding IRF services that will be paid for by Medicare in
their respective areas, have elicited objections from providers and their advocates as
Over objections from the Administration, the Consolidated Omnibus
Appropriations Act for 2005 (P.L. 108-447) delayed implementation of the IRF rule.
The Secretary of Health and Human Services (HHS) was required to review a
pending study by the Government Accountability Office (GAO) before the 75% rule
could be enforced for most IRFs. The GAO report was issued on April 22, 2005. On
June 21, 2005, CMS announced that the compliance thresholds will be implemented
as planned. Legislation that would hold the compliance threshold at 50% for two
years, among other provisions, has been introduced in the 109th Congress. The two-
year 50% compliance threshold was included in the Deficit Reduction Omnibus
Reconciliation Act of 2005 (S. 1932), passed by the Senate on November 3, 2005.
The conference report included a two-year 60% threshold that would postpone
enforcement of a 75% threshold at 75% from July 1, 2007 until July 1, 2008.
This report will be updated as events warrant.

Recent Developments..............................................1
Evolution of the “75% Rule” and Enforcement Standards..................2
Overview ....................................................2
Modifications to the “75% Rule”..................................4
Use of Local Coverage Determinations (LCDs)
by Medicare Contractors....................................6
Proposed Delay in Enforcement of the New 75% Rule
and Implementation of LCDs.................................8
Background on IRFs..............................................12
Overview of Rehabilitation Services and Post-Acute Care.............12
Medicare’s Conditions of Participation for IRFs.....................17
Effect of Medicare’s Prior Payment System for IRFs.................18
IRF Prospective Payment System Issues...............................19
Legislative Overview .........................................19
Description of IRF PPS and FY2006 Payment Adjustments............20
FY2006 IRF Payment Calculation................................24
Concluding Observations.......................................26
List of Tables
Table 1. Example of IRF-PPS Payment Calculation for CMG 0109
(for Certain Stroke Patients Without Comorbidities),
Including Facility-Level Adjustments, for FY2006...................25

Medicare Payment Issues Affecting
Inpatient Rehabilitation Facilities (IRFs)
Inpatient rehabilitation facilities (IRFs), either freestanding hospitals or distinct
part units of other hospitals, are exempt from Medicare’s prospective payment system
used to pay short-term, acute care hospitals. The majority of the IRFs that participate
in Medicare are distinct parts of other hospitals; in 2003, 971 of the 1,188 (or 82%)
Medicare participating IRFs were distinct part units, the majority of which were in
urban areas. In 2004, Medicare program payments to IRFs were estimated at $5.9
billion, making Medicare the largest single payer for inpatient rehabilitation services.
This report discusses recent developments affecting IRFs, and then turns to a more
detailed examination of issues attracting Congressional attention.
Recent Developments
Recent administrative actions by the Centers for Medicare and Medicaid
Services (CMS) have prompted congressional action within the FY2006 budget
reconciliation process. In May 2004, CMS published a final rule implementing
changes in its policies regarding the criteria used to determine which facilities qualify
for payment as IRFs. Simply put, this rule establishes that a certain proportion of
patients treated by an IRF must have specified medical conditions in order for the
facility to qualify as an IRF and receive higher Medicare payments. This proportion
increases from 50% to 75% over a three-year transition period.
Despite objections from the Bush Administration, the FY2005 Labor, Health
and Human Services, and Education and Related Agencies (Labor-HHS)
Appropriation delayed enforcement of the IRF compliance thresholds embodied in
the 75% rule. The HHS Secretary was required to review and respond to the issuance
of a previously mandated congressional report by the Government Accountability
Office (GAO) before the compliance thresholds for most IRFs (those certified before
June 30, 2004) could be enforced. The GAO report was published on April 22,
2005.1 On June 24, 2005, CMS announced that it was proceeding with the
implementation of the revised and expanded compliance criteria adopted in the May

2004 final rule, as these criteria are “not inconsistent” with GAO findings.2

1 U.S. Government Accountability Office, More Specific Criteria Needed to Classify
Inpatient Rehabilitation Facilities, GAO-05-366, Apr. 2005. (Hereafter cited as GAO IRF
2 According to industry press, during a House Ways and Means, Subcommittee on Health
hearing, GAO stated that the existing transition period provides CMS with sufficient time
to make necessary refinements to the rule: “Top CMS Official Hints Agency Will Pursue

Companion bills have been introduced in the House (H.R. 3373) and Senate (S.
1405) that will extend the 50% compliance threshold for two years, prevent the
Secretary from changing the designation of an IRF that is in compliance with the
50% threshold, and preclude medical necessity reviews that are based on other
criteria than those in the Medicare Benefits Policy Manual. A 17-member National
Advisory Council on Medical Rehabilitation would also be established. The council
would be charged with providing advice and recommendations regarding a variety
of topics, including the appropriate criteria for determining the clinical
appropriateness of IRF admissions and distinguishing an IRF from other providers.
The provision to extend the 50% compliance threshold for two years was included
in the Deficit Reduction Omnibus Reconciliation Act of 2005 (S. 1932), passed by
the Senate on November 3, 2005. The conference report, which was renamed the
Deficit Reduction Act of 2005, includes a provision that would extend the existing

60% compliance threshold for two years (until June 30, 2006), establish a 65%

threshold for a 12-month period starting July 1, 2007, and establish the 75%
threshold starting July 1, 2008. Under the current regulation, the 75% threshold
would start on July 1, 2007.
Another provision in the conference report has the potential to affect the
payment for IRF care. Under this provision, a three-year post-acute care payment
reform demonstration program would examine the costs and outcomes across
different post-acute care sites. A single standardized patient assessment instrument
would be used across all sites of care to measure the functional status during
treatment and at discharges. An additional assessment would be required at the end
of the episode of care. Program participants would be required to provide
information on the fixed and variable costs for each individual. A report to Congress,
including program results and recommendations, would be submitted no later than
six months after the completion of the demonstration. The costs of carrying out the
program would be funded by a $6 million transfer from the Part A trust fund.
Evolution of the “75% Rule”
and Enforcement Standards
The Medicare statute gives the Secretary of HHS discretion to establish the
criteria that facilities must meet in order to be exempt from the inpatient prospective
payment system (IPPS) used to pay acute care hospitals. Accordingly, the Secretary
established in regulation that an IRF must demonstrate that at least 75% of its
inpatients (all inpatients, not just Medicare beneficiaries) were treated for one or
more specified conditions during its most recently completed 12-month cost
reporting period. By January 1984, the 10 qualifying conditions were established as:
(1) stroke; (2) spinal cord injury; (3) congenital deformity; (4) amputations; (5) major
multiple trauma; (6) fracture of the femur (hip fracture); (7) brain injury; (8)

2 (...continued)
Full Implementation of 75 Percent Rule,” Inside CMS, June 17, 2005.

polyarthritis (including rheumatoid arthritis); (9) neurological disorders (including
multiple sclerosis, motor neuron diseases, polyneuropathy, muscular dystrophy, and
Parkinson’s disease); and (10) burns.3 The regulations established that when a
facility does not meet the 75% rule (and certain other conditions of participation
discussed later), it is no longer paid as an IRF, but will be paid as an short-term,
general hospital under IPPS.
Starting January 1, 2002, Medicare changed the payment system for IRFs, from
cost-based to prospective payments, but did not change the qualification criteria for
IRFs. In June 2002, CMS instructed its Medicare contractors (in this case, fiscal
intermediaries, or FIs) to defer enforcement of the 75% rule due to concerns that the
regulations had not been consistently applied among the different contractors. The
contractors were directed to continue their verification activities for existing IRFs,
but not change any facility’s status until a systematic assessment of the different
review procedures was completed and further guidance was issued.4
In addition to this review of FI administrative procedures, CMS analyzed IRF
claims data from the first eight months of 2002 (submitted under the new payment
system) to estimate the overall compliance with the existing 75% rule.5 Subject to
certain caveats, CMS estimated that only 13.35% of the 1,170 IRFs would meet the
75% threshold; the percentage in compliance would increase to 25.17% if the
threshold was lowered to 65%.6 The percentage of IRFs in compliance varied
significantly by region and by certain facility characteristics.7 CMS indicated that

3 Eight of these conditions were originally adopted in the Sept. 1, 1983 interim final rule.
The list was supplemented with two additional conditions in the Jan. 3, 1984 final
regulation; suggestions that chronic pain, pulmonary disorders, and cardiac disorders be
included were not accepted.
4 The temporary suspension did not appear to increase the number of nonqualifying patients
treated by IRFs. From 1996 to 2002, there has been a steady, substantial downward trend
in the percentage of Medicare cases counted in one of the ten conditions. However, the
decline was steeper from 1996 to 1999 than from 1999 to 2002. In 1996, 59.4% of the cases
were in the qualifying conditions; this percentage fell to 53% in 1999 and 50.9% in 2002.
Grace Carter, Orla Hayden, Susan Paddock, and Barbara Wynn, Case Mix Certification Rule
for Inpatient Rehabilitation Facilities, Draft Report (DRU-2981-CMS), Rand Health, May

2003, pp. 13-15. (Hereafter cited as Carter, et al., Case Mix Certification.)

5 68 Federal Register 26791, May 16, 2003.
6 Diagnosis data from administrative data sets were used to estimate compliance
percentages. In many cases, the diagnosis indicated that rehabilitation procedures were used,
not the specific condition. CMS indicated that compliance estimates would have likely been
higher if more detailed information from the medical record had been available. Carter, et
al., Case Mix Certification, p. 10.
7 For instance, almost half of the 121 IRFs in the Pacific region were estimated to be in
compliance with the 75% rule, and only 1.5% of the 66 IRFs in the East South Central
region were judged to meet that standard. Interestingly, the compliance rate was three times
higher in the IRF units (15.4%) than in the freestanding hospitals (4.7%). The compliance
rate in the 135 government-run IRFs (18.5%) and the 700 nonprofit IRFs (15.3) was more
than three times that in the 259 proprietary IRFs (5%). 68 Federal Register 26792, May

patients with lower extremity joint replacements, specifically knee and hip
replacements, are the largest group treated by IRFs that do not count toward
compliance with the 75% rule.8
Modifications to the “75% Rule”
CMS published proposed regulations to change the classification criteria for
IRFs in the September 9, 2003 Federal Register. Under this proposal, absent further
regulatory actions, the compliance threshold would be lowered to 65% until January
1, 2007 (when it would revert to its original 75% standard). Also, among other
changes, CMS proposed to replace the condition of polyarthritis with three other
arthritis-related conditions, which would bring the total number of qualifying primary
conditions to 12.9 CMS proposed two alternatives where patients with secondary
medical conditions in those 12 categories would also count toward the compliance
threshold until January 1, 2007.10 One alternative was limited to counting secondary
conditions only for patients with joint replacements; the other would count secondary
conditions for any admission. CMS declined the requests to add cancer, cardiac,
pulmonary, and pain conditions as qualifying criteria, in part because of a lack of
studies that demonstrate an improvement in patients’ outcomes when cared for in
IRFs as compared to other settings.
CMS issued the final rule on May 7, 2004 with an effective date of July 1, 2004.
In the final rule, CMS adopted several policy changes. First, it replaced polyarthritis
with four arthritis-related conditions for a total of 13 qualifying conditions.
Specifically, a patient with severe or advanced osteoarthritis involving two or more
major joints (not including a joint that has been replaced) will count toward a
facility’s compliance threshold. Second, certain beneficiaries with bilateral joint
replacements who are extremely obese or 85 years and older will count toward a
facility’s compliance. Third, CMS adopted its more expansive proposal to consider
secondary conditions for all patients (not just those who have had joint
replacements). This provision expires for cost-reporting periods on or after July 1,
2007. Fourth, CMS adopted a three-year transition period for the compliance
threshold as follows: at 50% from July 1, 2004 and before July 1, 2005; at 60% from
July 1, 2005 and before July 1, 2006; at 65 % from July 1, 2006 and before July 1,

2007; and at 75% from July 1, 2007 and thereafter. During this three-year period,

7 (...continued)

16, 2003.

8 According to the May 2003 proposed rule, nationally, less than 25% of Medicare
beneficiaries with joint replacements are admitted to IRFs after surgery.
9 The detailed description indicates that three or more major joints would need to be
significantly affected; these conditions should not have improved after an appropriate,
aggressive, sustained course of outpatient therapy preceding the admission. Also, a joint
replaced by a prosthesis is considered to no longer have arthritis even though that condition
was the reason for the joint replacement.
10 The secondary condition (or comorbidity) must cause a significant decline in the patient’s
functioning that, even in the absence of the admitting condition, the patient would require
intensive treatment unique to an IRF, rather than in another setting.

CMS pledged to convene a technical research panel with the assistance of the
National Institutes of Health to examine which are the most appropriate clinical
conditions for care in an IRF.
In developing the impact analysis for the regulation, CMS assumed that 10% of
the noncompliant cases would meet clinical criteria because of coding improvements
and that another 10% of the cases would become compliant when medical record data
(rather than more readily available administrative data) was examined. CMS also
assumed that 50% of the existing joint replacement cases would meet the clinical
criteria established in the rule. All in all, CMS projected that 0.1% of the 459,682
current Medicare IRF cases would not be admitted to that setting. The agency stated
that these cases would likely receive treatment in alternative settings. Since about
half of the IRFs were located in hospital complexes that include skilled nursing
facilities (SNF), CMS assumed that SNFs would have a higher probability of
absorbing cases no longer admitted to IRFs. CMS projected savings of
approximately $5,525 per case in FY2004; the savings represents the estimated
difference between IRF care and the cost of the other treatment. CMS projected
savings of $400,000 in program payments in FY2004, $10 million in FY2005, $30
million in 2006, $90 million in FY2007, and $190 million in FY2008.11
The changes to the proposed regulation adopted by CMS in the final rule,
including the three-year transition to the 75% threshold, did not satisfy industry
advocates. Although pleased by the lowering of the compliance standards, the
temporary relief is not seen as addressing overriding concerns with the regulation,
particularly a need to modernize the compliance standards. In the long run, they
perceive that facilities will be compelled to revise admission policies which will
result in large scale denial of access to IRF care.12 Also, industry advocates have
raised concerns with respect to the implementing instructions issued to the Medicare
contractors by CMS. Among other issues, the standard for providing appropriate,
aggressive and sustained therapy in another less intensive setting prior to an IRF
admission is seen as burdensome for providers and costly to beneficiaries. Other
objections about the recordkeeping and documentation requirements have been
expressed as well.13
With respect to implementing the regulations, on June 25, 2004, CMS issued
instructions on verification procedures that Medicare contractors should use to ensure
that IRFs meet Medicare’s new classification requirements.14 Generally, the

11 69 Federal Register 25772, May 7, 2004.
12 Statement of the American Medical Rehabilitation Providers Association, Apr. 30, 2004,
accessed on Oct. 7, 2004, at [], now available at
subscription website [].
13 Letter from the American Medical Rehabilitation Providers Association (AMRPA) to
Mark McClellan, Administrator of CMS, on CMS Program Transmittal No. 221 for
Inpatient Rehabilitation, July 29, 2004.
14 CMS Pub. 100-04, Medicare Claims Processing, Transmittal 221, Change Request 3332,
June 25, 2004. As indicated in footnote 17, these instructions have been subsequently

contractor will use the IRF’s patient assessment instrument (IRF-PAI) data from the
most recent, consecutive, and appropriate 12-month time period starting July 1, 2004
to verify compliance.15 The instruction includes lists of diagnoses and impairment
group codes that will be used to determine compliance with the specified conditions.
The contractor (and the regional office or RO) have the discretion to instruct the IRF
to submit specific sections of medical records from a random sample of inpatients (or
any selection of inpatients). Other procedures for verifying compliance with the
established threshold may apply to IRFs that have Medicare admissions that
constitute less than 50% of its total inpatient population or those whose Medicare
Part A fee-for-service admissions are not determined to be representative of the
patient population served by the IRF.16 A determination by the RO that a facility is
classified as an IRF is generally made at the start of a facility’s cost reporting period
and applies to the entire cost reporting period for which the determination is made.17
As of November 30, 2005, CMS contractors confirm that seven providers have been
reclassified as acute care hospitals since the new regulation has been enforced.
Use of Local Coverage Determinations
(LCDs) by Medicare Contractors
As discussed in the previous section, a facility may be subject to a threshold
compliance review to determine its eligibility for payment as an IRF. IRFs are also
subject to medical reviews to determine whether the care provided to an individual
was reasonable and necessary based on the patient’s condition as documented in the
medical record. Simply, Medicare contractors are required to ensure that payment
is made only for those services that are reasonable and necessary; the medical
documentation must support the patient’s need for an IRF level of care.
The medical review process is conducted according to both national and local
coverage policies.18 In the absence of national policy, Medicare contractors can
establish individual coverage policies, now known as local coverage determinations,
which clarify the existing national standards with respect to Medicare covered

14 (...continued)
clarified. All administrative publications affecting IRFs, including federal register notices,
manual instructions, and transmittals can be found at
[ ht t p: / / pr ovi der s / i r f pps/ pubs.asp] .
15 The contractor will use less than 12 months’ worth of data for certain compliance reviews
starting before July 1, 2005.
16 Generally, CMS presumes that if an IRF’s Medicare population meets the compliance
threshold, then the facility’s total population will satisfy this standard, particularly when the
IRF’s Medicare population represents at least a majority of its patients.
17 CMS Pub. 100-04, Medicare Claims Processing, Transmittal 347, Change Request 3503,
Oct. 29, 2004 and CMS Pub. 100-04, Transmittal 478, Change Request 3704, Feb. 18, 2005.
18 For more information, see CRS Report RL31711, Medicare: Coverage Policy, by Jennifer
O’ Sullivan.

services.19 Contractors develop local policies by considering medical literature, the
advice of local medical societies, and public comments. The policy only applies to
the geographic area served by the contractor. CMS strongly encourages multi-state
contractors to develop uniform policies across all of their jurisdictions. Generally,
Medicare’s IRF medical necessity standards for inpatient hospital services are
included in the Medicare Benefit Policy manual.20 The standards are based on
criteria finalized in 1980 by the American Academy of Physical Medicine and
Rehabilitation and the American Congress of Rehabilitation Medicine.
In November 2002, the Office of the Inspector General reported that IPPS-
exempt hospital inpatient services had not been routinely reviewed for medical
necessity since 1995. Although Quality Improvement Organizations (QIOs, formerly
peer review organizations, or PROs), FIs, and Medicare Integrity Program contractors
all had the authority to conduct medical review in hospitals, none were conducting
routine reviews prior to February 2002. At that point, CMS issued a program
memorandum to notify FIs that they may include PPS-exempt hospitals in their
reviews; no additional funding was provided for their expanded review responsibility,
however. 21
Certain of the draft local coverage determinations proposed by various Medicare
contractors have elicited some objections from the industry. Of particular concern
are those draft proposals that would use diagnosis-specific guidelines as initial
screens to determine the appropriateness of IRF admission and treatment. These
screens are viewed as restricting the ability of the referring and receiving
rehabilitation physician to make case-by-case determinations on the need for
inpatient rehabilitation care for each patient.22 Instead, providers (and their
advocates) are encouraging contractors to use broader, more flexible criteria to
determine medical necessity. Industry advocates prefer policy proposals that do not
use diagnosis-specific parameters for care, but instead cite the need for 24-hour
specialized nursing care and physician availability, which are the screening standards
included in the Medicare Benefit Policy Manual. Industry advocates object that all
proposed LCDs inappropriately establish a new condition of coverage for IRF care
by including a requirement that the services could not be provided in a less intensive

19 Effective Dec. 2003, Medicare contractors began issuing local coverage determinations
(LCD) instead of local medical review policies (LMRP). Generally, both policies support
decisions by contractors as to whether a particular service will be covered. A LCD consists
of only reasonable and necessary information while a LMRP may also contain statutory
exclusions. All existing LMRPs will be either retired or converted into LCDs no later than
Dec. 2005.
20 Chapter 1, Section 110 of the manual, covering inpatient rehabilitation services, can be
found at [].
21 Office of the Inspector General, Oversight of Medicare PPS-Exempt Hospital Services,
OEI-12-02-00170, Nov. 2002.
22 Comment letter from the American Hospital Association (AHA) to the Medical Director
of Palmetto GBA, May 3, 2004, p. 2.

setting.23 Skilled nursing facilities (SNFs) and their advocates, however, argue that
certain IRF cases could be appropriately treated in less intensive settings.
Proposed Delay in Enforcement of the
New 75% Rule and Implementation of LCDs
On July 14, 2004, the House Appropriations Committee approved an
amendment to the FY2005 Labor, HHS and Education appropriation bill that would
have prohibited any CMS funds from being used to implement the final rule
establishing the new IRF classification criteria. The amendment would also have
prevented Medicare contractors from using any existing or new local medical review
policies, local coverage determinations, or national coverage determinations
establishing medical necessity standards for IRFs. The amendment directed the
Secretary to contract with the Institute of Medicine (IOM) to study and make
recommendations on the IRF classification requirements and appropriate medical
necessity standards. The required report would have been due to Congress no later
than October 1, 2005. Nine months after the report’s submission, the prohibition on
spending to enforce the final rule and medical necessity standards would lapse. The
increased program expenditures associated with this amendment were offset by a $9
million reduction in CMS’s appropriation for program administration.
According to industry press, provider advocates were concerned that the
Chairman of the House Ways and Means Committee, Representative Thomas, would
try to block the amendment by asking the House Rules Committee (which
determines the procedures by which the House will consider specific legislation) to
exempt the amendment from point-of-order protection so it could be challenged
during the floor debate.24 Instead, the IRF provision in the bill (H.R. 5006) that was
approved in the House on September 8, 2004 was modified. It forbade HHS from
spending money to enforce the revised 75% rule for IRFs certified on or before June
30, 2004 (the day before the regulations became effective) until a GAO report is
published. GAO had been directed by the managers’ statement accompanying the
conference report for the Medicare Prescription Drug, Improvement and
Modernization Act of 2003 (MMA) to issue a report, in consultation with experts in
the field of physical medicine and rehabilitation, that looks at whether the current list
of conditions represents a clinically appropriate standard for defining IRF services.
MMA required the Secretary either to determine that the new 75% rule is not
inconsistent with GAO’s recommendations or to promulgate a regulation providing
for new criteria no later than 60 days after receiving this GAO report.

23 Analysis of Medicare Fiscal Intermediaries’ Use of the “Less Intensive Setting” Concept
in Local Coverage Determinations for Inpatient Rehabilitation, to CMS from AMRPA Joint
Coalition written by Powers, Pyles, Sutter, and Verville PC, Attorneys at Law, Sept. 20,


24 The Rules Committee’s procedures permit any committee chairman the right to request
that certain provisions of a bill be removed if those provisions pertain to issues that are
within the jurisdiction of the chairman’s committee. “Rehab Industry Urges Thomas Not
to Block Moratorium on 75 Percent Rule,” Inside CMS, July 29, 2004.

The increased program expenditures associated with this amended
appropriations provision was offset by a $12.5 million reduction in CMS’s funding
for program administration. In adopting this provision, the House disregarded
guidance from the Administration, which indicated that a delay in enforcing the rule
would result in inappropriate payments to hospitals that are not based on current
clinical practices. The Administration’s statement attributed savings of $10 million
in 2005 and $1.8 billion over 2005 to 2014 to timely enforcement of the regulation.25
On September 15, 2004, the Senate Appropriations Committee included a
provision in its Labor-HHS bill (S. 2810) that would have prohibited funds from
being spent by HHS or any Medicare contractor to apply the IRF compliance criteria
(the 75% rule) established in the Federal Register on May 7, 2004. The Committee
directed HHS to contract with IOM to study and make recommendations based on
the clinical consensus on how to modernize these criteria; the report is due no later
than October 1, 2005. Under the HHS contract, IOM was expected to use a
multidisciplinary panel of expert researchers and clinicians in the field of medical
rehabilitation. According to industry press, nursing home advocates urged the Senate
not to approve this moratorium on the 75% rule, in part because the delay would
continue perceived overpayments to IRF’s for care that skilled nursing facilities
(SNFs) can provide at half the cost.26
The Labor-HHS bill was included in the Consolidated Omnibus Appropriations
Act, 2005 (Division F, H.R. 4818, H.Rept. 108-792), which was signed on
December 8, 2004 as P.L. 108-447. The legislation contains language comparable
to the House passed enforcement delay of the 75% rule. Specifically, HHS cannot
spend money to enforce the revised 75% rule for IRFs certified on or before June 30,
2004 until a previously mandated GAO report is published. No later than 60 days
after receiving this GAO report, the Secretary is required either to determine that the
new 75% rule is not inconsistent with GAO’s recommendations or to promulgate a
regulation providing for new criteria. The legislation does not include $12.5 million
to offset the increased program expenditures attributed to the delay.27
The mandated GAO report was issued April 22, 2005. It analyzed FY2003 data
on Medicare patients admitted to IRFs using the conditions in the current regulation
that were not yet in effect. This analysis was supplemented by an examination of
data from July through December 2004. GAO found that less than 44% of the
Medicare admissions to IRFs had a primary condition that was included in the rule.28
The largest group of these patients admitted to IRFs had orthopedic conditions, with

25 Statement of Administration Policy, H.R. 5006, Labor, Health and Human Services, and
Education, and Related Agencies Appropriations Bill, FY2005, Sept. 8, 2004, pp. 3-4.
26 “Senate Appropriators Call for One-Year Delay of 75 Percent Rule,” Inside CMS, Sept.

23, 2004.

27 Section 219 can be found on p. H10325 of the Congressional Record, Nov. 19, 2004. The
discussion of the omission of the $12.5 million offset can be found on p. H10659
28 The percentage of Medicare patients increased to 62% when comorbid conditions
included in the rule were counted. GAO IRF report, pp. 13-14.

a predominant number of joint replacements.29 GAO suggested that although some
of these joint replacement patients may need intensive rehabilitation services, few
patients had comorbidities that suggested a need for IRF care.30 With respect to
compliance, looking at primary conditions only, 6% of the IRFs met the 75%
threshold required when the rule is fully phased-in; this percentage increased to 27%
when comorbid conditions were considered. The compliance threshold in effect from
July 1, 2004 though June 30, 2005 is 50%; looking at both primary and comorbid
conditions, over 80% of IRFs met that threshold.
Experts interviewed by GAO and by the IOM differed on whether additional
conditions should be added to the IRF compliance list, but agreed that condition
alone does not provide sufficient criteria to identify the types of patients
appropriately treated at an IRF; functional status should also be considered. In
addition, certain experts suggested that facility characteristics be used to classify
GAO recommended that Medicare contractors routinely review IRF admissions
for medical necessity; that CMS encourage research on the effectiveness of intensive
inpatient rehabilitation; and that CMS use the research and other information to
refine the rule to describe more thoroughly the subgroups of patients with a given
condition that are appropriately treated in an IRF. In responding to the GAO report,
CMS has generally concurred with these recommendations, but stated that
implementing other criteria to identify subgroups of appropriately treated patients
could result in more a restrictive policy than currently implemented and thus would
have to be carefully considered.
In the course of its mandated IRF study, GAO interviewed 10 officials
representing Medicare FIs and reported that over half were not conducting reviews
of patients admitted to IRFs. Those who were doing reviews used different
approaches for selecting records or facilities to assess. CMS estimated that less than
1% of admissions in facilities excluded from IPPS, such as IRFs, are reviewed. In
contrast, the major insurers interviewed by GAO indicated that private payers relied
on individual preauthorization to ensure that the most appropriate patients are
admitted to IRFs.31
In response to the GAO recommendation, CMS agreed that targeted reviews for
medical necessity are needed and indicated that it had expanded its efforts to provide
greater oversight of IRF admissions through local policies that have been

29 IRF patients with conditions not on the list were orthopedic cases (45.6%); about two-
thirds of those cases were joint replacements (or 30.6% of the IRF admissions). GAO IRF
report, p. 16.
30 Eighty-seven percent of the joint replacement patients in 2003 had unilateral procedures
and were younger than 85 years old and did not fit the criteria for joint replacement (based
on primary condition) in the subsequently adopted rule. GAO IRF report, p. 16
31 Officials from the three major insurers and one managed care plan interviewed by GAO
indicated that each admission to an IRF required preauthorization to determine whether a
specific patient should be admitted. Each case was judged individually on the basis of
various factors which differed from payer to payer. GAO IRF report, p. 21.

implemented or are being developed by the FIs. Selected FI’s have performed
different types of medical reviews (using either widespread, focused, or random
selection of cases) to examine the medical necessity of IRF patient admissions and,
in certain instances, to shape the development of LCDs. For example, one contractor
performed a widespread probe on patients who required rehabilitation therapy to
restore strength, increase range of motion, or upgrade the ability to function.
Examining 108 claims from 27 providers from June 1, 2003 to November 30, 2003,
the FI found that 50 claims (46%) could have been provided in a less intensive
rehabilitation setting. Another contractor reviewed 101 randomly selected claims
from January 1, 2004 to June 30 2004; 17 claims were denied because medical
records were not submitted on a timely basis; 72 of the 84 other claims were denied
as well. Twenty-five of those patients were considered to be inappropriate
admissions to the IRF.32 One other contractor had an overall denial rate of 72.3% for
IRF claims examined for medical necessity.33
On June 24, 2005, CMS announced that it was proceeding with the
implementation of the revised and expanded compliance criteria adopted in the May
2004 final rule as these criteria are “not inconsistent” with GAO findings. The
compliance criteria will be adopted over a transition period as planned (with the
threshold increasing from 50% to 75% for cost reporting periods beginning July 1,
2007). CMS has also requested that National Institute of Health (NIH) convene a
research panel to recommend future research on the types of patients that would most
benefit from intensive inpatient rehabilitation. The research recommendations are
now being evaluated; CMS intends to collaborate with NIH to determine how to best
promote research into the effectiveness of rehabilitation.34
In its press release accompanying the above announcement, CMS stated that its
monitoring efforts had detected no significant beneficiary problems in accessing IRF
services. It found that the number of IRF cases increased 1.2% when comparing
utilization in CY2003 to that in CY2004, when the revised regulations were
implemented. In contrast, an industry study projects a 7.7% decline in Medicare
discharges from IRFs from July 2004-June 2005, primarily in the diagnostic
categories most affected by enforcement of the compliance thresholds. However, this
study, based on data representing 77% of the IRF facilities and 66% of the Medicare
discharges, shows a 1.3% decline in Medicare cases when comparing utilization in
CY2003 to that in CY2004. The decline is not supported by CMS analysis of its
claims data, which, as indicated earlier, shows an 1.2% increase.
Still, industry advocate remain concerned about the effect of the new
compliance thresholds and enforcement efforts. CMS has released a November 2005
correspondence to Wall Street to discuss and refute these issues. Specifically, the
purpose and underlying assumptions of the impact analysis included in the 75% rule

32 Six of those cases were admitted to the IRF for cardiac rehabilitation and the remaining
19 were deemed to be too ill to participate in the intense level of therapy required by the
33 Information on contractors’ LCDs and different medical reviews can be found at
[], accessed Nov. 15, 2005.
34 70 Federal Register 36641, June 24, 2005.

is examined. According to CMS, the industry is concerned about differences in the
impact analyses and the actual provider experience since July 2004. This concern is
attributed to a misunderstanding of the mandated impact analyses, which are not
treated as expenditure targets. Moreover, while CMS predicted a moderate increase
in IRF expenditures based on historical growth rates, actual spending was
significantly higher. 35
Background on IRFs
The following section will provide a general overview of rehabilitation services
and post-acute care that may be provided by various Medicare entities. Medicare
beneficiaries can receive post-acute care from different types of providers in both
inpatient and outpatient settings. The availability of care from multiple sites
introduces concerns that Medicare may be paying different amounts for the different
types of post-acute care providers for patients with essentially similar needs for care.
After presenting that basic framework, the report will then discuss different Medicare
payment policies that are unique to IRFs. The remainder of this section will describe
Medicare conditions of participation for IRFs and discuss the effect of the former
cost-based reimbursement on IRFs. The last section of the report will discuss
legislative developments shaping the direction of the IRF-PPS. It will present
payment adjustments within the IRF-PPS and conclude with a example showing a
calculation for FY2006.
Overview of Rehabilitation Services and Post-Acute Care
Rehabilitation services consist of physical therapy, occupational therapy, and
speech and language services. These services are often furnished to patients
following a hospital stay or an ambulatory surgical procedure and can be provided
by a number of different Medicare-certified providers in either inpatient or outpatient
settings. The diversity in post-acute care providers that furnish rehabilitation services
provides for considerable variation and flexibility in the duration and intensity of
beneficiaries’ use of rehabilitation services and providers. Although the range of
possible service settings permits patients (and their physicians) some choice in where
beneficiaries receive the most appropriate care, it may also make rehabilitation
providers more sensitive to changes in Medicare’s payment policies and procedures.36
The term “post-acute care” is commonly used to refer to a continuum of service
settings where rehabilitation, nursing, and other services can be provided to persons
following treatment for an acute illness or injury. Eligible beneficiaries who are
referred from the community and use home health services without a prior

35 The CMS impact analyses projected that IRF-PPS expenditures would grow at a
compound annual growth rate of 5.6% from 2001-2004. Actual expenditures grew at 13.8%
compound annual growth rate from 2001-2004. See pp.7-8 of CMS Memorandum to
Interested Investors and Analysts ,by Lambert van der Walde, Nov. 30, 2005.
36 Medicare Payment Advisory Commission (MedPAC), Report to Congress: Context for
a Changing Medicare Program, June 1998, pp. 79, 89.

hospitalization also use post-acute care. Depending upon the context, post-acute care
may encompass more than vigorous rehabilitative services (often thought to be the
primary focus of inpatient rehabilitation facilities or IRF) and include convalescent
and palliative services, physical or speech therapy, wound care, skilled nursing care,
even pain management for terminal patients. Post-acute care also may be provided
on an inpatient or outpatient basis. IRFs are one of the inpatient settings where such
services may be provided. Other Medicare providers offering these services are long-
term care hospitals (LTCH), SNFs, and home health agencies (HHAs).
Medicare beneficiaries use post-acute care frequently. In 2001, almost one-third
of the beneficiaries discharged from acute care hospitals used post-acute care with
SNF care being the most common single care setting. Until the implementation of
Medicare’s acute care hospital inpatient prospective payment system (IPPS) in 1984,
however, follow-up care after hospital stays accounted for only a small part of
Medicare spending. Following implementation of IPPS and other policy changes
affecting SNFs and HHAs in the late 1980s, Medicare spending for post-acute care
began to grow rapidly. Total program spending for post-acute care increased an
average of about 21% per year from 1992 to 1997, from $14 billion to $35.7 billion;
of this total, IRF program spending increased from $2.8 billion in 1992 to $3.8
billion in 1997, an average increase of 6% per year.37 The change in Medicare’s
spending trends for post-acute care services was attributed to advances in technology
combined with the incentives provided by the fixed price payments under IPPS for
short-term general hospitals to discharge patients as quickly as possible to other
settings for continuing care, together with clarifications of coverage policies for
certain post-acute care settings.
In response to the rapid expenditure growth, the Balanced Budget Act of 1997
(BBA 97, P.L. 105-33) and subsequent legislation mandated development and use
of prospective payment systems for all post-acute care settings; these new payment
systems have been implemented gradually over the time period since passage of BBA

97.38 Between 1997 and 2001 (the year before implementation of IRF-PPS),

Medicare spending for post-acute care declined by more than 20%, from $35.7 billion
to $28.0 billion, due to a decline of more than 50% for home health care services;
Medicare spending in SNFs, IRFs, and long-term facilities increased by 12%, almost
11%, and 58% respectively in that time period. Between 2001 and 2004, Medicare
post-acute spending increased from $28.0 billion to $35.9 billion, slightly more than
its 1997 peak. Over the same time period, Medicare spending in IRFs increased from
$4.2 billion to $5.9 billion, more than 50% above 1997 IRF expenditures.

37 These figures reflect program spending and do not include beneficiary copayments.
MedPAC, Data Book: Healthcare Spending and the Medicare Program (Data Book), June
2004, p. 142 and MedPAC, Data Book, June 2005, p. 150. Medicare payments to IRFs
(including beneficiary copayments) grew 20% annually between 1985 and 1995, from $70
million to $430 million. MedPAC, Data Book, July 1998, p. 104.
38 Medicare’s payment reforms included establishing a case-mix adjusted per diem PPS for
SNFs using resource utilization groups (RUG-III) as a patient classification system, starting
in 1998; a case-mix adjusted PPS for home health services, starting in 1999; a per discharge
PPS for IRFs using function-related groups, starting in 2002; and a per discharge PPS for
long-term hospitals using modified diagnosis-related groups (DRGs), starting in 2002.

Implementation of the different payment systems has heightened concerns that
providers are shifting beneficiaries’ care in response to changing financial incentives
provided by the reimbursement methods rather than basing the decision for care on
the patient’s medical condition. Similarly, some view Medicare’s varying coverage
rules and eligibility criteria, as well as the different requirements that post-acute
providers must meet in order to participate in the program, as subject to
manipulation.39 For example, a three-day prior hospitalization is required to trigger
coverage for inpatient skilled nursing facility care, but is not required for other types
of inpatient post-acute care under Medicare. In addition, the beneficiary must require
daily skilled nursing or rehabilitation care. Beneficiaries who qualify for care in an
IRF must be medically capable of undergoing at least three hours of rehabilitation per
day that is expected to result in significant practical improvement within a reasonable
period of time. Medicare beneficiaries have no special eligibility requirements in
order to receive care in a LTCH; these facilities must only maintain an average
inpatient length of stay of at least 25 days. Medicare’s requirements for physician
involvement in the care provided in the different inpatient settings also varies.
Specifically, physicians must be integrally involved in care provided in IRFs and
LTCHs, but are required to visit a SNF patient only once every 30 days for the first

90 days and every 60 days thereafter.

Between 1992 and 2005, the supply of all major types of Medicare inpatient
post-acute care providers (as well as Medicare spending for these providers)
experienced significant growth as well. The number of SNFs increased from 12,303
to 15,632; the number of IRFs increased from 907 to 1,232; the number of long-term
hospitals increased from 97 to 365. Ownership of post-acute care providers has also
been shifting, with for-profit status becoming more common. Within these overall
trends, the regional distribution of different types of post-acute care providers has
remained uneven. The pattern of post-acute care provider use is determined in large
part by the supply of particular provider types in a given area. Because of wide
geographic variation in supply of provider types, utilization patterns, even for
patients with similar needs, may vary widely by geography.
The difference in the use of post-acute care services also likely reflects
variations in practice standards as well as availability. In this case, practice standards
are thought to include the inclinations of individual practitioners as affected by
regulation and the policies of Medicare contractors, such as fiscal intermediaries,
who influence the use of post-acute care services under Medicare (see earlier
discussion of the use of LCDs).40 Moreover, admissions to post-acute care are guided
by a hospital discharge planner who, in turn, may be affected by the knowledge of
which patients providers are willing to accept.41

39 MedPAC, Report to Congress, Variation and Innovation in Medicare, June 2003, p. 73.
40 Robert Kane, Wen-Chieh Lin, and Lynn Blewett, “Geographic Variation in the Use of
Post-Acute Care,” Health Services Research, vol. 37, no. 3 (June 2002), pp. 679-680.
41 Melinda Butin, Anita Garten, Susan Paddock Debra Saliba, Mark Totten, and José
Escarce, How Much Is Post-Acute Care Use Affected by Its Availability?, National Bureau
of Economic Research (NBER), Working Paper 10424, Apr. 2004, p. 4.

Also, as noted in a 1999 study sponsored by the Assistant Secretary of Planning
and Evaluation (ASPE), certain hospital characteristics appear to be associated with
the type of post-acute care to which patients are discharged. Larger or teaching
hospitals, for example, are more likely than other acute hospitals to discharge patients
to IRFs. Proprietary hospitals are more likely than non-profit hospitals to discharge
patients to home health care. The study found that there were some interactions
between types of post-acute care that were used by Medicare beneficiaries. For
example, IRF bed supply was positively associated with the rate of Medicare home
health care use, which indicated that these two types of care are used in sequence for
significant numbers of beneficiaries.42
Certain personal and health characteristics, in combination with some basic
differences between the types of post-acute care providers, have been found to
influence either the use of post-acute care or the propensity to use one type of
provider relative to others. The characteristics include the health or functional status
of the patient as well as the patient’s access to informal care (nonpaid care provided
by family or friends). For example, frail beneficiaries may not be able to withstand
the intensive therapy regimen (the minimum of three hours of daily therapy) required
in an IRF. Alternatively, severely disabled beneficiaries may be more easily cared
for in SNFs than in community settings with home health services. On the other
hand, availability of informal care increases the likelihood that post-acute care could
be provided in the community or in institutional settings where the goal is to return
to the community, rather than in institutional settings explicitly designed to provide
long-term care.43
The availability of multiple sites of post-acute care has led to concerns that the
care provided to beneficiaries is influenced by the different levels of payment offered
for similar services in the various settings. However, there is little definitive
information on the extent of patient overlap, differences in Medicare’s relative
payment levels in different settings for the same quality of care, and the appropriate
resource levels for the desired outcomes for patients with particular needs.44 These
shortcomings can be attributed, in part, to the fact that the existing administrative
data used for Medicare’s payment purposes (including patient assessment instruments
used to classify patients into the relevant payment groups in the different post-acute
care settings) do not contain information needed to measure the quality of care within
and across post-acute settings. Although the different data systems include
information on patients’ functional status (generally measured in terms of activities
of daily living, mobility, communication skills, and cognitive status), each of the
patient assessment instruments collect different measures recorded at different times

42 Korbin Liu, Barbara Gage, Jennie Harvell, David Stevenson, and Niall Brennan,
Medicare’s Post-Acute Care Benefit: Background, Trends, and Issues to Be Faced, Urban
Institute, Jan. 1999, p. 5.
43 Korbin Liu, et al., Medicare’s Post-Acute Care Benefit: Background, Trends, and Issues
to be Faced, Urban Institute, Jan. 1999, pp. 24-25, 41-43.
44 Marie Johnson, Danielle Hothaus, Jennie Harvell, Eric Coleman, Theresa Eilertsen and
Andrew Kramer, Medicare Post-Acute Care: Quality Measurement Final Report, University
of Colorado Health Sciences Center, Mar. 2001 (revised Mar. 2002), p. 7. (Hereafter cited
as Johnson, et al., Medicare Post-Acute Care.)

in the post-acute stay.45 These differences make it difficult to identify whether
similar patients are, in fact, treated in different settings and, if so, whether the
outcomes of care are comparable.46 Furthermore, patient assessment data are
collected only as long as a patient is treated in any particular post-acute care setting,
but the outcome of the care rendered may not be apparent until after the patient is
discharged. 47
Interestingly, policy makers (and legislators) have returned to a discussion of the
importance of using common patient assessment tools across post-acute settings. The
House Ways and Means, Subcommittee on Health held a hearing on this and related
topics on June 16, 2005.48 Recent MedPAC analysis indicates that the information
collected by Medicare’s current assessment tools cannot be easily integrated.49 Also,
CMS has not met a Congressional mandate report on the development of an
instrument to assess the health and functional status of beneficiaries who use post-
acute services.50 A seminal effort led by CMS to identify and establish consistent
terms and common measures for a clinical assessment of the quality of post-acute
care has not been successful. CMS has indicated that it will begin testing a patient
assessment tool for all post-acute care services by spring, 2006 as a first step toward
the possible creation of an integrated payment system across all settings.51
In support of this initiative, a provision in the Deficit Reduction Act of 2005
would establish a three-year post-acute care payment reform demonstration program
to examine the costs and outcomes across different post-acute care sites. A single

45 SNFs provide functional status information on all Medicare and Medicaid patients using
Minimum Data Set, version 2.0 (MDS 2.0) as the assessment tool; home health agencies
provide Outcome and Assessment Information Set (OASIS) as the assessment tool for theirTM
Medicare patients, IRFs have incorporated the Functional Improvement Measure (FIM)
as part of its patient assessment instrument (IRF-PAI) to report patient status information.
See MedPAC, Report to Congress: Issues in a Modernized Medicare Program, June 2005,
pp. 114-119 for a detailed comparison of the different patient assessment tools used in post-
acute settings.
46 Alan M. Jette, Stephen M. Haley, and Pengsheng Ni, “Comparison on Functional Status
Tools Used in Post-Acute Care,” Health Care Financing Review, spring 2003, vol. 24, no.
3, p. 13; Lisa I. Iezzoni and Marjorie S. Greenberg, “Capturing and Classifying Functional
Status Information in Administrative Databases,” Health Care Financing Review, spring

2003, vol. 24, no. 3, p. 61.

47 Johnson, et al., Medicare Post-Acute Care, p. 3.
48 Detailed information (testimonies, submissions for the record, and the hearing transcript
can be found at
[http://waysandmeans hearings .asp?formmode=detail&hearing=422&comm= 1 ] .
49 MedPAC, Report to Congress: Issues in a Modernized Medicare Program, June 2005,
p. 120.
50 The Medicare, Medicaid, and SCHIP Benefits Improvement Protection Act of 2000
(BIPA) instructed the Secretary to report on the development of an instrument to assess the
health and functional status of beneficiaries who use SNF services by Jan. 2005.
51 Bureau of National Affairs, “Medicare: CMS Says Patient Assessment Tool For Use in
Medicare Expected in 2006,” Health Care Daily, vol. 10, no. 116, June 17, 2005.

standardized patient assessment instrument would be used across all sites of care to
measure the functional status during treatment and at discharges. An additional
assessment would be required at the end of the episode of care. Program participants
would be required to provide information on the fixed and variable costs for each
individual. A report to Congress, including program results and recommendations,
would be submitted no later than six months after the completion of the
demonstration. The costs of carrying out the program would be funded by a $6
million transfer from the Part A trust fund.
Medicare’s Conditions of Participation for IRFs
IRFs and other specialty hospitals were excluded from IPPS when it was
implemented for short-term, general hospitals in 1984 because the patient
classification system for acute hospitals, diagnosis related groups or DRGs, was
thought not to adequately account for the costs associated with treating their patients.
As with other post-acute care services, functional and cognitive measures have been
judged to be better predictors of resource use in rehabilitation hospitals than
diagnoses. An IRF must perform basic hospital functions and also meet certain
requirements to be excluded from IPPS and paid as an IRF. As discussed earlier,
until recently, the exclusion required that at least 75% of a facility’s inpatient
discharges needed intensive rehabilitation services for one of 10 conditions. As of
July 2004, IRF qualification criteria have been modified and the qualifying
percentage has been lowered on a transition basis until January 1, 2007. In addition,
patients in IRFs are expected to improve as a result of therapy. Medicare patients
treated in an IRF must also be capable of receiving approximately three hours of
daily therapy (generally five days a week). Also, patients must require frequent
physician involvement, 24-hour rehabilitation nursing, and coordinated care by a
multidisciplinary group of professionals.
Medicare has established requirements (or conditions of participation) for IRFs
to receive payment from Medicare.52 Specifically, the facility must review each
prospective patient’s condition and medical history prior to admission to determine
whether the patient will benefit significantly from an intensive inpatient rehabilitation
program. IRFs must have a plan of treatment for each inpatient that is established,
reviewed and revised by a physician in consultation with other professional personnel
who provide services to the patient. As mentioned earlier, facilities must use a
coordinated multidisciplinary team approach documented by periodic clinical entries
in the medical record that discuss the patient’s progress toward a specified goal.53
Team conferences must be held at least every two weeks to determine the
appropriateness of treatment. IRFs must ensure that patients receive close medical
supervision by a physician with specialized training or experience in rehabilitation.
IRFs must assure 24-hour availability of such a physician as well as 24-hour
availability of a registered nurse with specialized training or rehabilitation

52 See 42 Code of Federal Regulations (C.F.R.) § 412.23(b)(3)(7).
53 A multidisciplinary team usually includes a physician, rehabilitation nurse, social worker
and/or psychologist as well as those therapists involved in the patient’s care. At a minimum,
a team must include a physician and, rehabilitation nurse, and one therapist. Medicare
Benefit Policy Manual, CMS Pub. 100-02, Section 110.4.4.

experience.54 Each facility must also have a physician who acts as the full-time
director of rehabilitation.55
IRFs that are distinct-part units of hospitals must meet additional conditions of
participation. Among other requirements, these units must have beds that are
physically separate from the hospital’s other beds, separately identified admission
and discharge records from those of the hospital, and policies that specify that
necessary clinical information is sent to the unit upon transfer of a hospital’s patient
to the unit.56
Effect of Medicare’s Prior Payment System for IRFs
Prior to implementation of the IRF-PPS, these facilities had been paid on a cost
related basis subject to per discharge limits as originally established by the Tax
Equity and Fiscal Responsibility Act of 1982 (TEFRA). Generally speaking, under
TEFRA, those facilities with operating costs below its payment ceiling received costs
plus an incentive payment; those with costs above their target were paid the ceiling
plus a relief payment. Each facility had a separate payment limit or target amount
established using its cost per discharge in its base year, subject to a cap; the target
amounts were subject to annual increases or updates. Capital costs were paid on a
pass-through basis, subject to certain limitations. New providers were exempt from
payment ceilings for the first three years of operation.
This payment system encouraged new exempt facilities to maximize their costs
in their base year to establish high, facility specific cost limits. Once subject to the
TEFRA constraints, a recent entrant could fairly readily reduce its costs below its
limit and receive Medicare payment for full costs. Older rehabilitation facilities
could not inflate their target amounts in this fashion, were more likely to incur costs
above their limits, and receive payments less than their costs.
Because of payment disparities between new and old IRFs, BBA 97 mandated
changes to Medicare’s existing IRF payment system and imposed national cost limits
(or national target amounts) on payments to specific IRFs. Accordingly, an IRF
would receive payments based on its costs per discharge, subject to the lower of
facility specific TEFRA limits or the national target amounts established by BBA 97.
The national target amount was set at the 75th percentile of the 1996 facility-specific
target amounts updated for inflation. Prior to BBA 97, payments to new IRFs were
based on their full Medicare allowable costs while their facility specific amount was
determined. With BBA 97, Medicare’s payments to new providers were limited to
the lesser of the provider’s costs or 110% of the wage adjusted, national median
target amount of established IRFs. Also, BBA 97 permitted long-established IRFs

54 The need is documented with frequent entries in the patient’s medical record of the direct,
medically necessary care by the physician at least every two or three days during the
patient’s stay. Medicare Benefit Policy Manual, CMS Pub. 100-02, Section 110.4.1.
55 The doctor must have at least two years of training or experience in medical management of
inpatients requiring rehabilitation services which is preceded by a one-year hospital internship.
56 See 42 C.F.R. § 412.25(a).

(those with base years beginning before October 1990) to rebase (or update) their
facility-specific target amount using averaged costs from certain of their three most
recent cost reporting periods. Among other payment changes (including a reduction
in capital payments), the legislation established a provider-specific update formula
in order to reduce existing payment disparities; facilities with costs above their target
amount received a larger update than those below their target amounts (which could
be no update). The amount and type of bonus and relief payments to IRFs were
modified as well.57
The TEFRA system was intended to be a temporary measure to control
Medicare hospital spending until prospective payment systems for the nonacute
hospitals could be implemented. It remained in effect longer than expected, in part
because of the difficulties in accounting for the variation in resource use across
patients in exempt facilities.58 Arguably, part of this unexplained cost variation may
have stemmed from providers’ responses to payment incentives within the TEFRA
IRF Prospective Payment System Issues
The following section will provide background on the legislative provisions
shaping the implementation of the IRF-PPS, then present information on the IRF-
PPS payment adjustments, and conclude with an example of a FY2006 payment
Legislative Overview
As well as the other provisions modifying the TEFRA payment system
discussed earlier, BBA 97 provided for the establishment of a PPS for IRFs with a
two-year transition period beginning by October 2000 and before October 2002. 59
In that legislation, Congress did not specify the unit of payment or the patient
classification system to be used with the IRF-PPS. Instead, the Secretary was given
discretion to establish classes of IRF patients (called “case mix groups”) based on
appropriate factors such as impairment, age, related prior hospitalization,
comorbidities, and functional capacity of the patient. The Secretary was required to
establish weighting factors for each case-mix group that would be adjusted from time
to time. These PPS amounts would be budget neutral, set at a rate that would equal

98% of the total payments that would have resulted without such changes, for

57 BBA 97 also expanded Medicare’s transfer policy beyond discharges from one acute
hospital to another to include certain transfers from acute hospitals to post-acute providers.
This change reduced payments to acute hospitals for certain types of patients discharged to
post-acute settings.
58 MedPAC, Report to Congress: Medicare Payment Policy, Mar. 1999, pp. 72-75.
59 As mandated, the payments would be based on two-thirds of the TEFRA payment and
one-third of the PPS payment from Oct. 1, 2000 and before Oct. 1, 2001; in the following
year, payments would be based on one-third of the TEFRA payment and two-thirds of the
PPS payment. Starting by Oct. 1, 2002, the IRF-PPS would be fully phased in.

FY2001 and FY2002. IRF payments would be subject to an area wage adjustment
which would vary depending upon where the facility was located. BBA 97 directed
that these relative wage values be updated every year in a fashion that does not
increase payments as a result of those changes. The legislation included provisions
establishing outlier payments that would be equal to no more than 5% of total IRF
In an attempt to move toward more uniform payment policies across different
post-acute care settings, the Health Care Financing Administration (HCFA, now
called CMS) began to consider modifications to the patient assessment instrument
(the Minimum Data Set or MDS) and the RUG-III classification system designed for
use with per diem payment in SNFs for use in IRFs.60 This effort was redirected by
specific provisions in the Balanced Budget Refinement Act of 1999 (BBRA, P.L.
106-33) that mandated certain characteristics of the IRF-PPS.61 Specifically, the
Secretary was directed to use discharges as the unit of payment. The legislation also
mandated use of a specific (and different) patient assessment and classification
system than being considered by HCFA. The legislation directed the IRF-PPS to use
case-mix groups based on impairment, age, comorbidities, and functional capability
of the patient and such other appropriate factors deemed to improve the explanatory
power of the functional independence measure-function related groups (FIM-FRG).62
The law also stated that the Secretary was not precluded from establishing an
adjustment in the IRF-PPS to account for early transfers of patients from IRFs to
other settings. Finally, the Secretary was directed to study the effect of the new IRF-
PPS on utilization and beneficiary access to services, with the study due to Congress
no later than January 2005. (The study was submitted in August 2005). Subsequent
changes in BIPA increased total payments in the IRF-PPS system by 2% in FY2002
and permitted facilities to make a one-time election before the start of PPS to be paid
based on a fully phased-in PPS rate (and skip the two-year transition period). The
IRF-PPS system began implementation as of January 1, 2002.
Description of IRF PPS and FY2006 Payment Adjustments
Generally speaking, under PPS, Medicare pays an IRF a predetermined, fixed
amount per discharge, depending upon a patient’s impairment level, functional status,
comorbid conditions and age. Certain adjustments are made for facility-level
characteristics to account for area wage variations, rural location, and the percentage
of low-income patients (LIPs) served. Starting in FY2006, the payment system
includes an adjustment to increase payments to teaching facilities. IRF-PPS also
includes case level adjustments. Specifically, reduced or additional amounts are paid
for early transfers, short-stay outliers, patients who die before transfer and patients

60 MedPAC, Report to Congress, Medicare Payment Policy, Mar. 1998, vol. I, p. 96.
61 This legislation was incorporated by reference into the conference agreement on H.R.

3194, the District of Columbia Appropriations Act.

62 The Functional-Related Groups (FRGs) system was developed by Dr. Margaret Stineman
and colleagues at the University of Pennsylvania and SUNY-Buffalo. This system is based
on a rehabilitation coding system, the Functional Independence Measure (FIM), developed
and owned by the Uniform Data System for Medical Rehabilitation (USDmr).

who are extraordinarily costly (outliers). These payments encompass inpatient
operating and capital costs of furnishing covered rehabilitation services, but not the
costs of approved educational activities, Medicare bad debts, and other services that
are paid outside of the IRF-PPS for which the providers receive additional payments.
Medicare’s IRF-PPS payment for any beneficiary will depend upon a clinician’s
comprehensive assessment of that patient upon admission and again at discharge.
These documented assessments must be based on the direct observation of and
communication with the patient; information may be supplemented with information
from other sources, including family members or other clinicians. The prescribed
patient assessment instrument (PAI) form, the Uniform Data Set for Medical
Rehabilitation (UDSmr), encompasses about 55 questions used to ascertain a
patient’s functional independence including motor skills and cognitive capacities and
to establish a patient’s comorbidities. A patient’s assessments (from both admission
and discharge) are transmitted to CMS electronically once at the end of the patient’s
stay. Failure to meet the IRF PAI transmission deadlines results in a 25% reduction
in Medicare’s payment in all but extraordinary circumstances.
Using data from the patient’s initial assessment, each Medicare patient is
classified into one of 92 mutually exclusive case-mix groups (CMGs). First, a
patient is placed into one of 21 rehabilitation impairment categories (RICs) that
encompass clinically similar conditions, such as stroke or traumatic brain injury, as
the primary cause of admission. Next, a patient is placed into a CMG within the
RIC; the CMG assignment depends upon the patient’s functional status and, in some
instances, age. Within a CMG, a patient is assigned to one of four categories or
comorbidity tiers using clinical information from the patient’s discharge assessment.
The presence of comorbidities was found to substantially increase the average cost
of a specific CMG. Patients with the most serious conditions are assigned to tier 1;
patients with the least serious conditions are assigned to tier 3; those without any
relevant comorbidities (or secondary conditions) are assigned to the “none” tier. The
21 RICs encompass the 87 CMGs; five other CMGs have been established for
patients with special circumstances; one of the five CMGs is for patients with very
short stays and the four remaining are for patients who die before treatment is
completed. Each of these five special CMGs have only one payment rate and no
comorbidity tiers.
Medicare pays a reduced amount for a patient who is an early transfer. The
patient has a length of stay that is greater than three days but less than the average for
the assigned CMG and is transferred to another rehabilitation facility (which has been
defined as a rehabilitation facility, a long-term hospital, a short-term hospital, or a
nursing home.) No payment reduction applies for patients who are discharged to a
home health agency or other outpatient therapy setting. Also, the IRF will receive
the full amount if the transfer occurs after the patient has been treated for the average
length of stay associated with the CMG. The payment rate for early transfers is based
on the per diem payment for the applicable CMG (to which the patient has been
assigned). The IRF will receive an additional one half day payment to recognize the
higher costs generally associated with the patient’s first day of care. The early
transfer payment would include any facility-level payment adjustments.

Medicare pays for short-stay outliers using one of the five special CMGs. These
are patients who are not transfers, but are discharged from the facility after being
hospitalized no more than three days. These short-stay outliers may occur because
the patient could not tolerate a full course of intensive inpatient rehabilitation
treatment, left against medical advice, or died within three days of admission. Also,
patients who are discharged from and return to the same IRF by midnight of the third
consecutive calendar day are considered interrupted stays. Medicare makes only one
IRF-PPS payment for these cases.63
Originally, CMS established relative or cost weights for the different CMGs
using cost report data from FY1996, FY1997, and FY1998 and charge data from
calendar year (CY) 1999.64 The weights were updated for FY2006 using the same
methodology and data after the IRF-PPS was implemented. Although updated for
FY2006, unlike those used in IPPS, these relative weights are not updated annually.
The relative weights account for a patient’s resource needs for each of the CMGs and
payment tiers; 353 relative weights are used to determine Medicare payment rates.
Within any given CMG, the cost weight for a patient with a high comorbidity is
greater than the cost weights for those patients with low or no comorbidities. CMS
did apply a budget neutrality factor to ensure that the estimated aggregate payments
due to FY2006 changes in the relative weights (and other changes to the CMGs) did
not increase.
This cost weight is multiplied by a standard payment conversion factor
(formerly known as the budget neutral conversion factor) to calculate the payment
for a given patient.65 The standard payment amount was originally constructed using
the facility-specific information from 508 facilities, including cost reports from
FY1995, FY1996, and FY1997; applicable target amounts, as well as Medicare
claims (including corresponding UDSmr data) from CY1996 and CY1997. CMS
reduced the standard payment amount by 1.9% in FY2006 to account for coding
changes that do not reflect real changes in case mix (or increases in the intensity of
the illness of patients who are treated). An analysis of CY2002 data indicated that
payments to IRFs were about $140 million more than expected because of changes
in patient classification.66

63 As mentioned earlier, in addition to PPS payments, Medicare will pay IRFs for certain
items such as Medicare beneficiaries’ bad debts, the costs of approved educational programs
and for blood clotting factors provided to Medicare inpatients who have hemophilia outside
of the PPS.
64 Centers for Medicare and Medicaid Services, “Medicare Program; Prospective Payment
System for Inpatient Rehabilitation Facilities,” 66 Federal Register 41351-41353, Aug. 7,


65 As mentioned earlier, BBA 97 specified that budget neutral payments were to be
established at 98% of what would have been spent under the prior system during FY2001
and FY2002. BIPA increased the amount of the IRF-PPS budget neutral payments to 100%
in FY2002. The overall IRF-PPS budget neutrality provision is no longer in effect.
66 RAND recommended decreasing the standard payment amount by between 1.9% and

5.8% to adjust for coding changes. 70 Federal Register 47904-47908, Aug. 15, 2005.

Until FY2006, each year the IRF-PPS standard payment amount is increased
based on the modified market basket (MB) for excluded hospitals (those not paid
under IPPS). This MB is based on cost report data from Medicare participating
inpatient rehabilitation and psychiatric facilities as well as long-term, children’s, and
cancer hospitals which were subject to TEFRA payment limitations. The TEFRA
MB only includes operating costs, so the IRF-PPS update had been based on a
modified TEFRA MB that reflects capital costs. Starting in FY2006, CMS is
increasing IRF payments using a market basket reflecting the operating and capital
cost structures for rehabilitation, psychiatric and long-term facilities (referred to as
the RPL MB). CMS revised and rebased the RPL MB to incorporate 2002 cost
report data starting in FY2006. The change in market basket and use of the 2002 cost
data will increase the labor related share (or effect of the IRF’s wage index) and
increase the relative influence of the capital to operating costs.67
IRF-PPS incorporates an adjustment to reflect the relative area wage levels of
a facility’s labor market. Generally speaking, an IRF’s labor market is defined using
standards established by the Office of Management and Budget (OMB). Counties
that are not included as part of a metropolitan area (by virtue of commuting patterns
and population density) are considered to be rural. An IRF is not required to submit
wage index data to Medicare. This adjustment uses data submitted by acute care
hospitals and is compiled for the labor market area where they are located without
taking into account any geographic reclassifications of those hospitals. Also, IRF
wage index values are not subject to the rural floor applied to IPPS hospitals (where
the wage index value in any urban area cannot be lower than the rural wage index in
that state). The IRF wage data is commonly referred to as a pre-reclassification, pre-
floor data.
In FY2006, CMS adopted revised OMB labor market definitions using the core
based statistical area (CBSA) classifications to determine urban and rural areas.
These labor market definitions were adopted in IPPS in FY2005. The change is
implemented with a budget neutral one-year transition period; the wage index of all
IRFs (not just those that are disadvantaged by the new labor market definitions) will
have half of their wage index based on the old labor market definitions and half
based on the new definitions. IRFs that had previously been in rural areas who are
now considered to be in urban areas will have their rural payment adjustment phased
out over a three-year period. Generally, these IRFs will receive two-thirds of the
FY2005 rural adjustment of 19.14% (or 12.76%) with a blended wage index in
FY2006; one-third of the adjustment (6.38%) with a CBSA wage index in FY2007
and a CBSA wage index in FY2008. This transition will be implemented on a
budget neutral basis. IRFs that had been in urban areas that are now considered to
be located in rural areas will receive the full FY2006 rural payment adjustment of


Other facility adjustments (aside from the rural adjustment) may apply to an
IRF’s payment calculation. Starting in FY2006, a teaching facility will receive

67 The labor related share in IRF-PPS will change from 72.024 in FY2005 to 75.865 in
FY2006. 70 Federal Register 47916, Aug. 15, 2005. A discussion of the RPL MB can be
found on the preceding six pages of that Federal Register.

additional payments. RAND’s regression analyses of CY2002 and CY2003 data
indicated that teaching facilities had higher costs than IRFs without teaching
programs. Subject to a cap on the number of residents, teaching facilities will receive
additional payments based on a logarithmic formula comparable to that used for the
indirect medical education (IME) adjustment used in capital IPPS. Teaching
intensity will be measured by dividing the number of approved resident full-time
equivalents (FTEs) by the average daily census of the IRF. This will be raised to the
0.9012 power. An IRF’s FTE resident cap will be based on the number shown in the
final settlement of the IRF’s most recent cost reporting period ending on or before
November 15, 2004.
An IRF may also receive additional Medicare payments depending on the
number of low-income patients (LIP) it serves. The LIP adjustment incorporate the
same measure used to establish the disproportionate share hospital (DSH) adjustment
in IPPS. Specifically, the adjustment uses the number of days provided to poor
Medicare beneficiaries (those who receive SSI) divided by the total number of
Medicare days plus the number of Medicaid (non-Medicare days) divided by the total
number of days in the hospital. In FY2005 (and previously) this measure was raised
to the 0.4836 power. Starting in FY2006, this is raised to the 0.6226 power.
FY2006 IRF Payment Calculation
To establish the FY2006 payment rates, CMS increased the FY2005 IRF
standard payment conversion factor by the update amount, reduced that by 1.9% to
account for coding changes, and applied various budget neutral adjustments (to
account for certain changes from the previous year). In FY2006, the update amount
equaled the market basket increase of 3.6%. The FY2006 budget neutral adjustment
factor for the labor related change and the CBSA transition period is 0.9995; that for
CMG changes is 0.9995; that for changes in the rural adjustment is 0.9957; that for
the LIP adjustment change is 0.9851; that for implementing the teaching adjustment
is 0.9889. The FY2005 standard payment was $12,958. After the MB increase and
the budget neutrality decreases, the FY2006 standard payment is $12,762.68
For FY2006 IRF-PPS payments, CMS uses FY2001 acute hospital wage data
to compute the IRF wage index values. As mentioned earlier, unlike IPPS, the IRF-
PPS does not permit geographic reassignments for facilities. The labor-related
portion (75.865%) of the federal payment rate is multiplied by the IPPS wage index
value for the IRF’s area.69 An IRF is either in a metropolitan statistical area (MSA)
or the rural area of the state (which is considered to be counties that have not been
assigned to MSAs). This wage-adjusted amount is added to the non-labor related
portion of the rate to determine the wage-adjusted federal payment rate. IRFs in rural
areas receive an additional 21.3% increase to the federal payment rate. Depending
upon the percentage of poor Medicare and Medicaid days in a given facility, a facility
will receive additional LIP adjustment.

68 70 Federal Register 47939, Aug. 15, 2005 as corrected by 70 Federal Register 57168,
Sept. 30, 2005
69 The labor-related share was 72.359 in prior years.

Table 1 shows the IRF-PPS adjusted payment calculation for CMG 0109
(without comorbidities) in three different facilities. CMG 0109 is used to establish
Medicare payments for stroke patients who are less than 84.5 years old who have
motor scores that range from 22.35 and 26.15. The relative weight used for these
patients who have no comorbidities is 1.8147; Medicare’s federal prospective
payment rate for this CMG is $23,159.20 ($12,762 * 1.8147 = $23,159.20). This
represents the federal rate before the relevant facility-level adjustments are applied.
IRF-PPS payments will be adjusted to account for a facility’s relative area wage, rural
location, low- income percentage, and teaching status. The example will use three
facilities. Facility A is a non-teaching IRF located in Duke County, MA. It has a
DSH patient share percentage of 5% which translates into a LIP adjustment of 1.031.
Facility B is a teaching IRF in urban Queens County, NY with a DSH patient share
percentage of 10% which translates into a LIP adjustment of 1.0612. It will also
receive a teaching adjustment of 1.0910. Facility C is a non-teaching IRF located in
Kings County, CA, which was considered to be a rural area prior to FY2006. This
IRF will receive a hold harmless rural adjustment of 12.76%.
Table 1. Example of IRF-PPS Payment Calculation for CMG

0109 (for Certain Stroke Patients Without Comorbidities),

Including Facility-Level Adjustments, for FY2006
IRF A inIRF B inIRF C in
ComponentCounty, MACounty, NYCounty, CA
Federal prospective payment rate$23,159.20$23,159.20$23,159.20
for CMG 0109
Labor portion of federal payment$17,569.73$17,569.73$17,569.73
($23,159.20 x 0.75865)
Blended transition wage index1.02161.34490.9797
for the IRF’s location
$17,949.23 $23,629.53 $17,213.06
Wage-adjusted amount($17,569.73 x($17,569.73 x($17,569.73 x

1.0216) 1.3449) 0.9797)

Nonlabor-related amount$5,589.47$5,589.47$5,589.47
($23,159.20 x 0.24135)
$23,538.70 $29,219 $22,802.53
Wage-adjusted federal payment($17,949.23 +($23,629.53 +($17,213.06 +
$5,589.47) $5,589.47) $5,589/47)
Rural adjustment1.21301.001.1276
$28,552.44 $29,219 $25,712.13
Subtotal(23,538.70 x($29,219 x($22,802.53 x

1.2130) 1.00) 1.1276)

LIP adjustment1.03101.06121.1203

IRF A inIRF B inIRF C in
ComponentCounty, MACounty, NYCounty, CA
$29,437.57 $31,007.20 $28,805.30
Subtotal($28,552.40 x($29,219 x($25,712.13 x

1.0310) 1.0612) 1.1203)

Teaching adjustment1.001.0901.00
Total FY2006 adjusted federal$29,437.57$33,797.85$28,805.30
prospective payment for CMG($29,437.57 x($31,007.20 x($28,805.30 x

0109 1.00) 1.09) 1.00)

Source: CRS calculation based on information in FY2006 IRF-PPS regulation published in the
Federal Register on August 15, 2005 and September 30, 2005.
In addition to facility-level adjustments, an IRF may receive additional or
reduced Medicare payment for any given case, depending upon the Medicare
patient’s circumstances. Additional payments are made for cases that are high cost
outliers. A patient will be considered to be an outlier if the estimated cost of the case
exceeds an adjusted threshold amount. This cost is calculated by multiplying the
charge by the facility’s overall cost-to-charge ratio obtained from the latest settled or
tentatively settled cost report.70 An IRF will receive 80% of the difference between
the estimated cost of the case and the outlier threshold (modified by facility-level
adjustments). For FY2006, the unadjusted threshold amount is $5,129 (down from
$11,211 in 2005), which CMS estimates will result in total estimated outlier
payments of approximately 3% of total IRF-PPS payments.
Concluding Observations
The magnitude of Medicare’s spending on post-acute care, as well as the variety
of post-acute providers, underscores the importance of developing policies that
ensure beneficiaries receive the appropriate level of care and service intensity.
Policymakers remain concerned that payment incentives in the Medicare program
may influence the type of post-acute care provided and unnecessarily increase
program spending. However, there is little definitive information on Medicare’s
relative payment levels in different settings for the same quality of care and desired
outcomes for patients with particular needs. With respect to IRFs, the Medicare
statute gives the Secretary of HHS discretion to establish the criteria that these
facilities must meet in order to be exempt from the IPPS used to pay acute hospitals.
Recent administrative actions by CMS and its contractors to develop and enforce
these criteria have prompted congressional actions to delay enforcement of the
criteria. At this point, absent further action, CMS will enforce the 60% compliance
threshold starting July 1, 2005. However, the Deficit Reduction Act of 2005 includes
a provision that would extend the existing 60% compliance threshold for two years

70 If a facility’s cost to charge ratio is three standard deviations above the applicable national
average cost to charge ratio, then a ceiling on this ratio is imposed. Separate national cost
to charge ratios apply for urban and rural IRFs.

(until June 30, 2006), establish a 65% threshold for a 12-month period starting July
1, 2007, and establish the 75% threshold starting July 1, 2008. Under the current
regulation, the 75% threshold would start on July 1, 2007.