Medicare: More Specific Criteria Needed to Classify Inpatient	 
Rehabilitation Facilities (22-APR-05, GAO-05-366).		 
                                                                 
Medicare classifies inpatient rehabilitation facilities (IRF)	 
using the "75 percent rule." If a facility can show that during 1
year at least 75 percent of its patients required intensive	 
rehabilitation for 1 of 13 specified conditions, it may be	 
classified as an IRF and paid at a higher rate than is paid for  
less intensive rehabilitation in other settings. Medicare	 
payments to IRFs have grown steadily over the past decade. In	 
this report, GAO (1) identifies the conditions--on and off the	 
list--that IRF Medicare patients have and the number of IRFs that
meet a 75 percent threshold, (2) describes IRF admission criteria
and Centers for Medicare & Medicaid Services (CMS) review of	 
admissions, and (3) evaluates use of a list of conditions in the 
rule. GAO analyzed data on Medicare patients (the majority of	 
patients in IRFs) admitted to IRFs in FY 2003, spoke to IRF	 
medical directors, and had the Institute of Medicine (IOM)	 
convene a meeting of experts.					 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-05-366 					        
    ACCNO:   A22417						        
  TITLE:     Medicare: More Specific Criteria Needed to Classify      
Inpatient Rehabilitation Facilities				 
     DATE:   04/22/2005 
  SUBJECT:   Data collection					 
	     Eligibility criteria				 
	     Eligibility determinations 			 
	     Health care facilities				 
	     Health statistics					 
	     Medical services rates				 
	     Medicare						 
	     Policy evaluation					 
	     Inpatient care services				 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO 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.                                         **
**                                                              **
******************************************************************
GAO-05-366

Report to the Senate Committee on Finance and the House Committee on Ways
and Means

United States Government Accountability Office

GAO

April 2005

MEDICARE

More Specific Criteria Needed to Classify Inpatient Rehabilitation
Facilities

GAO-05-366

Contents

Letter 1

Results in Brief 6
Background 8
Fewer Than Half of All IRF Medicare Patients in 2003 Were Admitted for
Conditions on List in Rule, and Few IRFs Were Able to Meet a 75 Percent
Threshold 11
IRFs Vary in the Criteria Used to Assess Patients for Admission, and CMS
Does Not Routinely Review IRFs' Admission Decisions 19
Experts Differed on Adding Conditions to List in Rule but Agreed That
Condition Alone Does Not Provide Sufficient Criteria 22
Conclusions 27
Recommendations for Executive Action 28
Agency Comments and Comments from National Associations and Our Evaluation
29
Appendix I List of Conditions in CMS's 75 Percent Rule 31
Appendix II Scope and Methodology 33
Appendix III Rates of IRF Medicare Admissions from Hospitals by Top 19
DRGs of Patients Admitted to IRFs, Fiscal Year 2003 38
Appendix IV Comments from the Centers for Medicare & Medicaid Services 39
Appendix V GAO Contact and Staff Acknowledgments 43
GAO Contact 43
Acknowledgments 43

Tables

Table 1: Proportion of All IRF Medicare Patients Who Had Condition on List
in Rule, by Condition as Defined by Impairment Group, Fiscal Year 2003 14
Table 2: IRFs That Met Varying Threshold Levels for Medicare Patients
Admitted with Any of 13 Conditions on List in Rule in Fiscal Year 2003 17
Table 3: Criteria That Characterize Appropriate Patients for Admission, as
Reported by 12 IRFs 19

Figures

Figure 1: Conditions of All Medicare Patients Admitted to IRFs, as Defined
by Impairment Group, Fiscal Year 2003 13
Figure 2: Distribution of IRF Medicare Patients Who Did Not Have Condition
on List in Rule, by Condition as Defined by Impairment Group, Fiscal Year
2003 16

Abbreviations

CMG case-mix group CMS Centers for Medicare & Medicaid Services DRG
diagnosis-related group FI fiscal intermediary ICD-9-CM International
Classification of Diseases, Ninth Revision, Clinical Modification IOM
Institute of Medicine IPPS inpatient prospective payment system IRF
inpatient rehabilitation facility IRF-PAI Inpatient Rehabilitation
Facility-Patient Assessment Instrument IRF PPS inpatient rehabilitation
facility prospective payment system MEDPAR Medicare Provider Analysis and
Review NIH National Institutes of Health PPS prospective payment system
SNF skilled nursing facility

This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.

United States Government Accountability Office

Washington, DC 20548

April 22, 2005

The Honorable Charles E. Grassley Chairman The Honorable Max Baucus
Ranking Minority Member Committee on Finance United States Senate

The Honorable William M. Thomas Chairman The Honorable Charles B. Rangel
Ranking Minority Member Committee on Ways and Means House of
Representatives

The number of inpatient rehabilitation facilities (IRF) and Medicare
payments to these facilities have grown steadily over the past decade.
IRFs are intended to serve patients recovering from medical conditions
that typically require an intensive level of rehabilitation in an
inpatient setting.1, 2 The number of IRFs grew from 907 in 1992 to 1,256
in 2003. Medicare payments to IRFs grew from $2.8 billion in 1992 to an
estimated $5.7 billion for the care of over 500,000 Medicare patients in
2003, and payments are projected to grow to almost $9 billion per year by
2015. Because patients treated at IRFs require more intensive
rehabilitation than is provided in other settings, such as an acute care
hospital or a skilled nursing facility (SNF),3 Medicare pays for treatment
in an IRF at a higher rate than it pays for treatment in other settings.
To distinguish IRFs from other settings for payment purposes and to ensure
that Medicare patients needing less intensive services are not in IRFs,
the Centers for Medicare & Medicaid Services (CMS) relies on a regulation
commonly known as the "75 percent rule," which was initially issued in
1983 and most recently revised in 2004.4 The 2004 rule, which is being
implemented over a 3-year transition period, states that if a facility can
show that during a 12-month period at least 75 percent of all its
patients, including its Medicare patients, required intensive
rehabilitation services for the treatment of at least 1 of the 13
conditions listed in the rule,5 it may be classified as an IRF.6 The rule
allows the remaining 25 percent of patients to have other conditions not
listed in the rule. An IRF that does not comply with the requirements of
the 75 percent rule may lose its classification as an IRF and therefore no
longer be eligible for payment at a higher rate. In addition to the 75
percent rule, IRFs must meet six other facility criteria to be classified
as an IRF.7

1Under authority provided in the Social Security Act, the Secretary
defines a rehabilitation hospital and unit. See 42 U.S.C.
S:1395ww(d)(1)(B) (2000).

2Not all patients with a given condition may require the level of
rehabilitation provided in an IRF. For example, although a subset of
patients who have had a stroke may require the intensive level of care
provided by an IRF, others may be less severely disabled and require less
intensive services.

3In addition to IRFs, acute care hospitals, and SNFs, other settings that
provide rehabilitation services include long-term care hospitals,
outpatient rehabilitation facilities, and home health care.

IRF compliance with the requirements of the rule has been problematic, and
some IRFs have questioned the requirements of the rule. CMS data indicate
that in 2002 only 13 percent of IRFs had at least 75 percent of patients
in 1 of the 10 conditions on the list at that time. CMS suspended
enforcement of the rule in 2002. IRF officials have contended that the
list of conditions in the rule should be updated because of changes in
medicine that have occurred since the list was established in 1983 and the
concomitant expansion of the population that could benefit from inpatient
rehabilitation services. They have noted that their patients are older
than the population served in 1983 and are surviving longer with
conditions they may not have survived in earlier years. CMS issued a final
rule-effective July 1, 2004-that increased the number of conditions from
10 to 13, adding, for example, certain hip and knee joint replacements.8
The 2004 final rule also laid out a 3-year transition period during which
enforcement of the rule is to be resumed, with the threshold for
percentage of patients meeting the condition requirements being lowered to
50 percent for the first year and rising in stages to reach 75 percent for
the IRF's cost reporting period starting on or after July 2007.

4See 42 C.F.R. S:412.23(b)(2) (2004).

5The 13 conditions listed in the 2004 rule are stroke; spinal cord injury;
congenital deformity; amputation; major multiple trauma; hip fracture;
brain injury; neurological disorders; burns; certain active polyarticular
rheumatoid arthritis, psoriatic arthritis, and seronegative arthropathies;
certain systemic vasculidities with joint inflammation; severe or advanced
osteoarthritis involving two or more major weight-bearing joints meeting
certain criteria; and knee or hip joint replacement meeting certain
specific criteria. The specific criteria for knee or hip joint replacement
are that the patient must have undergone a knee or hip joint replacement,
or both, during an acute care hospitalization immediately preceding the
inpatient rehabilitation stay and also have had a bilateral procedure, or
be at least 85 years of age or older, or be extremely obese with a body
mass index of at least 50. For an annotated list of these conditions, see
appendix I.

6The time period is defined by CMS or the CMS contractor.

7To be classified as an IRF, a facility would also have to meet six other
regulatory criteria showing that it had (1) a Medicare provider agreement;
(2) a preadmission screening procedure; (3) medical, nursing, and therapy
services; (4) a plan of treatment for each patient; (5) a coordinated
multidisciplinary team approach; and (6) a medical director of
rehabilitation with specified training or experience. IRFs must also meet
other criteria identified in 42 C.F.R. S:412.22 (2004) and 42 C.F.R.
S:412.25 (2004).

IRFs need to be correctly classified to be distinguished from settings in
which less intensive rehabilitation is provided because the difference in
payments to IRFs and payments to these other settings can be substantial.
For example, the estimated Medicare per case payment in 2004 for a patient
who underwent a major joint and limb replacement of a lower extremity was
$17,135 to an IRF and $6,165 to a SNF. Similarly, the estimated per case
payment for a patient with a stroke was $34,196 to an IRF and $8,905 to a
SNF.9 Therefore, if IRFs are not correctly classified, Medicare is at risk
of making large overpayments to incorrectly classified facilities.
Medicare is also at risk of overpayment for individual patients in an IRF
if patients are admitted who could be treated in a less intensive setting.
IRFs are required to assess patients prior to admission to ensure they
require the level of services provided in an IRF, and CMS is responsible
for evaluating the appropriateness of individual admissions after the
patient has been discharged through reviews for medical necessity
conducted under contract by its fiscal intermediaries (FI).

The Conference Report that accompanied the Medicare Prescription Drug,
Improvement, and Modernization Act of 2003 directed us to issue a report,
in consultation with experts in the field of physical medicine and
rehabilitation, to assess whether the current list of conditions
represents a clinically appropriate standard for defining IRF services
and, if not, to determine which additional conditions should be added to
the list.10 In this report, we (1) identify the conditions that IRF
Medicare patients have, the number of these patients considered to have 1
of the 13 conditions, and the number of IRFs that meet the requirements of
the 75 percent rule; (2) describe how IRFs assess patients for admission
and whether CMS reviews admission decisions; and (3) evaluate the approach
of using a list of conditions in the 75 percent rule to classify IRFs.11

8See 69 Fed. Reg. 25752 (May 7, 2004).

9See MedPAC, Report to the Congress: New Approaches in Medicare, Ch. 5,
"Defining Long-Term Care Hospitals" (Washington, D.C.: June 2004),123. CMS
officials also reported that preliminary data showed that IRF payments
exceeded costs by approximately 17 percent in 2002, the first year of IRF
prospective payment.

10See H.R. Rep. 108-391, at 649 (2003).

To identify the conditions that IRF Medicare patients have, the number of
patients considered to have 1 of the 13 conditions, and the number of IRFs
that meet the requirements of the 75 percent rule, we obtained the
Inpatient Rehabilitation Facility-Patient Assessment Instrument (IRF-PAI)
records from CMS for Medicare patients admitted to IRFs in fiscal year
2003. We conducted our analyses on Medicare patients only, because CMS
records contained data on the largest number of IRFs and the majority of
patients in IRFs are covered by Medicare.12, 13 The IRF-PAI records
contain, for each Medicare patient, the impairment group code14
identifying the patient's primary condition and the diagnostic code from
the International Classification of Diseases, Ninth Revision, Clinical
Modification (ICD-9-CM) identifying the patient's comorbid condition.15 We
used these codes to determine whether we considered the patient's primary
or comorbid condition to be linked to a condition on the list in the
rule.16 We also obtained and analyzed Medicare claims records for fiscal
year 2003 to identify patients that had been discharged from an acute care
hospital to an IRF. We assessed the reliability of the IRF-PAI data by
interviewing agency officials knowledgeable about the data and by
interviewing other researchers who had conducted analyses using the
IRF-PAI data. For both the IRF-PAI data and the claims data we performed
electronic testing of required data elements. We determined that the data
were sufficiently reliable for this analysis. Although we applied
different threshold levels to illustrate the impact of the transition
period on the number of IRFs that meet the requirements of the rule, we
did not assess the appropriateness of any threshold level. Our analyses
used administrative data only, and estimates could be different if medical
records were used.17

11The Consolidated Appropriations Act, 2005, effectively prohibits the
Secretary of Health and Human Services from enforcing the 75 percent rule
and reclassifying IRFs as hospitals subject to the inpatient prospective
payment system until he either (1) determines that the current rule is not
inconsistent with the recommendations contained in our report or (2)
issues an interim rule revising the 75 percent rule. The appropriations
act provides for the Secretary to take such action no later than 60 days
after our report is issued. See Pub. L. No. 108-447, Div. F., Tit. II,
S:219, 118 Stat. 2809, 3141-42.

12We analyzed the 2003 data-the most recent data available at the
time-using the 13 conditions in the current regulation even though in
fiscal year 2003 there were 10 conditions on the list.

13Other data sources contained data on only a subset of IRFs. In addition,
analyses by RAND using the 10 conditions on the list at that time found
that the percentage of Medicare patients with the conditions on the list
in the rule was a good predictor of the percentage of total patients with
the conditions on the list in the rule. See Grace M. Carter, O. Hayden, et
al., "Case Mix Certification Rule for Inpatient Rehabilitation
Facilities," DRU-2981-CMS (Santa Monica, Ca.: May 2003.)

14The impairment group code identifies the medical condition that caused
the patient to be admitted to an IRF, and its sole function is to
determine payment rates. As a result, the impairment group codes describe
every patient in an IRF and include medical conditions that are on the
list in the rule as well as those that are not on the list since IRFs may
treat patients with conditions not on the list. In contrast, the list of
conditions in the rule describes the patient population that is to be
treated in an IRF to ensure that a facility is appropriately classified to
justify payment for the level of services furnished.

15As used in this report, a primary condition is the first or foremost
medical condition for which the patient was admitted to an IRF, and other
medical conditions may coexist in the patient as comorbid conditions, or
comorbidities.

To determine how IRFs assess patients for admission and whether CMS
reviews admission decisions, we conducted structured interviews. We
interviewed the medical director at each of 12 IRFs selected to vary by
region and level of compliance with the 75 percent rule. We also
interviewed the medical director (or designee) at each of the 10 FIs that
covered the states in which the 12 IRFs are located (out of a total of 30
FIs). In addition, we interviewed an official representing each of CMS's
10 regional offices to determine whether any IRFs had ever been
declassified based on failure to comply with the 75 percent rule, and we
interviewed three insurers and one regional managed care organization
about their procedures for referring enrollees to IRFs.

To evaluate the approach of using a list of conditions in the 75 percent
rule to classify IRFs, we contracted with the Institute of Medicine (IOM)
of The National Academies to convene a 1-day meeting of clinical experts
in physical medicine and rehabilitation, including physicians,
rehabilitation nurses, physical therapists, occupational therapists, a
speech and language therapist, and clinical researchers in the field
(referred to in this report as "the experts IOM convened").

16Throughout this report, the "list in the rule" refers to the list of 13
conditions as specified in the 2004 75 percent rule, and when we say that
a condition is on (or off) the list, we mean that we have (or have not)
been able to link the condition as identified in the IRF-PAI record to a
condition on the list in the rule.

17We followed the instructions CMS provided to FIs for them to use as a
first step to "presumptively verify compliance" using the list of codes in
the manual to estimate how many patients have one of the conditions on the
list in the rule as recorded on the IRF-PAI instrument. (See CMS,
"Medicare Claims Processing,"CMS Manual System, pub. 100-04, Transmittal
347 (Baltimore, Md.: Oct. 29, 2004.))

In total, we talked with 106 individuals, of whom 65 were clinicians,
including the experts IOM convened. We conducted our work from May 2004
through April 2005 in accordance with generally accepted government
auditing standards. (For a complete description of our scope and
methodology, see app. II.)

                                Results in Brief

In fiscal year 2003, fewer than half of all IRF Medicare patients were
admitted for having a condition on the list in the 75 percent rule, and
few IRFs admitted at least 75 percent of their patients for one of those
conditions. The largest group of patients admitted to IRFs in 2003 had
orthopedic conditions, not all of which were on the list in the rule. In
addition, fewer than half of all IRF patients were admitted for a primary
condition that was on the list, with the proportion increasing to over
three-fifths when comorbid conditions on the list were counted, as they
would be during the rule's 3-year transition period. Almost half of
patients with conditions that were not on the list were admitted for
orthopedic conditions, and among those the largest group was joint
replacement patients. Although some joint replacement patients may need
admission to an IRF, our analysis showed that few of these patients had
comorbidities that suggested a possible need for the intensity of services
offered by an IRF. Additionally, we found that only 6 percent of IRFs in
fiscal year 2003 were able to meet a 75 percent threshold, and many IRFs
may not be able to meet the requirements of the rule as the threshold
increases to 75 percent during the transition period. CMS has not
generally declassified IRFs based on their failure to comply with the 75
percent rule.

IRFs varied in the criteria used to assess patients for admission, and CMS
has not routinely reviewed IRFs' admission decisions. Among the IRF
officials we interviewed, the criteria varied by facility and included
patient characteristics such as function in addition to condition.
Admission decisions may also be influenced by an IRF's level of compliance
with the 75 percent rule's list of conditions. The IRF officials we
interviewed reported that they tracked their facility's level of
compliance with the rule's list of conditions and that the decision to
admit a given patient could be affected by the IRF's compliance level at
that time. CMS, working through its FIs, has not routinely reviewed IRF
admission decisions, although it reported that such reviews could be used
to target problem areas.

The experts IOM convened and other experts we interviewed differed on
whether conditions should be added to the list in the 75 percent rule but
agreed that condition alone does not provide sufficient criteria to
identify types of patients appropriate for IRFs. The experts IOM convened
questioned the strength of the evidence for adding conditions to the list
in the rule. They reported that the evidence on the benefits of IRF
services is variable and the evidence on the benefits of such services for
certain orthopedic conditions is particularly weak, and they called for
further research to identify the types of patients that need inpatient
rehabilitation and to understand the effectiveness of IRFs in comparison
with other settings of care. Other experts we interviewed did not agree on
whether conditions, including a broader category of joint replacements,
should be added to the list in the rule. Experts, including those convened
by IOM, agreed that condition alone is insufficient for identifying
appropriate types of patients for inpatient rehabilitation, since within
any condition only a subgroup of patients require the level of services of
an IRF, and contended that functional status should also be considered.
The experts IOM convened suggested factors to use in classifying IRFs,
including both patient and facility characteristics.

To help ensure that IRFs can be classified appropriately and that only
patients needing the IRF level of services are admitted to them, we
recommend that CMS ensure that FIs routinely conduct targeted reviews for
medical necessity for IRF admissions; that CMS conduct additional
activities to encourage research on the effectiveness of intensive
inpatient rehabilitation and factors that predict patient need for these
services; and that CMS use the information obtained from reviews for
medical necessity, research activities, and other sources to refine the
rule to describe more thoroughly the subgroups of patients within a
condition that require IRF services, possibly using functional status or
other factors in addition to condition.

In commenting on a draft of this report, CMS stated that our work would be
of assistance to the agency in examining issues related to patient
coverage and the classification of IRFs. CMS generally agreed with our
recommendations. Although CMS indicated its intent to follow our
recommendation to more thoroughly describe subgroups of patients within a
condition, it said it wanted to carefully consider this action and
potentially base its descriptions on future research. We clarified
language in the recommendation to encourage CMS to obtain research for
this effort. CMS agreed on the need to encourage research and said it
would collaborate with the National Institutes of Health (NIH). CMS also
agreed that targeted reviews for medical necessity are necessary and said
that it expected resources to be directed toward areas of risk. In its
technical comments, CMS also noted we analyzed data from fiscal year 2003,
when the rule was not being enforced, and said that this could have
affected our findings. Other organizations that reviewed the report-the
American Hospital Association, the American Medical Rehabilitation
Providers Association, and the Federation of American Hospitals-also
raised concerns about our use of fiscal year 2003 data. We analyzed a
sample of data from July through December 2004, the first 6 months after
the rule took effect, and found no material difference for the same time
period in fiscal year 2003 data.

                                   Background

While the 75 percent rule has been in effect in one form or another for
over two decades, the current payment system and review procedures for
IRFs went into effect in recent years.

History of the 75 Percent Rule

The Social Security Amendments of 1983 changed the Medicare hospital
payment system from a cost-based retrospective reimbursement system to a
prospective system known as the inpatient prospective payment system
(IPPS), under which hospitals receive a per discharge payment for a
diagnosis-related group (DRG).18 However, the amendments excluded
"rehabilitation hospitals," and so IRFs continued to be paid under a
reasonable-cost-based retrospective system. Before the IPPS was
implemented, CMS consulted with the Joint Commission on Accreditation of
Healthcare Organizations (JCAHO)19 and other accrediting organizations to
determine how to classify IRFs, that is, distinguish them from other
facilities for payment purposes. The 75 percent rule was established for
that purpose in 1983.20 To develop the original list of conditions in the
75 percent rule, CMS relied, in part, on information from the American
Academy of Physical Medicine and Rehabilitation, the American Congress of
Rehabilitation Medicine, the National Association of Rehabilitation
Facilities, and the American Hospital Association.21 According to CMS, the
conditions on the list accounted for approximately 75 percent of the
admissions to IRFs when the original list was developed. In January 2002 a
prospective payment system (PPS) was implemented for IRFs-known as the
inpatient rehabilitation facility prospective payment system (IRF PPS).

18See Pub. L. No. 98-21, S:601(e), 97 Stat. 65, 152-162 (1983) (codified
at 42 U.S.C. S:1935ww(d) (2000)).

19At that time, JCAHO was known as the Joint Commission on Accreditation
of Hospitals.

20See 48 Fed. Reg. 39752 (Sept. 1, 1983).

On June 7, 2002, CMS suspended the enforcement of the 75 percent rule
after its study of FIs, which have responsibility under contract with CMS
for verifying compliance with the rule, revealed that they were using
inconsistent methods to determine whether an IRF was in compliance and
that in some cases IRFs were not being reviewed for compliance at all.
Specifically, CMS found that only 20 of the 29 FIs conducted reviews for
IRF compliance with the 75 percent rule and that the FIs that did these
reviews used different methods and data sources. In 2004, CMS standardized
the verification process that the FIs were to use to determine if an IRF
met the classification criteria, including how to determine whether a
patient is considered to have 1 of the 13 conditions.

The 2004 Final Rule

When the final rule was made effective on July 1, 2004, a transition
period was established for IRFs to meet the requirements of the rule. In
addition to lowering and then increasing the threshold, the transition
period allows a patient to be counted toward the required threshold if the
patient is admitted for either a primary or comorbid condition on the list
in the rule. But at the end of the transition period, a patient cannot be
counted toward the required threshold on the basis of a comorbidity on the
list in the rule. The requirements of the transition period are as
follows:22

           o  July 1, 2004, to June 30, 2005: 50 percent threshold, counting
           comorbidities
           o  July 1, 2005, to June 30, 2006: 60 percent threshold, counting
           comorbidities
           o  July 1, 2006, to June 30, 2007: 65 percent threshold, counting
           comorbidities

           Effective July 1, 2007, the threshold will be 75 percent, not
           counting comorbidities.

           During the 3-year transition period, CMS plans to analyze claims
           and patient assessment data to evaluate if and how the 75 percent
           threshold should be modified. In addition, the agency has
           announced its willingness to consider alternative policy proposals
           to the 75 percent rule submitted during this period. In the past,
           CMS has declined requests to modify the rule's threshold or list
           of conditions, citing a lack of supporting or objective data from
           the clinical community. However, in the final rule, the agency
           solicited "objective data or evidence from well-designed research
           studies" that would support a change in the rule's 75 percent
           threshold or list of conditions.23 Also, because of the relative
           absence of clinical research studies in the peer-reviewed medical
           literature, CMS contracted with NIH to convene one meeting of a
           research panel to review the current medical literature24 and
           identify priorities for conducting studies on inpatient
           rehabilitation.

           Beginning in January 2002, CMS implemented the IRF PPS to pay IRFs
           on a per-discharge basis. Payment is contingent on an IRF's
           completing a patient assessment after admission and transmitting
           the resulting data to CMS. The Inpatient Rehabilitation
           Facility-Patient Assessment Instrument (IRF-PAI) includes
           identification of an impairment group code that identifies the
           impairment group, or the condition that requires admission to
           rehabilitation. The patient's comorbidities are also recorded on
           the IRF-PAI.

           The impairment group code is combined with other information on
           the IRF-PAI to classify the patient into 1 of 100 case-mix groups
           (CMG). Patients are assigned to a CMG based on the impairment
           group code, age, and levels of functional and cognitive
           impairment. The CMG determines the payment the IRF will receive
           for a patient. Each CMG is weighted to account for the relative
           difference in resource use across all CMGs. Within each CMG, the
           weighting factors are "tiered" based on the estimated effect of
           comorbidities. Each CMG has four payment tiers reflecting the
           level of comorbidities.25

           CMS contracts with FIs, the entities that conduct compliance
           reviews, to conduct reviews for medical necessity to determine
           whether an individual admission to an IRF was covered under
           Medicare. FIs were specifically authorized to conduct reviews for
           medical necessity for inpatient rehabilitation services beginning
           in April 2002.26 According to the Medicare Benefit Policy Manual,
           two basic requirements must be met if inpatient hospital stays for
           rehabilitation services are to be covered: (1) the services must
           be reasonable and necessary, and (2) it must be reasonable and
           necessary to furnish the care on an inpatient hospital basis,
           rather than in a less intensive facility, such as a SNF, or on an
           outpatient basis.27 Determinations of whether hospital stays for
           rehabilitation services are reasonable and necessary must be based
           on an assessment of each beneficiary's individual care needs.

           Fewer than half of all IRF Medicare patients in fiscal year 2003
           were admitted for conditions on the list in the 75 percent rule.
           The patients admitted in 2003 had a variety of conditions, not all
           of which were on the list in the rule. Nearly half of the patients
           admitted for conditions not on the list were admitted for
           orthopedic conditions. The largest group of patients admitted for
           orthopedic conditions not on the list were admitted for joint
           replacements that did not meet the list's specific criteria for
           joint replacement. Relatively few of these patients had comorbid
           conditions that suggested a possible need for the intensive level
           of rehabilitation provided in IRFs. Additionally, we found that
           based on the fiscal year 2003 data few IRFs were able to meet a 75
           percent threshold.

           Medicare patients were admitted to IRFs in fiscal year 2003 with a
           variety of conditions, as defined by the impairment group codes we
           analyzed. Forty-two percent of the 506,662 Medicare patients
           admitted to IRFs in 2003 were admitted with orthopedic conditions,
           representing the largest category of patients.28 Figure 1 shows
           the distribution of all the conditions, based on impairment group
           codes, for which patients were admitted to IRFs in fiscal year
           2003. The largest impairment group consisted of patients admitted
           for joint replacement.29

21This information included Health Care Financing Administration Technical
Assistance Document No. 24, "Sample Screening Criteria for Review of
Admissions to Comprehensive Medical Rehabilitation Hospitals/Units,"
prepared by the Committee on Rehabilitation Criteria for the Professional
Standards Review Organization of the American Academy of Physical Medicine
and Rehabilitation and the American Congress of Rehabilitation Medicine.

22The threshold level applies to an IRF's cost reporting period beginning
on or after July 1 of each year.

Payment and Review for Medical Necessity

23See 69 Fed. Reg. 25752 (May 7, 2004).

24CMS contracted with the Agency for Healthcare Research and Quality to
prepare a literature review for the NIH meeting.

     Fewer Than Half of All IRF Medicare Patients in 2003 Were Admitted for
    Conditions on List in Rule, and Few IRFs Were Able to Meet a 75 Percent
                                   Threshold

25There are a total of 385 groups because five special CMGs do not have
tiers.

26Prior to this time, Quality Improvement Organizations had this
authority. CMS Transmittal 21 made clear that FIs have the authority to
review admissions to IRFs.

27Rehabilitative care in a hospital, rather than in a SNF or on an
outpatient basis, is considered to be reasonable and necessary when a
patient requires a more coordinated, intensive program of multiple
services than is generally found outside of a hospital (Medicare Benefit
Policy Manual, chapter 1, Section 110.1).

Medicare Patients Admitted to IRFs in 2003 Had Variety of Conditions

28Patients with orthopedic conditions include all patients with an
impairment group code related to unilateral or bilateral hip fracture,
femur fracture, pelvic fracture, unilateral or bilateral hip and/or knee
replacement, or other orthopedic patients.

29To determine whether admissions changed after enforcement of the rule,
we compared admissions for the largest group of patients, joint
replacement patients, between July through December 2003 and July through
December 2004. There was no material difference overall. Across all IRFs,
the percentage of Medicare patients admitted to an IRF whose primary
condition was joint replacement declined by 0.1 percentage point. Among
the top 10 percent of IRFs admitting the highest proportion of Medicare
joint replacement patients, the percentage of all Medicare patients
admitted for a joint replacement declined by about 6 percentage points.

Figure 1: Conditions of All Medicare Patients Admitted to IRFs, as Defined
by Impairment Group, Fiscal Year 2003

Fewer Than Half of All IRF Medicare Patients Were Admitted for Conditions on
List in Rule

Fewer than half of the Medicare patients (222,316 of the 506,662 patients)
admitted in fiscal year 2003 were admitted for a primary condition that
was on the list in the 75 percent rule. Using the impairment group codes
assigned to these patients at the time of their admission, we determined
that in fiscal year 2003 less than 44 percent of IRF admissions had a
primary condition that was on the list in the rule. However, when comorbid
conditions that were on the list were counted-as they would be during the
transition period-the number of patients having a listed condition rose to
311,740 (62 percent) of IRF patients in that year. (See table 1.)

Table 1: Proportion of All IRF Medicare Patients Who Had Condition on List
in Rule, by Condition as Defined by Impairment Group, Fiscal Year 2003

                                                           Patients whose
                                    Patients whose       primary or comorbid
                                 primary condition was  condition was on list
                                    on list in rule            in rule
                   Total number             Percentage             Percentage 
Condition, as    of patients            of patients            of patients 
defined by                in                     in                     in 
impairment        impairment             impairment             impairment 
group                  group    Number        group    Number        group 
Joint                                                                      
replacements         121,528    15,761         13.0    61,890         50.9
Stroke                85,516    85,516        100.0    85,516        100.0 
Hip fracture          51,467    51,467        100.0    51,467        100.0 
Other                                                                      
orthopedic                                                    
conditions            40,359         0          0.0    11,168         27.7
Medically                                                                  
complex               29,148         0          0.0     6,363         21.8
Cardiac               28,011         0          0.0     4,296         15.3 
Debility              27,208         0          0.0     5,784         21.3 
Neurologic                                                                 
conditions            23,422     9,933         42.4    16,846         71.9
Spinal cord                                                                
dysfunction           21,207    21,207        100.0    21,207        100.0
Brain                                                                      
dysfunction           17,733    15,694         88.5    16,885         95.2
Arthritis             16,195     5,372         33.2     7,874         48.6 
Amputation            14,448    13,165         91.1    13,652         94.5 
Pain syndromes        10,925         0          0.0     2,078         19.0 
Pulmonary                                                                  
disorders             10,009         0          0.0     1,393         13.9
Other disabling                                                            
impairments            5,258         0          0.0     1,113         21.2
Major multiple                                                             
trauma                 3,658     3,658        100.0     3,658        100.0
Burns                    345       345        100.0       345        100.0 
Congenital                                                                 
deformities              198       198        100.0       198        100.0
Developmental                                                              
disability                27         0          0.0         7         25.9
Total (overall                                                             
percentage)          506,662   222,316         43.9   311,740         61.5

Source: GAO analysis of CMS IRF-PAI data.

Note: CMS's Medicare Claims Processing Manual lists the specific codes
that we used to determine whether a patient's condition was on the list in
the rule. See CMS, "Medicare Claims Processing," CMS Manual System, pub.
100-04, Transmittal 347 (Baltimore, Md.: Oct. 29, 2004.)

The amount of increase that occurred when comorbid conditions were counted
varied by impairment group. For some impairment groups, the percentage of
patients who had a condition on the list in the rule substantially
increased when comorbidities were counted. For example, the percentage of
joint replacement patients having a listed condition increased from 13
percent to 51 percent by virtue of their comorbidities. The comorbidity
that qualified over 90 percent of this group was some form of arthritis.30
In contrast, the increase was lower for patients in the medically complex,
cardiac, debility, pain syndrome, and pulmonary disorder impairment
groups, increasing between 14 percentage points and 22 percentage points.
The comorbidity that qualified about one-third of cardiac and debility
patients was stroke, and the comorbidity that qualified over one-third of
pulmonary patients was a neurological condition.

Almost Half of IRF Medicare Patients That Did Not Have Condition on List in Rule
Were Admitted for Orthopedic Conditions

Almost half of the 194,922 IRF Medicare patients that did not have a
condition on the list in the rule, either as a primary condition or as a
comorbid condition, were admitted for orthopedic conditions. (See fig. 2.)
The single largest group of patients who did not have a condition on the
list were the joint replacement patients whose condition did not meet the
list's specific criteria for joint replacements.31 Over 30 percent of
patients who did not have a condition on the list had been admitted to
IRFs for joint replacement, with another 15 percent having been admitted
for "other orthopedic," that is, any orthopedic condition other than hip
fractures or joint replacements. The next largest group, cardiac patients,
represented 12 percent.

30The forms of arthritis include osteoarthritis, rheumatoid arthritis, and
systemic vasculidities. The extent to which these codes refer to arthritis
in the joint that was replaced as opposed to active arthritis following
the procedure cannot be determined from these data. The IRF-PAI training
manual generally encouraged coders to be comprehensive, instructing them
to list "ALL comorbid conditions, not just those conditions that may
affect Medicare payment." (CMS, IRF-PAI Training Manual, rev. Jan. 16,
2002 (Baltimore, Md.: 2002), II-17.))

31See footnote 5.

Figure 2: Distribution of IRF Medicare Patients Who Did Not Have Condition
on List in Rule, by Condition as Defined by Impairment Group, Fiscal Year
2003

aIncludes joint replacement patients who had a unilateral procedure and
were under age 85, and therefore did not meet the specific criteria for
joint replacements set out in the 75 percent rule. Codes from CMS for body
mass index were not available.

Relatively Few Medicare Joint Replacement Patients in IRFs Had Comorbid
Conditions That Suggested Possible Need for IRF Level of Services

Although some joint replacement patients may need the level of services of
an IRF, such as those who have a comorbid condition that significantly
affects their level of function, our analysis of the case-mix groups used
for payment purposes suggests that relatively few of the Medicare joint
replacement patients currently admitted by IRFs fit this description.32 In
particular, 87 percent of joint replacement patients admitted in fiscal
year 2003 had unilateral procedures and were less than 85 years of age,
and thus did not fit the criteria for joint replacement on the list in the
rule based on their primary condition. Of the joint replacement patients
who did not fit the criteria based on their primary condition, over 84
percent were in a payment tier with no comorbidities that affected
costs.33

32One of the experts at the meeting convened by IOM stated that the field
has suggested that joint replacement patients in the lowest comorbidity
tiers potentially could be treated in another setting.

Few IRFs Were Able to Meet a 75 Percent Threshold

Only 6 percent of IRFs were able to meet the requirements of full
implementation of the rule that would be in place at the end of the
transition period, that is, a 75 percent threshold not counting
comorbidities. Our analysis of fiscal year 2003 data for Medicare patients
admitted to IRFs, which used the current list of 13 conditions, showed
that as the threshold level increased from 50 percent to 75 percent and
both primary and comorbid conditions were counted, progressively fewer
IRFs were able to meet the higher threshold levels. (See table 2.) In
addition, when the count was based only on whether the patient's primary
condition was on the list in the rule, as it would be after the transition
period, even fewer IRFs met the requirements of the rule. However, many
IRFs were able to meet the lower thresholds that would be in place earlier
in the transition period. Over 80 percent of IRFs were able to meet a 50
percent threshold based on the primary conditions or comorbid conditions
of the patients they admitted in 2003.

Table 2: IRFs That Met Varying Threshold Levels for Medicare Patients
Admitted with Any of 13 Conditions on List in Rule in Fiscal Year 2003

                                                     Percentage of IRFs 
               Percentage of IRFs that met threshold that met threshold 
Compliance   based on either primary condition or    based solely on 
threshold             related comorbid conditions  primary condition 
50 percent                                     85                 39 
60 percent                                     62                 20 
65 percent                                     50                 14 
75 percent                                     27                  6 

Source: GAO analysis of CMS IRF-PAI data.

33The IRF PPS identifies three sets of comorbidities that past experience
has shown to be associated with either a low, medium, or high increment in
patient costs. IRF patients who have none of these comorbidities are
placed in a fourth payment category, or tier. These comorbidities affect
the payment rate to an IRF for a specific patient and are different from
the consideration of whether a patient has a comorbidity that is 1 of the
13 conditions on the list in the rule. Joint replacement patients without
these comorbidities still vary substantially in the degree of impairment
they present, as reflected in their placement among the different CMGs.
Across the six joint replacement-related CMGs, the proportion of patients
in the tier with no such comorbidities ranged from 74 percent to 91
percent.

Some IRF officials are concerned that they may have to limit admissions in
order to comply with the rule and that some IRFs may have to close or
reduce beds.34 Some of the IRF officials we interviewed reported that as
the threshold of the rule increases they expect to limit admissions for
patients with conditions not on the list in the rule. One IRF official
estimated that the facility's revenues would decrease by 40 percent by the
third year of the rule's transition period, severely harming the facility
financially and affecting access to care, and another IRF official
reported that the facility expected its census to drop by half, which
would affect the number of beds it could operate and staff it could
employ. An IRF official whose facility was meeting the 75 percent
threshold said that if the facility fell below the threshold, it would
limit admissions to remain in compliance.

IRFs have not generally been declassified based on the failure to comply
with the 75 percent rule, and CMS recently clarified instructions for FIs
to use to conduct compliance assessments. Officials from CMS's 10 regional
offices reported that no IRFs had been declassified in at least the past 5
years.35 When CMS found that FIs were using different approaches to
conduct compliance assessments, it determined that one cause was that the
CMS manuals did not detail the methodology FIs should use to perform the
reviews. Following CMS's modifications of the rule, it issued new
instructions in a program transmittal that defined and standardized the
procedures that FIs are to use to conduct compliance assessments, and some
FI officials we interviewed reported that instructions were clearer and
more detailed than in prior years.

34The American Hospital Association and the American Medical
Rehabilitation Providers Association, which represent IRFs, have also
reported concern with the impact of the rule on the field. They estimated
that in the first year almost 25 percent of IRFs would not meet the
requirements of the rule and that when the rule is fully implemented
following the transition period 80 percent of IRFs would not meet the
rule, which could force them to discontinue services or close.

35One CMS regional office official reported that five or six IRFs had been
declassified in the mid-1990s or earlier, but none since then.

 IRFs Vary in the Criteria Used to Assess Patients for Admission, and CMS Does
                 Not Routinely Review IRFs' Admission Decisions

The criteria IRFs used to assess patients for admission varied by
facility, and CMS has not routinely reviewed IRFs' admission decisions. In
particular, IRFs used a range of criteria in making admission decisions,
including patient characteristics such as function, in addition to
condition. Admission decisions may also be influenced by an IRF's level of
compliance with the 75 percent rule's list of conditions. CMS, working
through its FIs, has not routinely reviewed IRF admission decisions for
medical necessity, although the CMS officials reported that such reviews
could be used as a means to target problems.

IRFs Use Variety of Criteria, Including Functional Status, to Assess Patients
for Admission

The IRF officials we interviewed varied in the criteria they used to
characterize the patients that were appropriate for admission. (See table
3.) The number of criteria they reported using ranged from two to six,
with no IRF reporting that it relied on a single criterion for admission.

Table 3: Criteria That Characterize Appropriate Patients for Admission, as
Reported by 12 IRFs

                                               Number of IRFs reporting 
Criteria                                            use of criterion 
Potential to return to home/community,                               
discharge plan                                                     8
Need for/ability to tolerate 3 hours of                              
therapy daily                                                      8
Functional level/potential for functional                            
improvement                                                        6
Medical issues, requirement for inpatient                            
monitoring, level of medical stability                             6
Need for two types of therapies                                    3 
Cognitive ability to learn                                         3 
Patient willingness to participate in                                
therapy                                                            2
Need for 24-hour nursing care                                      2 
Family support, expectations                                       2 
Diagnosis                                                          2 
Need for multidisciplinary approach                                1 
3- to 4-week length of stay                                        1 
Age                                                                1 
Comorbidities that affect function                                 1 

Source: GAO analysis of IRF officials' interview responses.

Whereas some IRF officials reported that they used function to
characterize patients who were appropriate for admission (e.g., patients
with a potential for functional improvement), as shown in table 3, others
said they used function to characterize patients not appropriate for
admission (e.g., patients whose functional level was too high, indicating
that they could go home, or too low, indicating that they needed to be in
a SNF). In combination, all the IRF officials we interviewed evaluated a
patient's function when assessing whether a patient needed the level of
services of an IRF, and almost half of the IRF officials interviewed
stated that function was the main factor that should be considered in
assessing the need for IRF services.

The IRF officials we interviewed reported that they did not admit all the
patients they assessed. They estimated that the proportion of patients
they assessed but did not admit ranged from 5 percent to 58 percent.36
Most patients were admitted to IRFs from an acute care hospital, and the
IRF officials reported receiving referrals from as few as 1 hospital to as
many as 55 hospitals.37 The IRF typically received a request from a
physician in the acute care hospital requesting a medical consultation
from an IRF physician, or from a hospital discharge planner or social
worker indicating that they had a potential patient. An IRF staff
member-usually a physician and/or a nurse38-conducted an assessment prior
to admission to determine whether to admit a patient.

Admission Decisions May Also Be Influenced by IRF's Level of Compliance with
Rule's List of Conditions

In addition to individual patient characteristics, admission decisions may
also be influenced by an IRF's level of compliance with the 75 percent
rule's list of conditions. All the IRF officials we interviewed were able
to track their own facility's compliance level regularly, and said they
tracked it generally on a daily, weekly, or monthly basis. Some IRF
officials we interviewed reported that the admission decision for a given
patient may be affected by the IRF's compliance level at that time. For
example, on a day when the facility is at the required level of compliance
a patient with a certain condition that is not on the list in the rule may
be admitted, but on another day when the facility is below its compliance
level a patient with the same condition might not be admitted.39 Half of
the IRF officials said that when the rule is enforced they expect they
will try to admit more patients with conditions on the list in the rule.

36The most common response, by 7 of the 12 IRFs, was between 30 percent
and 40 percent.

37For hospital-based IRFs (10 of the 12 interviewed), the percentage of
referrals from the parent hospital ranged from 25 percent to 99 percent,
with 3 reporting that less than half their patients came from the parent
hospital.

38Most IRFs reported that the assessment was done by a physician and/or a
nurse, although one IRF reported that it was done by a recreational
therapist.

CMS Has Not Routinely Reviewed Admission Decisions

CMS, working through its FIs, has not routinely reviewed IRF admission
decisions for medical necessity. Among the 10 FI officials we interviewed,
over half were not conducting reviews of patients admitted to IRFs. Those
that were doing reviews used different approaches for selecting records to
be reviewed, such as focusing only on the largest IRFs that failed to
comply with the rule or requesting a few records from each IRF in its
service area. CMS officials estimated that less than 1 percent of
admissions in facilities excluded from IPPS, such as IRFs, are reviewed,
and reported that such reviews could be used as a means to target problems
or vulnerabilities.

Among the experts IOM convened and other experts we interviewed, it was
stated that because there has been no routine review for medical necessity
in IRFs, some IRFs have become "sloppy" in their admitting practices and
have taken a "laissez-faire attitude" toward admitting patients. This
perspective is borne out through ad hoc studies done by three FIs that
found inadequate justification for admission. For example, in one study an
FI official reviewed about 3,000 medical charts and reported that the need
for inpatient rehabilitation was unclear in about 30 percent to 40 percent
of the IRF patients' charts reviewed.40 The other two FIs reviewed fewer
cases, but found a higher proportion of patients in IRFs who did not
appear to need inpatient rehabilitation.

In contrast to CMS's approach, private payers rely on individual
preauthorization to ensure that the most appropriate patients are admitted
to IRFs. Of the three major insurers and one managed care plan whose
officials we interviewed, all required preauthorization for each admission
to an IRF when determining whether a specific patient should be admitted,
judging each case individually. In making their decisions, they relied on
a variety of factors, which differed from payer to payer, including
diagnosis, symptoms, treatment plan, the need for and the patient's
ability to participate in 3 hours of daily therapy, the need for care by a
physiatrist,41 and the potential for an IRF admission to provide an
earlier discharge from the acute care hospital (compared to a possibly
longer stay in the acute care hospital with discharge to home or a SNF).
Three private payers we spoke with indicated that IRFs are generally paid
on a per diem basis, and all said that patients are monitored by the
insurer or health plan throughout their IRF stay.

39Other experts also reported about the potential for the opposite to
happen. For example, a patient may have a condition on the list and not
need the intensity of services of an IRF, but still be admitted if the
facility wants to increase its compliance level.

40We did not conduct an independent review of these reported results.

Experts Differed on Adding Conditions to List in Rule but Agreed That Condition
                   Alone Does Not Provide Sufficient Criteria

The experts IOM convened and other experts we interviewed differed on
whether conditions should be added to the list in the 75 percent rule but
agreed that condition alone does not provide sufficient criteria to
identify the types of patients appropriate for IRFs. The experts IOM
convened questioned the strength of the evidence for adding conditions to
the list. They reported that the evidence on the benefits of IRF
services-particularly for certain orthopedic conditions-is variable, and
they called for further research. Other experts did not agree on whether
conditions, including a broader category of joint replacements, should be
added to the list. The experts IOM convened and other experts agreed that
condition alone is insufficient for identifying appropriate patients and
contended that functional status should also be considered. The experts
IOM convened suggested factors to use in classifying IRFs, including both
patient and facility characteristics.

Experts IOM Convened Questioned Evidence for Adding Conditions to List in Rule,
Finding Evidence for Certain Orthopedic Conditions Particularly Weak, and Called
for More Research

The experts IOM convened generally questioned the strength of the evidence
for the conditions suggested for addition to the list in the rule. Some of
them reported that there was little information available on the need for
inpatient rehabilitation for cardiac, transplant, pulmonary, or oncology
patients. One of them stated that inpatient rehabilitation may be the best
way of caring for patients who have weakened physically due to long
hospital stays but added that "we simply do not know." The same expert
also cited a study that showed that inpatient rehabilitation services made
a difference for patients with metastatic spine cancer and noted that this
result was unexpected and could indicate that "clinical intuition" on the
benefits of inpatient rehabilitation may not always be reliable.

41A physiatrist is a physician who specializes in physical medicine and
rehabilitation.

For conditions currently on the list in the rule, the experts IOM convened
reported varying degrees of strength in the evidence on the benefits of
IRF services. Although the experts IOM convened did not comment on every
condition on the list, the group generally agreed that the data on the
benefits of intensive inpatient rehabilitation for stroke are
"incontrovertible." For certain other conditions on the list, such as
spinal cord injury and traumatic brain injury, they reported that it is
reasonable to expect intensive inpatient rehabilitation to provide good
outcomes because these patients need intensive training about self-care
and patients with traumatic brain injury may also require behavioral
services. One expert questioned the strength of the evidence related to
hip fractures, saying it was unclear whether patients with a hip fracture
would be better served by sending them home right away, by putting them in
an IRF, or by giving them some combination of intensive inpatient
rehabilitation, home health care, or care in a SNF.

The condition the experts IOM convened discussed most was joint
replacement, which was the most common condition for patients admitted to
IRFs and is included on the list of conditions in the rule but only under
certain circumstances. In general, they reported that, except for a few
subpopulations, uncomplicated unilateral joint replacement patients rarely
need to be admitted to an IRF.42 For example, one of the experts said that
admission to an IRF of a healthy person with an uncomplicated joint
replacement is an example of a practice that is not evidence-based, and
others said that there are no data and little evidence on the
effectiveness of intensive inpatient rehabilitation for elective joint
replacement patients. Another expert stated that the evidence on the
benefits of IRF services for hip fracture and joint replacement patients
is "very, very weak," that orthopedics is the "heart of the issue" related
to the list of conditions in the rule, and that a panel of clinicians
should be convened to focus solely on the orthopedic conditions.

Most of the experts IOM convened called for more research in several
areas, including which types of patients can be treated best in IRFs and
the effectiveness of IRFs in comparison with other settings of care. CMS
has also identified questions for a future research agenda that can assess
the efficacy of rehabilitation services in various settings.43 CMS may
also undertake other activities, such as periodically holding additional
meetings with researchers or encouraging observational studies, as well as
soliciting comments from the public for additional studies.

42We interviewed a leading orthopedic physician who said that unilateral
joint replacement patients rarely require admission to an IRF following
surgery, the exceptions including patients with a surgical complication,
previous stroke, polio, or heart transplant because such patients need
close medical supervision. In addition, three of the four officials we
interviewed at major insurers and a managed care plan generally agreed
that unilateral joint replacement patients rarely require admission to an
IRF, unless there is an active comorbidity or accompanying complex medical
problem. One reported that an IRF referral for a unilateral joint
replacement patient was a "red flag" that called for closer review.

IRF Officials Differed on Whether Conditions, Including More Broadly Defined
Joint Replacement, Should Be Added to List in Rule

There was no general agreement among the IRF officials we consulted on
whether conditions should be added to the list in the rule, and if so,
which conditions. In our interviews with IRF officials, three-quarters
identified various conditions that should be added. Of these, all
suggested the addition of cardiac conditions, and some identified other
conditions, such as pulmonary conditions, transplants, and more joint
replacements than are currently on the list. The reasons these IRF
officials gave for adding these conditions included that these patients
can become weakened physically during a hospital stay and need services in
an IRF to regain their strength and also that their experience shows they
can achieve good outcomes for these patients. The remaining IRF officials
said no conditions should be added. Some reasons they cited were that
these patients can be treated in a less intensive setting, the conditions
are too broad to be meaningful, and using a list of conditions is the
wrong approach. IRF officials differed regarding the addition of joint
replacement patients. Half of them suggested that joint replacement be
more broadly defined to include more patients, saying, for example, that
the current requirements were too restrictive and arbitrary, and a couple
of them said that unilateral joint replacement patients are not generally
appropriate for IRFs.

43CMS Fact Sheet #1, "Inpatient Rehabilitation Facility Classification
Requirements," includes two specific questions with respect to IRFs: (1)
how better to identify those patients who are most appropriate for
intensive medical rehabilitation resources provided in the IRF setting as
opposed to alternative care settings, and (2) what conditions, in addition
to those on the list in the rule, typically require intensive
rehabilitation treatment available in IRFs but not in alternative care
settings.

Experts Contended That Functional Status, in Addition to Condition, Should Be
Used to Identify Appropriate Types of Patients for Intensive Inpatient
Rehabilitation

The experts IOM convened contended that condition alone was insufficient
for identifying which patients, or types of patients, required the level
of services available in an IRF and generally agreed that functional
status should also be used. A patient's condition was perceived as an
acceptable starting point to understanding patient needs and as a way to
characterize the patients served by IRFs. But the experts IOM convened
generally agreed that condition, by itself, was insufficient and that more
information was needed. They said that condition alone fails to identify
the subgroup within each condition that is most appropriate for intensive
inpatient rehabilitation. For example, one of them noted that although an
IRF could be filled with patients that have conditions on the list in the
rule, the patients could be completely inappropriate for that setting.
Another expert at the meeting reported general agreement among the group
that using diagnosis alone is not sufficient.44

In addition to the experts convened by IOM, other experts we interviewed
also said that condition alone was insufficient because having a condition
on the list in the rule does not automatically indicate the need for
intensive inpatient rehabilitation (e.g., even though stroke is on the
list, only a subgroup of stroke patients require IRF services) and having
a condition not on the list does not necessarily mean the patient does not
need IRF services (e.g., although there is no cardiac condition on the
list, a subgroup of cardiac patients need the level of services of an
IRF). In addition, the FI and IRF officials we interviewed generally
reported as well that condition alone was insufficient. Over half the FI
officials we interviewed said that condition is insufficient by itself to
determine the need for intensive inpatient rehabilitation, and some said
that diagnosis is only a starting point. As noted earlier, all the IRF
officials reported using a variety of criteria, beyond condition, to
assess patients for admission, including function.

Among the experts convened by IOM, functional status was identified most
frequently as the information required in addition to condition. Half of
the experts IOM convened commented on the need to add information about
functional status, such as functional need, functional decline, motor and
cognitive function, and functional disability. To measure both diagnosis
and function, one of them suggested using the case-mix groups because they
combine both dimensions.

44Our analysis of Medicare patients that had been discharged from
hospitals provides further indication that not all patients with a
condition on the list go to IRFs. The percentage of these patients who
went on to IRFs within 30 days for their postacute care varied across
selected diagnosis-related groups (DRG) and was in no case greater than 50
percent. The largest percentages of patients going to IRFs after hospital
discharge were bilateral joint replacement and unilateral joint
replacement patients. (See app. III.)

Experts we interviewed also raised some concerns, however, about using
function as a measure of need for intensive inpatient rehabilitation. The
concerns voiced by the FI officials we interviewed included the potential
for abuse by qualifying more patients for admissions and the potential for
difficulty in adjudicating claims. One FI official said that moving toward
an assessment of functional status would require a better instrument than
currently exists.45 Another expert we interviewed said that using only
functional status could lead to including custodial patients that are
currently in SNFs. Officials at CMS also expressed concerns regarding how
to measure the need for intensive inpatient rehabilitation based on
functional status because a patient can have a low functional status but
not need intensive inpatient rehabilitation.

Experts IOM Convened Suggested Factors to Consider Using to Classify IRFs

Almost all the experts IOM convened said that IRF classification should
include characteristics of the patients served, but a couple said that IRF
classification should not include patient characteristics. Among those
expressing the need to use patient characteristics, function was
identified most often, although it was mentioned that it would be hard to
operationalize. Some of the experts IOM convened also suggested that the
percentage threshold be set at a lower level than 75 percent (for example,
60 percent or 65 percent) as a compromise until more information becomes
available to modify the list in the rule.

The experts IOM convened who opposed using patient characteristics to
classify IRFs suggested that IRFs be classified with just the other six
facility criteria, potentially looking at state licensure requirements for
additional facility criteria that could be applied specifically to IRFs.
These experts (as well as others we interviewed) said that no other
facility is classified using both patient and facility characteristics and
that IRFs are unique in being subjected to this approach. However,
Medicare does classify other facilities that are exempt from IPPS using a
characteristic about the patients served in those facilities.46
Furthermore, other experts at the meeting did not agree that the six
certification criteria were sufficient for distinguishing IRFs since
long-term care hospitals could likely meet these criteria as well.47

45The FI official reported that the FIM(TM) instrument that is currently
used does not adequately measure progress in small increments, such as a
quadriplegic patient might experience. Another respondent also reported
that the FIM(TM) only measures functional status at a point in time, but
does not predict functional improvement.

                                  Conclusions

Our analysis of Medicare data shows that there are Medicare patients in
IRFs who may not need the intensive level of rehabilitation services these
facilities offer. Just over half of all Medicare patients admitted to IRFs
in fiscal year 2003 were admitted for a condition that was not on the list
in the 75 percent rule. Of those patients whose primary or comorbid
condition was not on the list, the largest group was joint replacement
patients whose condition did not fit the list's specific criteria for
joint replacement. The experts IOM convened and other experts we
interviewed reported that unilateral, uncomplicated joint replacement
patients rarely need to be in an IRF. These experts also reported that
patients who may not need to be in an IRF may have been admitted because
CMS has not been routinely reviewing the IRFs' admission decisions to
determine whether they were medically justified. Increased scrutiny of
individual admissions through routine reviews for medical necessity
following patient discharge could be used to target problems and
vulnerabilities and thereby reduce the number of inappropriate admissions
in the future.

While some patients do not need to be in an IRF, the need for IRF services
may be more difficult to determine for other patients. The experts
convened by IOM called for more research to understand the effectiveness
of intensive inpatient rehabilitation, reporting that the evidence for the
effectiveness of IRF services varied in strength for conditions on the
list and was particularly weak for certain orthopedic conditions. CMS has
also recognized the need for more research in this area and asked NIH to
convene one meeting to help identify research priorities for inpatient
rehabilitation. Research studies that can produce information on a timely
basis, such as observational studies or meetings of clinical experts with
specialized expertise, would be especially helpful in this effort.

46For example, generally, in cancer hospitals, 50 percent of patients must
have neoplastic diagnoses, and psychiatric hospitals must primarily
provide psychiatric services for the diagnosis and treatment of mentally
ill persons. See 42 C.F.R. S:412.23(f)(1)(iv) (cancer hospitals); 42
C.F.R. S:412.23(a)(1) (psychiatric hospitals).

47Long-term care hospitals use admission criteria to determine whether
patients require that level of care; have active daily involvement with
physicians; have licensed nurse staffing of 6 to 10 hours per day per
patient; employ specialist registered nurses; employ physical,
occupational, speech, and respiratory therapists; and have
multidisciplinary teams that prepare and carry out treatment plans. MedPAC
recommended that a combination of facility and patient criteria be used to
distinguish postacute settings of care. (MedPAC, Report to the Congress:
New Approaches in Medicare, Ch. 5, "Defining Long-Term Care Hospitals"
(Washington, D.C.: June 2004), 128-130.)

The presence of patients in IRFs who may not need that level of services
and the calls for more research on the effectiveness of inpatient
rehabilitation lead us to conclude that greater clarity is needed in the
rule about what types of patients are most appropriate for rehabilitation
in an IRF. There was general agreement among the experts we interviewed,
including the experts convened by IOM, that condition alone is not
sufficient to identify the most appropriate types of patients since within
any condition only a subgroup of patients require the level of services of
an IRF. We believe that if condition alone is not sufficient to identify
the most appropriate types of patients, it would not be useful to add more
conditions to the list at the present time. There was also general
agreement among the experts that more information is needed to
characterize appropriate types of patients, and the most commonly
identified factor was functional status. However, some of the experts
convened by IOM recognized the challenge of operationalizing a measure of
function, and some experts questioned the ability of current assessment
tools to predict which types of patients will improve if treated in an
IRF. Despite the challenge, more clearly delineating the most appropriate
types of patients would offer more direction to IRFs-and to the health
professionals that refer patients to them-about which types of patients
can be treated in IRFs.

We believe that action to conduct reviews for medical necessity and to
produce more information about the effectiveness of inpatient
rehabilitation could support future efforts to refine the rule over time
to increase its clarity about which types of patients are most appropriate
for IRFs. These actions could help to ensure that Medicare does not pay
IRFs for patients who could be treated in a less intensive setting and
does not misclassify facilities for payment.

                      Recommendations for Executive Action

To help ensure that IRFs can be classified appropriately and that only
patients needing intensive inpatient rehabilitation are admitted to IRFs,
we recommend that the CMS Administrator take three actions:

           o  CMS should ensure that FIs routinely conduct targeted reviews
           for medical necessity for IRF admissions.
           o  CMS should conduct additional activities to encourage research
           on the effectiveness of intensive inpatient rehabilitation and the
           factors that predict patient need for intensive inpatient
           rehabilitation.
           o  CMS should use the information obtained from reviews for
           medical necessity, research activities, and other sources to
           refine the rule to describe more thoroughly the subgroups of
           patients within a condition that are appropriate for IRFs rather
           than other settings, and may consider using other factors in the
           descriptions, such as functional status.

           In commenting on a draft of this report, CMS stated that our work
           would be of assistance to the agency in examining issues related
           to patient coverage and the classification of inpatient
           rehabilitation facilities. CMS generally agreed with our
           recommendations and provided technical comments, which were
           incorporated as appropriate. CMS agreed that targeted reviews for
           medical necessity are necessary and said that it expected its
           contractors to direct their scarce resources toward areas of risk.
           CMS said that it has expanded its efforts to provide greater
           oversight of IRF admissions through local policies that have been
           implemented or are being developed by the FIs. CMS also agreed
           with our recommendation to encourage additional research and noted
           that it has expanded its activities to guide future research
           efforts by encouraging government research organizations, academic
           institutions, and the rehabilitation industry to conduct both
           general and targeted research. CMS said that it would collaborate
           with NIH to determine how best to promote research. CMS also
           stated that, while it expected to follow our recommendation to
           describe subgroups of patients within a medical condition, it
           would need to give this action careful consideration because it
           could result in a more restrictive policy than the present
           regulations. CMS noted that future research could guide the
           agency's descriptions of subgroups. Although CMS indicated its
           intention to follow this recommendation, we clarified the language
           in the recommendation to encourage CMS to obtain research and
           other information to undertake this effort. CMS's written comments
           are reprinted in appendix IV.

           We also received oral comments on a draft of this report from
           representatives of the American Hospital Association, the American
           Medical Rehabilitation Providers Association, and the Federation
           of American Hospitals. All three groups noted that we applied the
           criteria for a rule that was effective July 1, 2004, to data from
           fiscal year 2003, when IRFs were operating under a different list
           of conditions. They stated that a difference between the lists of
           conditions in these 2 years was in the definition of
           polyarthritis, which affected the circumstances under which joint
           replacement patients were counted under the rule. They reported
           that in fiscal year 2003, IRFs admitted Medicare joint replacement
           patients who they believed were within the criteria of the rule in
           effect at that time, but may not have been within the criteria of
           the rule that took effect July 1, 2004. In its technical comments,
           CMS also raised concerns about our use of fiscal year 2003 data.
           We analyzed the admission of joint replacement patients to IRFs
           and found no material change between the same time periods in 2003
           and 2004, as noted in the report. In addition, all three groups
           supported the call for more research. The three groups also
           provided technical comments, which we incorporated where
           appropriate.

           We are sending copies of this report to the Administrator of CMS
           and other interested parties. We will also make copies available
           to others on request. In addition, the report will be available at
           no charge on the GAO Web site at http://www.gao.gov.

           If you or your staffs have any questions about this report, please
           call me at (202) 512-7114 or Linda T. Kohn at (202) 512-4371. The
           names of other staff members who made contributions to this report
           are listed in appendix V.

           Marjorie Kanof Managing Director, Health Care

           A facility may be classified as an IRF if it can show that, during
           a 12-month period1 at least 75 percent of all its patients,
           including its Medicare patients, required intensive rehabilitation
           services for the treatment of one or more of the following
           conditions:2

           In undertaking this work, we analyzed data on Medicare patients
           admitted to inpatient rehabilitation facilities (IRF) and also
           interviewed a wide variety of experts in the field to obtain
           various perspectives. We used several different sources of data,
           including data from the Centers for Medicare & Medicaid Services
           (CMS) about Medicare patients admitted to IRFs; interviews with
           officials at IRFs, fiscal intermediaries (FI), CMS regional
           offices, and private insurers; a 1-day meeting of clinical experts
           in the field of physical medicine and rehabilitation; and
           interviews with other clinical and nonclinical experts and
           researchers in the field of rehabilitation as well as officials
           from professional associations of various disciplines involved in
           inpatient rehabilitation. In total, during this engagement, we
           spoke with 106 individuals, of whom 65 were clinicians. We
           conducted our work from May 2004 through April 2005 in accordance
           with generally accepted government auditing standards.

           To identify the conditions that IRF patients have, we obtained
           from CMS the Inpatient Rehabilitation Facility-Patient Assessment
           Instrument (IRF-PAI) records for all IRF admissions of Medicare
           patients for fiscal year 2003 (October 1, 2002, to September 30,
           2003), which have data on patient age and sex, impairment group
           code and case-mix group (CMG) classification, and comorbid
           conditions. To assess whether individual patients were considered
           to have 1 of the 13 conditions defined by the list of conditions
           in CMS's 75 percent rule, we applied the criteria laid out in
           CMS's Medicare Claims Processing Manual.1 This document lists the
           specific impairment group codes and ICD-9-CM diagnostic codes for
           comorbid conditions entered into the patient's IRF-PAI record that
           were used to identify patients who belonged in the 13 conditions.2
           We conducted our analyses on Medicare patients only because CMS
           records contained data on the largest number of IRFs and the
           majority of patients in IRFs are covered by Medicare. Prior work
           by RAND found that the percentage of Medicare patients with the
           conditions on the list in the rule was a good predictor of the
           percentage of total patients in the conditions on the list in the
           rule.3 We analyzed these data at the patient level to compare
           compliance with the rule across impairment groups. To permit a
           discrete assignment of each patient to one impairment group, we
           gave priority to the impairment group code designated at
           admission.4 To assess the extent to which Medicare patients in
           IRFs with joint replacements had comorbidities, we examined their
           distribution among the four payment tiers assigned under the
           prospective payment system for IRFs. The assigned CMG in the
           IRF-PAI data set includes a letter prefix that indicates that the
           patient either had no comorbidities related to the cost of
           providing inpatient rehabilitation or had one or more
           comorbidities expected to have a low, medium, or high impact on
           those costs. We calculated the proportion of joint replacement
           patients that fell into the no-comorbidity group, both overall and
           within each of the six joint replacement CMGs. To do our
           supplementary analysis on a sample of 2004 data, we compared the
           proportion of Medicare patients admitted to an IRF whose primary
           condition was joint replacement from July through December 2003 to
           the proportion of such patients from July through December 2004,
           using data from IRF-PAI records. We computed the proportion of
           Medicare patients admitted to IRFs that were joint replacement
           patients, ranked the facilities according to the proportion of
           Medicare joint replacement patients in 2003, and calculated the
           difference across the two time periods.

           To determine the number of IRFs that met the requirements of the
           75 percent rule, we aggregated Medicare patients treated at the
           same IRF and calculated the total percentage of each IRF's
           patients that were admitted with a primary condition or a comorbid
           condition on the list in the rule. We examined the distribution of
           compliance levels across IRFs, applying the different thresholds
           that the rule phases in over several years, but we did not assess
           the appropriateness of any threshold level. To determine whether
           any IRFs had ever been declassified based on failure to comply
           with the 75 percent rule, we interviewed officials at CMS's 10
           regional offices.

           Our analyses rely on Medicare billing information, and we
           determined that these data were sufficiently reliable for this
           analysis. We followed the instructions CMS provided to FIs to
           "presumptively verify compliance" using the list of codes in the
           Medicare Claims Processing Manual to estimate how many patients
           have one of the conditions on the list in the rule as recorded on
           the IRF-PAI instrument. FIs use the list of codes in this manual
           as a first step to estimate how many patients have one of the
           conditions on the list in the rule. To assess the reliability of
           the IRF-PAI records for our data analyses, we interviewed two
           researchers who had experience using the IRF-PAI data set, and
           performed electronic testing of the required data elements,
           including impairment codes, comorbid conditions, and admission
           dates. We examined the IRF-PAI data set and found few missing or
           invalid entries for the variables we used. We did not compare the
           information entered on the IRF-PAI to medical records. All of
           these analyses encompassed services provided in facilities located
           in the 50 states and the District of Columbia.

           To determine how IRFs assess patients for admission and how CMS
           reviews admission decisions for medical necessity, we interviewed
           the medical directors at 12 IRFs and the medical director or
           designee at 10 FIs. We used data from the RAND Corporation's "Case
           Mix Certification Rule for Inpatient Rehabilitation Facilities"
           (2003), prepared under contract to CMS, to select our respondents
           out of a total of more than 1,200 IRFs. RAND had analyzed the
           level of compliance of each IRF with the rule using the 10
           conditions on the list at that time. We used RAND data to create a
           sampling frame to select IRFs to interview, but we did not rely on
           RAND's data for any findings or conclusions. We matched facilities
           with data from the IRF-PAI to identify them and sorted them by zip
           code according to the Northeast, Midwest, South, and West regions
           as defined by the U.S. Census Bureau. Within each region, we
           selected IRFs with a high, median, and low level of compliance
           with the 75 percent rule. We identified the median complier in
           each region, and if necessary adjusted the selection of IRFs to
           (1) avoid interviewing more than one IRF in the same state and (2)
           provide a selection of for-profit, freestanding, and rural
           facilities. If a selected provider was unwilling or unable to
           participate in the interview, we substituted the IRF next on the
           list that was most similar in characteristics to the facility
           originally chosen. We conducted a structured interview with the
           medical director of each facility, and provided unstructured time
           at the end of the interview for the respondent to raise other
           issues. For nonclinical questions that the medical directors were
           unable to answer, we spoke to a member of the administrative team.
           We identified the areas covered in the interviews through
           background interviews with professional associations, advocacy
           groups, CMS, and experts in inpatient rehabilitation and health
           policy research, and pretested the interview protocol with two
           IRFs not included in our sample.

           The FIs we selected to interview were those that serviced the
           states in which the IRFs we selected were located. Because some
           FIs serviced more than one state, our selection yielded 10 FIs
           (out of a total of 30). To facilitate our interviews, we spoke
           with the appropriate CMS regional office, which notified an
           official at each FI about this engagement. We conducted a
           structured interview with the medical director or designee
           regarding (1) appropriate patients for inpatient rehabilitation,
           (2) the list of conditions in the rule, (3) assessment for
           compliance, and (4) reviews for medical necessity. We pretested
           the interview protocol with three FIs that were not included in
           our sample. We also spoke with FI officials who had been
           identified as being interested in inpatient rehabilitation. All FI
           officials had the opportunity to discuss issues other than those
           we highlighted. To compare Medicare's approach to the approaches
           of other payers, we selected a convenience sample of three
           insurers and one regional managed care organization to learn about
           their activities regarding inpatient rehabilitation. We
           interviewed officials from these payers, asking how they
           identified facilities for intensive inpatient rehabilitation, and
           how they identified appropriate patients for such services.

           Our interviews do not represent all concerns or experiences of
           inpatient rehabilitation facilities, FIs, or private payers, and
           the answers to the structured interviews were not restricted to
           Medicare patients. Because we were directed to examine the 75
           percent rule and not directly to evaluate the relative value of
           inpatient rehabilitation, we did not ask questions about the full
           spectrum of postacute care.

           To evaluate the approach of using a list of conditions in the 75
           percent rule to classify IRFs, we contracted with the Institute of
           Medicine (IOM) of The National Academies to convene a 1-day
           meeting of clinical experts broadly representative of the field of
           physical medicine and rehabilitation. We identified for IOM the
           categories of participants preferred at the meeting. To identify
           specific participants, IOM obtained input from us, IOM members,
           advocacy groups, and individual experts in the field. It
           identified a pool of participants according to the preferred
           categories. In total, 14 experts participated: 4 practicing
           physicians, 2 physical therapists, 2 occupational therapists, 1
           speech therapist, 2 nurses, 1 physician/researcher in postacute
           care, 1 physician/researcher from a research institute, and 1
           health services researcher. The meeting was facilitated by a
           physician/researcher with expertise in Medicare payment policy.
           Invitations to participate were issued by IOM. Participants were
           invited as individual experts, not as organizational
           representatives. The group was not asked to reach consensus on any
           issues, and IOM was not asked to produce or publish a report of
           the meeting. We observed the meeting and subsequently reviewed the
           transcript and audiotape of the meeting, listed the individual
           comments made during the meeting, and grouped the comments around
           a limited number of themes. The comments from the meeting of the
           experts IOM convened represent their individual statements and not
           a consensus of the group as a whole. In convening the meeting, IOM
           was not able to get participation of clinical experts who were not
           employed in IRFs (such as referring physicians or therapists in
           acute care settings) and a private payer. The comments of
           participants should not be interpreted to represent the views of
           IOM or all clinical experts in the field of rehabilitation.

           To examine the proportion of Medicare patients discharged from
           hospitals with different diagnosis-related groups (DRG) who went
           to IRFs for postacute care, we obtained CMS's Medicare Provider
           Analysis and Review (MEDPAR) file that contained all Medicare
           inpatient discharges from both acute care hospitals and IRFs for
           fiscal year 2003. This file provided information on patient
           admission and discharge dates from acute care hospitals and
           rehabilitation facilities along with the DRG assigned for each
           acute care stay. We identified all the patients who entered IRFs
           within 30 days of their hospital discharge during fiscal year 2003
           and calculated the frequencies for each DRG among them. We then
           selected the 19 DRGs that represented at least 1 percent of IRF
           admissions from acute care hospitals. Next we determined the total
           number of hospital discharges with those DRGs and computed the
           proportion of patients in each of these DRGs that were admitted to
           an IRF within 30 days. The analysis of acute hospital discharges
           required that we use the separate MEDPAR file that had information
           on inpatient DRGs and on patients who did not enter IRFs as well
           as those who did. The MEDPAR analysis may therefore reflect a
           slightly different IRF patient population from that reflected in
           the analyses conducted with the IRF-PAI data set. Apparent
           variations in the admission dates recorded for IRF patients in the
           two sets of data prevented us from combining data from each into
           one consolidated data set.

           Source: GAO analysis of CMS MEDPAR data.

           aFor some DRG descriptions, we reworded the DRG definition for
           simplicity. We selected all DRGs that represented at least 1
           percent of IRF admissions from hospitals in fiscal year 2003.
           These 19 DRGs accounted for 59 percent of all such admissions.
           Over 94 percent of patients admitted to IRFs in fiscal year 2003
           came from acute care hospitals, while about 3 percent came from
           the community and 1 percent from SNFs. DRGs only partially
           coincide with the impairment group codes used to categorize
           patients admitted to IRFs. For example, patients with hip
           fractures are included in DRG 209 or 471 if they received one or
           more joint replacements. Hip fractures treated with other surgical
           procedures are coded under DRG 210 or 211, and those treated
           medically are in DRG 236.

           bContains two DRGs.

           Linda T. Kohn, (202) 512-4371

           Manuel Buentello, Behn Kelly, Ba Lin, Eric Peterson, Kristi
           Peterson, and Roseanne Price made key contributions to this
           report.

           The Government Accountability Office, the audit, evaluation and
           investigative arm of Congress, exists to support Congress in
           meeting its constitutional responsibilities and to help improve
           the performance and accountability of the federal government for
           the American people. GAO examines the use of public funds;
           evaluates federal programs and policies; and provides analyses,
           recommendations, and other assistance to help Congress make
           informed oversight, policy, and funding decisions. GAO's
           commitment to good government is reflected in its core values of
           accountability, integrity, and reliability.

           The fastest and easiest way to obtain copies of GAO documents at
           no cost is through GAO's Web site (www.gao.gov). Each weekday, GAO
           posts newly released reports, testimony, and correspondence on its
           Web site. To have GAO e-mail you a list of newly posted products
           every afternoon, go to www.gao.gov and select "Subscribe to
           Updates."

           The first copy of each printed report is free. Additional copies
           are $2 each. A check or money order should be made out to the
           Superintendent of Documents. GAO also accepts VISA and Mastercard.
           Orders for 100 or more copies mailed to a single address are
           discounted 25 percent. Orders should be sent to:

           U.S. Government Accountability Office 441 G Street NW, Room LM
           Washington, D.C. 20548

           To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax:
           (202) 512-6061

           Contact:

           Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail:
           [email protected] Automated answering system: (800) 424-5454 or
           (202) 512-7470

           Gloria Jarmon, Managing Director, [email protected] (202) 512-4400
           U.S. Government Accountability Office, 441 G Street NW, Room 7125
           Washington, D.C. 20548

           Paul Anderson, Managing Director, [email protected] (202)
           512-4800 U.S. Government Accountability Office, 441 G Street NW,
           Room 7149 Washington, D.C. 20548

                        1. Stroke.
                        2. Spinal cord injury.
                        3. Congenital deformity.
                        4. Amputation.
                        5. Major multiple trauma.
                        6. Fracture of femur (hip fracture).
                        7. Brain injury.
                        8. Neurological disorders (including multiple
                        sclerosis, motor neuron diseases, polyneuropathy,
                        muscular dystrophy, and Parkinson's disease).
                        9. Burns.
                        10. Active, polyarticular rheumatoid arthritis,
                        psoriatic arthritis, and seronegative arthropathies
                        resulting in significant functional impairment of
                        ambulation and other activities of daily living that
                        have not improved after an appropriate, aggressive,
                        and sustained course of outpatient therapy services
                        or services in other less intensive rehabilitation
                        settings immediately preceding the inpatient
                        rehabilitation admission or that result from a
                        systemic disease activation immediately before
                        admission, but have the potential to improve with
                        more intensive rehabilitation.
                        11. Systemic vasculidities with joint inflammation,
                        resulting in significant functional impairment of
                        ambulation and other activities of daily living that
                        have not improved after an appropriate, aggressive,
                        and sustained course of outpatient therapy services
                        or services in other less intensive rehabilitation
                        settings immediately preceding the inpatient
                        rehabilitation admission or that result from a
                        systemic disease activation immediately before
                        admission, but have the potential to improve with
                        more intensive rehabilitation.
                        12. Severe or advanced osteoarthritis (osteoarthritis
                        or degenerative joint disease) involving two or more
                        major weight bearing joints (elbow, shoulders, hips,
                        or knees, but not counting a joint with a prosthesis)
                        with joint deformity and substantial loss of range of
                        motion, atrophy of muscles surrounding the joint,
                        significant functional impairment of ambulation and
                        other activities of daily living that have not
                        improved after the patient has participated in an
                        appropriate, aggressive, and sustained course of
                        outpatient therapy services or services in other less
                        intensive rehabilitation settings immediately
                        preceding the inpatient rehabilitation admission but
                        have the potential to improve with more intensive
                        rehabilitation. (A joint replaced by a prosthesis no
                        longer is considered to have osteoarthritis, or other
                        arthritis, even though this condition was the reason
                        for the joint replacement.)
                        13. Knee or hip joint replacement, or both, during an
                        acute hospitalization immediately preceding the
                        inpatient rehabilitation stay and also meet one or
                        more of the following specific criteria:

                                     a. The patient underwent bilateral knee
                                     or bilateral hip joint replacement
                                     surgery during the acute hospital
                                     admission immediately preceding the IRF
                                     admission.

                        b. The patient is extremely obese, with a body mass
                        index of at least 50 at the time of admission to the
                        IRF.

                        c. The patient is age 85 or older at the time of
                        admission to the IRF.

Agency Comments and Comments from National Associations and Our Evaluation

Appendix I: List of Conditions in CMS's 75 Percent Rule  Appendix I: List
of Conditions in CMS's 75 Percent Rule

1The time period is defined by CMS or the CMS contractor.

2See 42 C.F.R. S:412.23(b)(2)(iii) (2004).

Appendix II: Scope and Methodology  Appendix II: Scope and Methodology

1CMS, "Medicare Claims Processing," CMS Manual System, pub. 100-04,
Transmittal 347  (Baltimore, Md.: Oct. 29, 2004).

2The procedure described by CMS counts comorbidities listed either as an
etiologic diagnosis or as a comorbid condition entered on the IRF-PAI
form. We followed the procedures CMS provided to FIs for them to
presumptively verify compliance.

3See Grace M. Carter, O. Hayden, et al., "Case Mix Certification Rule for
Inpatient Rehabilitation Facilities," DRU-2981-CMS (Santa Monica, Ca.: May
2003).

4Patients may have a different impairment group code assigned at discharge
(both are recorded in the IRF-PAI data set), but the IRF prospective
payment from Medicare is based on the admission impairment group code.

Appendix III: Rates of IRF Medicare Admissions from Hospitals by Top 19
DRGs of Patients Admitted to IRFs, Fiscal Year 2003  Appendix III: Rates
of IRF Medicare Admissions from Hospitals by Top 19 DRGs of Patients
Admitted to IRFs, Fiscal Year 2003

                                                                Percentage of 
                                                               total hospital 
            Medical condition or      Number of  Number of IRF     discharges 
            procedure described  total hospital     admissions    admitted to 
DRG      by DRGa                  discharges from hospitals           IRFs 
209      Unilateral joint            428,518        124,754           29.1 
            replacement of lower                               
            extremity                                          
14,15    Strokeb                     325,361         54,433           16.7 
210, 211 Hip or femur                155,366         30,381           19.6 
            proceduresb except                                 
            joint replacement                                  
127      Heart failure/shock         695,349         14,863            2.1 
243      Medical back                100,994          8,970            8.9 
            problems                                           
89       Pneumonia and               521,432          8,591            1.6 
            pleurisy                                           
88       Chronic obstructive         398,066          7,427            1.9 
            pulmonary disease                                  
113      Amputation for               38,656          7,200           18.6 
            circulatory                                        
            disorders except                                   
            upper limb and toe                                 
1        Craniotomy                   32,916          6,969           21.2 
471      Bilateral joint              14,420          6,941           48.1 
            replacement of lower                               
            extremity                                          
497      Spinal fusion except         25,714          6,613           25.7 
            cervical with                                      
            complication and                                   
            comorbidity                                        
107      Coronary artery              78,557          6,584            8.4 
            bypass surgery                                     
478      Vascular operations         110,609          5,881            5.3 
            except heart                                       
236      Hip or pelvis                42,231          5,863           13.9 
            fracture                                           
296      Nutritional and             262,387          5,588            2.1 
            metabolic disorders                                
121      Heart attack                164,548          5,440            3.3 
499      Back and neck                37,590          5,366           14.3 
            procedures except                                  
            spinal fusion                                      

Appendix IV: Comments from the Centers for Medicare & Medicaid Services 
Appendix IV: Comments from the Centers for Medicare & Medicaid Services

A  Appendix V: GAO Contact and Staff Acknowledgments

                                  GAO Contact

                                Acknowledgments

(290351)

GAO's Mission

Obtaining Copies of GAO Reports and Testimony

Order by Mail or Phone

To Report Fraud, Waste, and Abuse in Federal Programs

Congressional Relations

Public Affairs

www.gao.gov/cgi-bin/getrpt?GAO-05-366.

To view the full product, including the scope

and methodology, click on the link above.

For more information, contact Marjorie Kanof at (202) 512-7114.

Highlights of GAO-05-366, a report to Senate Committee on Finance and
House Committee on Ways and Means

April 2005

MEDICARE

More Specific Criteria Needed to Classify Inpatient Rehabilitation
Facilities

Medicare classifies inpatient rehabilitation facilities (IRF) using the
"75 percent rule." If a facility can show that during 1 year at least 75
percent of its patients required intensive rehabilitation for 1 of 13
specified conditions, it may be classified as an IRF and paid at a higher
rate than is paid for less intensive rehabilitation in other settings.
Medicare payments to IRFs have grown steadily over the past decade. In
this report, GAO (1) identifies the conditions-on and off the list-that
IRF Medicare patients have and the number of IRFs that meet a 75 percent
threshold, (2) describes IRF admission criteria and Centers for Medicare &
Medicaid Services (CMS) review of admissions, and (3) evaluates use of a
list of conditions in the rule. GAO analyzed data on Medicare patients
(the majority of patients in IRFs) admitted to IRFs in FY 2003, spoke to
IRF medical directors, and had the Institute of Medicine (IOM) convene a
meeting of experts.

What GAO Recommends

GAO recommends that CMS take several actions, including refining the rule
to describe more thoroughly the subgroups of patients within a condition
that require IRF services, possibly using functional status or other
factors in addition to condition, to help ensure that IRFs can be
classified appropriately and that only patients needing IRF services are
admitted. CMS generally agreed with the recommendations.

In fiscal year 2003, fewer than half of all IRF Medicare patients were
admitted for having a condition on the list in the 75 percent rule, and
few IRFs admitted at least 75 percent of their patients for one of those
conditions. The largest group of patients had orthopedic conditions, not
all of which were on the list in the rule, which had been suspended in
2002. Almost half of all patients with conditions not on the list were
admitted for orthopedic conditions, and among those the largest group was
joint replacement patients. Although some joint replacement patients may
need admission to an IRF, GAO's analysis showed that few of these patients
had comorbidities that suggested a possible need for the IRF level of
services. Additionally, GAO found that only 6 percent of IRFs in fiscal
year 2003 were able to meet a 75 percent threshold.

IRFs varied in the criteria used to assess patients for admission, and CMS
has not routinely reviewed IRF admission decisions. IRF officials reported
that the criteria they used to make admission decisions included patient
characteristics such as function, as well as condition. CMS, working
through its fiscal intermediaries, has not routinely reviewed IRF
admission decisions.

The experts IOM convened and other clinical and nonclinical experts GAO
interviewed differed on whether conditions should be added to the list in
the 75 percent rule but agreed that condition alone does not provide
sufficient criteria to identify the types of patients appropriate for
IRFs. The experts IOM convened questioned the strength of the evidence for
adding conditions to the list, finding the evidence for certain orthopedic
conditions particularly weak, and they called for further research to
identify the types of patients that need inpatient rehabilitation and to
understand the effectiveness of IRFs. Other experts did not agree on
whether conditions, including a broader category of joint replacements,
should be added to the list. Experts, including those IOM convened,
generally agreed that condition alone is insufficient for identifying
appropriate types of patients for inpatient rehabilitation, since within
any condition only a subgroup of patients require the level of services of
an IRF, and contended that functional status should also be considered.
*** End of document. ***