[Federal Register Volume 88, Number 147 (Wednesday, August 2, 2023)]
[Rules and Regulations]
[Pages 50956-51052]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 2023-16050]



[[Page 50955]]

Vol. 88

Wednesday,

No. 147

August 2, 2023

Part II





Department of Health and Human Services





-----------------------------------------------------------------------





Centers for Medicare & Medicaid Services





-----------------------------------------------------------------------





42 CFR Part 412





Medicare Program; Inpatient Rehabilitation Facility Prospective Payment 
System for Federal Fiscal Year 2024 and Updates to the IRF Quality 
Reporting Program; Final Rule

  Federal Register / Vol. 88 , No. 147 / Wednesday, August 2, 2023 / 
Rules and Regulations  

[[Page 50956]]


-----------------------------------------------------------------------

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Centers for Medicare & Medicaid Services

42 CFR Part 412

[CMS-1781-F]
RIN 0938-AV04


Medicare Program; Inpatient Rehabilitation Facility Prospective 
Payment System for Federal Fiscal Year 2024 and Updates to the IRF 
Quality Reporting Program

AGENCY: Centers for Medicare & Medicaid Services (CMS), HHS.

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: This final rule updates the prospective payment rates for 
inpatient rehabilitation facilities (IRFs) for Federal fiscal year (FY) 
2024. As required by statute, this final rule includes the 
classification and weighting factors for the IRF prospective payment 
system's case-mix groups and a description of the methodologies and 
data used in computing the prospective payment rates for FY 2024. It 
also rebases and revises the IRF market basket to reflect a 2021 base 
year. It also confirms when IRF units can become excluded and paid 
under the IRF PPS. This rule also includes updates for the IRF Quality 
Reporting Program (QRP).

DATES: 
    Effective date: These regulations are effective on October 1, 2023.
    Applicability dates: The updated IRF prospective payment rates are 
applicable for IRF discharges occurring on or after October 1, 2023, 
and on or before September 30, 2024 (FY 2024).

FOR FURTHER INFORMATION CONTACT: Kim Schwartz, (410) 786-2571, for 
general information.
    Catie Cooksey, (410) 786-0179, for information about the IRF 
payment policies and payment rates.
    Kim Schwartz, (410) 786-2571, for information about the IRF 
coverage policies.
    Ariel Cress, (410) 786-8571, for information about the IRF quality 
reporting program.

SUPPLEMENTARY INFORMATION:

Availability of Certain Information Through the Internet on the CMS 
Website

    The IRF prospective payment system (IRF PPS) Addenda along with 
other supporting documents and tables referenced in this final rule are 
available through the internet on the CMS website at https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS.
    We note that prior to 2020, each rule or notice issued under the 
IRF PPS has included a detailed reiteration of the various regulatory 
provisions that have affected the IRF PPS over the years. That 
discussion, along with detailed background information for various 
other aspects of the IRF PPS, is now available on the CMS website at 
https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS.

I. Executive Summary

A. Purpose

    This final rule updates the prospective payment rates for IRFs for 
FY 2024 (that is, for discharges occurring on or after October 1, 2023, 
and on or before September 30, 2024) as required under section 
1886(j)(3)(C) of the Social Security Act (the Act). As required by 
section 1886(j)(5) of the Act, this final rule includes the 
classification and weighting factors for the IRF PPS's case-mix groups 
(CMGs), and a description of the methodologies and data used in 
computing the prospective payment rates for FY 2024. It also rebases 
and revises the IRF market basket to reflect a 2021 base year. It also 
confirms when an IRF unit can be excluded and paid under the IRF PPS. 
This final rule includes several updates to the IRF QRP for the FY 2025 
IRF QRP and FY 2026 IRF QRP. This final rule will add two new measures 
to the IRF QRP, remove three measures from the IRF QRP, and modify one 
measure in the IRF QRP. This final rule also finalizes the public 
reporting schedule of four measures. In addition, this final rule 
includes a summary of the comments received on Centers for Medicare and 
Medicaid Services' (CMS') update on our efforts to close the health 
equity gap and on the request for information on principles CMS would 
use to select and prioritize IRF QRP quality measures in future years.

B. Summary of Major Provisions

    In this final rule, we use the methods described in the FY 2023 IRF 
PPS final rule (87 FR 47038) to update the prospective payment rates 
for FY 2024 using updated FY 2022 IRF claims and the most recent 
available IRF cost report data, which is FY 2021 IRF cost report data. 
It also rebases and revises the IRF market basket to reflect a 2021 
base year. It also modifies the regulation governing when an IRF unit 
can be excluded and paid under the IRF PPS.
    Beginning with the FY 2025 IRF QRP, IRFs will be required to submit 
data on a modified version of the COVID-19 Vaccination Coverage among 
Healthcare Personnel measure and the Discharge Function Score measure. 
Beginning with the FY 2025 IRF QRP, IRFs will no longer be required to 
submit data on the Application of Percent of Long-Term Care Hospital 
Patients with an Admission and Discharge Functional Assessment and a 
Care Plan That Addresses Function, the IRF Functional Outcome Measure: 
Change in Self-Care Score for Medical Rehabilitation Patients (CBE 
#2633), and the IRF Functional Outcome Measure: Change in Mobility 
Score for Medical Rehabilitation Patients (CBE #2634) measures. 
Beginning with the FY 2026 IRF QRP, IRFs will be required to submit 
data on the COVID-19 Vaccine: Percent of Patients/Residents Who Are Up 
to Date measure. This final rule also adopts policies to begin public 
reporting of the Transfer of Health Information to the Patient-Post-
Acute Care (PAC) and Transfer of Health Information to the Provider-PAC 
measures, the Discharge Function Score measure, and the COVID-19 
Vaccine: Percent of Patients/Residents Who Are Up to Date measure. 
Finally, we provide a summary of the comments received from interested 
parties on principles for selecting and prioritizing IRF QRP quality 
measures and concepts as well as a summary of the comments received on 
our continued efforts to close the health equity gap.

C. Summary of Impact

[[Page 50957]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.048

II. Background

A. Statutory Basis and Scope for IRF PPS Provisions

    Section 1886(j) of the Act provides for the implementation of a 
per-discharge PPS for inpatient rehabilitation hospitals and inpatient 
rehabilitation units of a hospital (collectively, hereinafter referred 
to as IRFs). Payments under the IRF PPS encompass inpatient operating 
and capital costs of furnishing covered rehabilitation services (that 
is, routine, ancillary, and capital costs), but not direct graduate 
medical education costs, costs of approved nursing and allied health 
education activities, bad debts, and other services or items outside 
the scope of the IRF PPS. A complete discussion of the IRF PPS 
provisions appears in the original FY 2002 IRF PPS final rule (66 FR 
41316) and the FY 2006 IRF PPS final rule (70 FR 47880) and we provided 
a general description of the IRF PPS for FYs 2007 through 2019 in the 
FY 2020 IRF PPS final rule (84 FR 39055 through 39057). A general 
description of the IRF PPS for FYs 2020 through 2023, along with 
detailed background information for various other aspects of the IRF 
PPS, is now available on the CMS website at https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS.
    Under the IRF PPS from FY 2002 through FY 2005, the prospective 
payment rates were computed across 100 distinct CMGs, as described in 
the FY 2002 IRF PPS final rule (66 FR 41316). We constructed 95 CMGs 
using rehabilitation impairment categories (RICs), functional status 
(both motor and cognitive), and age (in some cases, cognitive status 
and age may not be a factor in defining a CMG). In addition, we 
constructed five special CMGs to account for very short stays and for 
patients who expire in the IRF.
    For each of the CMGs, we developed relative weighting factors to 
account for a patient's clinical characteristics and expected resource 
needs. Thus, the weighting factors accounted for the relative 
difference in resource use across all CMGs. Within each CMG, we created 
tiers based on the estimated effects that certain comorbidities would 
have on resource use.
    We established the Federal PPS rates using a standardized payment 
conversion factor (formerly referred to as the budget-neutral 
conversion factor). For a detailed discussion of the budget-neutral 
conversion factor, please refer to our FY 2004 IRF PPS final rule (68 
FR 45684 through 45685). In the FY 2006 IRF PPS final rule (70 FR 
47880), we discussed in detail the methodology for determining the 
standard payment conversion factor.
    We applied the relative weighting factors to the standard payment 
conversion factor to compute the unadjusted prospective payment rates 
under the IRF PPS from FYs 2002 through 2005. Within the structure of 
the payment system, we then made adjustments to account for interrupted 
stays, transfers, short stays, and deaths. Finally, we applied the 
applicable adjustments to account for geographic variations in wages 
(wage index), the percentage of low-income patients, location in a 
rural area (if applicable), and outlier payments (if applicable) to the 
IRFs' unadjusted prospective payment rates.
    For cost reporting periods that began on or after January 1, 2002, 
and before October 1, 2002, we determined the final prospective payment 
amounts using the transition methodology prescribed in section 
1886(j)(1) of the Act. Under this provision, IRFs transitioning into 
the PPS were paid a blend of the Federal IRF PPS rate and the payment 
that the IRFs would have received had the IRF PPS not been implemented. 
This provision also allowed IRFs to elect to bypass this blended 
payment and immediately be paid 100 percent of the Federal IRF PPS 
rate. The transition methodology expired as of cost reporting periods 
beginning on or after October 1, 2002 (FY 2003), and payments for all 
IRFs now consist of 100 percent of the Federal IRF PPS rate.
    Section 1886(j) of the Act confers broad statutory authority upon 
the Secretary to propose refinements to the IRF PPS. In the FY 2006 IRF 
PPS final rule (70 FR 47880) and in correcting amendments to the FY 
2006 IRF PPS final rule (70 FR 57166), we finalized a number of 
refinements to the IRF PPS case-mix classification system (the CMGs and 
the corresponding relative weights) and the case-level and facility-
level adjustments. These refinements included the adoption of the 
Office of Management and Budget's (OMB's) Core-Based Statistical Area 
(CBSA) market definitions; modifications to the CMGs, tier 
comorbidities; and CMG relative weights, implementation of a new 
teaching status adjustment for IRFs; rebasing and revising the market 
basket used to update IRF payments, and updates to the rural, low-
income percentage (LIP), and high-cost outlier adjustments. Beginning 
with the FY 2006 IRF PPS final rule (70 FR 47908 through 47917), the 
market basket used to update IRF payments was a market basket 
reflecting the operating and capital cost structures for freestanding 
IRFs, freestanding inpatient psychiatric facilities (IPFs), and long-
term care hospitals (LTCHs) (hereinafter referred to as the 
rehabilitation, psychiatric, and long-term care (RPL) market basket). 
Any reference to the FY 2006 IRF PPS final rule in this final rule also 
includes the provisions effective in the correcting amendments. For a 
detailed discussion of the final key policy changes for FY 2006, please 
refer to the FY 2006 IRF PPS final rule.
    The regulatory history previously included in each rule or notice 
issued under the IRF PPS, including a general description of the IRF 
PPS for FYs 2007 through 2020, is available on the CMS website at 
https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS.
    In late 2019,\1\ the United States began responding to an outbreak 
of a virus named ``SARS-CoV-2'' and the disease it causes, which is 
named ``coronavirus disease 2019'' (abbreviated ``COVID-19''). Due to 
our prioritizing efforts in

[[Page 50958]]

support of containing and combatting the Public Health Emergency (PHE) 
for COVID-19, and devoting significant resources to that end, we 
published two interim final rules with comment period affecting IRF 
payment and conditions for participation. The interim final rule with 
comment period (IFC) entitled, ``Medicare and Medicaid Programs; Policy 
and Regulatory Revisions in Response to the COVID-19 Public Health 
Emergency,'' published on April 6, 2020 (85 FR 19230) (hereinafter 
referred to as the April 6, 2020 IFC), included certain changes to the 
IRF PPS medical supervision requirements at 42 CFR 412.622(a)(3)(iv) 
and 412.29(e) during the PHE for COVID-19. In addition, in the April 6, 
2020 IFC, we removed the post-admission physician evaluation 
requirement at Sec.  412.622(a)(4)(ii) for all IRFs during the PHE for 
COVID-19. In the FY 2021 IRF PPS final rule, to ease documentation and 
administrative burden, we also removed the post-admission physician 
evaluation documentation requirement at Sec.  412.622(a)(4)(ii) 
permanently beginning in FY 2021.
---------------------------------------------------------------------------

    \1\ Patel A, Jernigan DB. Initial Public Health Response and 
Interim Clinical Guidance for the 2019 Novel Coronavirus Outbreak--
United States, December 31, 2019-February 4, 2020. MMWR Morb Mortal 
Wkly Rep 2020;69:140-146. DOI http://dx.doi.org/10.15585/mmwr.mm6905e1.
---------------------------------------------------------------------------

    A second IFC entitled, ``Medicare and Medicaid Programs, Basic 
Health Program, and Exchanges; Additional Policy and Regulatory 
Revisions in Response to the COVID-19 Public Health Emergency and Delay 
of Certain Reporting Requirements for the Skilled Nursing Facility 
Quality Reporting Program'' was published on May 8, 2020 (85 FR 27550) 
(hereinafter referred to as the May 8, 2020 IFC). Among other changes, 
the May 8, 2020 IFC included a waiver of the ``3-hour rule'' at Sec.  
412.622(a)(3)(ii) to reflect the waiver required by section 3711(a) of 
the Coronavirus Aid, Relief, and Economic Security Act (CARES Act) 
(Pub. L. 116-136, enacted on March 27, 2020). In the May 8, 2020 IFC, 
we also modified certain IRF coverage and classification requirements 
for freestanding IRF hospitals to relieve acute care hospital capacity 
concerns in States (or regions, as applicable) experiencing a surge 
during the PHE for COVID-19. In addition to the policies adopted in our 
IFCs, we responded to the PHE with numerous blanket waivers \2\ and 
other flexibilities,\3\ some of which are applicable to the IRF PPS. 
CMS finalized these policies in the Calendar Year 2023 Hospital 
Outpatient Prospective Payment and Ambulatory Surgical Center Payment 
Systems final rule with comment period (87 FR 71748).
---------------------------------------------------------------------------

    \2\ CMS, ``COVID-19 Emergency Declaration Blanket Waivers for 
Health Care Providers,'' (updated Feb. 19 2021) (available at 
https://www.cms.gov/files/document/summary-covid-19-emergency-declaration-waivers.pdf).
    \3\ CMS, ``COVID-19 Frequently Asked Questions (FAQs) on 
Medicare Fee-for-Service (FFS) Billing,'' (updated March 5, 2021) 
(available at https://www.cms.gov/files/document/03092020-covid-19-faqs-508.pdf).
---------------------------------------------------------------------------

B. Provisions of the Patient Protection and the Affordable Care Act and 
the Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) 
Affecting the IRF PPS in FY 2012 and Beyond

    The Patient Protection and the Affordable Care Act (the Affordable 
Care Act or ACA) (Pub. L. 111-148) was enacted on March 23, 2010. The 
Health Care and Education Reconciliation Act of 2010 (Pub. L. 111-152), 
which amended and revised several provisions of the Patient Protection 
and Affordable Care Act, was enacted on March 30, 2010. In this final 
rule, we refer to the two statutes collectively as the ``Affordable 
Care Act'' or ``ACA''.
    The ACA included several provisions that affect the IRF PPS in FYs 
2012 and beyond. In addition to what was previously discussed, section 
3401(d) of the ACA also added section 1886(j)(3)(C)(ii)(I) of the Act 
(providing for a ``productivity adjustment'' for FY 2012 and each 
subsequent FY). The productivity adjustment for FY 2024 is discussed in 
section VI.D. of this final rule. Section 1886(j)(3)(C)(ii)(II) of the 
Act provides that the application of the productivity adjustment to the 
market basket update may result in an update that is less than 0.0 for 
a FY and in payment rates for a FY being less than such payment rates 
for the preceding FY.
    Section 3004(b) of the ACA and section 411(b) of the MACRA (Pub. L. 
114-10, enacted on April 16, 2015) also addressed the IRF PPS. Section 
3004(b) of ACA reassigned the previously designated section 1886(j)(7) 
of the Act to section 1886(j)(8) of the Act and inserted a new section 
1886(j)(7) of the Act, which contains requirements for the Secretary to 
establish a QRP for IRFs. Under that program, data must be submitted in 
a form and manner and at a time specified by the Secretary. Beginning 
in FY 2014, section 1886(j)(7)(A)(i) of the Act requires the 
application of a 2-percentage point reduction to the market basket 
increase factor otherwise applicable to an IRF (after application of 
paragraphs (C)(iii) and (D) of section 1886(j)(3) of the Act) for a FY 
if the IRF does not comply with the requirements of the IRF QRP for 
that FY. Application of the 2-percentage point reduction may result in 
an update that is less than 0.0 for a FY and in payment rates for a FY 
being less than such payment rates for the preceding FY. Reporting-
based reductions to the market basket increase factor are not 
cumulative; they only apply for the FY involved. Section 411(b) of the 
MACRA amended section 1886(j)(3)(C) of the Act by adding paragraph 
(iii), which required us to apply for FY 2018, after the application of 
section 1886(j)(3)(C)(ii) of the Act, an increase factor of 1.0 percent 
to update the IRF prospective payment rates.

C. Operational Overview of the Current IRF PPS

    As described in the FY 2002 IRF PPS final rule (66 FR 41316), upon 
the admission and discharge of a Medicare Part A fee-for-service (FFS) 
patient, the IRF is required to complete the appropriate sections of a 
Patient Assessment Instrument (PAI), designated as the IRF-PAI. In 
addition, beginning with IRF discharges occurring on or after October 
1, 2009, the IRF is also required to complete the appropriate sections 
of the IRF-PAI upon the admission and discharge of each Medicare 
Advantage (MA) patient, as described in the FY 2010 IRF PPS final rule 
(74 FR 39762) and the FY 2010 IRF PPS correction notice (74 FR 50712). 
All required data must be electronically encoded into the IRF-PAI 
software product. Generally, the software product includes patient 
classification programming called the Grouper software. The Grouper 
software uses specific IRF-PAI data elements to classify (or group) 
patients into distinct CMGs and account for the existence of any 
relevant comorbidities.
    The Grouper software produces a five-character CMG number. The 
first character is an alphabetic character that indicates the 
comorbidity tier. The last four characters are numeric characters that 
represent the distinct CMG number. A free download of the Grouper 
software is available on the CMS website at http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS/Software.html. The Grouper software is also embedded in the internet 
Quality Improvement and Evaluation System (iQIES) User tool available 
in iQIES at https://www.cms.gov/medicare/quality-safety-oversight-general-information/iqies.
    Once a Medicare Part A FFS patient is discharged, the IRF submits a 
Medicare claim as a Health Insurance Portability and Accountability Act 
of 1996 (HIPAA) (Pub. L. 104-191, enacted on August 21, 1996)--
compliant electronic claim or, if the Administrative Simplification 
Compliance Act of 2002 (ASCA) (Pub. L.

[[Page 50959]]

107-105, enacted on December 27, 2002) permits, a paper claim (a UB-04 
or a CMS-1450 as appropriate) using the five-character CMG number and 
sends it to the appropriate Medicare Administrative Contractor (MAC). 
In addition, once a MA patient is discharged, in accordance with the 
Medicare Claims Processing Manual, chapter 3, section 20.3 (Pub. 100-
04), hospitals (including IRFs) must submit to their MAC an 
informational-only bill (type of bill (TOB) 111) that includes 
Condition Code 04. This will ensure that the MA days are included in 
the hospital's Supplemental Security Income (SSI) ratio (used in 
calculating the IRF LIP adjustment) for FY 2007 and beyond. Claims 
submitted to Medicare must comply with both ASCA and HIPAA.
    Section 3 of the ASCA amended section 1862(a) of the Act by adding 
paragraph (22), which requires the Medicare program, subject to section 
1862(h) of the Act, to deny payment under Part A or Part B for any 
expenses for items or services for which a claim is submitted other 
than in an electronic form specified by the Secretary. Section 1862(h) 
of the Act, in turn, provides that the Secretary shall waive such 
denial in situations in which there is no method available for the 
submission of claims in an electronic form or the entity submitting the 
claim is a small provider. In addition, the Secretary also has the 
authority to waive such denial in such unusual cases as the Secretary 
finds appropriate. For more information, see the ``Medicare Program; 
Electronic Submission of Medicare Claims'' final rule (70 FR 71008). 
Our instructions for the limited number of Medicare claims submitted on 
paper are available at http://www.cms.gov/manuals/downloads/clm104c25.pdf.
    Section 3 of the ASCA operates in the context of the administrative 
simplification provisions of HIPAA, which include, among others, the 
requirements for transaction standards and code sets codified in 45 CFR 
part 160 and part 162, subparts A and I through R (generally known as 
the Transactions Rule). The Transactions Rule requires covered 
entities, including covered healthcare providers, to conduct covered 
electronic transactions according to the applicable transaction 
standards. (See the CMS program claim memoranda at http://www.cms.gov/ElectronicBillingEDITrans/ and listed in the addenda to the Medicare 
Intermediary Manual, Part 3, section 3600).
    The MAC processes the claim through its software system. This 
software system includes pricing programming called the ``Pricer'' 
software. The Pricer software uses the CMG number, along with other 
specific claim data elements and provider-specific data, to adjust the 
IRF's prospective payment for interrupted stays, transfers, short 
stays, and deaths, and then applies the applicable adjustments to 
account for the IRF's wage index, percentage of low-income patients, 
rural location, and outlier payments. For discharges occurring on or 
after October 1, 2005, the IRF PPS payment also reflects the teaching 
status adjustment that became effective as of FY 2006, as discussed in 
the FY 2006 IRF PPS final rule (70 FR 47880).

D. Advancing Health Information Exchange

    The Department of Health and Human Services (HHS) has a number of 
initiatives designed to encourage and support the adoption of 
interoperable health information technology and to promote nationwide 
health information exchange to improve health care and patient access 
to their digital health information.
    To further interoperability in post-acute care settings, CMS and 
the Office of the National Coordinator for Health Information 
Technology (ONC) participate in the Post-Acute Care Interoperability 
Workgroup (PACIO) to facilitate collaboration with interested parties 
to develop Health Level Seven International[supreg] (HL7) Fast 
Healthcare Interoperability Resource[supreg] (FHIR) standards. These 
standards could support the exchange and reuse of patient assessment 
data derived from the post-acute care (PAC) setting assessment tools, 
such as the minimum data set (MDS), inpatient rehabilitation facility-
patient assessment instrument (IRF-PAI), Long-Term Care Hospital (LTCH) 
continuity assessment record and evaluation (CARE) Data Set (LCDS), 
outcome and assessment information set (OASIS), and other 
sources.4 5 The PACIO Project has focused on HL7 FHIR 
implementation guides for: functional status, cognitive status and new 
use cases on advance directives, re-assessment timepoints, and Speech, 
language, swallowing, cognitive communication and hearing (SPLASCH) 
pathology.\6\ We encourage PAC provider and health IT vendor 
participation as the efforts advance.
---------------------------------------------------------------------------

    \4\ HL7 FHIR Release 4. Available at https://www.hl7.org/fhir/.
    \5\ HL7 FHIR. PACIO Functional Status Implementation Guide. 
Available at https://paciowg.github.io/functional-status-ig/.
    \6\ PACIO Project. Available at http://pacioproject.org/about/.
---------------------------------------------------------------------------

    The CMS Data Element Library (DEL) continues to be updated and 
serves as a resource for PAC assessment data elements and their 
associated mappings to health IT standards such as Logical Observation 
Identifiers Names and Codes (LOINC) and Systematized Nomenclature of 
Medicine Clinical Terms (SNOMED).\7\ The DEL furthers CMS' goal of data 
standardization and interoperability. Standards in the DEL can be 
referenced on the CMS website and in the ONC Interoperability Standards 
Advisory (ISA). The 2023 ISA is available at https://www.healthit.gov/sites/isa/files/inline-files/2023%20Reference%20Edition_ISA_508.pdf.
---------------------------------------------------------------------------

    \7\ Centers for Medicare & Medicaid Services. Newsroom. Fact 
sheet: CMS Data Element Library Fact Sheet. June 21, 2018. Available 
at https://www.cms.gov/newsroom/fact-sheets/cms-data-element-library-fact-sheet.
---------------------------------------------------------------------------

    We are also working with ONC to advance the United States Core Data 
for Interoperability (USCDI), a standardized set of health data classes 
and constituent data elements for nationwide, interoperable health 
information exchange.\8\ We are collaborating with ONC and other 
Federal agencies to define and prioritize additional data 
standardization needs and develop consensus on recommendations for 
future versions of the USCDI. We are also directly collaborating with 
ONC to build requirements to support data standardization and alignment 
with requirements for quality measurement. ONC has launched the USCDI+ 
initiative to support the identification and establishment of domain 
specific datasets that build on the core USCDI foundation.\9\ The 
USCDI+ quality measurement domain currently being developed aims to 
support defining additional data specifications for quality measurement 
that harmonize, where possible, with other Federal agency data needs 
and inform supplemental standards necessary to support quality 
measurement, including the needs of programs supporting quality 
measurement for long-term and post-acute care.
---------------------------------------------------------------------------

    \8\ USCDI. Available at https://www.healthit.gov/isa/united-states-core-data-interoperability-uscdi.
    \9\ USCDI+. Available at https://www.healthit.gov/topic/interoperability/uscdi-plus.
---------------------------------------------------------------------------

    The 21st Century Cures Act (Cures Act) (Pub. L. 114-255, enacted 
December 13, 2016) required HHS and ONC to take steps to promote 
adoption and use of electronic health record (EHR) technology.\10\ 
Specifically,

[[Page 50960]]

section 4003(b) of the Cures Act required ONC to take steps to advance 
interoperability through the development of a Trusted Exchange 
Framework and Common Agreement aimed at establishing full network-to-
network exchange of health information nationally. On January 18, 2022, 
ONC announced a significant milestone by releasing the Trusted Exchange 
Framework \11\ and Common Agreement Version 1.\12\ The Trusted Exchange 
Framework is a set of non-binding principles for health information 
exchange, and the Common Agreement is a contract that advances those 
principles. The Common Agreement and the Qualified Health Information 
Network Technical Framework Version 1 (incorporated by reference into 
the Common Agreement) establish the technical infrastructure model and 
governing approach for different health information networks and their 
users to securely share clinical information with each other, all under 
commonly agreed to terms. The technical and policy architecture of how 
exchange occurs under the Common Agreement follows a network-of-
networks structure, which allows for connections at different levels 
and is inclusive of many different types of entities at those different 
levels, such as health information networks, healthcare practices, 
hospitals, public health agencies, and Individual Access Services (IAS) 
Providers.\13\ On February 13, 2023, HHS marked a new milestone during 
an event at HHS headquarters,\14\ which recognized the first set of 
applicants accepted for onboarding to the Common Agreement as Qualified 
Health Information Networks (QHINs). QHINs will be entities that will 
connect directly to each other to serve as the core for nationwide 
interoperability.\15\ For more information, we refer readers to https://www.healthit.gov/topic/interoperability/trusted-exchange-framework-and-common-agreement.
---------------------------------------------------------------------------

    \10\ Sections 4001 through 4008 of Public Law 114-255. Available 
at https://www.govinfo.gov/content/pkg/PLAW-114publ255/html/PLAW-114publ255.htm.
    \11\ The Trusted Exchange Framework (TEF): Principles for 
Trusted Exchange (Jan. 2022). Available at https://www.healthit.gov/sites/default/files/page/2022-01/Trusted_Exchange_Framework_0122.pdf.
    \12\ Common Agreement for Nationwide Health Information 
Interoperability Version 1 (Jan. 2022). Available at https://www.healthit.gov/sites/default/files/page/2022-01/Common_Agreement_for_Nationwide_Health_Information_Interoperability_Version_1.pdf.
    \13\ The Common Agreement defines Individual Access Services 
(IAS) as ``with respect to the Exchange Purposes definition, the 
services provided utilizing the Connectivity Services, to the extent 
consistent with Applicable Law, to an Individual with whom the QHIN, 
Participant, or Subparticipant has a Direct Relationship to satisfy 
that Individual's ability to access, inspect, or obtain a copy of 
that Individual's Required Information that is then maintained by or 
for any QHIN, Participant, or Subparticipant.'' The Common Agreement 
defines ``IAS Provider'' as: ``Each QHIN, Participant, and 
Subparticipant that offers Individual Access Services.'' See Common 
Agreement for Nationwide Health Information Interoperability Version 
1, at 7 (Jan. 2022), https://www.healthit.gov/sites/default/files/page/2022-01/Common_Agreement_for_Nationwide_Health_Information_Interoperability_Version_1.pdf.
    \14\ ``Building TEFCA,'' Micky Tripathi and Mariann Yeager, 
Health IT Buzz Blog. February 13, 2023. https://www.healthit.gov/buzz-blog/electronic-health-and-medical-records/interoperability-electronic-health-and-medical-records/building-tefca.
    \15\ The Common Agreement defines a QHIN as ``to the extent 
permitted by applicable SOP(s), a Health Information Network that is 
a U.S. Entity that has been Designated by the RCE and is a party to 
the Common Agreement countersigned by the RCE.'' See Common 
Agreement for Nationwide Health Information Interoperability Version 
1, at 10 (Jan. 2022), https://www.healthit.gov/sites/default/files/page/2022-01/Common_Agreement_for_Nationwide_Health_Information_Interoperability_Version_1.pdf.
---------------------------------------------------------------------------

    We invited providers to learn more about these important 
developments and how they are likely to affect IRFs.

III. Summary of Provisions of the Proposed Rule

    In the FY 2024 IRF PPS proposed rule, we proposed to update the IRF 
PPS for FY 2024 and the IRF QRP for FY 2025 and FY 2026.
    The proposed policy changes and updates to the IRF prospective 
payment rates for FY 2024 are as follows:
     Update the CMG relative weights and average length of stay 
values for FY 2024, in a budget neutral manner, as discussed in section 
IV. of the FY 2024 IRF PPS proposed rule (88 FR 20954 through 20959).
     Update the IRF PPS payment rates for FY 2024 by the market 
basket increase factor, based upon the most current data available, 
with a productivity adjustment required by section 1886(j)(3)(C)(ii)(I) 
of the Act, as described in section V. of the FY 2024 IRF PPS proposed 
rule (88 FR 20959, 20973 through 20974).
     Rebase and revise the IRF market basket to reflect a 2021 
base year, as discussed in section V. of the FY 2024 IRF PPS proposed 
rule (88 FR 20959 through 20973).
     Update the FY 2024 IRF PPS payment rates by the FY 2024 
wage index and the labor-related share in a budget-neutral manner, as 
discussed in section V. of the FY 2024 IRF PPS proposed rule (88 FR 
20974 through 20977).
     Describe the calculation of the IRF standard payment 
conversion factor for FY 2024, as discussed in section V. of the FY 
2024 IRF PPS proposed rule (88 FR 20977).
     Update the outlier threshold amount for FY 2024, as 
discussed in section VI. of the FY 2024 IRF PPS proposed rule (88 FR 
20980 through 20981).
     Update the cost-to-charge ratio (CCR) ceiling and urban/
rural average CCRs for FY 2024, as discussed in section VI. of the FY 
2024 IRF PPS proposed rule (88 FR 20981).
     Describe the proposed modification to the regulation for 
IRF units to become excluded and paid under the IRF PPS as discussed in 
section VII. of the FY 2024 IRF PPS proposed rule (88 FR 20981 through 
20984).
    We also proposed updates to the IRF QRP and requested information 
in section VIII. of the FY 2024 IRF PPS proposed rule as follows:
     Modify the COVID-19 Vaccination Coverage among Healthcare 
Personnel measure beginning with the FY 2025 IRF QRP.
     Adopt the Discharge Function Score measure beginning with 
the FY 2025 IRF QRP.
     Remove the Application of Percent of Long-Term Care 
Hospital Patients with an Admission and Discharge Functional Assessment 
and a Care Plan That Addresses Function measure beginning with the FY 
2025 IRF QRP.
     Remove the IRF Functional Outcome Measure: Change in Self-
Care Score for Medical Rehabilitation Patients (NQF #2633) measure 
beginning with the FY 2025 IRF QRP.
     Remove the IRF Functional Outcome Measure: Change in 
Mobility Score for Medical Rehabilitation Patients (NQF #2634) measure 
beginning with the FY 2025 IRF QRP.
     Adopt the COVID-19 Vaccine: Percent of Patients/Residents 
Who Are Up to Date measure beginning with the FY 2026 IRF QRP.
     Request information on principles for selecting and 
prioritizing IRF QRP quality measures and concepts.
     Provide an update on our continued efforts to close the 
health equity gap.

IV. Analysis of and Responses to Public Comments

    We received 45 timely responses from the public, many of which 
contained multiple comments on the FY 2024 IRF PPS proposed rule (88 FR 
20950). We received comments from various trade associations, inpatient 
rehabilitation facilities, individual physicians, therapists, 
clinicians, health care industry organizations, and health care 
consulting firms. The following sections, arranged by subject area, 
include a summary of the public comments that we received, and our 
responses.

[[Page 50961]]

A. General Comments on the FY 2024 IRF PPS Proposed Rule

    In addition to the comments, we received on specific proposals 
contained within the proposed rule (which we address later in this 
final rule), commenters also submitted more general observations on the 
IRF PPS and IRF care generally.
    Comment: We received several comments that were outside the scope 
of the FY 2024 IRF PPS proposed rule. Specifically, we received 
comments regarding the inclusion of recreational therapy in the IRF 
intensity of therapy requirement, disclosures of ownership and 
additional disclosable parties' information in the skilled nursing 
facility setting, the ``low wage index policy,'' Medicare Advantage 
rules, waiving the ``three-hour rule,'' and the IRF Review Choice 
Demonstration. We also received comments about making refinements to 
our measures to address the impact of COVID-19 and social determinants 
of health, to change the HCP COVID-19 measure specifications to annual 
data submission, and concerns of being inappropriately penalized for 
NHSN technical errors.
    Response: We thank the commenters for bringing these issues to our 
attention and will take these comments into consideration for potential 
policy refinements or direct the comments to the appropriate subject 
matter experts.

V. Update to the Case-Mix Group (CMG) Relative Weights and Average 
Length of Stay (ALOS) Values for FY 2024

    As specified in Sec.  412.620(b)(1), we calculate a relative weight 
for each CMG that is proportional to the resources needed by an average 
inpatient rehabilitation case in that CMG. For example, cases in a CMG 
with a relative weight of 2, on average, will cost twice as much as 
cases in a CMG with a relative weight of 1. Relative weights account 
for the variance in cost per discharge due to the variance in resource 
utilization among the payment groups, and their use helps to ensure 
that IRF PPS payments support beneficiary access to care, as well as 
provider efficiency.
    In the proposed rule, we proposed to update the CMG relative 
weights and ALOS values for FY 2024. Typically, we use the most recent 
available data to update the CMG relative weights and ALOS values. For 
FY 2024, we proposed to use the FY 2022 IRF claims and FY 2021 IRF cost 
report data. These data are the most current and complete data 
available at this time. Currently, only a small portion of the FY 2022 
IRF cost report data are available for analysis, but the majority of 
the FY 2022 IRF claims data are available for analysis. We also 
proposed that if more recent data became available after the 
publication of the proposed rule and before the publication of the 
final rule, we would use such data to determine the FY 2024 CMG 
relative weights and ALOS values in the final rule.
    We proposed to apply these data using the same methodologies that 
we have used to update the CMG relative weights and ALOS values each FY 
since we implemented an update to the methodology. The detailed CCR 
data from the cost reports of IRF provider units of primary acute care 
hospitals is used for this methodology, instead of CCR data from the 
associated primary care hospitals, to calculate IRFs' average costs per 
case, as discussed in the FY 2009 IRF PPS final rule (73 FR 46372). In 
calculating the CMG relative weights, we use a hospital-specific 
relative value method to estimate operating (routine and ancillary 
services) and capital costs of IRFs. The process to calculate the CMG 
relative weights for this final rule is as follows:
    Step 1. We estimate the effects that comorbidities have on costs.
    Step 2. We adjust the cost of each Medicare discharge (case) to 
reflect the effects found in the first step.
    Step 3. We use the adjusted costs from the second step to calculate 
CMG relative weights, using the hospital-specific relative value 
method.
    Step 4. We normalize the FY 2024 CMG relative weights to the same 
average CMG relative weight from the CMG relative weights implemented 
in the FY 2023 IRF PPS final rule (87 FR 47038).
    Consistent with the methodology that we have used to update the IRF 
classification system in each instance in the past, we proposed to 
update the CMG relative weights for FY 2024 in such a way that total 
estimated aggregate payments to IRFs for FY 2024 are the same with or 
without the changes (that is, in a budget-neutral manner) by applying a 
budget neutrality factor to the standard payment amount. We note that, 
as we typically do, we updated our data between the FY 2024 IRF PPS 
proposed and final rules to ensure that we use the most recent 
available data in calculating IRF PPS payments. This updated data 
reflects a more complete set of claims for FY 2022 and additional cost 
report data for FY 2021. To calculate the appropriate budget neutrality 
factor for use in updating the FY 2024 CMG relative weights, we use the 
following steps:
    Step 1. Calculate the estimated total amount of IRF PPS payments 
for FY 2024 (with no changes to the CMG relative weights).
    Step 2. Calculate the estimated total amount of IRF PPS payments 
for FY 2024 by applying the changes to the CMG relative weights (as 
discussed in this final rule).
    Step 3. Divide the amount calculated in step 1 by the amount 
calculated in step 2 to determine the budget neutrality factor of 
1.0002 that would maintain the same total estimated aggregate payments 
in FY 2024 with and without the changes to the CMG relative weights.
    Step 4. Apply the budget neutrality factor from step 3 to the FY 
2024 IRF PPS standard payment amount after the application of the 
budget-neutral wage adjustment factor.
    In section VI.G. of this final rule, we discuss the use of the 
existing methodology to calculate the standard payment conversion 
factor for FY 2024.
    In Table 2, ``Relative Weights and Average Length of Stay Values 
for Case-Mix Groups,'' we present the CMGs, the comorbidity tiers, the 
corresponding relative weights, and the ALOS values for each CMG and 
tier for FY 2024. The ALOS for each CMG is used to determine when an 
IRF discharge meets the definition of a short-stay transfer, which 
results in a per diem case level adjustment.

[[Page 50962]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.049


[[Page 50963]]


[GRAPHIC] [TIFF OMITTED] TR02AU23.050


[[Page 50964]]


[GRAPHIC] [TIFF OMITTED] TR02AU23.051


[[Page 50965]]


[GRAPHIC] [TIFF OMITTED] TR02AU23.052

    Generally, updates to the CMG relative weights result in some 
increases and some decreases to the CMG relative weight values. Table 2 
shows how we estimate that the application of the revisions for FY 2024 
would affect particular CMG relative weight values, which would affect 
the overall distribution of payments within CMGs and tiers. We note 
that, because we implement the CMG relative weight revisions in a 
budget-neutral manner (as previously described), total estimated 
aggregate payments to IRFs for FY 2024 are not affected as a result of 
the CMG relative weight revisions. However, the revisions affect the 
distribution of payments within CMGs and tiers.
[GRAPHIC] [TIFF OMITTED] TR02AU23.053

    As shown in Table 3, 99.4 percent of all IRF cases are in CMGs and 
tiers that would experience less than a 5 percent change (either 
increase or decrease) in the CMG relative weight value as a result of 
the revisions for FY 2024. The changes in the ALOS values for FY 2024, 
compared with the FY 2023 ALOS values, are small and do not show any 
particular trends in IRF length of stay patterns.
    We invited public comment on our proposed updates to the CMG 
relative weights and ALOS values for FY 2024.
    The following is a summary of the public comments received on the 
proposed revisions to update the CMG relative weights and ALOS values 
for FY 2024 and our responses.
    Comment: Commenters were generally supportive of the proposed 
updates to the relative weights and ALOS values and encouraged CMS to 
use the latest available data to update these values in the final rule. 
A few commenters expressed concern regarding reductions in certain 
relative weight values associated with traumatic spinal cord injury, 
major multiple traumas with brain or spinal cord injury, and Guillain-
Barr[eacute]. A few commenters also expressed concerns related to the 
increase of the ALOS for CMG 0404. These commenters noted that CMS did 
not propose a similar increase in reimbursement for this CMG and 
suggested the change may be due to distortions in the data rather than 
actual care changes.
    Response: We appreciate these commenters' support for updating the 
relative weights and ALOS values for FY 2024. The CMG relative weights 
are updated each year in a budget neutral manner, thus leading to 
increases in some CMG relative weights and corresponding decreases in 
other CMG relative weights. We note that, as we typically do, we have 
updated our data between the FY 2024 IRF PPS proposed and this final 
rule to ensure that we use the most recent available data in 
calculating IRF PPS payments. The relative weights associated with 
these CMGs include both increases and decreases, and the variation for 
FY 2024 is similar to the typical year-to-year variation that we 
observe. The relative weight values are updated each year to ensure 
that the IRF case mix system is as reflective as possible of the 
current IRF population, thereby ensuring that IRF payments 
appropriately reflect the relative costs of caring for all types of IRF 
patients.
    Additionally, the ALOS values are updated annually to be as 
reflective as possible of recent IRF utilization. The ALOS values are 
only used to determine which cases qualify for the short-stay transfer 
policy and are not used to determine payments for the non-short-stay 
transfer cases.
    Comment: A commenter expressed concern that decreases to the CMG 
relative weights and ALOS values do not reflect the medical complexity 
of the patients and suggested that CMS should revise the CMG relative 
weights and ALOS values to ensure adequate coverage and reimbursement 
for the services required to treat patients in IRF settings.
    Response: We believe that these data accurately reflect the 
severity of the IRF patient population and the associated costs of 
caring for these patients in the IRF setting. The CMG relative weights 
are updated each year based on the most recent available data for the 
full population of IRF Medicare fee-for-service beneficiaries. This 
ensures that the IRF case mix system is as reflective

[[Page 50966]]

as possible of changes in the IRF patient populations and the 
associated coding practices.
    After consideration of the comments we received, we are finalizing 
our proposal to update the CMG relative weights and ALOS values for FY 
2024, as shown in Table 2 of this final rule. These updates are 
effective for FY 2024, that is, for discharges occurring on or after 
October 1, 2023, and on or before September 30, 2024.

VI. FY 2024 IRF PPS Payment Update

A. Background

    Section 1886(j)(3)(C) of the Act requires the Secretary to 
establish an increase factor that reflects changes over time in the 
prices of an appropriate mix of goods and services for which payment is 
made under the IRF PPS. According to section 1886(j)(3)(A)(i) of the 
Act, the increase factor shall be used to update the IRF prospective 
payment rates for each FY. Section 1886(j)(3)(C)(ii)(I) of the Act 
requires the application of a productivity adjustment described in 
section 1886(b)(3)(B)(xi)(II) of the Act. Thus, we proposed to update 
the IRF PPS payments for FY 2024 by a market basket increase percentage 
as required by section 1886(j)(3)(C) of the Act based upon the most 
current data available, with a productivity adjustment as required by 
section 1886(j)(3)(C)(ii)(I) of the Act.
    We have utilized various market baskets through the years in the 
IRF PPS. For a discussion of these market baskets, we refer readers to 
the FY 2016 IRF PPS final rule (80 FR 47046).
    In FY 2016, we finalized the use of a 2012-based IRF market basket, 
using Medicare cost report data for both freestanding and hospital-
based IRFs (80 FR 47049 through 47068). In FY 2020, we finalized a 
rebased and revised IRF market basket to reflect a 2016 base year. The 
FY 2020 IRF PPS final rule (84 FR 39071 through 39086) contains a 
complete discussion of the development of the 2016-based IRF market 
basket. Beginning with FY 2024, we proposed to rebase and revise the 
IRF market basket to reflect a 2021 base year. In the following 
discussion, we provide an overview of the market basket and describe 
the methodologies used to determine the operating and capital portions 
of the 2021-based IRF market basket.

B. Overview of the 2021-Based IRF Market Basket

    The 2021-based IRF market basket is a fixed-weight, Laspeyres-type 
price index. A Laspeyres price index measures the change in price, over 
time, of the same mix of goods and services purchased in the base 
period. Any changes in the quantity or mix of goods and services (that 
is, intensity) purchased over time relative to the base period are not 
measured.
    The index itself is constructed in three steps. First, a base 
period is selected (for the proposed IRF market basket in the proposed 
rule, we proposed to use 2021 as the base period) and total base period 
costs are estimated for a set of mutually exclusive and exhaustive cost 
categories. Each category is calculated as a proportion of total costs. 
These proportions are called cost weights. Second, each cost category 
is matched to an appropriate price or wage variable, referred to as a 
price proxy. In almost every instance, these price proxies are derived 
from publicly available statistical series that are published on a 
consistent schedule (preferably at least on a quarterly basis). 
Finally, the cost weight for each cost category is multiplied by the 
level of its respective price proxy. The sum of these products (that 
is, the cost weights multiplied by their price index levels) for all 
cost categories yields the composite index level of the market basket 
in a given time period. Repeating this step for other periods produces 
a series of market basket levels over time. Dividing an index level for 
a given period by an index level for an earlier period produces a rate 
of growth in the input price index over that timeframe.
    As noted, the market basket is described as a fixed-weight index 
because it represents the change in price over time of a constant mix 
(quantity and intensity) of goods and services needed to provide IRF 
services. The effects on total costs resulting from changes in the mix 
of goods and services purchased subsequent to the base period are not 
measured. For example, an IRF hiring more nurses after the base period 
to accommodate the needs of patients would increase the volume of goods 
and services purchased by the IRF but would not be factored into the 
price change measured by a fixed-weight IRF market basket. Only when 
the index is rebased would changes in the quantity and intensity be 
captured, with those changes being reflected in the cost weights. 
Therefore, we rebase the market basket periodically so that the cost 
weights reflect recent changes in the mix of goods and services that 
IRFs purchase to furnish inpatient care between base periods.

C. Rebasing and Revising of the IRF PPS Market Basket

    As discussed in the FY 2020 IRF PPS final rule (84 FR 39071 through 
39086), the 2016-based IRF market basket cost weights reflect the 2016 
Medicare cost report data submitted by both freestanding and hospital-
based facilities.
    Beginning with FY 2024, we proposed to rebase and revise the 2016-
based IRF market basket cost weights to a 2021 base year reflecting the 
2021 Medicare cost report data submitted by both freestanding and 
hospital-based IRFs. Below we provide a detailed description of our 
methodology used to develop the 2021-based IRF market basket. This 
proposed methodology is generally similar to the methodology used to 
develop the 2016-based IRF market basket.
    We invited public comment on our proposed methodology for 
developing the 2021-based IRF market basket.
    Comment: Many commenters supported the rebasing and revising of the 
IRF market basket from a 2016 base year to a 2021 base year as 
proposed. Some of these commenters encouraged CMS to focus greater 
attention on the costs and data needed to support payment changes in 
the future.
    Several commenters, while supporting moving forward with a 2021 
base year, requested that CMS consider rebasing the IRF market basket 
to a later base year, such as 2022 or 2023, when the data become 
available, to more fully incorporate changes to IRF cost structures. 
One commenter stated that inflationary pressures and cost increases 
seem to have moderated somewhat during FY 2023 and therefore, using FY 
2023 in future rulemaking would better align permanent changes that 
have occurred in more recent years. One commenter stated that they 
believe that using FY 2023 data, when available, may more accurately 
capture costs being incurred by IRFs and they requested that CMS update 
the IRF market basket cost weights with the most recently available 
data in the final rule.
    Response: We appreciate the commenters' support to rebase and 
revise the IRF market basket. As discussed in section VI.A of this 
final rule, the market basket used to update IRF PPS payments has been 
periodically rebased and revised over the history of the IRF PPS to 
reflect more recent data on IRF cost structures. For the FY 2024 IRF 
PPS proposed rule, we proposed to rebase and revise the IRF market 
basket using 2021 Medicare cost reports, the most recent year of 
complete data available at the time of rulemaking, which showed an 
increase in the Compensation cost weight from 2016 to 2021. Data for 
2022 and 2023 are incomplete at this time. Because complete 2022 IRF 
cost report data are

[[Page 50967]]

currently unavailable, we believe it is more appropriate to update the 
base year cost weights to 2021 to reflect changes over this period 
rather than to delay the rebasing. It has been our longstanding 
practice to rebase the market basket on a regular basis to ensure it 
reflects the input cost structure of IRFs. As stated in the FY 2024 IRF 
PPS proposed rule (88 FR 20960), given the potential impact of the PHE 
on the Medicare cost report data, we will continue to monitor the 
Medicare cost report data as they become available and, if appropriate, 
propose any changes to the IRF market basket in future rulemaking.
    We provide a summary of the more detailed public comments received 
on our proposed methodology for developing the 2021-based IRF market 
basket and our responses in the following sections.
1. Development of Cost Categories and Weights for the 2021-Based IRF 
Market Basket
a. Use of Medicare Cost Report Data
    We proposed a 2021-based IRF market basket that consists of seven 
major cost categories and a residual derived from the 2021 Medicare 
cost reports (CMS Form 2552-10, OMB No. 0938-0050) for freestanding and 
hospital-based IRFs. The seven major cost categories are Wages and 
Salaries, Employee Benefits, Contract Labor, Pharmaceuticals, 
Professional Liability Insurance (PLI), Home Office/Related 
Organization Contract Labor, and Capital. The residual category 
reflects all remaining costs not captured in the seven cost categories. 
The 2021 cost reports include providers whose cost reporting period 
began on or after October 1, 2020, and before October 1, 2021. As noted 
previously, the current IRF market basket is based on 2016 Medicare 
cost reports and, therefore, reflects the 2016 cost structure for IRFs. 
As described in the FY 2023 IRF PPS final rule (87 FR 47049 through 
47050), we received comments on the FY 2023 IRF PPS proposed rule where 
interested parties expressed concern that the proposed market basket 
update was inadequate relative to input price inflation experienced by 
IRFs, particularly as a result of the COVID-19 PHE. These commenters 
stated that the PHE, along with inflation, has significantly driven up 
operating costs. Specifically, some commenters noted changes to the 
labor markets that led to the use of more contract labor, a trend that 
we verified in analyzing the Medicare cost reports through 2021. 
Therefore, we believe it is appropriate to incorporate more recent data 
to reflect updated cost structures for IRFs, and so we proposed to use 
2021 as the base year because we believe that the Medicare cost reports 
for this year represent the most recent, complete set of Medicare cost 
report data available for developing the proposed IRF market basket at 
the time of this rulemaking. Given the potential impact of the PHE on 
the Medicare cost report data, we will continue to monitor these data 
going forward and any changes to the IRF market basket will be proposed 
in future rulemaking.
    Since our goal is to establish cost weights that are reflective of 
case mix and practice patterns associated with the services IRFs 
provide to Medicare beneficiaries, as we did for the 2016-based IRF 
market basket, we proposed to limit the cost reports used to establish 
the 2021-based IRF market basket to those from facilities that had a 
Medicare ALOS that was relatively similar to their facility ALOS. We 
believe that this requirement eliminates statistical outliers and 
ensures a more accurate market basket that reflects the costs generally 
incurred during a Medicare-covered stay. The Medicare ALOS for 
freestanding IRFs is calculated from data reported on line 14 of 
Worksheet S-3, part I. The Medicare ALOS for hospital-based IRFs is 
calculated from data reported on line 17 of Worksheet S-3, part I. We 
proposed to include the cost report data from IRFs with a Medicare ALOS 
within 15 percent (that is, 15 percent higher or lower) of the facility 
ALOS to establish the sample of providers used to estimate the 2021-
based IRF market basket cost weights. We proposed to apply this ALOS 
edit to the data for IRFs to exclude providers that serve a population 
whose ALOS would indicate that the patients served are not consistent 
with an ALOS of a typical Medicare patient. We note that this is the 
same ALOS edit that we applied to develop the 2016-based IRF market 
basket. This process resulted in the exclusion of about nine percent of 
the freestanding and hospital-based IRF Medicare cost reports. Of those 
excluded, about 15 percent were freestanding IRFs and 85 percent were 
hospital-based IRFs. This ratio is relatively consistent with the 
universe of freestanding and hospital-based IRF cost reports where 
freestanding IRFs represent about 30 percent of the total.
    We then proposed to use the cost reports for IRFs that met this 
ALOS edit requirement to calculate the costs for the seven major cost 
categories (Wages and Salaries, Employee Benefits, Contract Labor, 
Professional Liability Insurance, Pharmaceuticals, Home Office/Related 
Organization Contract Labor, and Capital) for the market basket. These 
are the same categories used for the 2016-based IRF market basket. 
Also, as described in section V.C.1.d. of the proposed rule, and as 
done for the 2016-based IRF market basket, we also proposed to use the 
Medicare cost report data to calculate the detailed capital cost 
weights for the Depreciation, Interest, Lease, and Other Capital-
Related cost categories. We note that we proposed to rename the Home 
Office Contract Labor cost category to the Home Office/Related 
Organization Contract Labor cost category to be more consistent with 
the Medicare cost report instructions.
    Similar to the 2016-based IRF market basket major cost weights, for 
the majority of the 2021-based IRF market basket cost weights, we 
proposed to divide the 2021 costs for each cost category by the 2021 
total Medicare allowable costs (routine, ancillary and capital) that 
are eligible for reimbursement through the IRF PPS (we note that we use 
total facility medical care costs as the denominator to derive both the 
PLI and Home Office/Related Organization Contract Labor cost weights). 
We next describe our proposed methodology for deriving the cost levels 
used to derive the 2021-based IRF market basket.
(1) Total Medicare Allowable Costs
    For freestanding IRFs, we proposed that total Medicare allowable 
costs would be equal to the sum of total costs for the Medicare 
allowable cost centers as reported on Worksheet B, part I, column 26, 
lines 30 through 35, 50 through 76 (excluding 52 and 75), 90 through 
91, and 93.
    For hospital-based IRFs, we proposed that total Medicare allowable 
costs would be equal to the total costs for the IRF inpatient unit 
after the allocation of overhead costs (Worksheet B, part I, column 26, 
line 41) and a proportion of total ancillary costs reported on 
Worksheet B, part I, column 26, lines 50 through 76 (excluding 52 and 
75), 90 through 91, and 93.
    We proposed to calculate total ancillary costs attributable to the 
hospital-based IRF by first deriving an ``IRF ancillary ratio'' for 
each ancillary cost center. The IRF ancillary ratio is defined as the 
ratio of IRF Medicare ancillary costs for the cost center (as reported 
on Worksheet D-3, column 3 for hospital-based IRFs) to total Medicare 
ancillary costs for the cost center (equal to the sum of Worksheet D-3, 
column 3 for all relevant prospective payment systems (PPS) [that is, 
inpatient prospective payment

[[Page 50968]]

system (IPPS), IRF PPS, inpatient psychiatric facilities (IPF) PPS and 
skilled nursing facility (SNF) PPS]). For example, if hospital-based 
IRF Medicare physical therapy costs represent about 30 percent of the 
total Medicare physical therapy costs for the entire facility, then the 
IRF ancillary ratio for physical therapy costs would be 30 percent. We 
believe it is appropriate to use only a portion of the ancillary costs 
in the market basket cost weight calculations since the hospital-based 
IRF only utilizes a portion of the facility's ancillary services. We 
believe the ratio of reported IRF Medicare costs to reported total 
Medicare costs provides a reasonable estimate of the ancillary services 
utilized, and costs incurred, by the hospital-based IRF. We proposed 
that this IRF ancillary ratio for each cost center also be used to 
calculate Wages and Salaries and Capital costs, as described in section 
VI.C.1.a.(2) of this final rule.
    Then for each ancillary cost center, we proposed to multiply the 
IRF ancillary ratio for the given cost center by the total facility 
ancillary costs for that specific cost center (as reported on Worksheet 
B, part I, column 26) to derive IRF ancillary costs. For example, the 
30 percent IRF ancillary ratio for physical therapy cost center would 
be multiplied by the total ancillary costs for physical therapy 
(Worksheet B, part I, column 26, line 66). The IRF ancillary costs for 
each cost center are then added to total costs for the IRF inpatient 
unit after the allocation of overhead costs (Worksheet B, part I, 
column 26, line 41) to derive total Medicare allowable costs.
    We proposed to use these methods to derive levels of total Medicare 
allowable costs for IRF providers. This is the same methodology used 
for the 2016-based IRF market basket. We proposed that these total 
Medicare allowable costs for the IRF will be the denominator for the 
cost weight calculations for the Wages and Salaries, Employee Benefits, 
Contract Labor, Pharmaceuticals, and Capital cost weights. With this 
work complete, we then set about deriving cost levels for the seven 
major cost categories and then derive a residual cost weight reflecting 
all other costs not classified.
(2) Wages and Salaries Costs
    For freestanding IRFs, we proposed to derive Wages and Salaries 
costs as the sum of routine inpatient salaries (Worksheet A, column 1, 
lines 30 through 35), ancillary salaries (Worksheet A, column 1, lines 
50 through 76 (excluding 52 and 75), 90 through 91, and 93), and a 
proportion of overhead (or general service cost centers in the Medicare 
cost reports) salaries. Since overhead salary costs are attributable to 
the entire IRF, we only include the proportion attributable to the 
Medicare allowable cost centers. We proposed to estimate the proportion 
of overhead salaries that are attributed to Medicare allowable costs 
centers by multiplying the ratio of Medicare allowable area salaries 
(Worksheet A, column 1, lines 30 through 35, 50 through 76 (excluding 
52 and 75), 90 through 91, and 93) to total non-overhead salaries 
(Worksheet A, column 1, line 200 less Worksheet A, column 1, lines 4 
through 18) times total overhead salaries (Worksheet A, column 1, lines 
4 through 18). This is a similar methodology as used in the 2016-based 
IRF market basket.
    For hospital-based IRFs, we proposed to derive Wages and Salaries 
costs as the sum of the following salaries attributable to the 
hospital-based IRF: inpatient routine salary costs (Worksheet A, column 
1, line 41); overhead salary costs; ancillary salary costs; and a 
portion of overhead salary costs attributable to the ancillary 
departments.
(a) Overhead Salary Costs
    We proposed to calculate the portion of overhead salary costs 
attributable to hospital-based IRFs by first calculating an IRF 
overhead salary ratio, which is equal to the ratio of total facility 
overhead salaries (as reported on Worksheet A, column 1, lines 4-18) to 
total facility noncapital overhead costs (as reported on Worksheet A, 
column 1 and 2, lines 4-18). We then proposed to multiply this IRF 
overhead salary ratio by total noncapital overhead costs (sum of 
Worksheet B, part I, columns 4 through 18, line 41, less Worksheet B, 
part II, columns 4 through 18, line 41). This methodology assumes the 
proportion of total costs related to salaries for the overhead cost 
center is similar for all inpatient units (that is, acute inpatient or 
inpatient rehabilitation).
(b) Ancillary Salary Costs
    We proposed to calculate hospital-based IRF ancillary salary costs 
for a specific cost center (Worksheet A, column 1, lines 50 through 76 
(excluding 52 and 75), 90 through 91, and 93) as salary costs from 
Worksheet A, column 1, multiplied by the IRF ancillary ratio for each 
cost center as described in section V.C.1.a.(1) of the proposed rule. 
The sum of these costs represents hospital-based IRF ancillary salary 
costs.
(c) Overhead Salary Costs for Ancillary Cost Centers
    We proposed to calculate the portion of overhead salaries 
attributable to each ancillary department (lines 50 through 76 
(excluding 52 and 75), 90 through 91, and 93) by first calculating 
total noncapital overhead costs attributable to each specific ancillary 
department (sum of Worksheet B, part I, columns 4-18 less, Worksheet B, 
part II, column 26). We then identify the portion of these total 
noncapital overhead costs for each ancillary department that is 
attributable to the hospital-based IRF by multiplying these costs by 
the IRF ancillary ratio as described in section V.C.1.a.(1) of the 
proposed rule. We then sum these estimated IRF Medicare allowable 
noncapital overhead costs for all ancillary departments (cost centers 
50 through 76, 90 through 91, and 93). Finally, we then identify the 
portion of these IRF Medicare allowable noncapital overhead costs that 
are attributable to Wages and Salaries by multiplying these costs by 
the IRF overhead salary ratio as described in section V.C.1.a.(2)(a) of 
the proposed rule. This is the same methodology used to derive the 
2016-based IRF market basket.
(3) Employee Benefits Costs
    Effective with the implementation of CMS Form 2552-10, we began 
collecting Employee Benefits and Contract Labor data on Worksheet S-3, 
part V.
    For the 2021 Medicare cost report data, 54 percent of providers 
reported Employee Benefits data on Worksheet S-3, part V; particularly, 
approximately 57 percent of freestanding IRFs and 53 percent of 
hospital-based IRFs reported Employee Benefits data on Worksheet S-3, 
part V. For comparison, for 2016, about 45 percent of providers 
reported Employee Benefits data on Worksheet S-3, part V. Again, we 
continue to encourage all providers to report these data on the 
Medicare cost report.
    For freestanding IRFs, we proposed Employee Benefits costs would be 
equal to the data reported on Worksheet S-3, part V, column 2, line 2. 
We note that while not required to do so, freestanding IRFs also may 
report Employee Benefits data on Worksheet S-3, part II, which is 
applicable to only IPPS providers. Similar to the method for the 2016-
based IRF market basket, for those freestanding IRFs that report 
Worksheet S-3, part II, data, but not Worksheet S-3, part V, we 
proposed to use the sum of Worksheet S-3, part II, lines 17, 18, 20, 
and 22, to derive Employee Benefits costs.

[[Page 50969]]

    For hospital-based IRFs, we proposed to calculate total benefit 
costs as the sum of inpatient unit benefit costs, a portion of 
ancillary departments benefit costs, and a portion of overhead benefits 
attributable to both the routine inpatient unit and the ancillary 
departments. For those hospital-based IRFs that report Worksheet S-3, 
part V data, we proposed inpatient unit benefit costs be equal to 
Worksheet S-3, part V, column 2, line 4. Given the limited reporting on 
Worksheet S-3, part V, we proposed that for those hospital-based IRFs 
that do not report these data, we calculate inpatient unit benefits 
costs using a portion of benefits costs reported for Excluded areas on 
Worksheet S-3, part II. We proposed to calculate the ratio of inpatient 
unit salaries (Worksheet A, column 1, line 41) to total excluded area 
salaries (sum of Worksheet A, column 1, lines 20, 23, 40 through 42, 
44, 45, 46, 94, 95, 98 through 101, 105 through 112, 114, 115 through 
117, 190 through 194). We then proposed to apply this ratio to Excluded 
area benefits (Worksheet S-3, part II, column 4, line 19) to derive 
inpatient unit benefits costs for those providers that do not report 
benefit costs on Worksheet S-3, part V.
    We proposed the ancillary departments benefits and overhead 
benefits (attributable to both the inpatient unit and ancillary 
departments) costs are derived by first calculating the sum of 
hospital-based IRF overhead salaries as described in section 
V.C.1.a.(2)(a) of the proposed rule, hospital-based IRF ancillary 
salaries as described in section V.C.1.a.(2)(b) of the proposed rule 
and hospital-based IRF overhead salaries for ancillary cost centers as 
described in section V.C.1.a.(2)(c) of the proposed rule. This sum is 
then multiplied by the ratio of total facility benefits to total 
facility salaries, where total facility benefits is equal to the sum of 
Worksheet S-3, part II, column 4, lines 17-25, and total facility 
salaries is equal to Worksheet S-3, part II, column 4, line 1.
(4) Contract Labor Costs
    Contract Labor costs are primarily associated with direct patient 
care services. Contract labor costs for other services such as 
accounting, billing, and legal are calculated separately using other 
government data sources as described in section V.C.1.c. of the 
proposed rule. To derive contract labor costs using Worksheet S-3, part 
V, data, for freestanding IRFs, we proposed Contract Labor costs be 
equal to Worksheet S-3, part V, column 1, line 2. As we noted for 
Employee Benefits, freestanding IRFs also may report Contract Labor 
data on Worksheet S-3, part II, which is applicable to only IPPS 
providers. For those freestanding IRFs that report Worksheet S-3, part 
II data, but not Worksheet S-3, part V, we proposed to use the sum of 
Worksheet S-3, part II, column 4, lines 11 and 13, to derive Contract 
Labor costs.
    For hospital-based IRFs, we proposed that Contract Labor costs 
would be equal to Worksheet S-3, part V, column 1, line 4. For 2021 
Medicare cost report data, 30 percent of providers reported Contract 
Labor data on Worksheet S-3, part V; particularly, approximately 56 
percent of freestanding IRFs and 18 percent of hospital-based IRFs 
reported data on Worksheet S-3, part V. For comparison, for the 2016-
based IRF market basket, about 26 percent of providers reported 
Contract Labor data on Worksheet S-3, part V. We continue to encourage 
all providers to report these data on the Medicare cost report.
    Given the limited reporting on Worksheet S-3, part V, we proposed 
that for those hospital-based IRFs that do not report these data, we 
calculate Contract Labor costs using a portion of contract labor costs 
reported on Worksheet S-3, part II. We proposed to calculate the ratio 
of contract labor costs (Worksheet S-3, part II, column 4, lines 11 and 
13) to PPS salaries (Worksheet S-3, part II, column 4, line 1 less the 
sum of Worksheet S-3, part II, column 4, lines 3, 401, 5, 6, 7, 701, 8, 
9, 10 less Worksheet A, column 1, line 20 and 23). We then proposed to 
apply this ratio to total inpatient routine salary costs (Worksheet A, 
column 1, line 41) to derive contract labor costs for those providers 
that do not report contract labor costs on Worksheet S-3, part V.
(5) Pharmaceuticals Costs
    For freestanding IRFs, we proposed to calculate pharmaceuticals 
costs using non-salary costs reported on Worksheet A, column 7, less 
Worksheet A, column 1, for the pharmacy cost center (line 15) and drugs 
charged to patients cost center (line 73).
    For hospital-based IRFs, we proposed to calculate pharmaceuticals 
costs as the sum of a portion of the non-salary pharmacy costs and a 
portion of the non-salary drugs charged to patient costs reported for 
the total facility. We proposed that non-salary pharmacy costs 
attributable to the hospital-based IRF would be calculated by 
multiplying total pharmacy costs attributable to the hospital-based IRF 
(as reported on Worksheet B, part I, column 15, line 41) by the ratio 
of total non-salary pharmacy costs (Worksheet A, column 2, line 15) to 
total pharmacy costs (sum of Worksheet A, columns 1 and 2 for line 15) 
for the total facility. We proposed that non-salary drugs charged to 
patient costs attributable to the hospital-based IRF would be 
calculated by multiplying total non-salary drugs charged to patient 
costs (Worksheet B, part I, column 0, line 73 plus Worksheet B, part I, 
column 15, line 73 less Worksheet A, column 1, line 73) for the total 
facility by the ratio of Medicare drugs charged to patient ancillary 
costs for the IRF unit (as reported on Worksheet D-3 for hospital-based 
IRFs, column 3, line 73) to total Medicare drugs charged to patient 
ancillary costs for the total facility (equal to the sum of Worksheet 
D-3, column 3, line 73 for all relevant PPS (that is, IPPS, IRF, IPF 
and SNF).
(6) Professional Liability Insurance Costs
    For freestanding and hospital-based IRFs, we proposed that 
Professional Liability Insurance (PLI) costs (often referred to as 
malpractice costs) would be equal to premiums, paid losses and self-
insurance costs reported on Worksheet S-2, columns 1 through 3, line 
118--the same data used for the 2016-based IRF market basket. For 
hospital-based IRFs, we proposed to assume that the PLI weight for the 
total facility is similar to the hospital-based IRF unit since the only 
data reported on this worksheet is for the entire facility, as we 
currently have no means to identify the proportion of total PLI costs 
that are only attributable to the hospital-based IRF. However, when we 
derive the cost weight for PLI for both hospital-based and freestanding 
IRFs, we use the total facility medical care costs as the denominator 
as opposed to total Medicare allowable costs. For freestanding IRFs and 
hospital-based IRFs, we proposed to derive total facility medical care 
costs as the sum of total costs (Worksheet B, part I, column 26, line 
202) less non-reimbursable costs (Worksheet B, part I, column 26, lines 
190 through 201).
(7) Home Office/Related Organization Contract Labor Costs
    For freestanding and hospital-based IRFs, we proposed to calculate 
the home office/related organization contract labor costs using data 
reported on Worksheet S-3, part II, column 4, lines 1401, 1402, 2550, 
and 2551. Similar to the PLI costs, these costs are for the entire 
facility. Therefore, when we derive the cost weight for Home Office/
Related Organization Contract Labor costs, we use the total facility 
medical care costs as the denominator (reflecting the total facility 
costs less the non-reimbursable costs reported on lines 190 through 
201). Our assumption is that the

[[Page 50970]]

same proportion of expenses are used among each unit of the hospital.
(8) Capital Costs
    For freestanding IRFs, we proposed that capital costs would be 
equal to Medicare allowable capital costs as reported on Worksheet B, 
part II, column 26, lines 30 through 35, 50 through 76 (excluding 52 
and 75), 90 through 91, and 93.
    For hospital-based IRFs, we proposed that capital costs would be 
equal to IRF inpatient capital costs (as reported on Worksheet B, part 
II, column 26, line 41) and a portion of IRF ancillary capital costs. 
We calculate the portion of ancillary capital costs attributable to the 
hospital-based IRF for a given cost center by multiplying total 
facility ancillary capital costs for the specific ancillary cost center 
(as reported on Worksheet B, part II, column 26) by the IRF ancillary 
ratio as described in section V.C.1.a.(1) of the proposed rule. For 
example, if hospital-based IRF Medicare physical therapy costs 
represent 30 percent of the total Medicare physical therapy costs for 
the entire facility, then 30 percent of total facility physical therapy 
capital costs (as reported in Worksheet B, part II, column 26, line 66) 
would be attributable to the hospital-based IRF.
b. Final Major Cost Category Computation
    After we derive costs for each of the major cost categories and 
total Medicare allowable costs for each provider using the Medicare 
cost report data as previously described, we proposed to address data 
outliers using the following steps. First, for the Wages and Salaries, 
Employee Benefits, Contract Labor, Pharmaceuticals, and Capital cost 
weights, we first divide the costs for each of these five categories by 
total Medicare allowable costs calculated for the provider to obtain 
cost weights for the universe of IRF providers. We then proposed to 
trim the data to remove outliers (a standard statistical process) by: 
(1) requiring that major expenses (such as Wages and Salaries costs) 
and total Medicare allowable operating costs be greater than zero; and 
(2) excluding the top and bottom 5 percent of the major cost weight 
(for example, Wages and Salaries costs as a percent of total Medicare 
allowable operating costs). We note that missing values are assumed to 
be zero consistent with the methodology for how missing values were 
treated in the 2016-based IRF market basket. After these outliers have 
been excluded, we sum the costs for each category across all remaining 
providers. We then divide this by the sum of total Medicare allowable 
costs across all remaining providers to obtain a cost weight for the 
2021-based IRF market basket for the given category.
    The proposed trimming methodology for the Home Office/Related 
Organization Contract Labor and PLI cost weights is slightly different 
than the proposed trimming methodology for the other five cost 
categories as described previously in this final rule. For these cost 
weights, since we are using total facility medical care costs rather 
than Medicare allowable costs associated with IRF services, we proposed 
to trim the freestanding and hospital-based IRF cost weights 
separately.
    For the PLI cost weight, for each of the providers, we first divide 
the PLI costs by total facility medical care costs to obtain a PLI cost 
weight for the universe of IRF providers. We then proposed to trim the 
data to remove outliers by: (1) requiring that PLI costs are greater 
than zero and are less than total facility medical care costs; and (2) 
excluding the top and bottom 5 percent of the major cost weight 
trimming freestanding and hospital-based providers separately. After 
removing these outliers, we are left with a trimmed data set for both 
freestanding and hospital-based providers. We then proposed to 
separately sum the costs for each category (freestanding and hospital-
based) across all remaining providers. We next divide this by the sum 
of total facility medical care costs across all remaining providers to 
obtain both a freestanding cost weight and hospital-based cost weight. 
Lastly, we proposed to weight these two cost weights together using the 
Medicare allowable costs from the sample of freestanding and hospital-
based IRFs that passed the PLI trim (59 percent for hospital-based and 
41 percent for freestanding IRFs) to derive a PLI cost weight for the 
2021-based IRF market basket.
    For the Home Office/Related Organization Contract Labor cost 
weight, for each of the providers, we first divide the home office/
related organization contract labor costs by total facility medical 
care costs to obtain a Home Office/Related Organization Contract Labor 
cost weight for the universe of IRF providers. We then proposed to trim 
only the top 1 percent of providers to exclude outliers while also 
allowing providers who have reported zero home office costs to remain 
in the Home Office/Related Organization Contract Labor cost weight 
calculations as not all providers will incur home office/relation 
organization contract labor costs. After removing these outliers, we 
are left with a trimmed data set for both freestanding and hospital-
based providers. We then proposed to separately sum the costs for each 
category (freestanding and hospital-based) across all remaining 
providers. We next divide this by the sum of total facility medical 
care costs across all remaining providers to obtain a freestanding cost 
weight and hospital-based cost weight. Lastly, we proposed to weight 
these two cost weights together using the Medicare allowable costs from 
the sample of freestanding and hospital-based IRFs that passed the Home 
Office/Related Organization Contract Labor cost weight trim (68 percent 
for hospital-based and 32 percent for freestanding IRFs) to derive a 
Home Office/Related Organization Contract Labor cost weight for the 
2021-based IRF market basket.
    Finally, we proposed to calculate the residual ``All Other'' cost 
weight that reflects all remaining costs that are not captured in the 
seven cost categories listed. See Table 4 for the resulting cost 
weights for these major cost categories that we obtain from the 
Medicare cost reports.

[[Page 50971]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.054

    As we did for the 2016-based IRF market basket, we proposed to 
allocate the Contract Labor cost weight to the Wages and Salaries and 
Employee Benefits cost weights based on their relative proportions 
under the assumption that contract labor costs are comprised of both 
wages and salaries and employee benefits. The Contract Labor allocation 
proportion for Wages and Salaries is equal to the Wages and Salaries 
cost weight as a percent of the sum of the Wages and Salaries cost 
weight and the Employee Benefits cost weight. For the proposed rule, 
the rounded percentage is 80 percent; therefore, we proposed to 
allocate 80 percent of the Contract Labor cost weight to the Wages and 
Salaries cost weight and 20 percent to the Employee Benefits cost 
weight. This allocation was 81/19 in the 2016-based IRF market basket 
(84 FR 39076). Table 5 shows the Wages and Salaries and Employee 
Benefit cost weights after Contract Labor cost weight allocation for 
both the 2021-based IRF market basket and 2016-based IRF market basket.
[GRAPHIC] [TIFF OMITTED] TR02AU23.055

    The following is a summary of the public comments received on our 
proposed methodology for developing the major cost weights of the 2021-
based IRF market basket and our responses.
    Comment: A few commenters noted that their review of the market 
basket cost categories shows only modest increases, including with 
respect to labor and capital-related costs, despite their members 
experiencing much more significant actual increases in expenditures 
compared to 2016. One commenter requested that CMS consider increases 
in wages, salaries, benefits, and contract labor, among other 
categories, in its methodology.
    One commenter supported the increase in proposed weights given the 
sustained labor increases and market challenges. However, the commenter 
stated that labor and supplies are significant stressors and requested 
CMS review pharmaceuticals and capital-related costs more closely 
before the final rule. The commenter stated that while they recognize 
that not all categories can increase, these components have all 
contributed to financial strain on the industry and stated that a 
decrease in their cost weights in the market basket does not reflect 
their current contribution to overall costs.
    Response: As discussed previously, the major cost weights 
calculated from the Medicare cost reports for the 2021-based IRF market 
basket represent each cost category's share of total costs. Therefore, 
any changes in the cost weight from a prior base period will reflect 
the growth in the costs for that specific category relative to the 
growth in the costs for other categories. As a result, while costs for 
a particular category may have increased from 2016 to 2021 (such as 
capital-related costs as stated by the commenters), the Capital-Related 
cost weight would only increase if capital-related costs increased 
faster than the increase in total costs from 2016 to 2021. In response 
to the commenters' request that CMS consider increases in wages, 
salaries, benefits, and contract labor, among other categories, in its 
methodology, we believe that the proposed methodology to derive the 
major cost categories is detailed and robust. To allow for interested 
parties to evaluate this methodology, we have provided all of the 
detailed calculations and Medicare cost report fields so that 
commenters are able to replicate the methodology and provide specific 
comments on the derivation of these cost weights. We will continue to 
monitor the Medicare cost reports as new data becomes available for all 
of the major cost weights, including the categories mentioned by the 
commenter, and any changes to the IRF market basket will be proposed in 
future rulemaking.
    We appreciate the commenter's request to review the pharmaceuticals 
and capital-related costs used in the proposed 2021-based IRF market 
basket more closely. We note that each of the cost weights in the 
market basket reflect a distribution and will change over time only 
when costs grow differently (either

[[Page 50972]]

higher or lower) than other costs. The Pharmaceuticals cost weight in 
the 2021-based IRF market basket is 4.7 percent compared to the 2016-
based IRF market basket with 5.1 percent. We examined the Medicare cost 
report data in more detail and found that the Pharmaceuticals cost 
weight decreased, in aggregate, for both urban and rural IRFs, 
government and for-profit IRFs, and for freestanding and hospital-based 
IRFs. The median Pharmaceuticals cost weight also decreased from 5.0 
percent to 4.4 percent. Therefore, we believe that the proposed 
Pharmaceuticals cost weight is appropriate and reflects its share of 
overall costs.
    The Capital-Related cost weight in the 2021-based IRF market basket 
is 8.6 percent compared to the 2016-based IRF market basket with 9.0 
percent. We examined the Medicare cost report data in more detail and 
found that the Capital-Related cost weight decreased, in aggregate, for 
both urban and rural IRFs and for all ownership-types. The median 
Capital-Related cost weight also decreased from 8.8 percent to 8.1 
percent. We note that both pharmaceuticals and capital-related costs 
per day increased from 2016 to 2021; however, they increased at a 
slower rate than total Medicare allowable costs per day (which is the 
denominator in the cost weight calculation) resulting in slightly lower 
cost weights in 2021 compared to 2016. Therefore, we believe that the 
proposed Capital-Related cost weight is appropriate and reflects its 
share of overall costs.
    Comment: A few commenters requested that CMS educate interested 
parties on the importance of reporting accurate and robust data on the 
Medicare cost reports. One commenter recognized that CMS is relying on 
the Medicare cost report data for the market basket cost weights, but 
noted that such data may not always be adequately recorded or 
prioritized for input. One commenter specifically noted that not all 
IRFs are properly reporting data for Employee Benefits and Contract 
Labor on the Medicare cost reports. The commenter stated that while all 
of their hospitals have reported these cost report line items, they 
urged CMS to emphasize their importance to ensure that the IRF sector 
understands the importance of accurately and fully reporting these line 
items to reduce data gaps for future updates.
    Response: We recognize the commenters' concerns and reiterate that 
accurate and complete reporting of all data on the Medicare cost 
reports by IRFs help to ensure that the cost weights for the IRF market 
basket are reflective of the cost structure of IRFs. We also note that 
we analyze the Medicare cost report data to evaluate their 
representativeness; for example, we reweight the data reported by 
ownership type and urban/rural so that it reflects the universe of 
providers and compare it to the proposed cost weights that are based on 
reported data. Our analysis shows the proposed cost weights are 
representative across these dimensions. In addition, we also trim the 
data to eliminate outliers as described in section VI.C.1.b. of this 
final rule. As stated in the FY 2024 IRF PPS proposed rule (88 FR 
20961) and previous IRF PPS rules, we continue to encourage all 
providers to report the Employee Benefits and Contract Labor data on 
the Medicare cost report. Going forward, we will continue to work with 
interested parties to communicate the importance of all providers 
filling out the Medicare cost report with accurate and complete data.
    After consideration of the public comments, we are finalizing our 
methodology for developing the major cost weights and therefore, we are 
finalizing these major cost weights as proposed.
c. Derivation of the Detailed Operating Cost Weights
    To further divide the ``All Other'' residual cost weight estimated 
from the 2021 Medicare cost report data into more detailed cost 
categories, we proposed to use the 2012 Benchmark Input-Output (I-O) 
``Use Tables/Before Redefinitions/Purchaser Value'' for North American 
Industry Classification System (NAICS) 622000, Hospitals, published by 
the Bureau of Economic Analysis (BEA). This data is publicly available 
at http://www.bea.gov/industry/io_annual.htm. For the 2016-based IRF 
market basket, we also used the 2012 Benchmark I-O data, the most 
recent data available at the time (84 FR 39076).
    The BEA Benchmark I-O data are scheduled for publication every 5 
years with the most recent data available for 2012. The 2012 Benchmark 
I-O data are derived from the 2012 Economic Census and are the building 
blocks for BEA's economic accounts. Thus, they represent the most 
comprehensive and complete set of data on the economic processes or 
mechanisms by which output is produced and distributed.\16\ BEA also 
produces Annual I-O estimates; however, while based on a similar 
methodology, these estimates reflect less comprehensive and less 
detailed data sources and are subject to revision when benchmark data 
becomes available. Instead of using the less detailed Annual I-O data, 
we proposed to inflate the 2012 Benchmark I-O data forward to 2021 by 
applying the annual price changes from the respective price proxies to 
the appropriate market basket cost categories that are obtained from 
the 2012 Benchmark I-O data. We repeat this practice for each year. We 
then proposed to calculate the cost shares that each cost category 
represents of the inflated 2012 data. These resulting 2021 cost shares 
are applied to the All Other residual cost weight to obtain the 
detailed cost weights for the 2021-based IRF market basket. For 
example, the cost for Food: Direct Purchases represents 5.0 percent of 
the sum of the ``All Other'' 2012 Benchmark I-O Hospital Expenditures 
inflated to 2021; therefore, the Food: Direct Purchases cost weight 
represents 5.0 percent of the 2021-based IRF market basket's ``All 
Other'' cost category (20.4 percent), yielding a ``final'' Food: Direct 
Purchases cost weight of 1.0 percent in the 2021-based IRF market 
basket (0.05 * 20.4 percent = 1.0 percent).
---------------------------------------------------------------------------

    \16\ http://www.bea.gov/papers/pdf/IOmanual_092906.pdf.
---------------------------------------------------------------------------

    Using this methodology, we proposed to derive seventeen detailed 
IRF market basket cost category weights from the 2021-based IRF market 
basket residual cost weight (20.4 percent). These categories are: (1) 
Electricity and Other Non-Fuel Utilities, (2) Fuel: Oil and Gas (3) 
Food: Direct Purchases, (4) Food: Contract Services, (5) Chemicals, (6) 
Medical Instruments, (7) Rubber and Plastics, (8) Paper and Printing 
Products, (9) Miscellaneous Products, (10) Professional Fees: Labor-
Related, (11) Administrative and Facilities Support Services, (12) 
Installation, Maintenance, and Repair Services, (13) All Other Labor-
Related Services, (14) Professional Fees: Nonlabor-Related, (15) 
Financial Services, (16) Telephone Services, and (17) All Other 
Nonlabor-Related Services.
    We did not receive any comments on our methodology to use the BEA 
I-O data to derive the detailed operating cost weights. We are 
finalizing this methodology as we proposed. We note that we did receive 
one comment on the derivation of the Professional Fees: Labor-Related 
cost weight which we discuss in section VI.E. of this final rule.
d. Derivation of the Detailed Capital Cost Weights
    As described in section V.C.1.b. of the proposed rule, we proposed 
a Capital-Related cost weight of 8.6 percent as obtained from the 2021 
Medicare cost reports for freestanding and hospital-based IRF 
providers. We proposed to

[[Page 50973]]

then separate this total Capital-Related cost weight into more detailed 
cost categories.
    Using 2021 Medicare cost reports, we are able to group Capital-
Related costs into the following categories: Depreciation, Interest, 
Lease, and Other Capital-Related costs. For each of these categories, 
we proposed to determine separately for hospital-based IRFs and 
freestanding IRFs what proportion of total capital-related costs the 
category represents.
    For freestanding IRFs, using Medicare cost report data on Worksheet 
A-7 part III, we proposed to derive the proportions for Depreciation 
(column 9), Interest (column 11), Lease (column 10), and Other Capital-
Related costs (column 12 through 14), which is similar to the 
methodology used for the 2016-based IRF market basket.
    For hospital-based IRFs, data for these four categories are not 
reported separately for the hospital-based IRF; therefore, we proposed 
to derive these proportions using data reported on Worksheet A-7 for 
the total facility. We assumed the cost shares for the overall hospital 
are representative for the hospital-based IRF unit. For example, if 
depreciation costs make up 60 percent of total capital costs for the 
entire facility, we believe it is reasonable to assume that the 
hospital-based IRF would also have a 60 percent proportion because it 
is a unit contained within the total facility. This is the same 
methodology used for the 2016-based IRF market basket (84 FR 39077).
    To combine each detailed capital cost weight for freestanding and 
hospital-based IRFs into a single capital cost weight for the 2021-
based IRF market basket, we proposed to weight together the shares for 
each of the categories (Depreciation, Interest, Lease, and Other 
Capital-Related costs) based on the share of total capital costs each 
provider type represents of the total capital costs for all IRFs for 
2021. Applying this methodology results in proportions of total 
capital-related costs for Depreciation, Interest, Lease and Other 
Capital-Related costs that are representative of the universe of IRF 
providers. This is the same methodology used for the 2016-based IRF 
market basket (84 FR 39077).
    Lease costs are unique in that they are not broken out as a 
separate cost category in the 2021-based IRF market basket. Rather, we 
proposed to proportionally distribute these costs among the cost 
categories of Depreciation, Interest, and Other Capital-Related costs, 
reflecting the assumption that the underlying cost structure of leases 
is similar to that of capital-related costs in general. As was done 
under the 2016-based IRF market basket, we proposed to assume that 10 
percent of the lease costs as a proportion of total capital-related 
costs represents overhead and assign those costs to the Other Capital-
Related cost category accordingly. We proposed to distribute the 
remaining lease costs proportionally across the three cost categories 
(Depreciation, Interest, and Other Capital-Related) based on the 
proportion that these categories comprise of the sum of the 
Depreciation, Interest, and Other Capital-Related cost categories 
(excluding lease expenses). This would result in three primary capital-
related cost categories in the 2021-based IRF market basket: 
Depreciation, Interest, and Other Capital-Related costs. This is the 
same methodology used for the 2016-based IRF market basket (84 FR 
39077). The allocation of these lease expenses is shown in Table 6.
    Finally, we proposed to further divide the Depreciation and 
Interest cost categories. We proposed to separate Depreciation into the 
following two categories: (1) Building and Fixed Equipment and (2) 
Movable Equipment. We proposed to separate Interest into the following 
two categories: (1) Government/Nonprofit and (2) For-profit.
    To disaggregate the Depreciation cost weight, we need to determine 
the percent of total Depreciation costs for IRFs that is attributable 
to Building and Fixed Equipment, which we hereafter refer to as the 
``fixed percentage.'' For the 2021-based IRF market basket, we proposed 
to use slightly different methods to obtain the fixed percentages for 
hospital-based IRFs compared to freestanding IRFs.
    For freestanding IRFs, we proposed to use depreciation data from 
Worksheet A-7 of the 2021 Medicare cost reports. However, for hospital-
based IRFs, we determined that the fixed percentage for the entire 
facility may not be representative of the hospital-based IRF unit due 
to the entire facility likely employing more sophisticated movable 
assets that are not utilized by the hospital-based IRF. Therefore, for 
hospital-based IRFs, we proposed to calculate a fixed percentage using: 
(1) building and fixture capital costs allocated to the hospital-based 
IRF unit as reported on Worksheet B, part I, column 1, line 41, and (2) 
building and fixture capital costs for the top five ancillary cost 
centers utilized by hospital-based IRFs accounting for 78 percent of 
hospital-based IRF ancillary total costs: Physical Therapy (Worksheet 
B, part I, column 1, line 66), Drugs Charged to Patients (Worksheet B, 
part I, column 1, line 73), Occupational Therapy (Worksheet B, part I, 
column 1, line 67), Laboratory (Worksheet B, part I, column 1, line 60) 
and Clinic (Worksheet B, part I, column 1, line 90). We proposed to 
weight these two fixed percentages (inpatient and ancillary) using the 
proportion that each capital cost type represents of total capital 
costs in the 2021-based IRF market basket. We proposed to then weight 
the fixed percentages for hospital-based and freestanding IRFs together 
using the proportion of total capital costs each provider type 
represents. For both freestanding and hospital-based IRFs, this is the 
same methodology used for the 2016-based IRF market basket (84 FR 
39077).
    To disaggregate the Interest cost weight, we determined the percent 
of total interest costs for IRFs that are attributable to government 
and nonprofit facilities, which is hereafter referred to as the 
``nonprofit percentage,'' as price pressures associated with these 
types of interest costs tend to differ from those for for-profit 
facilities. For the 2021-based IRF market basket, we proposed to use 
interest costs data from Worksheet A-7 of the 2021 Medicare cost 
reports for both freestanding and hospital-based IRFs. We proposed to 
determine the percent of total interest costs that are attributed to 
government and nonprofit IRFs separately for hospital-based and 
freestanding IRFs. We then proposed to weight the nonprofit percentages 
for hospital-based and freestanding IRFs together using the proportion 
of total capital costs that each provider type represents.
    Table 6 provides the detailed capital cost share composition 
estimated from the 2021 IRF Medicare cost reports. These detailed 
capital cost share composition percentages are applied to the total 
Capital-Related cost weight of 8.6 percent calculated using the 
methodology described in section V.C.1.a.(8) of the proposed rule.

[[Page 50974]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.056

    We did not receive any comments on our proposed methodology for 
developing the detailed capital cost weights of the 2021-based IRF 
market basket. We are finalizing these detailed capital cost weights as 
proposed.
e. 2021-Based IRF Market Basket Cost Categories and Weights
    Table 7 compares the cost categories and weights for the 2021-based 
IRF market basket compared to the 2016-based IRF market basket.

[[Page 50975]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.057

2. Selection of Price Proxies
    After developing the cost weights for the 2021-based IRF market 
basket, we proposed to select the most appropriate wage and price 
proxies currently available to represent the rate of price change for 
each expenditure category. For the majority of the cost weights, we 
base the price proxies on U.S. Bureau of Labor Statistics (BLS) data 
and group them into one of the following BLS categories:
     Employment Cost Indexes. Employment Cost Indexes (ECIs) 
measure the rate of change in employment wage rates and employer costs 
for employee benefits per hour worked. These indexes are fixed-weight 
indexes and strictly measure the change in wage rates and employee 
benefits per hour. ECIs are superior to Average Hourly Earnings (AHE) 
as price proxies for input price indexes because they are not affected 
by shifts in occupation or industry mix, and because they measure pure 
price change and are available by both occupational group and by 
industry. The industry ECIs are based on the NAICS and the occupational 
ECIs are based on the Standard Occupational Classification System 
(SOC).
     Producer Price Indexes. Producer Price Indexes (PPIs) 
measure the average change over time in the selling prices received by 
domestic producers for their output. The prices included in the PPI are 
from the first commercial

[[Page 50976]]

transaction for many products and some services (https://www.bls.gov/ppi/).
     Consumer Price Indexes. Consumer Price Indexes (CPIs) 
measure the average change over time in the prices paid by urban 
consumers for a market basket of consumer goods and services (https://www.bls.gov/cpi/). CPIs are only used when the purchases are similar to 
those of retail consumers rather than purchases at the producer level, 
or if no appropriate PPIs are available.
    We evaluated the price proxies using the criteria of reliability, 
timeliness, availability, and relevance:
     Reliability. Reliability indicates that the index is based 
on valid statistical methods and has low sampling variability. Widely 
accepted statistical methods ensure that the data were collected and 
aggregated in a way that can be replicated. Low sampling variability is 
desirable because it indicates that the sample reflects the typical 
members of the population. (Sampling variability is variation that 
occurs by chance because only a sample was surveyed rather than the 
entire population.)
     Timeliness. Timeliness implies that the proxy is published 
regularly, preferably at least once a quarter. The market baskets are 
updated quarterly, and therefore, it is important for the underlying 
price proxies to be up-to-date, reflecting the most recent data 
available. We believe that using proxies that are published regularly 
(at least quarterly, whenever possible) helps to ensure that we are 
using the most recent data available to update the market basket. We 
strive to use publications that are disseminated frequently, because we 
believe that this is an optimal way to stay abreast of the most current 
data available.
     Availability. Availability means that the proxy is 
publicly available. We prefer that our proxies are publicly available 
because this will help ensure that our market basket updates are as 
transparent to the public as possible. In addition, this enables the 
public to be able to obtain the price proxy data on a regular basis.
     Relevance. Relevance means that the proxy is applicable 
and representative of the cost category weight to which it is applied. 
The CPIs, PPIs, and ECIs that we have selected to propose in this 
regulation meet these criteria. Therefore, we believe that they 
continue to be the best measure of price changes for the cost 
categories to which they would be applied.
    Below is a detailed explanation of the price proxies we proposed 
for each cost category weight.
a. Price Proxies for the Operating Portion of the 2021-Based IRF Market 
Basket
(1) Wages and Salaries
    We proposed to continue to use the ECI for Wages and Salaries for 
All Civilian workers in Hospitals (BLS series code CIU1026220000000I) 
to measure the wage rate growth of this cost category. This is the same 
price proxy used in the 2016-based IRF market basket (84 FR 39080).
(2) Benefits
    We proposed to continue to use the ECI for Total Benefits for All 
Civilian workers in Hospitals to measure price growth of this category. 
This ECI is calculated using the ECI for Total Compensation for All 
Civilian workers in Hospitals (BLS series code CIU1016220000000I) and 
the relative importance of wages and salaries within total 
compensation. This is the same price proxy used in the 2016-based IRF 
market basket (84 FR 39080).
(3) Electricity and Other Non-Fuel Utilities
    We proposed to continue to use the PPI Commodity Index for 
Commercial Electric Power (BLS series code WPU0542) to measure the 
price growth of this cost category (which we proposed to rename from 
Electricity to Electricity and Other Non-Fuel Utilities). This is the 
same price proxy used in the 2016-based IRF market basket (84 FR 
39080).
(4) Fuel: Oil and Gas
    Similar to the 2016-based IRF market basket, for the 2021-based IRF 
market basket, we proposed to use a blend of the PPI for Petroleum 
Refineries and the PPI Commodity for Natural Gas. Our analysis of the 
Bureau of Economic Analysis' 2012 Benchmark Input-Output data (use 
table before redefinitions, purchaser's value for NAICS 622000 
[Hospitals]), shows that Petroleum Refineries expenses account for 
approximately 90 percent and Natural Gas expenses account for 
approximately 10 percent of Hospitals' (NAICS 622000) total Fuel: Oil 
and Gas expenses. Therefore, we proposed to use a blend of 90 percent 
of the PPI for Petroleum Refineries (BLS series code PCU324110324110) 
and 10 percent of the PPI Commodity Index for Natural Gas (BLS series 
code WPU0531) as the price proxy for this cost category. This is the 
same blend that was used for the 2016-based IRF market basket (84 FR 
39080).
(5) Professional Liability Insurance
    We proposed to continue to use the CMS Hospital Professional 
Liability Index to measure changes in PLI premiums. To generate this 
index, we collect commercial insurance premiums for a fixed level of 
coverage while holding non-price factors constant (such as a change in 
the level of coverage). This is the same proxy used in the 2016-based 
IRF market basket (84 FR 39080).
(6) Pharmaceuticals
    We proposed to continue to use the PPI for Pharmaceuticals for 
Human Use, Prescription (BLS series code WPUSI07003) to measure the 
price growth of this cost category. This is the same proxy used in the 
2016-based IRF market basket (84 FR 39080).
(7) Food: Direct Purchases
    We proposed to continue to use the PPI for Processed Foods and 
Feeds (BLS series code WPU02) to measure the price growth of this cost 
category. This is the same proxy used in the 2016-based IRF market 
basket (84 FR 39080).
(8) Food: Contract Purchases
    We proposed to continue to use the CPI for Food Away From Home (BLS 
series code CUUR0000SEFV) to measure the price growth of this cost 
category. This is the same proxy used in the 2016-based IRF market 
basket (84 FR 39080).
(9) Chemicals
    Similar to the 2016-based IRF market basket, we proposed to use a 
four-part blended PPI as the proxy for the chemical cost category in 
the 2021-based IRF market basket. The blend is composed of the PPI for 
Industrial Gas Manufacturing, Primary Products (BLS series code 
PCU325120325120P), the PPI for Other Basic Inorganic Chemical 
Manufacturing (BLS series code PCU32518-32518-), the PPI for Other 
Basic Organic Chemical Manufacturing (BLS series code PCU32519-32519-), 
and the PPI for Other Miscellaneous Chemical Product Manufacturing (BLS 
series code PCU325998325998). For the 2021-based IRF market basket, we 
proposed to derive the weights for the PPIs using the 2012 Benchmark I-
O data.
    Table 8 shows the weights for each of the four PPIs used to create 
the blended Chemical proxy for the 2021 IRF market basket. This is the 
same blend that was used for the 2016-based IRF market basket (84 FR 
39080).

[[Page 50977]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.058

(10) Medical Instruments
    We proposed to use a blended price proxy for the Medical 
Instruments category, as shown in Table 9. The 2012 Benchmark I-O data 
shows the majority of medical instruments and supply costs are for 
NAICS 339112--Surgical and medical instrument manufacturing costs 
(approximately 56 percent) and NAICS 339113--Surgical appliance and 
supplies manufacturing costs (approximately 43 percent). Therefore, we 
proposed to use a blend of these two price proxies. To proxy the price 
changes associated with NAICS 339112, we proposed using the PPI for 
Surgical and medical instruments (BLS series code WPU1562). This is the 
same price proxy we used in the 2016-based IRF market basket. To proxy 
the price changes associated with NAICS 339113, we proposed to use a 
50/50 blend of the PPI for Medical and surgical appliances and supplies 
(BLS series code WPU1563) and the PPI for Miscellaneous products, 
Personal safety equipment and clothing (BLS series code WPU1571). We 
proposed to include the latter price proxy as it would reflect personal 
protective equipment including but not limited to face shields and 
protective clothing. The 2012 Benchmark I-O data does not provide 
specific expenses for these products; however, we recognize that this 
category reflects costs faced by IRFs.
[GRAPHIC] [TIFF OMITTED] TR02AU23.059

(11) Rubber and Plastics
    We proposed to continue to use the PPI for Rubber and Plastic 
Products (BLS series code WPU07) to measure price growth of this cost 
category. This is the same proxy used in the 2016-based IRF market 
basket (84 FR 39081).
(12) Paper and Printing Products
    We proposed to continue to use the PPI for Converted Paper and 
Paperboard Products (BLS series code WPU0915) to measure the price 
growth of this cost category. This is the same proxy used in the 2016-
based IRF market basket (84 FR 39081).
(13) Miscellaneous Products
    We proposed to continue to use the PPI for Finished Goods Less Food 
and Energy (BLS series code WPUFD4131) to measure the price growth of 
this cost category. This is the same proxy used in the 2016-based IRF 
market basket (84 FR 39081).
(14) Professional Fees: Labor-Related
    We proposed to continue to use the ECI for Total Compensation for 
Private Industry workers in Professional and Related (BLS series code 
CIU2010000120000I) to measure the price growth of this category. This 
is the same proxy used in the 2016-based IRF market basket (84 FR 
39081).
(15) Administrative and Facilities Support Services
    We proposed to continue to use the ECI for Total Compensation for 
Private Industry workers in Office and Administrative Support (BLS 
series code CIU2010000220000I) to measure the price growth of this 
category. This is the same proxy used in the 2016-based IRF market 
basket (84 FR 39081).
(16) Installation, Maintenance, and Repair Services
    We proposed to continue to use the ECI for Total Compensation for 
Civilian workers in Installation, Maintenance, and Repair (BLS series 
code CIU1010000430000I) to measure the price growth of this cost 
category. This is the same proxy used in the 2016-based IRF market 
basket (84 FR 39081).
(17) All Other: Labor-Related Services
    We proposed to continue to use the ECI for Total Compensation for 
Private Industry workers in Service Occupations (BLS series code 
CIU2010000300000I) to measure the price growth of this cost category. 
This is the same proxy used in the 2016-based IRF market basket (84 FR 
39081).
(18) Professional Fees: Nonlabor-Related
    We proposed to continue to use the ECI for Total Compensation for 
Private Industry workers in Professional and Related (BLS series code 
CIU2010000120000I) to measure the price growth of this category. This 
is the same proxy used in the 2016-based IRF market basket (84 FR 
39081).
(19) Financial Services
    We proposed to continue to use the ECI for Total Compensation for 
Private Industry workers in Financial Activities (BLS series code 
CIU201520A000000I) to measure the price growth of this cost category. 
This is the same proxy used in the 2016-based IRF market basket (84 FR 
39081).
(20) Telephone Services
    We proposed to continue to use the CPI for Telephone Services (BLS 
series code CUUR0000SEED) to measure the price growth of this cost 
category. This is the same proxy used in the 2016-based IRF market 
basket (84 FR 39081).

[[Page 50978]]

(21) All Other: Nonlabor-Related Services
    We proposed to continue to use the CPI for All Items Less Food and 
Energy (BLS series code CUUR0000SA0L1E) to measure the price growth of 
this cost category. This is the same proxy used in the 2016-based IRF 
market basket (84 FR 39081).
    The following is a summary of the public comments received on our 
proposed price proxies for the operating portion of the 2021-based IRF 
market basket and our responses.
    Comment: A few commenters expressed concern that CMS's use of the 
IHS Global Inc. (IGI) forecast for determining the market basket update 
does not capture the specialized nature of IRF costs. The commenters 
stated that IGI's general forecasts for hospital goods and services 
likely are not accounting for the fact that IRFs are providing more 
specialized services compared to other hospital settings such as 
specialized staff, equipment, and drugs.
    Response: As described previously, the IRF market basket measures 
price changes (including changes in the prices for wages and salaries) 
over time and would not reflect increases in costs associated with 
changes in the volume or intensity of input goods and services until 
the market basket is rebased. In this final rule, we are rebasing and 
revising the current 2016-based IRF market basket to reflect a 2021 
base year. As stated previously, we believe the 2021-based IRF market 
basket appropriately reflects IRF cost structures. To reflect expected 
price growth for each of the cost categories in the IRF market basket, 
we rely on impartial economic forecasts of the price proxies used in 
the market basket from IGI; as previously discussed, we use the best 
available price proxies that would measure expected price growth of the 
goods and services purchased by IRFs. We have consistently used the IGI 
economic price proxy forecasts in the market baskets used to update the 
IRF PPS payments since the implementation of the IRF PPS. For example, 
to measure price growth for IRF wages and salaries costs in the IRF 
market basket, since IRF-specific information is unavailable, we 
proposed to use the ECI for Wages and Salaries for All Civilian workers 
in Hospitals. We believe that this ECI is the best available price 
proxy to account for the occupational skill mix within IRFs. We note 
that we reviewed the Bureau of Labor Statistics Occupational Employment 
and Wage Statistics (OEWS) data for NAICS 622100 (General Medical and 
Surgical Hospitals)--one of the primary data sources used to derive the 
weights for the ECI for Wages and Salaries for All Civilian workers in 
Hospitals--and found that in 2021, the updated base year of the IRF 
market basket, approximately 56 percent of total estimated salaries 
(total employment multiplied by mean annual wage) for NAICS 622100 was 
attributed to Health Professional and Technical occupations, and 
approximately 20 percent was attributed to Health Service occupations. 
Therefore, in the absence of an IRF-specific ECI, we believe that the 
highly skilled hospital workforce captured by the ECI for Wages and 
Salaries for All Civilian workers in Hospitals (inclusive of 
therapists, nurses, other clinicians, etc.) is a reasonable proxy for 
the compensation component of the IRF market basket. We would welcome 
any publicly available IRF-specific data that the commenters could 
provide regarding wage, benefits, or supplies prices.
    Comment: One commenter encouraged CMS to explore other changes to 
the composition of the market basket to better capture evolving 
dynamics in the labor force. The commenter provided as an example that 
the ECI may no longer accurately capture the changing composition and 
cost structure of the hospital labor market given the large increases 
in short-term contract labor use and its growing costs.
    Response: The purpose of the market basket is to measure the 
average change in the price of goods and services hospitals purchase in 
order to provide IRF medical services. We believe the ECI is an 
appropriate index to measure the price changes for Compensation costs 
as it holds occupational distribution constant. We note that the 2021-
based IRF market basket cost weights show that contract labor costs 
account for about 3 percent of total compensation costs (reflecting 
employed and contract labor staff) for IRFs in 2021. In addition, an 
analysis of Medicare cost report data for IPPS hospitals shows that 
contract labor hours accounted for about 4 percent of total 
compensation hours (reflecting employed and contract labor staff) in 
2021. Therefore, while we acknowledge that the ECI measures only 
reflect price changes for employed staff, we believe that the ECI for 
hospital workers is accurately reflecting the price change associated 
with the labor used to provide hospital care (as employed workers' 
hours account for 97 percent of hospital compensation hours). We will 
continue to monitor the trends in the ECI as well as the increased use 
of contract labor as a result of the PHE. We welcome any additional 
publicly available data that commenters can provide regarding 
alternative price indexes.
    After consideration of the public comments, we are finalizing the 
price proxies for the operating portion of the 2021-based IRF market 
basket as proposed.
    Table 11 lists all price proxies that we are finalizing for the 
2021-based IRF market basket.
b. Price Proxies for the Capital Portion of the 2021-Based IRF Market 
Basket
(1) Capital Price Proxies Prior to Vintage Weighting
    We proposed to continue to use the same price proxies for the 
capital-related cost categories in the 2021-based IRF market basket as 
were used in the 2016-based IRF market basket, which are provided in 
Table 11 and described below. Specifically, we proposed to proxy:
     Depreciation: Building and Fixed Equipment cost category 
by BEA's Chained Price Index for Nonresidential Construction for 
Hospitals and Special Care Facilities (BEA Table 5.4.4. Price Indexes 
for Private Fixed Investment in Structures by Type).
     Depreciation: Movable Equipment cost category by the PPI 
for Machinery and Equipment (BLS series code WPU11).
     Nonprofit Interest cost category by the average yield on 
domestic municipal bonds (Bond Buyer 20-bond index).
     For-profit Interest cost category by the iBoxx AAA 
Corporate Bond Yield index
     Other Capital-Related cost category by the CPI-U for Rent 
of Primary Residence (BLS series code CUUS0000SEHA).
    We believe these are the most appropriate proxies for IRF capital-
related costs that meet our selection criteria of relevance, 
timeliness, availability, and reliability. We also proposed to continue 
to vintage weight the capital price proxies for Depreciation and 
Interest to capture the long-term consumption of capital. This vintage 
weighting method is similar to the method used for the 2016-based IRF 
market basket (84 FR 39082) and is described below.
(2) Vintage Weights for Price Proxies
    Because capital is acquired and paid for over time, capital-related 
expenses in any given year are determined by both past and present 
purchases of physical and financial capital. The vintage-weighted 
capital-related portion of the 2021-based IRF market basket is intended 
to capture the long-term

[[Page 50979]]

consumption of capital, using vintage weights for depreciation 
(physical capital) and interest (financial capital). These vintage 
weights reflect the proportion of capital-related purchases 
attributable to each year of the expected life of building and fixed 
equipment, movable equipment, and interest. We proposed to use vintage 
weights to compute vintage-weighted price changes associated with 
depreciation and interest expenses.
    Capital-related costs are inherently complicated and are determined 
by complex capital-related purchasing decisions, over time, based on 
such factors as interest rates and debt financing. In addition, capital 
is depreciated over time instead of being consumed in the same period 
it is purchased. By accounting for the vintage nature of capital, we 
are able to provide an accurate and stable annual measure of price 
changes. Annual non-vintage price changes for capital are unstable due 
to the volatility of interest rate changes, and therefore, do not 
reflect the actual annual price changes for IRF capital-related costs. 
The capital-related component of the 2021-based IRF market basket 
reflects the underlying stability of the capital-related acquisition 
process.
    The methodology used to calculate the vintage weights for the 2021-
based IRF market basket is the same as that used for the 2016-based IRF 
market basket (84 FR 39082 through 39083) with the only difference 
being the inclusion of more recent data. To calculate the vintage 
weights for depreciation and interest expenses, we first need a time 
series of capital-related purchases for building and fixed equipment 
and movable equipment. We found no single source that provides an 
appropriate time series of capital-related purchases by hospitals for 
all of the above components of capital purchases. The early Medicare 
cost reports did not have sufficient capital-related data to meet this 
need. Data we obtained from the American Hospital Association (AHA) do 
not include annual capital-related purchases. However, we are able to 
obtain data on total expenses back to 1963 from the AHA. Consequently, 
we proposed to use data from the AHA Panel Survey and the AHA Annual 
Survey to obtain a time series of total expenses for hospitals. We then 
proposed to use data from the AHA Panel Survey supplemented with the 
ratio of depreciation to total hospital expenses obtained from the 
Medicare cost reports to derive a trend of annual depreciation expenses 
for 1963 through 2020, which is the latest year of AHA data available. 
We proposed to separate these depreciation expenses into annual amounts 
of building and fixed equipment depreciation and movable equipment 
depreciation as determined earlier. From these annual depreciation 
amounts, we derive annual end-of-year book values for building and 
fixed equipment and movable equipment using the expected life for each 
type of asset category. While data is not available that is specific to 
IRFs, we believe this information for all hospitals serves as a 
reasonable alternative for the pattern of depreciation for IRFs.
    To continue to calculate the vintage weights for depreciation and 
interest expenses, we also need to account for the expected lives for 
Building and Fixed Equipment, Movable Equipment, and Interest for the 
2021-based IRF market basket. We proposed to calculate the expected 
lives using Medicare cost report data from Worksheet A-7 part III for 
freestanding and hospital-based IRFs. The expected life of any asset 
can be determined by dividing the value of the asset (excluding fully 
depreciated assets) by its current year depreciation amount. This 
calculation yields the estimated expected life of an asset if the rates 
of depreciation were to continue at current year levels, assuming 
straight-line depreciation. We proposed to determine the expected life 
of building and fixed equipment separately for hospital-based IRFs and 
freestanding IRFs, and then weight these expected lives using the 
percent of total capital costs each provider type represents. We 
proposed to apply a similar method for movable equipment. Using these 
methods, we determined the average expected life of building and fixed 
equipment to be equal to 25 years, and the average expected life of 
movable equipment to be equal to 12 years. For the expected life of 
interest, we believe vintage weights for interest should represent the 
average expected life of building and fixed equipment because, based on 
previous research described in the FY 1997 IPPS final rule (61 FR 
46198), the expected life of hospital debt instruments and the expected 
life of buildings and fixed equipment are similar. We note that for the 
2016-based IRF market basket, the expected life of building and fixed 
equipment is 22 years, and the expected life of movable equipment is 11 
years (84 FR 39082) using the 2016 Medicare cost report data for 
freestanding and hospital-based IRFs.
    Multiplying these expected lives by the annual depreciation amounts 
results in annual year-end asset costs for building and fixed equipment 
and movable equipment. We then calculate a time series, beginning in 
1964, of annual capital purchases by subtracting the previous year's 
asset costs from the current year's asset costs.
    For the building and fixed equipment and movable equipment vintage 
weights, we proposed to use the real annual capital-related purchase 
amounts for each asset type to capture the actual amount of the 
physical acquisition, net of the effect of price inflation. These real 
annual capital-related purchase amounts are produced by deflating the 
nominal annual purchase amount by the associated price proxy as 
provided earlier in the proposed rule. For the interest vintage 
weights, we proposed to use the total nominal annual capital-related 
purchase amounts to capture the value of the debt instrument 
(including, but not limited to, mortgages and bonds). Using these 
capital-related purchase time series specific to each asset type, we 
proposed to calculate the vintage weights for building and fixed 
equipment, for movable equipment, and for interest.
    The vintage weights for each asset type are deemed to represent the 
average purchase pattern of the asset over its expected life (in the 
case of building and fixed equipment and interest, 25 years, and in the 
case of movable equipment, 12 years). For each asset type, we used the 
time series of annual capital-related purchase amounts available from 
2020 back to 1964. These data allow us to derive thirty-three 25-year 
periods of capital-related purchases for building and fixed equipment 
and interest, and 46 12-year periods of capital-related purchases for 
movable equipment. For each 25-year period for building and fixed 
equipment and interest, or 12-year period for movable equipment, we 
calculate annual vintage weights by dividing the capital-related 
purchase amount in any given year by the total amount of purchases over 
the entire 25-year or 12-year period. This calculation is done for each 
year in the 25-year or 12-year period and for each of the periods for 
which we have data. We then calculate the average vintage weight for a 
given year of the expected life by taking the average of these vintage 
weights across the multiple periods of data. The vintage weights for 
the capital-related portion of the 2021-based IRF market basket and the 
2016-based IRF market basket are presented in Table 10.
BILLING CODE 4120-01-P

[[Page 50980]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.060

    The process of creating vintage-weighted price proxies requires 
applying the vintage weights to the price proxy index where the last 
applied vintage weight in Table 10 is applied to the most recent data 
point. We have provided on the CMS website an example of how the 
vintage weighting price proxies are calculated, using example vintage 
weights and example price indices. The example can be found at http://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/MedicareProgramRatesStats/MarketBasketResearch.html in the zip 
file titled ``Weight Calculations as described in the IPPS FY 2010 
Proposed Rule.''
    We did not receive any comments on our proposed price proxies for 
the capital portion of the 2021-based IRF market basket. We are 
finalizing these price proxies as proposed.
c. Summary of Price Proxies of the 2021-Based IRF Market Basket
    Table 11 shows both the operating and capital price proxies that we 
are finalizing for the 2021-based IRF market basket.

[[Page 50981]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.061


[[Page 50982]]


BILLING CODE 4120-01-C
    After consideration of public comments, we are finalizing the 2021-
based IRF market basket as proposed.

D. FY 2024 Market Basket Update and Productivity Adjustment

1. FY 2024 Market Basket Update
    For FY 2024 (that is, beginning October 1, 2023, and ending 
September 30, 2024), we proposed to use an estimate of the 2021-based 
IRF market basket increase percentage to update the IRF PPS base 
payment rate as required by section 1886(j)(3)(C)(i) of the Act. 
Consistent with historical practice, we proposed to estimate the market 
basket update for the IRF PPS based on IHS Global Inc.'s (IGI's) 
forecast using the most recent available data. IGI is a nationally 
recognized economic and financial forecasting firm with which CMS 
contracts to forecast the components of the market baskets.
    Based on IGI's fourth quarter 2022 forecast with historical data 
through the third quarter of 2022, the proposed 2021-based IRF market 
basket percentage increase for FY 2024 was 3.2 percent. Therefore, 
consistent with our historical practice of estimating market basket 
increases based on the best available data, we proposed a market basket 
increase percentage of 3.2 percent for FY 2024. We also proposed that 
if more recent data were subsequently available (for example, a more 
recent estimate of the market basket) we would use such data, if 
appropriate, to determine the FY 2024 update in the final rule.
    Based on IGI's second quarter 2023 forecast with historical data 
through the first quarter of 2023, the 2021-based IRF market basket 
increase percentage for FY 2024 is 3.6 percent. Therefore, consistent 
with our historical practice of estimating market basket increases 
based on the best available data, we are finalizing a market basket 
increase percentage of 3.6 percent for FY 2024. For comparison, the 
current 2016-based IRF market basket is also projected to increase by 
3.6 percent in FY 2024 based on IGI's second quarter 2023 forecast. 
Table 12 compares the 2021-based IRF market basket and the 2016-based 
IRF market basket percent changes. On average, the two indexes produce 
similar updates to one another, with the 4-year average historical 
growth rates (for FY 2019-FY 2022) of the 2021-based IRF market basket 
being equal to 3.2 percent compared to the 2016-based IRF market basket 
with 3.1 percent.
[GRAPHIC] [TIFF OMITTED] TR02AU23.062

2. Productivity Adjustment
    According to section 1886(j)(3)(C)(i) of the Act, the Secretary 
shall establish an increase factor based on an appropriate percentage 
increase in a market basket of goods and services. Section 
1886(j)(3)(C)(ii) of the Act then requires that, after establishing the 
increase factor for a FY, the Secretary shall reduce such increase 
factor for FY 2012 and each subsequent FY, by the productivity 
adjustment described in section 1886(b)(3)(B)(xi)(II) of the Act. 
Section 1886(b)(3)(B)(xi)(II) of the Act sets forth the definition of 
this productivity adjustment. The statute defines the productivity 
adjustment to be equal to the 10-year moving average of changes in 
annual economy-wide, private nonfarm business multifactor productivity 
(as projected by the Secretary for the 10-year period ending with the 
applicable FY, year, cost reporting period, or other annual period) 
(the ``productivity adjustment''). The U.S. Department of Labor's 
Bureau of Labor Statistics (BLS) publishes the official measures of 
productivity for the U.S. economy. We note that previously the 
productivity measure referenced in section 1886(b)(3)(B)(xi)(II) of the 
Act, was published by BLS as private nonfarm business multifactor 
productivity. Beginning with the November 18, 2021 release of 
productivity data, BLS replaced the term multifactor productivity (MFP) 
with total factor productivity (TFP). BLS noted that this is a change 
in terminology only and will not affect the data or methodology. As a 
result of the BLS name change, the productivity measure referenced in 
section 1886(b)(3)(B)(xi)(II) is now published by BLS as private 
nonfarm business total factor productivity. However, as mentioned 
above, the data and methods are unchanged. Please see www.bls.gov for 
the BLS historical published TFP data. A complete description of IGI's 
TFP projection methodology is available on the CMS website at https://www.cms.gov/Research-Statistics-Dataand-Systems/Statistics-Trends-andReports/MedicareProgramRatesStats/MarketBasketResearch. In addition, 
in

[[Page 50983]]

the FY 2022 IRF final rule (86 FR 42374), we noted that effective with 
FY 2022 and forward, CMS changed the name of this adjustment to refer 
to it as the productivity adjustment rather than the MFP adjustment.
    Using IGI's fourth quarter 2022 forecast, the 10-year moving 
average growth of TFP for FY 2024 was projected to be 0.2 percent. 
Thus, in accordance with section 1886(j)(3)(C) of the Act, we proposed 
to calculate the FY 2024 market basket update, which is used to 
determine the applicable percentage increase for the IRF payments, 
using IGI's fourth quarter 2022 forecast of the proposed 2021-based IRF 
market basket. We proposed to then reduce this percentage increase by 
the estimated productivity adjustment for FY 2024 of 0.2 percentage 
point (the 10-year moving average growth of TFP for the period ending 
FY 2024 based on IGI's fourth quarter 2022 forecast). Therefore, the 
proposed FY 2024 IRF update was equal to 3.0 percent (3.2 percent 
market basket update reduced by the 0.2 percentage point productivity 
adjustment). Furthermore, we proposed that if more recent data became 
available after the publication of the proposed rule and before the 
publication of the final rule (for example, a more recent estimate of 
the market basket and/or productivity adjustment), we would use such 
data, if appropriate, to determine the FY 2024 market basket update and 
productivity adjustment in the final rule.
    Using IGI's second quarter 2023 forecast, the 10-year moving 
average growth of TFP for FY 2024 is projected to be 0.2 percent. Thus, 
in accordance with section 1886(j)(3)(C) of the Act, we calculate the 
FY 2024 market basket update, which is used to determine the applicable 
percentage increase for the IRF payments, using IGI's second quarter 
2023 forecast of the 2021-based IRF market basket. We then reduce this 
percentage increase by the estimated productivity adjustment for FY 
2024 of 0.2 percentage point (the 10-year moving average growth of TFP 
for the period ending FY 2024 based on IGI's second quarter 2023 
forecast). Therefore, the FY 2024 IRF update is equal to 3.4 percent 
(3.6 percent market basket update reduced by the 0.2 percentage point 
productivity adjustment).
    For FY 2024, the Medicare Payment Advisory Commission (MedPAC) 
recommends that we reduce IRF PPS payment rates by 3 percent. As 
discussed, and in accordance with sections 1886(j)(3)(C) and 
1886(j)(3)(D) of the Act, the Secretary proposed to update the IRF PPS 
payment rates for FY 2024 by a productivity-adjusted IRF market basket 
increase percentage of 3.0 percent. Section 1886(j)(3)(C) of the Act 
does not provide the Secretary with the authority to apply a different 
update factor to IRF PPS payment rates for FY 2024.
    We invited public comment on our proposals for the FY 2024 market 
basket update and productivity adjustment.
    The following is a summary of the public comments received on the 
proposed FY 2024 market basket update and productivity adjustment:
    Comment: Several commenters supported the proposed payment update 
for FY 2024 and the use of the latest available data. Many commenters 
expressed concern that the FY 2024 payment update does not adequately 
factor in the effects of many challenges faced by IRFs such as the 
impact of the PHE, inflationary pressure, higher patient acuity, 
sequestration, increasing labor costs due to labor shortages, and other 
increased costs such as PPE, drugs, and supplies. One commenter 
expressed concern over the accuracy of the forecast underlying the 
proposed 3.2 percent market basket update for FY 2024.
    A few commenters requested that CMS reexamine the forecasting 
approach or consider other methods and data sources to calculate the 
final rule market basket update that better reflects the rapidly 
increasing input prices and costs facing IRFs. One commenter requested 
that CMS discuss in the final rule how the agency will account for the 
increased costs to hospitals that are not reflected in the recent 
market basket adjustments.
    Response: We acknowledge and appreciate commenters' concerns 
regarding recent trends in inflation. We are required to update IRF PPS 
payments by the market basket update adjusted for productivity, as 
directed by section 1886(j)(3)(C) of the Act. Specifically, section 
1886(j)(3)(C)(i) states that the increase factor shall be based on an 
appropriate percentage increase in a market basket of goods and 
services comprising services for which payment is made. In the FY 2024 
IRF PPS proposed rule, we proposed to rebase and revise the current 
2016-based IRF market basket to reflect a 2021 base year. See section 
VI.C. of this final rule for a description of this proposal, the 
comments received, and the final 2021-based IRF market basket. We 
believe the increase in the 2021-based IRF market basket adequately 
reflects the average change in the price of goods and services 
hospitals purchase in order to provide IRF medical services and is 
technically appropriate to use as the IRF payment update factor. The 
IRF market basket is a fixed-weight, Laspeyres-type index that measures 
the change in price over time of the same mix of goods and services 
purchased by IRFs in the base period. As we discussed in response to 
similar comments in the FY 2023 IRF PPS final rule, the IRF market 
basket update would reflect the prospective price pressures described 
by the commenters as increasing during a high inflation period (such as 
faster wage growth or higher energy prices) but would inherently not 
reflect other factors that might increase the level of costs, such as 
the quantity of labor used or any shifts between contract and staff 
nurses. We note that cost changes (that is, the product of price and 
quantities) would only be reflected when a market basket is rebased, 
and the base year weights are updated to a more recent time period. As 
stated previously, we are finalizing an IRF market basket that reflects 
a 2021 base year and therefore, any change in the cost structure for 
IRFs that occurred between 2016 and 2021 is now captured in the cost 
weights for this rebased market basket.
    In response to the commenter's request that we reexamine the 
current forecasting approach for determining the IRF PPS market basket 
update, we provide the following information. As stated previously, IGI 
is a nationally recognized economic and financial forecasting firm with 
which CMS contracts to forecast the components of the market baskets. 
At the time of the FY 2024 IRF PPS proposed rule, based on IGI's fourth 
quarter 2022 forecast with historical data through the third quarter of 
2022, the 2021-based IRF market basket update was forecasted to be 3.2 
percent for FY 2024, reflecting forecasted compensation price growth of 
3.9 percent (by comparison, compensation price growth in the IRF market 
basket averaged 2.4 percent from 2013-2022). In the FY 2024 IRF PPS 
proposed rule, we proposed that if more recent data became available, 
we would use such data, if appropriate, to derive the final FY 2024 IRF 
market basket update for the final rule. For this final rule, we now 
have an updated forecast of the price proxies underlying the market 
basket that incorporates more recent historical data and reflects a 
revised outlook regarding the U.S. economy and expected price inflation 
for FY 2024. Based on IGI's second quarter 2023 forecast with 
historical data through the first quarter of 2023, we are projecting a 
FY 2024 IRF market basket update of 3.6 percent (reflecting forecasted 
compensation price growth of 4.3 percent) and a productivity

[[Page 50984]]

adjustment of 0.2 percentage point. Therefore, for FY 2024 a final IRF 
productivity-adjusted market basket update of 3.4 percent (3.6 percent 
less 0.2 percentage point) will be applicable, compared to the 3.0 
percent market basket update that was proposed.
    We do acknowledge that FY 2022 compensation price growth for the 
2016-based IRF market basket was higher (5.3 percent) than was 
forecasted at the time of the FY 2022 IRF PPS final rule (2.7 percent). 
We note that the lower projected FY 2024 IRF market basket percent 
increase relative to the FY 2022 historical increase and the FY 2023 
projected increase reflects the expectation that wage and price 
pressures will lessen in FY 2024 relative to recent history.
    Comment: Several commenters expressed concern about the continued 
application of the productivity adjustment to IRFs. The commenters 
noted that the PHE has resulted in further productivity challenges for 
IRFs and other healthcare providers. One commenter cited an article and 
data reporting declines in overall productivity in the economy and 
requested that CMS consider these developments in the update to the 
productivity adjustment in the IRF PPS final rule. A few commenters 
requested that CMS carefully monitor the impact that these productivity 
adjustments will have on the rehabilitation hospital sector, provide 
feedback to Congress as appropriate, and reduce the productivity 
adjustment. One commenter requested that CMS explore ways to use its 
authority to offset or waive these adjustments. One commenter requested 
that CMS suspend at least temporarily the productivity adjustment that 
reduces the market basket update due to recent declines in hospital 
productivity. One commenter requested that CMS use its exceptions and 
adjustments authority under section 1886(j)(3)(A)(v) of the Act to 
remove the productivity adjustment for any fiscal year that was covered 
under PHE determination, that is, 2020 (0.4 percent), 2021 (0.0 
percent), 2022 (0.7 percent), and 2023 (0.3 percent), from the 
calculation of the market basket for FY 2024 and any year thereafter.
    Response: Section 1886(j)(3)(C)(ii)(I) of the Act requires the 
application of the productivity adjustment, described in section 
1886(b)(3)(xi)(II), to the IRF PPS market basket increase factor. As 
required by statute, the FY 2024 productivity adjustment is derived 
based on the 10-year moving average growth in economy-wide productivity 
for the period ending FY 2024. We recognize the concerns of the 
commenters regarding the appropriateness of the productivity 
adjustment; however, we are required pursuant to section 
1886(j)(3)(C)(ii)(I) to apply the specific productivity adjustment 
described here. In addition, with respect to providing feedback to 
Congress, we note that MedPAC annually monitors various factors for 
Medicare providers in terms of profitability and beneficiary access to 
care and reports the findings to Congress on an annual basis. MedPAC 
did a full analysis of payment adequacy for IRF providers in its March 
2023 Report to Congress (https://www.medpac.gov/document/march-2023-report-to-the-congress-medicare-payment-policy/). MedPAC stated that 
given the positive payment adequacy indicators for IRFs, they 
recommended that the IRF base payment rate be reduced by 3 percent for 
FY 2024. Additionally, we note that we did not propose to use our 
authority under section 1886(d)(5)(I)(i) of the Act to remove or offset 
the application of the productivity adjustment for FY 2024. As 
previously noted, we are required pursuant to section 
1886(j)(3)(C)(ii)(I) of the Act to apply the productivity adjustment to 
the IRF PPS market basket increase factor.
    Comment: A number of commenters requested that CMS deviate from its 
usual update and consider making one-time adjustments to the market 
basket update or applying a forecast error adjustment. One commenter 
stated CMS should apply a temporary payment adjustment or add-on 
payment to the IRF PPS in FY 2024 of 10 to 20 percent per discharge. 
Another commenter requested an adjustment to account for what the 
commenter described as CMS' ``underpayment'' of IRFs since 2020.
    Response: As most recently discussed in the FY 2023 IRF PPS final 
rule, the IRF PPS market basket updates are set prospectively, which 
means that the market basket update relies on a mix of both historical 
data for part of the period for which the update is calculated and 
forecasted data for the remainder. For instance, the FY 2024 market 
basket update in this final rule reflects historical data through the 
first quarter of CY 2023 and forecasted data through the third quarter 
of CY 2024. While there is currently no mechanism to adjust for market 
basket forecast error in the IRF payment update, the forecast error for 
a market basket update is calculated as the actual market basket 
increase for a given year less the forecasted market basket increase. 
Due to the uncertainty regarding future price trends, forecast errors 
can be both positive and negative. In evaluating the difference between 
the forecast increase and later acquired actual data for the period 
from FY 2012 through FY 2020, we found the forecasted market basket 
updates for each payment year for IRFs were higher than the actual 
market basket updates. Therefore, we disagree with the suggestion that 
the FY 2024 base rates are too low based solely on the calculation of a 
forecast error over a short period of time (instead of considering 
forecast errors over longer periods). For this final rule, we have 
incorporated more recent historical data and forecasts to capture the 
price and wage pressures facing IRFs and believe it is the best 
available projection of inflation to determine the applicable 
percentage increase for the IRF payments in FY 2024.
    After consideration of public comments, we are finalizing a FY 2024 
IRF productivity-adjusted market basket increase of 3.4 percent based 
on the most recent data available.

E. Labor-Related Share for FY 2024

    Section 1886(j)(6) of the Act specifies that the Secretary is to 
adjust the proportion (as estimated by the Secretary from time to time) 
of inpatient rehabilitation facilities' costs that are attributable to 
wages and wage-related costs, of the prospective payment rates computed 
under section 1886(j)(3) of the Act for area differences in wage levels 
by a factor (established by the Secretary) reflecting the relative 
hospital wage level in the geographic area of the rehabilitation 
facility compared to the national average wage level for such 
facilities. The labor-related share is determined by identifying the 
national average proportion of total costs that are related to, 
influenced by, or vary with the local labor market. We proposed to 
continue to classify a cost category as labor-related if the costs are 
labor-intensive and vary with the local labor market. As stated in the 
FY 2020 IRF PPS final rule (84 FR 39087), the labor-related share was 
defined as the sum of the relative importance of Wages and Salaries, 
Employee Benefits, Professional Fees: Labor-Related Services, 
Administrative and Facilities Support Services, Installation, 
Maintenance, and Repair Services, All Other: Labor-Related Services, 
and a portion of the Capital-Related Costs from the 2016-based IRF 
market basket.
    Based on our definition of the labor-related share and the cost 
categories in the 2021-based IRF market basket, we proposed to include 
in the labor-related share for FY 2024 the sum of the FY 2024 relative 
importance of Wages and Salaries, Employee Benefits, Professional Fees: 
Labor-Related,

[[Page 50985]]

Administrative and Facilities Support Services, Installation, 
Maintenance, and Repair Services, All Other: Labor-Related Services, 
and a portion of the Capital-Related cost weight from the 2021-based 
IRF market basket.
    Similar to the 2016-based IRF market basket (84 FR 39087), the 
2021-based IRF market basket includes two cost categories for 
nonmedical Professional Fees (including, but not limited to, expenses 
for legal, accounting, and engineering services). These are 
Professional Fees: Labor-Related and Professional Fees: Nonlabor-
Related. For the 2021-based IRF market basket, we proposed to estimate 
the labor-related percentage of non-medical professional fees (and 
assign these expenses to the Professional Fees: Labor-Related services 
cost category) based on the same method that was used to determine the 
labor-related percentage of professional fees in the 2016-based IRF 
market basket.
    As was done in the 2016-based IRF market basket (84 FR 39087), we 
proposed to determine the proportion of legal, accounting and auditing, 
engineering, and management consulting services that meet our 
definition of labor-related services based on a survey of hospitals 
conducted by us in 2008, a discussion of which can be found in the FY 
2010 IPPS/LTCH PPS final rule (74 FR 43850 through 43856). Based on the 
weighted results of the survey, we determined that hospitals purchase, 
on average, the following portions of contracted professional services 
outside of their local labor market:
     34 percent of accounting and auditing services.
     30 percent of engineering services.
     33 percent of legal services.
     42 percent of management consulting services.
    We proposed to apply each of these percentages to the respective 
Benchmark I-O cost category underlying the professional fees cost 
category to determine the Professional Fees: Nonlabor-Related costs. 
The Professional Fees: Labor-Related costs were determined to be the 
difference between the total costs for each Benchmark I-O category and 
the Professional Fees: Nonlabor-Related costs. This is the same 
methodology that we used to separate the 2016-based IRF market basket 
professional fees category into Professional Fees: Labor-Related and 
Professional Fees: Nonlabor-Related cost categories (84 FR 39087).
    Effective for transmittal 18 (https://www.cms.gov/Regulations-and-Guidance/Guidance/Transmittals/Transmittals/r18p240i), the hospital 
Medicare Cost Report (CMS Form 2552-10, OMB No. 0938-0050) is 
collecting information on whether a hospital purchased professional 
services (for example, legal, accounting, tax preparation, bookkeeping, 
payroll, advertising, and/or management/consulting services) from an 
unrelated organization and if the majority of these expenses were 
purchased from unrelated organizations located outside of the main 
hospital's local area labor market. We encourage all providers to 
provide this information so we can potentially use in future rulemaking 
to determine the labor-related share.
    In the 2021-based IRF market basket, nonmedical professional fees 
that are subject to allocation based on these survey results represent 
4.0 percent of total costs (and are limited to those fees related to 
Accounting & Auditing, Legal, Engineering, and Management Consulting 
services). Based on our survey results, we proposed to apportion 
approximately 2.6 percentage points of the 4.0 percentage point figure 
into the Professional Fees: Labor-Related share cost category and the 
remaining 1.4 percentage point into the Professional Fees: Nonlabor-
Related cost category.
    In addition to the professional services listed, for the 2021-based 
IRF market basket, we proposed to allocate a proportion of the Home 
Office/Related Organization Contract Labor cost weight, calculated 
using the Medicare cost reports as stated previously in this final 
rule, into the Professional Fees: Labor-Related and Professional Fees: 
Nonlabor-Related cost categories. We proposed to classify these 
expenses as labor-related and nonlabor-related as many facilities are 
not located in the same geographic area as their home office, and 
therefore, do not meet our definition for the labor-related share, 
which requires the services to be purchased in the local labor market.
    Similar to the 2016-based IRF market basket, we proposed for the 
2021-based IRF market basket to use the Medicare cost reports for both 
freestanding IRF providers and hospital-based IRF providers to 
determine the home office labor-related percentages. The Medicare cost 
report requires a hospital to report information regarding its home 
office provider. For the 2021-based IRF market basket, we proposed to 
start with the sample of IRF providers that passed the top 1 percent 
trim used to derive the Home Office/Related Organization Contract Labor 
cost weight as described in section V.C.1.b. of the proposed rule. 
Using information on the Medicare cost report, for freestanding and 
hospital-based providers separately, we first compare the location of 
the IRF with the location of the IRF's home office and classify an IRF 
based on whether its home office is located in the hospital facility's 
same Metropolitan Statistical Area. For both freestanding and hospital-
based providers, we proposed to multiply each provider's Home Office/
Related Organization Contract Labor cost weight (calculated using data 
from the total facility) by Medicare allowable total costs. We then 
calculate the proportion of Medicare allowable home office compensation 
costs that these IRFs represent of total Medicare allowable home office 
compensation costs. We proposed to multiply this percentage (45 
percent) by the Home Office/Related Organization Contract Labor cost 
weight (5.4 percent) to determine the proportion of costs that should 
be allocated to the labor-related share. Therefore, we proposed to 
allocate 2.4 percentage points of the Home Office/Related Organization 
Contract Labor cost weight (5.4 percent times 45 percent) to the 
Professional Fees: Labor-Related cost weight and 3.0 percentage points 
of the Home Office/Related Organization Contract Labor cost weight to 
the Professional Fees: Nonlabor-Related cost weight (5.4 percent times 
55 percent). For the 2016-based IRF market basket, we used a similar 
methodology (84 FR 39088) and determined that 42 percent of the 2016-
based Home Office/Related Organization Contract Labor cost weight 
should be allocated to the labor-related share.
    In summary, we apportioned 2.6 percentage points of the non-medical 
professional fees and 2.4 percentage points of the Home Office/Related 
Organization Contract Labor cost weight into the Professional Fees: 
Labor-Related cost category. This amount was added to the portion of 
professional fees that was identified to be labor-Related using the I-O 
data such as contracted advertising and marketing costs (approximately 
0.6 percentage point of total costs) resulting in a Professional Fees: 
Labor-Related cost weight of 5.6 percent.
    Comment: A few commenters supported the proposal to increase the 
labor-related share using data that better reflects increased labor 
costs as a percentage of IRFs' overall cost structure.
    One commenter disagreed with CMS' proposal to exclude from the 
labor-related share the proportion of non-medical professional services 
fees presumed to have been purchased outside of the hospital's labor 
market. The commenter disagreed with CMS' assumption that services 
purchased

[[Page 50986]]

from national firms are not affected by the local labor market. The 
commenter stated that when hospitals seek professional services, the 
services they are seeking (for example accounting, engineering, 
management consulting) typically are not so unique that they could only 
be provided by regional or national firms. The commenter stated that 
CMS' own survey data support this conclusion, as approximately 65 
percent of these services are sourced from firms in the local market. 
The commenter stated that costs of services purchased from firms 
outside the hospital's labor market should be included with the labor-
related share of costs.
    The commenter requested that CMS provide evidence that pricing for 
professional services provided by regional and national firms to 
hospitals is offered in a national market that is not subject to 
geographic cost variation. The commenter requested that CMS restore the 
1.4 percentage points it proposes to reclassify to Professional 
Services: Nonlabor-Related to the Professional Services: Labor-Related 
category, if the agency cannot produce strong evidence that prices for 
professional services provided by firms outside of a hospital's local 
labor market are homogenous.
    Response: We disagree with the commenter and believe it is 
appropriate that a proportion of Accounting & Auditing, Legal, 
Engineering, and Management Consulting services costs purchased by 
hospitals should be excluded from the labor-related share. Section 
1886(j)(6) of the Act specifies that the Secretary is to adjust the 
proportion (as estimated by the Secretary from time to time) of IRFs' 
costs that are attributable to wages and wage-related costs, of the 
prospective payment rates computed under section 1886(j)(3) of the Act 
for area differences in wage levels by a factor (established by the 
Secretary) reflecting the relative hospital wage level in the 
geographic area of the rehabilitation facility compared to the national 
average wage level for such facilities.
    The purpose of the labor-related share is to reflect the proportion 
of the national PPS base payment rate that is adjusted by the 
hospital's wage index (representing the relative costs of their local 
labor market to the national average). Therefore, we include a cost 
category in the labor-related share if the costs are labor intensive 
and vary with the local labor market.
    As acknowledged by the commenter and confirmed by the survey of 
hospitals conducted by CMS in 2008 (as stated previously in this final 
rule), professional services can be purchased from local firms as well 
as national and regional professional services firms. It is not 
necessarily the case, as asserted by the commenter, that these national 
and regional firms have fees that match those in the local labor market 
even though providers have the option to utilize those firms. That is, 
fees for services purchased from firms outside the local labor market 
may differ from those that would be purchased in the local labor market 
for any number of reasons (including but not limited to, the skill 
level of the contracted personnel, higher capital costs, etc.). As 
noted earlier in this section of this final rule, the definition for 
the labor-related share requires the services to be purchased in the 
local labor market; therefore, CMS' allocation of approximately 65 
percent (2.6 percentage points of 4.0 percentage points) of the 
Professional Fees cost weight to Professional Fees: Labor-Related costs 
based on the 2008 survey results \17\ is consistent with the 
commenter's assertion that not all Professional Fees services are 
purchased in the local labor market. We believe it is reasonable to 
conclude that the costs of those Professional Fees services purchased 
directly within the local labor market are directly related to local 
labor market conditions and, thus, should be included in the labor-
related share. The remaining approximately 35 percent of Professional 
Fees costs, which are purchased outside the local labor market, reflect 
different and additional factors outside the local labor market and, 
thus, should be excluded from the labor-related share. In addition, we 
note the compensation costs of professional services provided by 
hospital employees (which would reflect the local labor market) are 
included in the labor-related share as they are included in the Wages 
and Salaries and Employee Benefits cost weights.
---------------------------------------------------------------------------

    \17\ The 65 percent is based on a survey conducted by CMS in 
2008 as detailed in the FY 2010 IPPS/LTCH PPS final rule (74 FR 
43850 through 43856). This was also used to determine the 
Professional Fees: Labor-related cost weight in the 2016-based IRF 
market basket.
---------------------------------------------------------------------------

    Therefore, for the reasons discussed, we believe our proposed 
methodology of continuing to allocate only a portion of Professional 
Fees to the Professional Fees: Labor-Related cost category is 
appropriate. As stated previously, effective for transmittal 18 
(https://www.cms.gov/Regulations-and-Guidance/Guidance/Transmittals/Transmittals/r18p240i), the hospital Medicare Cost Report (CMS Form 
2552-10, OMB No. 0938-0050) is collecting information on whether a 
hospital purchased professional services (for example, legal, 
accounting, tax preparation, bookkeeping, payroll, advertising, and/or 
management/consulting services) from an unrelated organization and if 
the majority of these expenses were purchased from unrelated 
organizations located outside of the main hospital's local area labor 
market. We encourage all providers to provide this information so we 
can potentially use in future rulemaking to determine the labor-related 
share.
    Comment: One commenter disagreed with the assumption that home 
office compensation costs that occur outside of a hospital's labor 
market are not subject to geographic wage variation and stated that 
they do not believe that the proposed reclassification to the 
Professional Fees: Non-Labor-Related cost category is justified. The 
commenters stated that the proposed methodology fails to consider that 
the home office is essentially a part of the hospital, and thus the 
hospital, along with its home office, is operating in multiple labor 
markets. The commenters stated that the home office's portion of the 
hospital's labor costs should not be excluded from the labor-related 
share simply because they are not in the same labor market as the 
hospital.
    The commenter conducted their own analysis of the Medicare cost 
report data showing that providers with a home office outside of their 
local labor market had a wage index both below 1 as well as greater 
than 1. The commenter stated that those hospitals in a labor market 
with a wage index greater than 1 had mean home office average hourly 
wage costs that were greater than the mean home office average hourly 
wage costs of those hospitals in a labor market with a wage index less 
than 1. The commenter claimed that these data indicate that, contrary 
to CMS' assertion, home office salary, wage, and benefit costs for 
hospitals with home offices outside of their labor market are subject 
to geographic wage variation.
    The commenter requested that CMS allocate the full 5.4 percentage 
points of the Home Office/Related Organization cost weight to the 
labor-related share.
    Response: As previously stated, the purpose of the labor-related 
share is to determine the proportion of the national PPS base payment 
rate that is adjusted by the hospital's wage index (representing the 
relative costs of their local labor market to the national average). 
Therefore, we include a cost category in the labor-related share if the 
costs are labor intensive and vary with the local labor market.
    As the commenter stated and as validated with the Medicare cost 
report, a hospital's home office can be located

[[Page 50987]]

outside the hospital's local labor market. The proposed methodology for 
allocating 45 percent of the Home Office/Related Organization cost 
weight (reflecting compensation costs) is consistent with the intent of 
the statute to identify the proportion of costs likely to directly vary 
with the hospital's local labor market. Our methodology relies on the 
Medicare cost report data for hospitals reporting home office 
information to determine whether their home office is located in the 
same local labor market (which we define as the hospital's Metropolitan 
Statistical Area). As with professional services, we believe it is 
reasonable to conclude that costs of those home office services 
purchased directly within the local labor market are directly related 
to local labor market conditions while the remaining 55 percent of home 
office costs which are purchased outside the local labor market would 
reflect different and additional factors and, thus, should be excluded 
from the labor-related share.
    Therefore, we believe our proposed methodology of continuing to 
allocate only a portion of the Home Office/Related Organization cost 
weight into the Professional Fees: Labor-Related cost weight is 
appropriate. In addition, we would note that the compensation costs for 
hospital employees (which would reflect the local labor market) 
performing the same tasks as home office personnel are included in the 
labor-related share as they are included in the Wages and Salaries and 
Employee Benefits cost weights.
    As stated previously, we proposed to include in the labor-related 
share the sum of the relative importance of Wages and Salaries, 
Employee Benefits, Professional Fees: Labor-Related, Administrative and 
Facilities Support Services, Installation, Maintenance, and Repair 
Services, All Other: Labor-Related Services, and a portion of the 
Capital-Related cost weight from the 2021-based IRF market basket. The 
relative importance reflects the different rates of price change for 
these cost categories between the base year (2021) and FY 2024. Based 
on IGI's fourth quarter 2022 forecast for the proposed 2021-based IRF 
market basket, the sum of the FY 2024 relative importance for Wages and 
Salaries, Employee Benefits, Professional Fees: Labor-related, 
Administrative and Facilities Support Services, Installation 
Maintenance & Repair Services, and All Other: Labor-Related Services is 
70.3 percent. The portion of Capital-Related costs that is influenced 
by the local labor market is estimated to be 46 percent, which is the 
same percentage applied to the 2016-based IRF market basket (84 FR 
39088 through 39089). Since the relative importance of Capital-Related 
costs is 8.2 percent of the proposed 2021-based IRF market basket in FY 
2024, we took 46 percent of 8.2 percent to determine the proposed 
labor-related share of Capital-Related costs for FY 2024 of 3.8 
percent. Therefore, we proposed a total labor-related share for FY 2024 
of 74.1 percent (the sum of 70.3 percent for the operating costs and 
3.8 percent for the labor-related share of Capital-Related costs).
    After consideration of public comments, we are finalizing the 2021-
based IRF market basket labor-related cost categories and base year 
cost weights as proposed.
    Based on IGI's second quarter 2023 forecast for the 2021-based IRF 
market basket, the sum of the FY 2024 relative importance for Wages and 
Salaries, Employee Benefits, Professional Fees: Labor-related, 
Administrative and Facilities Support Services, Installation 
Maintenance & Repair Services, and All Other: Labor-Related Services is 
70.3 percent. The portion of Capital-Related costs that is influenced 
by the local labor market is estimated to be 46 percent, which is the 
same percentage applied to the 2016-based IRF market basket (84 FR 
39088 through 39089). Since the relative importance for Capital is 8.2 
percent of the 2021-based IRF market basket in FY 2024, we took 46 
percent of 8.2 percent to determine the labor-related share of Capital-
Related costs for FY 2024 of 3.8 percent. Therefore, the total labor-
related share for FY 2024 based on more recent data is 74.1 percent 
(the sum of 70.3 percent for the operating costs and 3.8 percent for 
the labor-related share of Capital-Related costs).
    Table 13 shows the FY 2024 labor-related share using the 2021-based 
IRF market basket relative importance and the FY 2023 labor-related 
share using the 2016-based IRF market basket relative importance.
[GRAPHIC] [TIFF OMITTED] TR02AU23.063


[[Page 50988]]


    The FY 2024 labor-related share using the 2021-based IRF market 
basket is 1.2 percentage point higher than the FY 2023 labor-related 
share using the 2016-based IRF market basket. This higher labor-related 
share is primarily due to the incorporation of the 2021 Medicare cost 
report data, which increased the Compensation cost weight by 
approximately 0.8 percentage point compared to the 2016-based IRF 
market basket as shown in Tables 4 and 5.

F. Wage Adjustment for FY 2024

1. Background
    Section 1886(j)(6) of the Act requires the Secretary to adjust the 
proportion of rehabilitation facilities' costs attributable to wages 
and wage-related costs (as estimated by the Secretary from time to 
time) by a factor (established by the Secretary) reflecting the 
relative hospital wage level in the geographic area of the 
rehabilitation facility compared to the national average wage level for 
those facilities. The Secretary is required to update the IRF PPS wage 
index on the basis of information available to the Secretary on the 
wages and wage-related costs to furnish rehabilitation services. Any 
adjustment or updates made under section 1886(j)(6) of the Act for a FY 
are made in a budget-neutral manner.
    In the FY 2023 IRF PPS final rule (87 FR 47054 through 47056) we 
finalized a policy to apply a 5-percent cap on any decrease to a 
provider's wage index from its wage index in the prior year, regardless 
of the circumstances causing the decline. Additionally, we finalized a 
policy that a new IRF would be paid the wage index for the area in 
which it is geographically located for its first full or partial FY 
with no cap applied because a new IRF would not have a wage index in 
the prior FY. Also, in the FY 2023 IRF PPS final rule, we amended the 
regulations at Sec.  412.624(e)(1)(ii) to reflect this permanent cap on 
wage index decreases. A full discussion of the adoption of this policy 
is found in the FY 2023 IRF PPS final rule.
    For FY 2024, we proposed to maintain the policies and methodologies 
described in the FY 2023 IRF PPS final rule (87 FR 47038) related to 
the labor market area definitions and the wage index methodology for 
areas with wage data. Thus, we proposed to use the core based 
statistical areas (CBSAs) labor market area definitions and the FY 2024 
pre-reclassification and pre-floor hospital wage index data. In 
accordance with section 1886(d)(3)(E) of the Act, the FY 2024 pre-
reclassification and pre-floor hospital wage index is based on data 
submitted for hospital cost reporting periods beginning on or after 
October 1, 2019, and before October 1, 2020 (that is, FY 2020 cost 
report data).
    The labor market designations made by the OMB include some 
geographic areas where there are no hospitals and, thus, no hospital 
wage index data on which to base the calculation of the IRF PPS wage 
index. We proposed to continue to use the same methodology discussed in 
the FY 2008 IRF PPS final rule (72 FR 44299) to address those 
geographic areas where there are no hospitals and, thus, no hospital 
wage index data on which to base the calculation for the FY 2024 IRF 
PPS wage index.
    We invited public comment on our proposals regarding the Wage 
Adjustment for FY 2024.
    The following is a summary of the public comments received on the 
proposals regarding the Wage Adjustment for FY 2024, with our 
responses:
    Comment: Commenters stated support of the permanent 5-percent cap 
on wage index decreases. One commenter encouraged CMS to implement 
these caps in a non-budget neutral manner to mitigate volatility caused 
by wage index shifts.
    Response: We appreciate the commenters' support of the permanent 
cap on wage index decreases. As for budget neutrality, we do not 
believe that the permanent 5-percent cap policy for the IRF wage index 
should be applied in a non-budget-neutral manner. Any adjustment or 
updates made under section 1886(j)(6) of the Act for a FY must be made 
in a manner that assures that the aggregated payments under this 
subsection in the FY are not greater or less than those that would have 
been made in the year without such adjustments. In accordance with 
section 1186(j)(6) of the Act, our longstanding historical practice has 
been to implement updates to the wage index under the IRF PPS in a 
budget neutral manner. We refer readers to the FY 2023 IRF PPS final 
rule (87 FR 47054 through 47056) for a detailed discussion and for 
responses to these and other comments relating to the wage index cap 
policy.
    Comment: One commenter encouraged CMS to release provider-level 
wage index tables in the final rule that would indicate what wage index 
value each IRF would receive, including whether or not the IRF would 
receive a capped wage index value, in order to avoid errors in the 
payment rates established by the MACs. Commenters also requested that 
CMS release the necessary wage index tables and data to enable IRFs to 
crosswalk the IPPS values after application of the low-wage index 
adjustment to the IRF PPS wage indices. These commenters also requested 
that CMS detail what data it believes is necessary to enable use of the 
post-reclassification and post-floor IPPS wage index data in the IRF 
PPS.
    Response: The wage index tables for IRF PPS are provided at the 
CBSA level. The 5-percent cap policy is applied at the provider level. 
Hence, when the 5-percent cap is applicable, each IRF should work 
directly with its MAC to understand how the 5-percent cap is applied. 
MACs have more detailed information about the location of each IRF and 
the applicability of the 5-percent cap to each IRF's situation, and CMS 
has provided careful instructions to the MACs on applying the 5-percent 
cap policy (see publication 100-04 Medicare Claims Processing Manual, 
chapter 3). Further, we are unable to provide crosswalk tables or data 
related to IPPS wage index policies. Data pertaining to the FY 2024 
IPPS proposed rule is available at https://www.cms.gov/medicare/medicare-fee-for-service-payment/acuteinpatientpps. We do not have any 
additional data on this for the IRF PPS.
    Comment: Commenters encouraged CMS to continue to reform the wage 
index policies. Commenters suggested that CMS revise the IRF wage index 
to adopt the IPPS policies such as geographic reclassification, rural 
floor, low wage adjustment, and the Outpatient PPS (OPPS) outmigration 
adjustments.
    Response: We appreciate the commenters' suggestion to adopt the 
IPPS reclassification and rural floor policies, low wage, and the OPPS 
outmigration adjustments for the IRF wage index. The OPPS outmigration 
adjustment policy is a longstanding policy for that setting, and it 
should be noted that the wage index applied to the OPPS also includes 
the rural floor and any policies and adjustments applied to the IPPS 
wage index. As we do not have an IRF-specific wage index, we are unable 
to determine the degree, if any, to which these IPPS/OPPS policies 
under the IRF PPS would be appropriate. Data pertaining to any IPPS 
policies that are applied to the pre-reclassification/pre-floor wage 
index is available in the FY 2024 IPPS proposed rule at https://www.cms.gov/medicare/medicare-fee-for-service-payment/acuteinpatientpps. The rationale for our current wage index policies 
was most recently published in the FY 2022 IRF PPS final rule (86 FR 
42377 through 42378) and fully described in the FY 2006 IRF PPS final 
rule (70 FR 47880, 47926 through 47928).

[[Page 50989]]

    After consideration of the comments we received, we are finalizing 
our proposals regarding the Wage Adjustment for FY 2024.
2. Core-Based Statistical Areas (CBSAs) for the FY 2024 IRF Wage Index
    The wage index used for the IRF PPS is calculated using the pre-
reclassification and pre-floor inpatient PPS (IPPS) wage index data and 
is assigned to the IRF on the basis of the labor market area in which 
the IRF is geographically located. IRF labor market areas are 
delineated based on the CBSAs established by the OMB. The CBSA 
delineations (which were implemented for the IRF PPS beginning with FY 
2016) are based on revised OMB delineations issued on February 28, 
2013, in OMB Bulletin No. 13-01. OMB Bulletin No. 13-01 established 
revised delineations for Metropolitan Statistical Areas, Micropolitan 
Statistical Areas, and Combined Statistical Areas in the United States 
and Puerto Rico based on the 2010 Census, and provided guidance on the 
use of the delineations of these statistical areas using standards 
published in the June 28, 2010 Federal Register (75 FR 37246 through 
37252). We refer readers to the FY 2016 IRF PPS final rule (80 FR 47068 
through 47076) for a full discussion of our implementation of the OMB 
labor market area delineations beginning with the FY 2016 wage index.
    Generally, OMB issues major revisions to statistical areas every 10 
years, based on the results of the decennial census. Additionally, OMB 
occasionally issues updates and revisions to the statistical areas in 
between decennial censuses to reflect the recognition of new areas or 
the addition of counties to existing areas. In some instances, these 
updates merge formerly separate areas, transfer components of an area 
from one area to another, or drop components from an area. On July 15, 
2015, OMB issued OMB Bulletin No. 15-01, which provides minor updates 
to and supersedes OMB Bulletin No. 13-01 that was issued on February 
28, 2013. The attachment to OMB Bulletin No. 15-01 provides detailed 
information on the update to statistical areas since February 28, 2013. 
The updates provided in OMB Bulletin No. 15-01 are based on the 
application of the 2010 Standards for Delineating Metropolitan and 
Micropolitan Statistical Areas to Census Bureau population estimates 
for July 1, 2012 and July 1, 2013.
    In the FY 2018 IRF PPS final rule (82 FR 36250 through 36251), we 
adopted the updates set forth in OMB Bulletin No. 15-01 effective 
October 1, 2017, beginning with the FY 2018 IRF wage index. For a 
complete discussion of the adoption of the updates set forth in OMB 
Bulletin No. 15-01, we refer readers to the FY 2018 IRF PPS final rule. 
In the FY 2019 IRF PPS final rule (83 FR 38527), we continued to use 
the OMB delineations that were adopted beginning with FY 2016 to 
calculate the area wage indexes, with updates set forth in OMB Bulletin 
No. 15-01 that we adopted beginning with the FY 2018 wage index.
    On August 15, 2017, OMB issued OMB Bulletin No. 17-01, which 
provided updates to and superseded OMB Bulletin No. 15-01 that was 
issued on July 15, 2015. The attachments to OMB Bulletin No. 17-01 
provide detailed information on the update to statistical areas since 
July 15, 2015, and are based on the application of the 2010 Standards 
for Delineating Metropolitan and Micropolitan Statistical Areas to 
Census Bureau population estimates for July 1, 2014 and July 1, 2015. 
In the FY 2020 IRF PPS final rule (84 FR 39090 through 39091), we 
adopted the updates set forth in OMB Bulletin No. 17-01 effective 
October 1, 2019, beginning with the FY 2020 IRF wage index.
    On April 10, 2018, OMB issued OMB Bulletin No. 18-03, which 
superseded the August 15, 2017 OMB Bulletin No. 17-01, and on September 
14, 2018, OMB issued OMB Bulletin No. 18-04, which superseded the April 
10, 2018 OMB Bulletin No. 18-03. These bulletins established revised 
delineations for Metropolitan Statistical Areas, Micropolitan 
Statistical Areas, and Combined Statistical Areas, and provided 
guidance on the use of the delineations of these statistical areas. A 
copy of this bulletin may be obtained at https://www.whitehouse.gov/wp-content/uploads/2018/09/Bulletin-18-04.pdf.
    To this end, as discussed in the FY 2021 IRF PPS proposed (85 FR 
22075 through 22079) and final (85 FR 48434 through 48440) rules, we 
adopted the revised OMB delineations identified in OMB Bulletin No. 18-
04 (available at https://www.whitehouse.gov/wp-content/uploads/2018/09/Bulletin-18-04.pdf) beginning October 1, 2020, including a 1-year 
transition for FY 2021 under which we applied a 5-percent cap on any 
decrease in an IRF's wage index compared to its wage index for the 
prior fiscal year (FY 2020). The updated OMB delineations more 
accurately reflect the contemporary urban and rural nature of areas 
across the country, and the use of such delineations allows us to 
determine more accurately the appropriate wage index and rate tables to 
apply under the IRF PPS. OMB issued further revised CBSA delineations 
in OMB Bulletin No. 20-01, on March 6, 2020 (available on the web at 
https://www.whitehouse.gov/wp-content/uploads/2020/03/Bulletin-20-01.pdf). However, we determined that the changes in OMB Bulletin No. 
20-01 do not impact the CBSA-based labor market area delineations 
adopted in FY 2021. Therefore, CMS did not propose to adopt the revised 
OMB delineations identified in OMB Bulletin No. 20-01 for FY 2022 or 
2023, and for these reasons CMS is likewise not making such a proposal 
for FY 2024.
3. IRF Budget-Neutral Wage Adjustment Factor Methodology
    To calculate the wage-adjusted facility payment for the payment 
rates set forth in this final rule, we multiply the unadjusted Federal 
payment rate for IRFs by the FY 2024 labor-related share based on the 
2021-based IRF market basket relative importance (74.1 percent) to 
determine the labor-related portion of the standard payment amount. (A 
full discussion of the calculation of the labor-related share appears 
in section VI.E. of this final rule.) We would then multiply the labor-
related portion by the applicable IRF wage index. The wage index tables 
are available on the CMS website at https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS/IRF-Rules-and-Related-Files.html.
    Adjustments or updates to the IRF wage index made under section 
1886(j)(6) of the Act must be made in a budget-neutral manner. We 
calculate a budget-neutral wage adjustment factor as established in the 
FY 2004 IRF PPS final rule (68 FR 45689) and codified at Sec.  
412.624(e)(1), as described in the steps below. We use the listed steps 
to ensure that the FY 2024 IRF standard payment conversion factor 
reflects the update to the wage indexes (based on the FY 2020 hospital 
cost report data) and the update to the labor-related share, in a 
budget-neutral manner:
    Step 1. Calculate the total amount of estimated IRF PPS payments 
using the labor-related share and the wage indexes from FY 2023 (as 
published in the FY 2023 IRF PPS final rule (87 FR 47038)).
    Step 2. Calculate the total amount of estimated IRF PPS payments 
using the FY 2024 wage index values (based on updated hospital wage 
data and considering the permanent cap on wage index decreases policy) 
and the FY 2024 labor-related share of 74.1 percent.
    Step 3. Divide the amount calculated in step 1 by the amount 
calculated in step 2. The resulting quotient is the FY

[[Page 50990]]

2024 budget-neutral wage adjustment factor of 1.0028.
    Step 4. Apply the budget neutrality factor from step 3 to the FY 
2024 IRF PPS standard payment amount after the application of the 
increase factor to determine the FY 2024 standard payment conversion 
factor.
    We discuss the calculation of the standard payment conversion 
factor for FY 2024 in section VI.G. of this final rule.
    We invited public comment on the proposed IRF wage adjustment for 
FY 2024.
    We did not receive any comments on the proposed IRF budget-neutral 
wage adjustment factor methodology for FY 2024. Comments related to the 
budget neutral wage index cap policy are addressed in the Wage 
Adjustment section (VI.F) above.
    We are finalizing our proposals regarding the IRF budget neutral 
wage adjustment factor methodology for FY 2024.

G. Description of the IRF Standard Payment Conversion Factor and 
Payment Rates for FY 2024

    To calculate the standard payment conversion factor for FY 2024, as 
illustrated in Table 14, we begin by applying the increase factor for 
FY 2024, as adjusted in accordance with sections 1886(j)(3)(C) of the 
Act, to the standard payment conversion factor for FY 2023 ($17,878). 
Applying the 3.4 percent increase factor for FY 2024 to the standard 
payment conversion factor for FY 2023 of $17,878 yields a standard 
payment amount of $18,486. Then, we apply the budget neutrality factor 
for the FY 2024 wage index (taking into account the permanent cap on 
wage index decreases policy), and labor-related share of 1.0028, which 
results in a standard payment amount of $18,538. We next apply the 
budget neutrality factor for the CMG relative weights of 1.0002, which 
results in the standard payment conversion factor of $18,541 for FY 
2024.
    We invited public comment on the proposed FY 2024 standard payment 
conversion factor.
    We did not receive any comments on the FY 2024 standard payment 
conversion factor, and therefore, we are finalizing the revisions as 
proposed.
BILLING CODE 4120-01-P
[GRAPHIC] [TIFF OMITTED] TR02AU23.064

    After the application of the CMG relative weights described in 
section V. of this final rule to the FY 2024 standard payment 
conversion factor ($18,541), the resulting unadjusted IRF prospective 
payment rates for FY 2024 are shown in Table 15.

[[Page 50991]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.065


[[Page 50992]]


[GRAPHIC] [TIFF OMITTED] TR02AU23.066

H. Example of the Methodology for Adjusting the Prospective Payment 
Rates

    Table 16 illustrates the methodology for adjusting the prospective 
payments (as described in section VI. of this final rule). The 
following examples are based on two hypothetical Medicare 
beneficiaries, both classified into CMG 0104 (without comorbidities). 
The unadjusted prospective payment rate for CMG 0104 (without 
comorbidities) appears in Table 16.
    Example: One beneficiary is in Facility A, an IRF located in rural 
Spencer County, Indiana, and another beneficiary is in Facility B, an 
IRF located in urban Harrison County, Indiana. Facility A, a rural non-
teaching hospital has a Disproportionate Share Hospital (DSH) 
percentage of 5 percent (which would result in a LIP adjustment of 
1.0156), a wage index of 0.8347, and a rural adjustment of 14.9 
percent. Facility B, an urban teaching hospital, has a DSH percentage 
of 15 percent (which would result in a LIP adjustment of 1.0454 
percent), a wage index of 0.8793, and a teaching status adjustment of 
0.0784.
    To calculate each IRF's labor and non-labor portion of the 
prospective

[[Page 50993]]

payment, we begin by taking the unadjusted prospective payment rate for 
CMG 0104 (without comorbidities) from Table 16. Then, we multiply the 
labor-related share for FY 2024 (74.1 percent) described in section 
VI.E. of this final rule by the unadjusted prospective payment rate. To 
determine the non-labor portion of the prospective payment rate, we 
subtract the labor portion of the Federal payment from the unadjusted 
prospective payment.
    To compute the wage-adjusted prospective payment, we multiply the 
labor portion of the Federal payment by the appropriate wage index 
located in the applicable wage index table. This table is available on 
the CMS website at https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS/IRF-Rules-and-Related-Files.html.
    The resulting figure is the wage-adjusted labor amount. Next, we 
compute the wage-adjusted Federal payment by adding the wage-adjusted 
labor amount to the non-labor portion of the Federal payment.
    Adjusting the wage-adjusted Federal payment by the facility-level 
adjustments involves several steps. First, we take the wage-adjusted 
prospective payment and multiply it by the appropriate rural and LIP 
adjustments (if applicable). Second, to determine the appropriate 
amount of additional payment for the teaching status adjustment (if 
applicable), we multiply the teaching status adjustment (0.0784, in 
this example) by the wage-adjusted and rural-adjusted amount (if 
applicable). Finally, we add the additional teaching status payments 
(if applicable) to the wage, rural, and LIP-adjusted prospective 
payment rates. Table 16 illustrates the components of the adjusted 
payment calculation.
[GRAPHIC] [TIFF OMITTED] TR02AU23.067

BILLING CODE 4120-01-C
    Thus, the adjusted payment for Facility A would be $29,568.51, and 
the adjusted payment for Facility B would be $29,548.23.

VII. Update to Payments for High-Cost Outliers Under the IRF PPS for FY 
2024

A. Update to the Outlier Threshold Amount for FY 2024

    Section 1886(j)(4) of the Act provides the Secretary with the 
authority to make payments in addition to the basic IRF prospective 
payments for cases incurring extraordinarily high costs. A case 
qualifies for an outlier payment if the estimated cost of the case 
exceeds the adjusted outlier threshold. We calculate the adjusted 
outlier threshold by adding the IRF PPS payment for the case (that is, 
the CMG payment adjusted by all of the relevant facility-level 
adjustments) and the adjusted threshold amount (also adjusted by all of 
the relevant facility-level adjustments). Then, we calculate the 
estimated cost of a case by multiplying the IRF's overall CCR by the 
Medicare allowable covered charge. If the estimated cost of the case is 
higher than the adjusted outlier threshold, we make an outlier payment 
for the case equal to 80 percent of the difference between the 
estimated cost of the case and the outlier threshold.
    In the FY 2002 IRF PPS final rule (66 FR 41362 through 41363), we 
discussed our rationale for setting the outlier threshold amount for 
the IRF PPS so that estimated outlier payments would equal 3 percent of 
total estimated payments. For the FY 2002 IRF PPS final rule, we 
analyzed various outlier policies using 3, 4, and 5 percent of the 
total estimated payments, and we concluded that an outlier policy set 
at 3 percent of total estimated payments would optimize the extent to 
which we could reduce the financial risk to IRFs of caring for high-
cost patients, while still providing for adequate payments for all 
other (non-high cost outlier) cases.
    Subsequently, we updated the IRF outlier threshold amount in the 
FYs 2006 through 2023 IRF PPS final rules and the FY 2011 and FY 2013 
notices (70 FR 47880, 71 FR 48354, 72 FR 44284, 73 FR 46370, 74 FR 
39762, 75 FR 42836, 76 FR 47836, 76 FR 59256, 77 FR 44618, 78 FR 47860, 
79 FR 45872, 80 FR 47036, 81 FR 52056, 82 FR 36238, 83 FR 38514, 84 FR 
39054, 85 FR 48444, 86 FR 42362, and 87 FR 47038, respectively) to 
maintain estimated outlier payments at 3 percent of total estimated 
payments. We also stated in the FY 2009 final rule (73 FR 46370 at 
46385) that we would continue to analyze the estimated outlier payments 
for subsequent years and adjust the outlier threshold amount as 
appropriate to maintain the 3 percent target.

[[Page 50994]]

    To update the IRF outlier threshold amount for FY 2024, we proposed 
to use FY 2022 claims data and the same methodology that we used to set 
the initial outlier threshold amount in the FY 2002 IRF PPS final rule 
(66 FR 41362 through 41363), which is also the same methodology that we 
used to update the outlier threshold amounts for FYs 2006 through 2023. 
The outlier threshold is calculated by simulating aggregate payments 
and using an iterative process to determine a threshold that results in 
outlier payments being equal to 3 percent of total payments under the 
simulation. To determine the outlier threshold for FY 2024, we 
estimated the amount of FY 2024 IRF PPS aggregate and outlier payments 
using the most recent claims available (FY 2022) and the proposed FY 
2024 standard payment conversion factor, labor-related share, and wage 
indexes, incorporating any applicable budget-neutrality adjustment 
factors. The outlier threshold is adjusted either up or down in this 
simulation until the estimated outlier payments equal 3 percent of the 
estimated aggregate payments. Based on an analysis of the preliminary 
data used for the proposed rule, we estimated that IRF outlier payments 
as a percentage of total estimated payments would be approximately 2.3 
percent in FY 2023. Therefore, we proposed to update the outlier 
threshold amount from $12,526 for FY 2023 to $9,690 for FY 2024 to 
maintain estimated outlier payments at approximately 3 percent of total 
estimated aggregate IRF payments for FY 2024.
    We note that, as we typically do, we updated our data between the 
FY 2024 IRF PPS proposed and final rules to ensure that we use the most 
recent available data in calculating IRF PPS payments. This updated 
data includes a more complete set of claims for FY 2022. Based on our 
analysis using this updated data, we estimate that IRF outlier payments 
as a percentage of total estimated payments are approximately 2.5 
percent in FY 2023. Therefore, we will update the outlier threshold 
amount from $12,526 for FY 2023 to $10,423 for FY 2024 to account for 
the increases in IRF PPS payments and estimated costs and to maintain 
estimated outlier payments at approximately 3 percent of total 
estimated aggregate IRF payments for FY 2024.
    The following is a summary of the public comments received on the 
proposed update to the FY 2024 outlier threshold amount and our 
responses.
    Comment: Commenters were supportive of the update to the outlier 
threshold for FY 2024; however, some commenters recommended that CMS 
implement a new methodology to set the outlier fixed loss amount using 
a 3-year average approach to promote stability in the outlier threshold 
value. One commenter suggested that changes in the outlier threshold 
should be limited to no more than plus or minus the market basket 
amount in any given year.
    Response: We thank the commenters for their suggestions regarding 
the outlier threshold. We appreciate the suggestion to modify the 
outlier threshold methodology to use a 3-year average; however, it has 
been our long-standing practice to utilize the most recent full fiscal 
year of data to update the prospective payment rates and determine the 
outlier threshold amount, as this data is generally considered to be 
the best overall predictor of experience in the upcoming fiscal year. 
Additionally, we do not believe it would be appropriate to limit 
changes in the outlier threshold to changes in the market basket as 
constraining adjustments to the outlier threshold may result in a 
threshold that generates outlier payments above or below the 3 percent 
target. We appreciate the commenters' suggestions and will take them 
into consideration as we continue to consider revisions to our outlier 
threshold methodology in future rulemaking.
    Comment: Commenters suggested that CMS should consider policies to 
better target outlier payments, such as placing a cap on the amount of 
outlier payments any IRF could receive, lowering the 3 percent outlier 
pool, and including historical outlier reconciliation dollars in the 
outlier projections. Additionally, commenters encouraged CMS to monitor 
the increasing concentration of outlier payments and provide additional 
information on outlier payments for the public.
    Response: We appreciate the various suggestions regarding the 
outlier threshold methodology. As most recently discussed in the FY 
2023 IRF PPS Final Rule (87 FR 47038) our outlier policy is intended to 
reimburse IRFs for treating extraordinarily costly cases. Any future 
consideration given to imposing a limit on outlier payments or 
adjusting the outlier threshold to account for historical outlier 
reconciliation dollars would need to be carefully assessed and take 
into consideration the effect on access to IRF care for certain high-
cost populations. We continue to believe that maintaining the outlier 
pool at 3 percent of aggregate IRF payments optimizes the extent to 
which we can reduce financial risk to IRFs of caring for highest-cost 
patients, while still providing for adequate payments for all other 
non-outlier cases. We appreciate the commenters' suggestions for 
refinements to the outlier methodology as well as the suggested areas 
of analysis and will take them into consideration as we continue to 
assess our outlier threshold methodology. We will continue to monitor 
our outlier policy to ensure it continues to compensate IRFs 
appropriately.
    After consideration of the comments received and considering the 
most recent available data, we are finalizing the outlier threshold 
amount of $10,423 to maintain estimated outlier payments at 
approximately 3 percent of total estimated aggregate IRF payments for 
FY 2024.

B. Update to the IRF Cost-to-Charge Ratio Ceiling and Urban/Rural 
Averages for FY 2024

    CCRs are used to adjust charges from Medicare claims to costs and 
are computed annually from facility-specific data obtained from MCRs. 
IRF specific CCRs are used in the development of the CMG relative 
weights and the calculation of outlier payments under the IRF PPS. In 
accordance with the methodology stated in the FY 2004 IRF PPS final 
rule (68 FR45692 through 45694), we proposed to apply a ceiling to 
IRFs' CCRs. Using the methodology described in that final rule, we 
proposed to update the national urban and rural CCRs for IRFs, as well 
as the national CCR ceiling for FY 2024, based on analysis of the most 
recent data available. We apply the national urban and rural CCRs in 
the following situations:
     New IRFs that have not yet submitted their first MCR.
     IRFs whose overall CCR is in excess of the national CCR 
ceiling for FY 2024, as discussed below in this section.
     Other IRFs for which accurate data to calculate an overall 
CCR are not available.
    Specifically, for FY 2024, we proposed to estimate a national 
average CCR of 0.487 for rural IRFs, which we calculated by taking an 
average of the CCRs for all rural IRFs using their most recently 
submitted cost report data. Similarly, we proposed to estimate a 
national average CCR of 0.398 for urban IRFs, which we calculated by 
taking an average of the CCRs for all urban IRFs using their most 
recently submitted cost report data. We apply weights to both of these 
averages using the IRFs' estimated costs, meaning that the CCRs of IRFs 
with higher total costs factor more heavily into the averages than the 
CCRs of IRFs with lower total costs. For this

[[Page 50995]]

final rule, we have used the most recent available cost report data (FY 
2021). This includes all IRFs whose cost reporting periods begin on or 
after October 1, 2020, and before October 1, 2021. If, for any IRF, the 
FY 2021 cost report was missing or had an ``as submitted'' status, we 
used data from a previous FY's (that is, FY 2004 through FY 2020) 
settled cost report for that IRF. We do not use cost report data from 
before FY 2004 for any IRF because changes in IRF utilization since FY 
2004 resulting from the 60 percent rule and IRF medical review 
activities suggest that these older data do not adequately reflect the 
current cost of care. Using updated FY 2021 cost report data for this 
final rule, we estimate a national average CCR of 0.491 for rural IRFs, 
and a national average CCR of 0.402 for urban IRFs.
    In accordance with past practice, we proposed to set the national 
CCR ceiling at 3 standard deviations above the mean CCR. Using this 
method, we proposed a national CCR ceiling of 1.45 for FY 2024. This 
means that, if an individual IRF's CCR were to exceed this ceiling of 
1.45 for FY 2024, we will replace the IRF's CCR with the appropriate 
proposed national average CCR (either rural or urban, depending on the 
geographic location of the IRF). We calculated the proposed national 
CCR ceiling by:
    Step 1. Taking the national average CCR (weighted by each IRF's 
total costs, as previously discussed) of all IRFs for which we have 
sufficient cost report data (both rural and urban IRFs combined).
    Step 2. Estimating the standard deviation of the national average 
CCR computed in step 1.
    Step 3. Multiplying the standard deviation of the national average 
CCR computed in step 2 by a factor of 3 to compute a statistically 
significant reliable ceiling.
    Step 4. Adding the result from step 3 to the national average CCR 
of all IRFs for which we have sufficient cost report data, from step 1.
    We also proposed that if more recent data become available after 
the publication of this proposed rule and before the publication of the 
final rule, we would use such data to determine the FY 2024 national 
average rural and urban CCRs and the national CCR ceiling in the final 
rule. Using the updated FY 2021 cost report data for this final rule, 
we estimate a national average CCR ceiling of 1.48, using the same 
methodology.
    We invited public comment on the proposed update to the IRF CCR 
ceiling and the urban/rural averages for FY 2024.
    We did not receive any comments on the proposed revisions to the 
IRF CCR ceiling and the urban/rural averages for FY 2024. Consistent 
with the methodology outlined in the proposed rule, and using the most 
recent cost report data, we are finalizing a national average urban CCR 
at 0.402, the national average rural CCR at 0.491, and the national 
average CCR ceiling at 1.48 for FY 2024.

VIII. Modification to the Regulation for Excluded Inpatient 
Rehabilitation Facility Units Paid Under the IRF PPS

A. Background

    Under current regulation, to be excluded from the IPPS, and to be 
paid under the IRF PPS or the IPF PPS, an IRF or IPF unit of a hospital 
must meet a number of requirements under Sec.  412.25. Both this 
regulation and the policies applying to excluded units (which include 
excluded IRF units and excluded IPF units) have been in effect since 
before both the IRF PPS and IPF PPS were established, as discussed in 
the following paragraphs of this section. Before the IRF PPS and the 
IPF PPS were established, excluded units were paid based on their 
costs, as reported on their Medicare cost reports, subject to certain 
facility-specific cost limits. These cost-based payments were 
determined separately for operating and capital costs. Thus, under 
cost-based payments, the process of allocating costs to an IRF or IPF 
unit for reimbursement created significant administrative complexity. 
This administrative complexity necessitated strict regulations that 
allowed hospitals to open a new IPPS-excluded unit only at the start of 
a cost reporting period.
    In the January 3, 1984, final rule (49 FR 235), CMS (then known as 
the Health Care Financing Administration) established policies and 
regulations for hospitals and units subject to and excluded from the 
IPPS. In that rule, we explained that section 1886(d) of the Act 
requires that the prospective payment system apply to inpatient 
hospital services furnished by all hospitals participating in the 
Medicare program except those hospitals or units specifically excluded 
by the law. We further explained our expectation that a hospital's 
status (that is, whether it is subject to, or excluded from, the 
prospective payment system) would generally be determined at the 
beginning of each cost reporting period. We also stated that this 
status would continue throughout the period, which is normally 1 year. 
Accordingly, we stated that changes in a hospital's (or unit's) status 
that result from meeting or failing to meet the criteria for exclusion 
would be implemented only at the start of a cost reporting period. 
However, we also acknowledged that under some circumstances involving 
factors external to the hospital, status changes could be made at times 
other than the beginning of the cost reporting period. For example, a 
change in status could occur if a hospital is first included under the 
prospective payment system and, after the start of its cost reporting 
period, is excluded because of its participation in an approved 
demonstration project or State reimbursement control program that 
begins after the hospital's cost reporting period has begun.
    In the FY 1993 IPPS final rule (57 FR 39798 through 39799), we 
codified our longstanding policies regarding when a hospital unit can 
change its status from not excluded to excluded. We explained in that 
final rule that since the inception of the prospective payment system 
for operating costs of hospital inpatient services in October 1983, 
certain types of specialty-care hospitals and hospital units have been 
excluded from that system under section 1888(d)(1)(B) of the Act. We 
noted that these currently include psychiatric and rehabilitation 
hospitals and distinct part units, children's hospitals, and long-term 
care hospitals. We further explained that section 6004(a)(1) of the 
Omnibus Budget Reconciliation Act of 1989, (Pub. L. 101-239, enacted 
December 19, 1989) amended section 1886(d)(1)(B) of the Act to provide 
that certain cancer hospitals are also excluded. We noted that the 
preamble to the January 3,1984 final rule implementing the prospective 
payment system for operating costs (49 FR 235) stated that the status 
of a hospital or unit (that is, whether it is subject to, or excluded 
from, the prospective payment system) will be determined at the 
beginning of each cost reporting period. We noted that that same 1984 
final rule also provided that changes in a hospital's or unit's status 
that result from meeting or failing to meet the criteria for exclusion 
will be implemented prospectively only at the start of a cost reporting 
period, that is, starting with the beginning date of the next cost 
reporting period (49 FR 243). However, we noted that this policy was 
not set forth in the regulations. In the FY 1993 final rule, we stated 
that we proposed revising Sec. Sec.  412.22 and 412.25 to specify that 
changes in the status of each hospital or hospital unit would be 
recognized only at the start of a cost reporting period. We stated that 
except in the case of retroactive payment

[[Page 50996]]

adjustments for excluded rehabilitation units described in Sec.  
412.30(c), any change in a hospital's or unit's compliance with the 
exclusion criteria that occurs after the start of a cost reporting 
period would not be considered until the start of the following period. 
We noted that this policy would also apply to any unit that is added to 
a hospital during the hospital's cost reporting period. We also stated 
that we proposed revising Sec.  412.25(a) to specify that as a 
requirement for exclusion, a hospital unit must be fully equipped and 
staffed, and be capable of providing inpatient psychiatric or 
rehabilitation care, as of the first day of the first cost reporting 
period for which all other exclusion requirements are met. We explained 
that a unit that meets this requirement would be considered open 
regardless of whether there are any inpatients in the unit.
    In the same FY 1993 IPPS final rule, we responded to commenters who 
objected to this policy, stating that it unnecessarily penalizes 
hospitals for factors beyond their control, such as construction 
delays, that it discourages hospitals from making changes in their 
programs to meet community needs, or that it can place undue workload 
demands on regulatory agencies during certain time periods. In 
response, we explained that we believed that regulatory agencies, 
hospitals, and the public generally would benefit from policies that 
are clearly stated, can be easily understood by both hospitals and 
intermediaries, and can be simply administered. We stated that 
recognizing changes in status only at the beginning of cost reporting 
periods is consistent with these goals, while recognizing changes in 
the middle of cost reporting periods would introduce added complexity 
to the administration of the exclusion provisions. Therefore, we did 
not revise the proposed changes based on these comments.
    In the FY 2000 IPPS final rule (64 FR 41531 through 41532), we 
amended the regulations at Sec.  412.25(c) to allow a hospital unit to 
change from excluded to not excluded at any time during the cost 
reporting period. We explained the statutory basis and rationale for 
this change in the FY 2000 IPPS proposed rule (64 FR 24740), and noted 
that a number of hospitals suggested that we consider a change in our 
policy to recognize, for purposes of exclusion from the IPPS, 
reductions in number of beds in, or entire closure of, units at any 
time during a cost reporting period. In that FY 2000 IPPS proposed 
rule, we explained that hospitals indicated that the bed capacity made 
available as a result of these changes could be used, as they need 
them, to provide additional services to meet patient needs in the acute 
care part of the hospital that is paid under the IPPS. We further 
explained that we evaluated the concerns of the hospitals and the 
effect on the administration of the Medicare program and the health 
care of beneficiaries of making these payment changes. As a result of 
that evaluation, we stated that we believed it was reasonable to adopt 
a more flexible policy in recognition of hospitals' changes in the use 
of their facilities. However, we noted that whenever a hospital 
establishes an excluded unit within the hospital, our Medicare fiscal 
intermediary would need to be able to determine costs of the unit 
separately from costs of the part of the hospital paid under the 
prospective payment system. At that time, we stated that the proper 
determination of costs ensured that the hospital was paid the correct 
amount for services in each part of the facility, and that payments 
under the IPPS did not duplicate payments made under the rules that 
were applicable to excluded hospitals and units, or vice versa. For 
this reason, we stated that we did not believe it would be appropriate 
to recognize, for purposes of exclusion from the IPPS, changes in the 
bed size or status of an excluded unit that are so frequent that they 
interfere with the ability of the intermediary to accurately determine 
costs. Moreover, we explained that section 1886(d)(1)(B) of the Act 
authorizes exclusion from the IPPS of specific types of hospitals and 
units, but not of specific admissions or stays, such as admissions for 
rehabilitation or psychiatric care, in a hospital paid under the IPPS. 
We stated that without limits on the frequency of changes in excluded 
units for purposes of proper Medicare payment, there was the potential 
for some hospitals to adjust the status or size of their excluded units 
so frequently that the units would no longer be distinct entities and 
the exclusion would effectively apply only to certain types of care.
    In the FY 2012 IRF PPS final rule (76 FR 47870), we began further 
efforts to increase flexibilities for excluded IPF and IRF units. In 
that rule, we explained that cost-based reimbursement methodologies 
that were in place before the IPF PPS and IRF PPS meant that the 
facilities' capital costs were determined, in part, by their bed size 
and square footage. Changes in the bed size and square footage would 
complicate the facilities' capital cost allocation. Thus, the 
regulations at Sec.  412.25 limited the situations under which an IRF 
or IPF could change its bed size and square footage. In the FY 2012 IRF 
PPS final rule, we revised Sec.  412.25(b) to enable IRFs and IPFs to 
more easily adjust to beneficiary changes in demand for IRF or IPF 
services and improve beneficiary access to these services. We believed 
that the first requirement (that beds can only be added at the start of 
a cost reporting period) was difficult, and potentially costly, for 
IRFs and IPFs that were expanding through new construction because the 
exact timing of the end of a construction project is often difficult to 
predict.
    In that same FY 2012 IRF PPS final rule, commenters suggested that 
CMS allow new IRF units or new IPF units to open and begin being paid 
under their respective IRF PPS or IPF PPS at any time during a cost 
reporting period, rather than requiring that they could only begin 
being paid under the IRF PPS or the IPF PPS at the start of a cost 
reporting period. In response, we stated that we believed that this 
suggestion was outside the scope of the FY 2012 IRF PPS proposed rule 
(76 FR 24214) because we did not propose any changes to the regulations 
in Sec.  412.25(c). However, we stated that we would consider this 
suggestion for possible inclusion in future rulemaking. Within the FY 
2018 IRF PPS proposed rule (82 FR 20690, 20742 through 20743), CMS 
published a request for information (RFI) on ways to reduce burden for 
hospitals, physicians, and patients; improve the quality of care; 
decrease costs; and ensure that patients and their providers and 
physicians are making the best health care choices possible. In 
response to the RFI, we received comments from IRF industry 
associations, State and national hospital associations, industry groups 
representing hospitals, and individual IRF providers. One of the 
comments we received in response to the RFI suggested allowing new IRF 
units to become excluded and be paid under the IRF PPS at any time 
during the cost reporting period, rather than only at the start of a 
cost reporting period, which the commenter believed would increase 
flexibility and eliminate a policy that may impose higher costs for 
providers while harmonizing an IRF payment system versus the IPPS 
payment system across all new IRF units.

B. Current Challenges Related to Excluded Hospital Units (Sec.  
412.25(c)(1) and (c)(2))

    Currently, under Sec.  412.25(c)(1), a hospital can only start 
being paid under the IRF PPS or the IPF PPS for services provided in an 
excluded unit at the start

[[Page 50997]]

of a cost reporting period. Specifically, Sec.  412.25(c) limits when 
the status of hospital units may change for purposes of exclusion from 
the IPPS, as specified in Sec.  412.25(c)(1) and Sec.  412.25(c)(2). 
Section 412.25(c)(1) states that the status of a hospital unit may be 
changed from not excluded to excluded only at the start of the cost 
reporting period. If a unit is added to a hospital after the start of a 
cost reporting period, it cannot be excluded from the IPPS before the 
start of a hospital's next cost reporting period. Under Sec.  
412.25(c)(2), the status of a hospital unit may be changed from 
excluded to not excluded at any time during a cost reporting period, 
but only if the hospital notifies the fiscal intermediary and the CMS 
Regional Office in writing of the change at least 30 days before the 
date of the change, and maintains the information needed to accurately 
determine costs that are or are not attributable to the excluded unit. 
A change in the status of a unit from excluded to not excluded that is 
made during a cost reporting period must remain in effect for the rest 
of that cost reporting period.
    In recent years, interested parties, such as hospitals, have 
written to CMS to express concerns about what they see as the 
unnecessary restrictiveness of the requirements of Sec.  412.25(c). 
Based on this feedback, we continued to explore opportunities to reduce 
burden for providers and clinicians, while keeping patient-centered 
care a priority. For instance, we considered whether this regulation 
might create unnecessary burden for hospitals and could potentially 
delay necessary rehabilitation beds from opening and being paid under 
the IRF PPS. As we continued to review and reconsider regulations to 
identify ways to improve policy, we recognized that the requirement at 
Sec.  412.25(c)(1) that hospital units can only be excluded at the 
start of a cost reporting period, may be challenging to meet and 
potentially costly for facilities under some circumstances, for 
example, those that are expanding through new construction. Hospitals 
have indicated it is often difficult to predict the exact timing of the 
end of a construction project and construction delays may hamper a 
hospital's ability to have the construction of an excluded unit 
completed exactly at the start of a cost reporting period, which 
hospitals stated can lead to significant revenue loss if they are 
unable to be paid under the IRF PPS or IPF PPS until the start of the 
next cost reporting period.
    As discussed, the requirements of Sec.  412.25(c) were established 
to manage the administrative complexity associated with cost-based 
reimbursement for excluded IRF and IPF units. Today, however, because 
IRF units are paid under the IRF PPS, and IPF units are paid under the 
IPF PPS, cost allocation is not used for payment purposes. Because 
advancements in technology since the inception of the IRF PPS and IPF 
PPS have simplified the cost reporting process and enhanced 
communication between providers, CMS, and Medicare contractors, we are 
reconsidering whether it is necessary to continue to allow hospital 
units to become excluded only at the start of a cost reporting period.

C. Changes to Excluded Hospital Units (Sec.  412.25(c)(1) and (c)(2))

    We are committed to continuing to transform the health care 
delivery system--and the Medicare program--by putting additional focus 
on patient-centered care and working with providers, physicians, and 
patients to improve outcomes, while meeting relevant health care 
priorities and reducing burden.
    In response to the need for availability of inpatient 
rehabilitation beds we are finalizing changes to Sec.  412.25(c) to 
allow greater flexibility for hospitals to open excluded units, while 
minimizing the amount of effort Medicare contractors would need to 
spend administering the regulatory requirements. Although we are 
cognizant that there is a need for rehabilitative health services and 
support for providers along a continuum of care, including a robust 
investment in community-based rehabilitative services, this rule is 
focused on inpatient rehabilitation facility settings.
    We note that Sec.  412.25(c) applies to both IRFs and IPFs; 
therefore, revisions to Sec.  412.25(c) will also affect IPFs in 
similar ways. Readers should refer to the FY 2024 IPF PPS final rule 
for discussion of revisions to Sec.  412.25(c) and unique 
considerations applicable to IPF units.
    As discussed, the current requirements of Sec.  412.25(c)(1) were 
originally established to manage the administrative complexity 
associated with cost-based reimbursement for excluded IPF and IRF 
units. Because IPF and IRF units are no longer paid under cost-based 
reimbursement, but rather under the IPF PPS and IRF PPS respectively, 
we believe that the restriction that limits an IPF or IRF unit to being 
excluded only at the start of a cost reporting period is no longer 
necessary.
    We amended our regulations in the FY 2012 IRF PPS final rule to 
address a regulation that similarly was previously necessary for cost-
based reimbursement, but was not material to payment under the IRF PPS 
and IPF PPS. In that final rule, we explained that under cost-based 
payments, the facilities' capital costs were determined, in part, by 
their bed size and square footage. Changes in the bed size and square 
footage would complicate the facilities' capital cost allocation. We 
explained that under the IRF PPS and IPF PPS, however, a facility's bed 
size and square footage were not relevant for determining the 
individual facility's Medicare payment. Therefore, we believed it was 
appropriate to modify some of the restrictions on a facility's ability 
to change its bed size and square footage. Accordingly, we relaxed the 
restrictions on a facility's ability to increase its bed size and 
square footage. Under the revised requirements that we adopted in the 
FY 2012 IRF PPS final rule in Sec.  412.25(b), an IRF or IPF can change 
(either increase or decrease) its bed size or square footage one time 
at any point in a given cost reporting period as long as it notifies 
the CMS Regional Office at least 30 days before the date of the 
proposed change, and maintains the information needed to accurately 
determine costs that are attributable to the excluded units.
    Similarly, in the case of the establishment of a new excluded IPF 
and IRF units, we do not believe that the timing of the establishment 
of the new unit is material for determining the individual facility's 
level of Medicare payment under the IRF PPS or IPF PPS. We believe it 
would be appropriate to allow a unit to become excluded at any time in 
the cost reporting year. However, we also believe it is important to 
minimize the potential administrative complexity associated with units 
changing their excluded status.
    Accordingly, we amend the requirements currently in regulation at 
Sec.  412.25(c)(1) to allow a hospital to open a new IRF unit anytime 
within the cost reporting year, as long as the hospital notifies the 
CMS Regional Office and Medicare Administrative Contractor (MAC) in 
writing of the change at least 30 days before the date of the change. 
Additionally, if a unit becomes excluded during a cost reporting year, 
this change would remain in effect for the rest of that cost reporting 
year. We maintain the current requirements of Sec.  412.25(c)(2), which 
specify that, if an excluded unit becomes not excluded during a cost 
reporting year, the hospital must notify the MAC and the CMS Regional 
Office in writing of the change at least 30 days before the change, and 
this change would remain in effect for the rest of that cost reporting 
year.

[[Page 50998]]

Finally, we consolidate the requirements for Sec.  412.25(c)(1) and 
Sec.  412.25(c)(2) into a new Sec.  412.25(c)(1) that would apply to 
IRF units and specify the requirements for an IRF unit to become 
excluded or not excluded.
    We believe this will provide IRFs greater flexibility when 
establishing an excluded unit at a time other than the start of a cost 
reporting period.
    As noted, we proposed an identical policy for inpatient psychiatric 
units of hospitals in Sec.  412.25(c)(2) in the FY 2024 IPF PPS 
proposed rule.
    We proposed discrete regulation text for each of the hospital unit 
types (that is, IRF units and IPF units) to solicit comment on issues 
that might affect one hospital unit type and not the other. However, we 
stated that we may consider adopting one consolidated regulation text 
for both IRF and IPF units in either the IRF or IPF final rules for 
both unit types if we finalize both of our proposals. We requested 
public comments on finalizing a consolidated provision that would 
pertain to both IRF and IPF units.
    The following is a summary of the public comments received on 
finalizing a consolidated provision that would pertain to both IRF and 
IPF units and our responses.
    Comment: Commenters expressed broad support for the revision to the 
excluded hospital unit regulation at Sec.  412.25(c). Many commenters 
stated that amending the excluded unit regulation improves access to 
critical rehabilitative services. One commenter appreciated CMS' 
recognition that the prior policy at Sec.  412.25(c) created burden and 
complexity when attempting to open a new IRF unit amid construction, 
State agencies and certificate of need constraints, sometimes resulting 
in missing the start of the new cost reporting period.
    Response: We appreciate the commenters' support of the modification 
to the excluded unit regulation allowing the opening of a new IRF unit 
to occur at any time during the cost reporting period. We agree with 
the commenters that the proposed amendments to Sec.  412.25(c) will 
reduce burden and complexity and make it easier to open a new IRF unit.
    After consideration of the comments we received, we are finalizing 
the consolidated provision that pertains to both IRF and IPF units. The 
amendments to Sec.  412.25(c) for this consolidated provision will be 
finalized in the IPF final rule published elsewhere in this issue of 
the Federal Register.

IX. Inpatient Rehabilitation Facility (IRF) Quality Reporting Program 
(QRP)

A. Background and Statutory Authority

    The Inpatient Rehabilitation Facility Quality Reporting Program 
(IRF QRP) is authorized by section 1886(j)(7) of the Act, and it 
applies to freestanding IRFs, as well as inpatient rehabilitation units 
of hospitals or Critical Access Hospitals (CAHs) paid by Medicare under 
the IRF PPS. Section 1886(j)(7)(A)(i) of the Act requires the Secretary 
to reduce by 2 percentage points the annual increase factor for 
discharges occurring during a fiscal year (FY) for any IRF that does 
not submit data in accordance with the IRF QRP requirements set forth 
in subparagraphs (C) and (F) of section 1886(j)(7) of the Act. Section 
1890A of the Act requires that the Secretary establish and follow a 
pre-rulemaking process, in coordination with the consensus-based entity 
(CBE) with a contract under section 1890 of the Act, to solicit input 
from certain groups regarding he selection of quality and efficiency 
measures for the IRF QRP. We have codified our program requirements in 
our regulations at Sec.  412.634.
    In the FY 2024 IRF PPS proposed rule, we proposed to adopt two new 
measures, remove three existing measures, and modify one existing 
measure. Second, we sought information on principles we could use to 
select and prioritize IRF QRP quality measures in future years. Third, 
we provided an update on our efforts to close the health equity gap. 
Finally, we proposed to begin public reporting of four measures.

B. General Considerations Used for the Selection of Measures for the 
IRF QRP

    For a detailed discussion of the considerations we use for the 
selection of IRF QRP quality, resource use, or other measures, we refer 
readers to the FY 2016 IRF PPS final rule (80 FR 47083 through 47084).
1. Quality Measures Currently Adopted for the FY 2024 IRF QRP
    The IRF QRP currently has 18 measures for the FY 2024 IRF QRP, 
which are listed in Table 17. For a discussion of the factors used to 
evaluate whether a measure should be removed from the IRF QRP, we refer 
readers to Sec.  412.634(b)(2).

[[Page 50999]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.068

C. Overview of IRF QRP Quality Measure Proposals

    In the FY 2024 IRF PPS proposed rule, we proposed to adopt two new 
measures, remove three existing measures, and modify one existing 
measure for the FY 2025 IRF QRP and the FY 2026 IRF QRP. Beginning with 
the FY 2025 IRF QRP we proposed to (1) modify the COVID-19 Vaccination 
Coverage among Healthcare Personnel (HCP) measure, (2) adopt the 
Discharge Function Score measure,\18\ which we specified under sections 
1886(j)(7)(F) and 1899B(c)(1) of the Act, and (3) remove three current 
measures: (i) the Application of Percent of Long-Term Care Hospital 
(LTCH) Patients with an Admission and Discharge Functional Assessment 
and a Care Plan That Addresses Function measure, (ii) the IRF 
Functional Outcome Measure: Change in Self-Care Score for Medical 
Rehabilitation Patients measure, and (iii) the IRF Functional Outcome 
Measure: Change in Mobility Score for Medical Rehabilitation Patients 
measure.
---------------------------------------------------------------------------

    \18\ This measure was submitted to the Measures Under 
Consideration (MUC) List as the Cross-Setting Discharge Function 
Score. Subsequent to the MAP Workgroup meetings, the measure 
developer modified the name. Discharge Function Score for Inpatient 
Rehabilitation Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    We proposed to add one new measure beginning with the FY 2026 IRF 
QRP, the COVID-19 Vaccine: Percent of Patients/Residents Who Are Up to 
Date measure, which we are specifying under sections 1886(j)(7)(F) and 
1899B(d)(1) of the Act.
1. IRF QRP Quality Measures Beginning With the FY 2025 IRF QRP
a. Modification of the COVID-19 Vaccination Coverage Among Healthcare 
Personnel (HCP) Measure Beginning With the FY 2025 IRF QRP
(1) Background
    On January 31, 2020, the Secretary declared a public health 
emergency (PHE) for the United States in response to the global 
outbreak of SARS-CoV-2, a novel (new) coronavirus that causes 
``coronavirus disease 2019'' (COVID-19).\19\ Subsequently, in the FY 
2022 IRF PPS final rule (86 FR 42385 through 42396), we adopted the 
COVID-19 Vaccination Coverage among Healthcare Personnel (HCP COVID-19 
Vaccine)

[[Page 51000]]

measure for the IRF QRP. The HCP COVID-19 Vaccine measure requires each 
IRF to submit data on the number of healthcare personnel (HCP) eligible 
to work in the IRF for at least one day during the reporting period, 
excluding persons with contraindications to the COVID-19 vaccine, who 
have received a complete vaccination course against SARS-CoV-2 (86 FR 
42389 through 42396).
---------------------------------------------------------------------------

    \19\ U.S. Department of Health and Human Services, Office of the 
Assistant Secretary for Preparedness and Response. Determination 
that a Public Health Emergency Exists. January 31, 2020. https://aspr.hhs.gov/legal/PHE/Pages/2019-nCoV.aspx.
---------------------------------------------------------------------------

    Since that time, COVID-19 has continued to spread domestically and 
around the world with more than 103.8 million cases and 1.1 million 
deaths in the United States as of March 21, 2023.\20\ In recognition of 
the ongoing significance and complexity of COVID-19, the Secretary has 
renewed the PHE on April 21, 2020, July 23, 2020, October 2, 2020, 
January 7, 2021, April 15, 2021, July 19, 2021, October 15, 2021, 
January 14, 2022, April 12, 2022, July 15, 2022, October 13, 2022, 
January 11, 2023, and February 9, 2023.\21\ The Department of Health 
and Human Services (HHS) let the PHE expire on May 11, 2023. However, 
HHS stated that the public health response to COVID-19 remains a public 
health priority with a whole-of-government approach to combatting the 
virus, including through vaccination efforts.\22\
---------------------------------------------------------------------------

    \20\ Centers for Disease Control and Prevention. COVID Data 
Tracker. March 21, 2023. https://covid.cdc.gov/covid-data-tracker/#datatracker-home.
    \21\ U.S. Department of Health and Human Services. Office of the 
Assistant Secretary for Preparedness and Response. Renewal of 
Determination that a Public Health Emergency Exists. February 9, 
2023. https://aspr.hhs.gov/legal/PHE/Pages/COVID19-9Feb2023.aspx.
    \22\ U.S. Department of Health and Human Services. Fact Sheet: 
COVID-19 Public Health Emergency Transition Roadmap. February 9, 
2023. https://www.hhs.gov/about/news/2023/02/09/fact-sheet-covid-19-public-health-emergency-transition-roadmap.html.
---------------------------------------------------------------------------

    In the FY 2022 IRF PPS final rule (86 FR 42386 through 42396) and 
in the Revised Guidance for Staff Vaccination Requirements,\23\ we 
stated that vaccination is a critical part of the nation's strategy to 
effectively counter the spread of COVID-19. We continue to believe it 
is important to incentivize and track HCP vaccination in IRFs through 
quality measurement in order to protect healthcare workers, patients, 
and caregivers, and to help sustain the ability of IRFs to continue 
serving their communities after the PHE. At the time we issued the FY 
2022 IRF PPS final rule where we adopted the HCP COVID-19 Vaccine 
measure, the Food and Drug Administration (FDA) had issued emergency 
use authorizations (EUAs) for COVID-19 vaccines manufactured by Pfizer-
BioNTech,\24\ Moderna,\25\ and Janssen.\26\ The populations for which 
all three vaccines were authorized at that time included individuals 18 
years of age and older. Shortly following the publication of the FY 
2022 IRF PPS final rule on August 23, 2021, the FDA issued an approval 
for the Pfizer-BioNTech vaccine, marketed as Comirnaty.\27\ The FDA 
issued approval for the Moderna vaccine, marketed as Spikevax, on 
January 31, 2022 \28\ and an EUA for the Novavax vaccine, on July 13, 
2022.\29\ The FDA also issued EUAs for single booster doses of the then 
authorized COVID-19 vaccines. As of November 19, 
2021,30 31 32 a single booster dose of each COVID-19 vaccine 
was authorized for all eligible individuals 18 years of age and older. 
EUAs were subsequently issued for a second booster dose of the Pfizer-
BioNTech and Moderna vaccines in certain populations in March 2022.\33\ 
The FDA first authorized the use of a booster dose of bivalent or 
``updated'' COVID-19 vaccines from Pfizer-BioNTech and Moderna in 
August 2022.\34\
---------------------------------------------------------------------------

    \23\ Centers for Medicare & Medicaid Services. Revised Guidance 
for Staff Vaccination Requirements QSO-23-02-ALL. October 26, 2022. 
https://www.cms.gov/files/document/qs0-23-02-all.pdf.
    \24\ Food and Drug Administration. FDA Takes Key Action in Fight 
Against COVID-19 By Issuing Emergency Use Authorization for First 
COVID-19 Vaccine. December 11, 2020. https://www.fda.gov/news-events/press-announcements/fda-takes-key-action-fight-against-covid-19-issuing-emergency-use-authorization-first-covid-19.
    \25\ Food and Drug Administration. FDA Takes Additional Action 
in Fight Against COVID-19 By Issuing Emergency Use Authorization for 
Second COVID-19 Vaccine. December 18, 2020. https://www.fda.gov/news-events/press-announcements/fda-takes-additional-action-fight-against-covid-19-issuing-emergency-use-authorization-second-covid.
    \26\ Food and Drug Administration. FDA Issues Emergency Use 
Authorization for Third COVID-19 Vaccine. February 27, 2021. https://www.fda.gov/news-events/press-announcements/fda-issues-emergency-use-authorization-third-covid-19-vaccine.
    \27\ Food and Drug Administration. FDA Approves First COVID-19 
Vaccine. August 23, 2021. https://www.fda.gov/news-events/press-announcements/fda-approves-first-covid-19-vaccine.
    \28\ Food and Drug Administration. Coronavirus (COVID-19) 
Update: FDA Takes Key Action by Approving Second COVID-19 Vaccine. 
January 21, 2022. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-takes-key-action-approving-second-covid-19-vaccine.
    \29\ Food and Drug Administration. Coronavirus (COVID-19) 
Update: FDA Authorizes Emergency Use of Novavax COVID-19 Vaccine, 
Adjuvanted. July 13, 2022. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-authorizes-emergency-use-novavax-covid-19-vaccine-adjuvanted.
    \30\ Food and Drug Administration. FDA Authorizes Booster Dose 
of Pfizer-BioNTech COVID-19 Vaccine for Certain Populations. 
September 22, 2021. https://www.fda.gov/news-events/press-announcements/fda-authorizes-booster-dose-pfizer-biontech-covid-19-vaccine-certain-populations.
    \31\ Food and Drug Administration. Coronavirus (COVID-19) 
Update: FDA Takes Additional Actions on the Use of a Booster Dose 
for COVID-19 Vaccines. October 20, 2021. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-takes-additional-actions-use-booster-dose-covid-19-vaccines.
    \32\ Food and Drug Administration. Coronavirus (COVID-19) 
Update: FDA Expands Eligibility for COVID-19 Vaccine Boosters. 
November 19, 2021. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-expands-eligibility-covid-19-vaccine-boosters.
    \33\ Food and Drug Administration. Coronavirus (COVID-19) 
Update: FDA Authorizes Second Booster Dose of Two COVID-19 Vaccines 
for Older and Immunocompromised Individuals. March 29, 2022. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-authorizes-second-booster-dose-two-covid-19-vaccines-older-and.
    \34\ Food and Drug Administration. (August 2022). Coronavirus 
(COVID-19) Update: FDA Authorizes Moderna, Pfizer-BioNTech Bivalent 
COVID-19 Vaccines for Use as a Booster Dose. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-authorizes-moderna-pfizer-biontech-bivalent-covid-19-vaccines-use.
---------------------------------------------------------------------------

(a) Measure Importance
    In the FY 2022 IRF PPS final rule (86 FR 42401), we acknowledged 
that we were still learning how effective the vaccines were against new 
variants of the virus that cause COVID-19. While the impact of COVID-19 
vaccines on asymptomatic infection and transmission is not yet fully 
known, there are now robust data available across multiple populations 
on COVID-19 vaccine effectiveness against severe illness, 
hospitalization, and death. Two-dose COVID-19 vaccines from Pfizer-
BioNTech and Moderna were found to be 88 percent and 93 percent 
effective against hospitalization for COVID-19, respectively, over 6 
months for adults over age 18 without immunocompromising 
conditions.\35\ During a SARS-CoV-2 surge in the spring and summer of 
2021, 92 percent of COVID-19 hospitalizations and 91 percent of COVID-
19-associated deaths were reported among persons not fully 
vaccinated.\36\ Real-world studies of population-level vaccine 
effectiveness indicated similarly high rates of efficacy

[[Page 51001]]

in preventing SARS-CoV-2 infection among frontline workers in multiple 
industries, with a 90 percent effectiveness in preventing symptomatic 
and asymptomatic infection from December 2020 through August 2021.\37\ 
Vaccines have also been highly effective in real-world conditions at 
preventing COVID-19 in HCP with up to 96 percent efficacy for fully 
vaccinated HCP, including those at risk for severe infection and those 
in racial and ethnic groups disproportionately affected by COVID-
19.\38\ Overall, data demonstrate that COVID-19 vaccines are effective 
and prevent severe disease, hospitalization, and death.
---------------------------------------------------------------------------

    \35\ Self WH, Tenforde MW, Rhoads JP, et al. Comparative 
Effectiveness of Moderna, Pfizer-BioNTech, and Janssen (Johnson & 
Johnson) Vaccines in Preventing COVID-19 Hospitalizations Among 
Adults Without Immunocompromising Conditions--United States, March-
August 2021. MMWR Morb Mortal Wkly Rep 2021;70:1337-1343. doi: 
10.15585/mmwr.mm7038e1. https://cdc.gov/mmwr/volumes/70/wr/mm7038e1.htm?s_cid=mm7038e1_w.
    \36\ Scobie HM, Johnson AG, Suthar AB, et al. Monitoring 
Incidence of COVID-19 Cases, Hospitalizations, and Deaths, by 
Vaccination Status--13 U.S. Jurisdictions, April 4-July 17, 2021. 
MMWR Morb Mortal Wkly Rep 2021;70:1284-1290. doi: 10.15585/
mmwr.mm7037e1. https://www.cdc.gov/mmwr/volumes/70/wr/mm7037e1.htm.
    \37\ Fowlkes A, Gaglani M, Groover K, et al. Effectiveness of 
COVID-19 Vaccines in Preventing SARS-CoV-2 Infection Among Frontline 
Workers Before and During B.1.617.2 (Delta) Variant Predominance--
Eight U.S. Locations, December 2020-August 2021. MMWR Morb Mortal 
Wkly Rep 2021;70:1167-1169. doi: 10.15585/mmwr.mm7034e4. https://www.cdc.gov/mmwr/volumes/70/wr/mm7034e4.htm.
    \38\ Pilishvili T, Gierke R, Fleming-Dutra KE, et al. 
Effectiveness of mRNA Covid-19 Vaccine among U.S. Health Care 
Personnel. N Engl J Med. 2021 Dec 16;385(25):e90. doi: 10.1056/
NEJMoa2106599. PMID: 34551224; PMCID: PMC8482809. https://pubmed.ncbi.nlm.nih.gov/34551224/.
---------------------------------------------------------------------------

    As SARS-CoV-2 persists and evolves, our COVID-19 vaccination 
strategy must remain responsive. When we adopted the HCP COVID-19 
Vaccine measure in the FY 2022 IRF PPS final rule, we stated that the 
need for additional/booster doses of COVID-19 vaccines had not been 
established and no additional doses had been recommended (86 FR 42390). 
We also stated that we believed the numerator was sufficiently broad to 
include potential future additional/booster doses as part of a 
``complete vaccination course'' and that the measure was sufficiently 
specified to address boosters (86 FR 42390). Since we adopted the HCP 
COVID-19 Vaccine measure in the FY 2022 IRF PPS final rule, new 
variants of SARS-CoV-2 have emerged around the world and within the 
United States. Specifically, the Omicron variant (and its related 
subvariants) is listed as a variant of concern by the Centers for 
Disease Control and Prevention (CDC) because it spreads more easily 
than earlier variants.\39\ Vaccine manufacturers have responded to the 
Omicron variant by developing bivalent COVID-19 vaccines, which include 
a component of the original virus strain, to provide broad protection 
against COVID-19 and a component of the Omicron variant, to provide 
better protection against COVID-19 caused by the Omicron variant.\40\ 
These booster doses of the bivalent COVID-19 vaccines have been shown 
to increase immune response to SARS-CoV-2 variants, including Omicron, 
particularly in individuals that are more than 6 months removed from 
receipt of their primary series.\41\ The FDA issued EUAs for booster 
doses of two bivalent COVID-19 vaccines, one from Pfizer-BioNTech \42\ 
and one from Moderna \43\ and strongly encourages anyone who is 
eligible to consider receiving a booster dose with a bivalent COVID-19 
vaccine to provide better protection against currently circulating 
variants.\44\ COVID-19 booster doses are associated with a greater 
reduction in infections among HCP relative to those who only received 
primary series vaccination, with a rate of breakthrough infections 
among HCP who received only a two-dose regimen of 21.4 percent compared 
to a rate of 0.7 percent among HCP who received booster doses of the 
COVID-19 vaccine.45 46
---------------------------------------------------------------------------

    \39\ Centers for Disease Control and Prevention. COVID-19: 
Variants. https://www.cdc.gov/coronavirus/2019-ncov/variants/index.html.
    \40\ Food and Drug Administration. COVID-19 Bivalent Vaccines. 
https://www.fda.gov/emergency-preparedness-and-response/coronavirus-disease-2019-covid-19/covid-19-bivalent-vaccines.
    \41\ Chalkias S, Harper C, Vrbicky K, et al. A Bivalent Omicron-
Containing Booster Vaccine Against COVID-19. N Engl J Med. 2022 Oct 
6;387(14):1279-1291. doi: 10.1056/NEJMoa2208343. PMID: 36112399; 
PMCID: PMC9511634.
    \42\ Food and Drug Administration. Pfizer-BioNTech COVID-19 
Vaccines. https://www.fda.gov/emergency-preparedness-and-response/coronavirus-disease-2019-covid-19/pfizer-biontech-covid-19-vaccines.
    \43\ Food and Drug Administration. Moderna COVID-19 Vaccines. 
https://www.fda.gov/emergency-preparedness-and-response/coronavirus-disease-2019-covid-19/moderna-covid-19-vaccines.
    \44\ Food and Drug Administration. Coronavirus (COVID-19) 
Update: FDA Authorizes Moderna, Pfizer-BioNTech Bivalent COVID-19 
Vaccines for Use as a Booster Dose. August 31, 2022. https://www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-authorizes-moderna-pfizer-biontech-bivalent-covid-19-vaccines-use.
    \45\ Oster Y, Benenson S, Nir-Paz R, Buda I, Cohen MJ. The 
effect of a third BNT162b2 vaccine on breakthrough infections in 
health care workers: a cohort analysis. Clin Microbiol Infect. 2022 
May;28(5):735.e1-735.e3. https://pubmed.ncbi.nlm.nih.gov/35143997/.
    \46\ Prasad N, Derado G, Acharya Nanduri S, et al. Effectiveness 
of a COVID-19 Additional Primary or Booster Vaccine Dose in 
Preventing SARS-CoV-2 Infection Among Nursing Home Residents During 
Widespread Circulation of the Omicron Variant--United States, 
February 14-March 27, 2022. MMWR Morb Mortal Wkly Rep. 2022 May 
6;71(18):633-637. doi: 10.1016/j.cmi.2022.01.019. PMID: 35143997; 
PMCID: PMC8820100.
---------------------------------------------------------------------------

    We believe that vaccination remains the most effective means to 
prevent the severe consequences of COVID-19, including severe illness, 
hospitalization, and death. Given the availability of vaccine efficacy 
data, EUAs issued by the FDA for bivalent boosters, the continued 
presence of SARS-CoV-2 in the United States, and variance among rates 
of booster dose vaccination, it is important to update the 
specifications of the HCP COVID-19 Vaccine measure to refer to HCP who 
receive primary series and additional/booster doses in a timely manner. 
Given the persistent spread of COVID-19, we continue to believe that 
monitoring and surveillance of vaccination rates among HCP is important 
and provides patients, beneficiaries, and their caregivers with 
information to support informed decision making. We proposed to modify 
the HCP COVID-19 Vaccine measure to replace the term ``complete 
vaccination course'' with the term ``up to date'' in the HCP 
vaccination definition. We also proposed to update the numerator to 
specify the time frames within which an HCP is considered up to date 
with recommended COVID-19 vaccines, including additional/booster doses, 
beginning with the FY 2025 IRF QRP.
(b) Measure Testing
    The CDC conducted beta testing of the proposed modified HCP COVID-
19 Vaccine measure by assessing if the collection of information on 
additional/booster doses received by HCP was feasible, as information 
on receipt of additional/booster doses is required for determining if 
HCP are up to date with the current COVID-19 vaccination 
recommendations. Feasibility was assessed by calculating the proportion 
of facilities that reported additional/booster doses of the COVID-19 
vaccine. The assessment was conducted in various facility types, 
including IRFs, using vaccine coverage data for the first quarter of 
calendar year (CY) 2022 (January-March), which was reported through the 
CDC's National Healthcare Safety Network (NHSN). Feasibility of 
reporting additional/booster doses is evident by the fact that 63.9 
percent of IRFs reported vaccination additional/booster dose coverage 
data to the NHSN for the first quarter of 2022.\47\ Additionally, HCP 
COVID-19 Vaccine measure scores calculated using January 1-March 31, 
2022 data had a median of 20.3 percent and an interquartile range of 
8.9 to 37.7 percent, indicating a measure performance gap as there are 
clinically significant differences in

[[Page 51002]]

additional/booster dose vaccination coverage rates among IRFs.\48\
---------------------------------------------------------------------------

    \47\ National Quality Forum. Measure Applications Partnership 
(MAP) Post-Acute Care/Long-Term Care: 2022-2023 Measures Under 
Consideration (MUC) Cycle Measure Specifications. December 1, 2022. 
https://mmshub.cms.gov/sites/default/files/map-pac-muc-measure-specifications-2022-2023.pdf.
    \48\ National Quality Forum. Measure Applications Partnership 
(MAP) Post-Acute Care/Long-Term Care: 2022-2023 Measures Under 
Consideration (MUC) Cycle Measure Specifications. December 1, 2022. 
https://mmshub.cms.gov/sites/default/files/map-pac-muc-measure-specifications-2022-2023.pdf.
---------------------------------------------------------------------------

(2) Competing and Related Measures
    Section 1886(j)(7)(D)(i) of the Act and section 1899B(e)(2)(A) of 
the Act require that, absent an exception under section 
1886(j)(7)(D)(i) and section 1899B(e)(2)(B) of the Act, measures 
specified under section 1899B of the Act must be endorsed by a CBE with 
a contract under section 1890(a) of the Act. In the case of a specified 
area or medical topic determined appropriate by the Secretary for which 
a feasible and practical measure has not been endorsed, section 
1886(j)(7)(D)(i) of the Act and section 1899B(e)(2)(B) of the Act 
permit the Secretary to specify a measure that is not so endorsed, as 
long as due consideration is given to measures that have been endorsed 
or adopted by a consensus organization identified by the Secretary.
    The current version of the HCP COVID-19 Vaccine measure recently 
received endorsement by the CBE on July 26, 2022 under the name 
``Quarterly Reporting of COVID-19 Vaccination Coverage Among Healthcare 
Personnel.'' \49\ However, this measure received endorsement based on 
its specifications depicted in the FY 2022 IRF PPS final rule (86 FR 
42386 through 42396) and does not capture information about whether HCP 
are up to date with their COVID-19 vaccinations. The proposed 
modification of this measure utilizes the term up to date in the HCP 
vaccination definition and updates the numerator to specify the time 
frames within which an HCP is considered up to date with recommended 
COVID-19 vaccines. We were unable to identify any measures endorsed or 
adopted by a consensus organization for IRFs that captured information 
on whether HCP are up to date with their COVID-19 vaccinations, and we 
found no other feasible and practical measure on this topic.
---------------------------------------------------------------------------

    \49\ Partnership for Quality Measurement. Quarterly Reporting of 
COVID-19 Vaccination Coverage among Healthcare Personnel. July 26, 
2022. https://p4qm.org/measures/3636.
---------------------------------------------------------------------------

    Therefore, after consideration of other available measures, we 
found that the exception under sections 1886(j)(7)(D)(ii) and 
1899B(e)(2)(B) of the Act applies and proposed the modified measure, 
HCP COVID-19 Vaccine beginning with the FY 2025 IRF QRP. The CDC, the 
measure developer, is pursuing CBE endorsement for the modified version 
of the measure and is considering an expedited review process as the 
current version of the measure has already received endorsement.
(3) Measure Applications Partnership (MAP) Review
    We refer readers to the FY 2022 IRF PPS final rule (86 FR 42387 
through 42388) for more information on the initial review of the HCP 
COVID-19 Vaccine measure by the Measure Applications Partnership (MAP).
    The pre-rulemaking process includes making publicly available a 
list of quality and efficiency measures, called the Measures Under 
Consideration (MUC) List, that the Secretary is considering adopting 
for use in the Medicare program, including our quality reporting 
programs. This allows interested parties to provide recommendations to 
the Secretary on the measures included on the list. We included an 
updated version of the HCP COVID-19 Vaccine measure on the MUC List, 
entitled ``List of Measures under Consideration for December 1, 2022'' 
\50\ for the 2022-2023 pre-rulemaking cycle for consideration by the 
MAP. Interested parties submitted three comments during the pre-
rulemaking process on the proposed modifications of the HCP COVID-19 
Vaccine measure, and support was mixed. One commenter noted the 
importance for HCP to be vaccinated against COVID-19 and supported 
measurement and reporting as an important strategy to help healthcare 
organizations assess their performance in achieving high rates of up to 
date vaccination of their HCP, while also noting that the measure would 
provide valuable information to the government as part of its ongoing 
response to the pandemic. This commenter also recommended the measure 
be used for internal quality improvement purposes rather than being 
publicly reported on Care Compare. Finally, this commenter also 
suggested that the measure should be stratified by social risk factors. 
However, two commenters supported less specific criteria for 
denominator and numerator inclusion. Specifically, one such commenter 
did not support the inclusion of unpaid volunteers in the measure 
denominator and found the measure's denominator to be unclear. Two 
commenters expressed concerns regarding burden of data collection, data 
lag, staffing challenges, and reportedly ``high rates of providers 
contesting penalties tied to the existing HCP COVID-19 Vaccine measure 
adopted in the FY 2022 IRF PPS final rule.'' One commenter recommended 
that the measure be recharacterized as a surveillance measure given 
what they referred to as a tenuous relationship between collected data 
and quality of care provided by IRFs. Finally, all three commenters 
raised concern about the difficulty of defining up to date for purposes 
of the measure.
---------------------------------------------------------------------------

    \50\ Centers for Medicare & Medicaid Services. Overview of the 
List of Measures Under Consideration for December 1, 2022. CMS.gov. 
https://mmshub.cms.gov/sites/default/files/2022-MUC-List-Overview.pdf.
---------------------------------------------------------------------------

    Shortly after publication of the MUC List, several MAP workgroups 
met to provide input on the modification we proposed for the current 
HCP COVID-19 Vaccine measure. First, the MAP Health Equity Advisory 
Group convened on December 6-7, 2022. The MAP Health Equity Advisory 
Group questioned whether the measure excludes patients with 
contraindications to FDA authorized or approved COVID-19 vaccines, and 
whether the measure will be stratified by demographic factors. The 
measure developer (that is, the CDC) confirmed that HCP with 
contraindications to the vaccines are excluded from the measure 
denominator and responded that the measure will not be stratified by 
demographic factors since the data are submitted at an aggregate rather 
than an individual level.
    The MAP Rural Health Advisory Group met on December 8-9, 2022, 
during which a few members expressed concerns about data collection 
burden, given that small rural hospitals may not have employee health 
software. The measure developer acknowledged the challenge of getting 
adequate documentation and emphasized their goal is to ensure the 
measures do not present a burden on the provider. The measure developer 
also noted that the model used for the HCP COVID-19 Vaccine measure is 
based on the Influenza Vaccination Coverage among HCP measure (CBE 
#0431), and it intends to utilize a similar approach to the modified 
HCP COVID-19 Vaccine measure if vaccination strategy becomes seasonal. 
The measure developer acknowledged that if COVID-19 becomes seasonal, 
the measure model could evolve to capture seasonal vaccination.
    Next, the MAP Post-Acute Care/Long-Term Care (PAC/LTC) workgroup 
met on December 12, 2022, and provided input on the modification we 
proposed for the HCP COVID-19 Vaccine measure. The MAP PAC/LTC 
workgroup noted that the previous version of the measure received 
endorsement from the CBE (CBE

[[Page 51003]]

#3636),\51\ and that the CDC intends to submit the updated measure for 
endorsement. The PAC/LTC workgroup voted to support the staff 
recommendation of conditional support for rulemaking pending testing 
indicating the measure is reliable and valid, and endorsement by the 
CBE.
---------------------------------------------------------------------------

    \51\ Partnership for Quality Measurement. Quarterly Reporting of 
COVID-19 Vaccination Coverage among Healthcare Personnel. July 26, 
2022. https://p4qm.org/measures/3636.
---------------------------------------------------------------------------

    Following the PAC/LTC workgroup meeting, a public comment period 
was held in which interested parties commented on the PAC/LTC 
workgroup's preliminary recommendations, and the MAP received three 
comments. Two supported the proposed modification of the HCP COVID-19 
Vaccine measure, one of which strongly supported the vaccination of HCP 
against COVID-19. Although these commenters supported the measure, one 
commenter recommended seeking CBE \52\ endorsement for the updated 
measure and encouraged CMS to monitor any unintended consequences from 
the measure. Two commenters raised concerns with the measure's 
specifications. Specifically, one noted the denominator included a 
broad number of HCP, and another recommended a vaccination exclusion or 
exception for sincerely held religious beliefs. Finally, one commenter 
raised issues related to the time lag between data collection and 
public reporting on Care Compare and encouraged CMS to provide 
information as to whether the measure is reflecting vaccination rates 
accurately and encouraging HCP vaccination.
---------------------------------------------------------------------------

    \52\ We emphasize that any references to NQF in the proposed 
rule were intended to refer to the CBE contracted by CMS at that 
time.
---------------------------------------------------------------------------

    The MAP Coordinating Committee convened on January 24-25, 2023, 
during which the proposed measure was placed on the consent calendar 
and received a final recommendation of conditional support for 
rulemaking pending testing indicating the measure is reliable and 
valid, and endorsement by the CBE. We refer readers to the final MAP 
recommendations, titled 2022-2023 MAP Final Recommendations.\53\
---------------------------------------------------------------------------

    \53\ 1 Measure Applications Partnership. 2022-2023 MAP Final 
Recommendations. https://mmshub.cms.gov/sites/default/files/2022-2023-MAP-Final-Recommendations-508.xlsx.
---------------------------------------------------------------------------

(4) Quality Measure Calculation
    The HCP COVID-19 Vaccine measure is a process measure developed by 
the CDC to track COVID-19 vaccination coverage among HCP in facilities 
such as IRFs. The HCP COVID-19 Vaccine measure is a process measure and 
is not risk-adjusted.
    The denominator would be the number of HCP eligible to work in the 
facility for at least one day during the reporting period, excluding 
persons with contraindications to COVID-19 vaccination that are 
described by the CDC.\54\ We believe it is necessary to allow IRFs to 
include all HCP within the facility in the reporting because all HCP 
would have access to and may interact with IRF patients. IRFs report 
the following four categories of HCP to NHSN; the first three are 
included in the measure denominator:
---------------------------------------------------------------------------

    \54\ Centers for Disease Control and Prevention. 
Contraindications and precautions. https://www.cdc.gov/vaccines/covid-19/clinical-considerations/interim-considerations-us.html#contraindications.
---------------------------------------------------------------------------

     Employees: Includes all persons who receive a direct 
paycheck from the reporting facility (that is, on the facility's 
payroll), regardless of clinical responsibility or patient contact.
     Licensed independent practitioners (LIPs): This includes 
physicians (MD, DO), advanced practice nurses, and physician assistants 
only who are affiliated with the reporting facility but are not 
directly employed by it (that is, they do not receive a direct paycheck 
from the facility), regardless of clinical responsibility or patient 
contact. Post-residency fellows are also included in this category if 
they are not on the facility's payroll.
     Adult students/trainees and volunteers: This includes all 
medical, nursing, or other health professional, students, interns, 
medical residents and volunteers aged 18 or over who are affiliated 
with the healthcare facility, but are not directly employed by it (that 
is, they do not receive a direct paycheck from the facility) regardless 
of clinical responsibility or patient contact.
     Other contract personnel: Contract personnel are defined 
as persons providing care, treatment, or services at the facility 
through a contract who do not fall into any of the above-mentioned 
denominator categories. This also includes vendors providing care, 
treatment, or services at the facility who may or may not be paid 
through a contract. Facilities are required to enter data on other 
contract personnel for submission in the NHSN application, but data for 
this category are not included in the HCP COVID-19 Vaccine measure.
    The denominator excludes denominator-eligible individuals with 
contraindications as defined by the CDC.\55\ We did not propose any 
changes to the denominator exclusions.
---------------------------------------------------------------------------

    \55\ Centers for Disease Control and Prevention. 
Contraindications and precautions. https://www.cdc.gov/vaccines/covid-19/clinical-considerations/interim-considerations-us.html#contraindications.
---------------------------------------------------------------------------

    The numerator would be the cumulative number of HCP in the 
denominator population who are considered up to date with CDC-
recommended COVID-19 vaccines. Providers would refer to the definition 
of up to date as of the first day of the quarter, which can be found at 
https://www.cdc.gov/nhsn/pdfs/hps/covidvax/UpToDateGuidance-508.pdf. 
For the purposes of NHSN surveillance, individuals would have been 
considered up to date during the Quarter 4 CY 2022 reporting period 
(surveillance period September 26, 2022--December 25, 2022) for the IRF 
QRP if they meet one of the following criteria in place at the time:
    1. Individuals who received an updated bivalent \56\ booster dose, 
or
---------------------------------------------------------------------------

    \56\ The updated (bivalent) Moderna and Pfizer-BioNTech boosters 
target the most recent Omicron subvariants. The updated (bivalent) 
boosters were recommended by the CDC on September 2, 2022. As of 
this date, the original, monovalent mRNA vaccines are no longer 
authorized as a booster dose for people ages 12 years and older.
---------------------------------------------------------------------------

    2a. Individuals who received their last booster dose less than 2 
months ago, or
    2b. Individuals who completed their primary series \57\ less than 2 
months ago.
---------------------------------------------------------------------------

    \57\ Completing a primary series means receiving a two-dose 
series of a COVID-19 vaccine or a single dose of Janssen/J&J COVID-
19 vaccine.
---------------------------------------------------------------------------

    We refer readers to https://www.cdc.gov/nhsn/pdfs/nqf/covid-vax-hcpcoverage-rev-2023-508.pdf for more details on the measure 
specifications.\58\
---------------------------------------------------------------------------

    \58\ We highlight that the hyperlink included in the FY 2024 IRF 
PPS proposed rule has been retired as the CDC has uploaded a new 
measure specification document to the NHSN. Therefore, the hyperlink 
has been updated in this FY 2024 IRF PPS final rule.
---------------------------------------------------------------------------

    While we did not propose any changes to the data submission or 
reporting process for the HCP COVID-19 Vaccine measure, we proposed 
that for purposes of meeting FY 2025 IRF QRP compliance, IRFs would 
report HCP who are up to date beginning in quarter four of CY 2023. 
Under the data submission and reporting process, IRFs would collect the 
numerator and denominator for the modified HCP COVID-19 Vaccine measure 
for at least one self-selected week during each month of the reporting 
quarter. IRFs would submit the data to the NHSN Healthcare Personnel 
Safety (HPS) Component before the quarterly deadline. If an IRF submits 
more than 1 week of data in a month, the CDC would use the most recent 
week's data to

[[Page 51004]]

calculate the measure. Each quarter, the CDC would calculate a single 
quarterly COVID-19 HCP vaccination coverage rate for each IRF, which 
would be calculated by taking the average of the data from the three 
weekly rates submitted by the IRF for that quarter. Beginning with the 
FY 2026 IRF QRP, we proposed that IRFs would be required to submit data 
for the entire calendar year.
    We also proposed that public reporting of the modified version of 
the HCP COVID-19 Vaccine measure would begin by the September 2024 Care 
Compare refresh or as soon as technically feasible.
    We invited public comment on our proposal to modify the HCP COVID-
19 Vaccine measure beginning with the FY 2025 IRF QRP. The following is 
a summary of the comments we received on our proposal to modify the HCP 
COVID-19 Vaccine measure beginning with the FY 2025 IRF QRP and our 
responses.
    Comment: Several commenters supported our proposal to modify the 
numerator definition for the HCP COVID-19 Vaccine measure and to update 
the numerator to specify the time frames within which an HCP is 
considered up to date with recommended COVID-19 vaccines. One of these 
commenters said they continue to believe COVID-19 vaccination among HCP 
in all healthcare settings is the most effective infection prevention 
tool to protect staff, patients, and visitors against severe illness, 
hospitalization, and death. Another one of these commenters stated they 
recognized that vaccinations play a critical role in the nation's 
strategy to counter the spread of COVID-19, but still encouraged CMS to 
continue to monitor the measure.
    Response: We thank the commenters for their support. We agree that 
vaccination is a critical part of the nation's strategy to effectively 
counter the spread of COVID-19. We continue to believe it is important 
to incentivize and track HCP vaccination through quality measurement 
across care settings, including IRFs, in order to protect HCP, 
patients, and caregivers, and to help sustain the ability of HCP in 
each of these care settings to continue serving their communities. We 
will continue to monitor all measures to identify any concerning trends 
as part of our routine monitoring activities to regularly assess 
measure performance, reliability, and reportability for all data 
submitted for the IRF QRP.
    Comment: Several commenters were concerned that the measure has not 
undergone full reliability and validity testing, and they believe the 
CBE endorsement process will allow a full evaluation of a range of 
issues affecting measure reliability, accuracy, and feasibility. Two of 
these commenters, however, stated that the current version of the HCP 
COVID-19 Vaccine measure has not had a holistic evaluation to determine 
whether it is working as intended since it never went through a CBE 
endorsement process and is relatively new to the CMS quality reporting 
programs.
    Response: We refer commenters to section IX.C.1.a.2. of this final 
rule where we point out that the current version of the HCP COVID-19 
Vaccine measure received endorsement by the CBE on July 26, 2022, under 
the name ``Quarterly Reporting of COVID-19 Vaccination Coverage among 
Healthcare Personnel.'' \59\ However, this measure received endorsement 
based on its specifications in the FY 2022 IRF PPS final rule (86 FR 
42386 through 42396). Even though the current, endorsed version does 
not capture information about whether HCP are up to date with their 
COVID-19 vaccinations, we believe its endorsement speaks to the quality 
of the measure design as we proposed that many components of the 
measure remain intact in this modified version. Since we were unable to 
identify any CBE-endorsed measures for IRFs that captured information 
on whether HCP are up to date with their COVID-19 vaccinations, and we 
found no other feasible and practical measure on this topic, we find 
the modification to the HCP COVID-19 Vaccine measure reasonable for IRF 
QRP adoption and implementation. The CDC, the measure developer, is 
pursuing CBE endorsement for the modified version of the measure.
---------------------------------------------------------------------------

    \59\ Partnership for Quality Measurement. Quarterly Reporting of 
COVID-19 Vaccination Coverage among Healthcare Personnel. July 26, 
2022. https://p4qm.org/measures/3636.
---------------------------------------------------------------------------

    In terms of measure testing, as mentioned in section IX.C.1.a.1.b. 
of this final rule, we reiterate that the CDC conducted beta testing of 
the modified HCP COVID-19 Vaccine measure and concluded that the 
collection of information on additional/booster doses received by HCP 
was feasible with 63.9 percent of IRFs reported vaccination additional/
booster dose coverage data to the NHSN for the first quarter of 2022. 
Additionally, the measure score displayed a performance gap indicating 
clinically significant differences in additional/booster dose 
vaccination coverage rates among IRFs. We will continue to monitor all 
our measures to identify any concerning trends as part of our routine 
monitoring activities to regularly assess measure performance, 
reliability, and reportability for all data submitted for the IRF QRP.
    Comment: Several commenters opposed the proposed modifications to 
the HCP COVID-19 Vaccine measure. The most frequently cited reasons 
were that the COVID-19 PHE ended on May 11, 2023, and subsequently CMS 
removed the staff vaccination requirement under the Hospital Conditions 
of Participation (CoP) at Sec.  482.42(g) established by the Omnibus 
COVID-19 Health Care Staff Vaccination Interim Final Rule (86 FR 
61555). Two of these commenters questioned why the HCP COVID-19 Vaccine 
measure would still be used as a metric for quality of care in the IRF 
QRP at the same time CMS is removing the requirement that covered 
providers and suppliers establish policies and procedures for staff 
vaccination for COVID-19 and removing the COVID-19 vaccination 
requirements from the hospital conditions of participation. One of 
these commenters suggested that if CMS plans to require providers 
report staff vaccination status, it would be more appropriate to 
implement the requirement through the CoPs rather than the IRF QRP. One 
of these commenters highlighted that facilities will no longer have any 
Federal authority to require staff to receive any COVID-19 vaccines and 
demand vaccination status from staff. One commenter suggested the 
proposed revision to the measure would be inconsistent with Federal and 
State mandates which require only a primary vaccination series, and 
since the PHE is ending, many (if not all) of these mandates are being 
lifted. They point out that the Federal and State mandates did not 
extend the HCP vaccination requirement to include the bivalent booster 
or any other booster. Given the Administration's announcement that the 
COVID-19 PHE has ended, they believe the need for HCP to be up to date 
with vaccinations will be diminished, and the benefit of this measure 
may be compromised.
    Response: We appreciate the commenters' feedback, but disagree. We 
continue to believe that it is important to measure vaccination status 
regardless of whether the COVID-19 PHE is in effect. We also believe 
this measure continues to align with our goals to promote wellness and 
disease prevention. Under CMS' Meaningful Measures Framework 2.0, the 
HCP COVID-19 Vaccine measure addresses the quality priorities of 
``Immunizations'' and ``Public Health'' through the Meaningful Measures 
Area

[[Page 51005]]

of ``Wellness and Prevention.'' \60\ Under the National Quality 
Strategy, the measure addresses the goal of Safety under the priority 
area Safety and Resiliency.\61\ While we removed vaccination 
requirements from the Hospital CoP at the end of the PHE as discussed 
previously, we note that the reporting requirements of the IRF QRP for 
the proposed modified version of the HCP COVID-19 Vaccine measure are 
distinct from those cited by the commenter. Specifically, the IRF QRP 
is a pay-for-reporting program, and therefore the inclusion of this 
measure does not require that HCP actually receive these additional/
booster vaccine doses. The Administration's continued response to 
COVID-19 is not fully dependent on the emergency declaration for the 
COVID-19 PHE, and even beyond the end of the COVID-19 PHE, we will 
continue to work to protect individuals and communities from the virus 
and its worst impacts by supporting access to COVID-19 vaccines, 
treatments, and tests.\62\
---------------------------------------------------------------------------

    \60\ Centers for Medicare & Medicaid Services. June 17, 2022. 
Meaningful Measures 2.0: Moving from Measure Reduction to 
Modernization. https://www.cms.gov/medicare/meaningful-measures-framework/meaningful-measures-20-moving-measure-reduction-modernization.
    \61\ Centers for Medicare & Medicaid Services. May 1, 2023. CMS 
National Quality Strategy. https://www.cms.gov/medicare/quality-initiatives-patient-assessment-instruments/value-based-programs/cms-quality-strategy.
    \62\ U.S. Department of Health and Human Services. May 9, 2023. 
Fact Sheet: End of the COVID-19 Public Health Emergency. https://
www.hhs.gov/about/news/2023/05/09/fact-sheet-end-of-the-covid-19-
public-health-
emergency.html#:~:text=That%20means%20with%20the%20COVID,the%20expira
tion%20of%20the%20PHE.
---------------------------------------------------------------------------

    Comment: One commenter requested that CMS clarify whether the 
elimination of vaccine ``mandates'' will impact the adoption or use of 
the proposed HCP COVID-19 Vaccine measure.
    Response: We clarify that the vaccination requirements under Sec.  
482.42(g) (which have now been lifted), are separate from IRF QRP 
requirements to report HCP COVID-19 vaccination data. Even though the 
PHE has ended and vaccination requirements have been lifted, CMS 
intends to encourage ongoing COVID-19 vaccination through use of its 
quality reporting programs (88 FR 36487). One way to encourage patient 
safety and COVID-19 vaccination is through adoption of the modified up 
to date numerator definition of the HCP COVID-19 Vaccine measure. 
Despite the White House's announcement,\63\ the IRF QRP still requires 
data submission of the HCP COVID-19 Vaccine measure to the NHSN for 
IRFs to remain in compliance with the IRF QRP. However, since the IRF 
QRP is a pay-for-reporting program, HCP COVID-19 vaccination is not 
mandated by this measure.
---------------------------------------------------------------------------

    \63\ The White House. May 1, 2023. The Biden-Harris 
Administration Will End COVID-19 Vaccination Requirements for 
Federal Employees, Contractors, International Travelers, Head Start 
Educators, and CMS-Certified Facilities. https://www.whitehouse.gov/briefing-room/statements-releases/2023/05/01/the-biden-administration-will-end-covid-19-vaccination-requirements-for-federal-employees-contractors-international-travelers-head-start-educators-and-cms-certified-facilities/.
---------------------------------------------------------------------------

    Comment: A number of commenters expressed concerns with the 
evolving nature of the measure's up to date numerator definition, and 
believe that the reliability and validity of the measure may be 
negatively impacted if the up to date definition were to change 
frequently. Several of these commenters raised concerns with the 
potential inaccuracy of the measure since the term up to date could be 
revised between reporting periods or in the middle of a reporting 
period. One of these commenters suggested the definition will quickly 
and frequently become outdated, and another commenter believes the 
science is still emerging and it is too soon to adopt a revised 
definition for the HCP COVID-19 vaccine. Finally, several commenters 
believed that the current specifications are flawed given the lack of a 
stable definition of the up to date numerator definition.
    Response: We recognize that the up to date COVID-19 vaccination 
definition may evolve due to the changing nature of the virus. Since 
the adoption of the current version of the measure, the public health 
response to COVID-19 has necessarily adapted to respond to the changing 
nature of the virus's transmission and community spread. As mentioned 
in the FY 2022 IRF PPS final rule (86 FR 42362), we received several 
public comments during the current measure's pre-rulemaking process 
encouraging us to continue to update the measure as new evidence on 
COVID-19 continues to arise and we stated our intention to continue to 
work with partners including FDA and CDC to consider any updates to the 
measure in future rulemaking as appropriate. We believe that the 
proposed modification to this measure aligns with our responsive 
approach to COVID-19 and will continue to support vaccination as the 
most effective means to prevent the worst consequences of COVID-19, 
including severe illness, hospitalization, and death.
    In response to the commenter's concerns that the up to date 
numerator definition may evolve, we refer commenters to section 
IX.C.1.a.4. of this final rule where we explained that providers would 
refer to the definition of up to date as the first day of the quarter, 
which can be found at the following CDC NHSN web page: https://www.cdc.gov/nhsn/pdfs/hps/covidvax/UpToDateGuidance-508.pdf. The CDC 
notes that this aforementioned document will be updated quarterly to 
reflect any changes as COVID-19 guidance evolves, and notes that 
providers should use the definitions for the reporting period 
associated with the reporting weeks included in data submission. At the 
beginning of each reporting period and before collecting or submitting 
data on this modified measure, IRFs must refer to the aforementioned 
document to determine the then-applicable definition of up to date to 
apply when collecting data on the vaccination status of HCP for that 
quarterly reporting period. As such, the up to date vaccination 
definition during a particular reporting period would not change, and 
each provider will be measured against the same criteria within the 
same quarter. If the requirements do change from one quarter to the 
next, IRFs would have the up to date definition at the beginning of the 
quarter (using the aforementioned CDC NHSN web page) and have a minimum 
of 3 weeks to assess whether their HCP meet the definition of up to 
date before submitting HCP COVID-19 Vaccine measure data during the 
self-selected week of a corresponding month. We will continue to 
monitor all measures to identify any concerning trends as part of our 
routine monitoring activities to regularly assess measures performance, 
reliability, and reportability for all data submitted for the IRF QRP.
    Comment: Several commenters also suggested that the proposed 
modification to the measure numerator would be administratively 
burdensome due to the time it will take to (1) stay abreast of the 
current definition of up to date and (2) track whether their HCP met 
that definition at a time when IRFs are dealing with workforce issues. 
One commenter stated that given the current workforce shortage, adding 
more requirements on the healthcare workforce and health care systems 
will only exacerbate the situation. Another commenter said that 
healthcare facilities that are currently voluntarily reporting data to 
the CDC using the new up to date definition find the collection process 
quite administratively burdensome. Many commenters were concerned that 
frequent changes to the

[[Page 51006]]

definition of up to date would increase administrative burden for IRFs 
because they would have to alter their data collection processes to 
ensure that they report the proper data on HCP vaccination.
    Response: We appreciate commenters' concerns regarding the 
reporting of the measure, but disagree that the proposed up to date 
numerator definition for the HCP COVID-19 Vaccine measure may 
exacerbate workforce shortages. We believe that the risks associated 
with COVID-19 warrant direct attention, especially because HCP are 
working directly with, and in close proximity to, patients. IRFs have 
been reporting the current version of the measure since the measure's 
initial data submission period (October 1, 2021 through December 31, 
2021), and we believe that there has been sufficient time to allocate 
the necessary resources required to report this measure. We note that 
for purposes of NHSN surveillance, the CDC used the up to date 
numerator definition during the Quarter 4 2022 surveillance period 
(September 26, 2022 through December 25, 2022) (88 FR 20905) and IRFs 
have been successfully reporting the measure in alignment with the 
proposed modifications.
    The CDC provides frequent communications and education to support 
IRFs' understanding of the latest guidelines. CDC posts an updated 
document approximately 2 weeks before the start of a new reporting 
quarter. If there are any changes to the definition, forms, etc., CDC 
will host a webinar in the 1-2 weeks before the beginning of a new 
reporting quarter. If IRFs have any concerns they would like to address 
with CMS regarding the data submission of this measure, they can voice 
their concerns during CMS' Hospitals Open Door Forums (ODFs). For more 
information on ODFs and to sign up for email notifications, we refer 
readers to the following CMS web page: https://www.cms.gov/outreach-and-education/outreach/opendoorforums/odf_hospitals.
    Comment: One commenter questioned whether HCP without booster(s) 
would be mandated to get booster(s) if the proposed measure were 
adopted. Two commenters were concerned that because the proposed 
reporting requirements are inconsistent with internal, State, and 
Federal policies for vaccination, it will lead to inaccurate reporting.
    Response: The current HCP COVID-19 Vaccine measure in the IRF QRP 
does not require HCP to receive a COVID-19 vaccine and the proposed 
modification to the measure numerator definition would not mandate HCP 
to receive an additional/booster dose under the up to date definition 
for this measure. It is an IRF's responsibility to determine its own 
personnel policies. The HCP COVID-19 Vaccine measure only requires 
reporting of vaccination rates for an IRF to successfully participate 
in the IRF QRP. As we have described previously, the CDC posts an 
updated document approximately 2 weeks before the start of a new 
reporting quarter. If there are any changes to the definition, forms, 
etc., CDC will host a webinar in the 1-2 weeks before the beginning of 
a new reporting quarter. It is the IRF's responsibility to accurately 
report vaccination status of HCP in accordance with this measure's 
specifications.
    Comment: One commenter noted that the CDC's vaccination guidance 
suggests that some individuals with certain risk factors should 
consider receiving an additional booster dose within four months of 
receiving their first bivalent dose. Yet, the commenter noted that IRFs 
usually do not have routine access to data to know which of their HCP 
may need an additional booster. The commenter was concerned that, in 
order to collect accurate data, IRFs would have to obtain permission to 
inquire and attain information on each individual HCP's underlying 
health risk factors and a mechanism to keep the data fully secure. As a 
result, they express concern that the resource intensiveness of 
collecting data under the CDC's current definitions for the HCP COVID-
19 Vaccine measure may outweigh its value.
    Response: IRFs have been engaging with their staff since October 1, 
2022 when the data collection for the HCP COVID-19 Vaccine measure 
began. This proposed modification to the HCP COVID-19 Vaccine measure 
should not require any changes to how IRFs currently engage with their 
staff and administer a comprehensive vaccine administration strategy. 
Specifically, we note that considerations for individuals with certain 
risk factors, such as those who are immunocompromised, are not impacted 
by the modification proposed to this measure as these considerations 
are present with the primary vaccination series for the current HCP 
COVID-19 Vaccine measure. As emphasized in the CDC NHSN ``COVID-19 
Vaccination Modules: Understanding Key Terms and Up to Date 
Vaccination'' web page https://www.cdc.gov/nhsn/pdfs/hps/covidvax/UpToDateGuidance-508.pdf referred to in section IX.C.1.a.4. of this 
final rule, the NHSN surveillance definition for up to date is 
currently the same for all HCP regardless of immunocompromised status.
    Comment: One commenter acknowledged that even though the proposed 
modification to this measure does not mandate HCP become up to date 
with their COVID-19 vaccine, it may affect how providers approach 
vaccination requirements for their workforce. They are concerned that 
entry-level workers will choose to work in other areas of commerce 
without similar COVID-19 vaccination requirements.
    Response: We clarify that the HCP COVID-19 Vaccine measure does not 
require providers to adopt mandatory vaccination policies, and note 
that it is an IRFs' responsibility to determine its own personnel 
policies. The proposed modified HCP COVID-19 Vaccine measure would only 
require reporting of HCP vaccination rates, which would then be 
publicly reported on CMS' Care Compare website. We believe that the 
risks associated with COVID-19 warrant direct attention, especially 
because HCP are working directly with, and in close proximity to, 
patients. To support a comprehensive vaccine administration strategy, 
we encourage IRFs to voluntarily engage in the provision of appropriate 
and accessible education and vaccine-offering activities. Many IRFs 
across the country are educating staff, patients, and patients' 
representatives, participating in vaccine distribution programs, and 
voluntarily reporting up to date vaccine administration.
    Comment: One commenter questioned whether the measure would be a 
comparison of the number of HCP with a primary series only and the 
number of HCP with a primary series and booster doses.
    Response: We interpret the commenter's response as asking whether 
the measure would compare an IRF's HCP's primary series vaccination 
rate to an IRF's performance on the modified version of the HCP COVID-
19 Vaccine measure. The modification to the HCP COVID-19 Vaccine 
measure does not make a comparison between the two HCP groups. Rather, 
the measure assesses the ratio between the number of HCP who are 
considered up to date on their COVID-19 vaccinations with the total 
number of HCP eligible to work in the facility for at least one day 
during the reporting period.
    Comment: Several commenters did not support the HCP COVID-19 
quality measure since it does not exclude HCP who choose not to receive 
up to date vaccinations due to personal or religious beliefs. Four of 
these commenters suggested we align the measure's exclusion criteria 
with the Hospital Conditions of Participation (CoPs)

[[Page 51007]]

requirement from the interim final rule ``Medicare and Medicaid 
Programs; Omnibus COVID-19 Health Care Staff Vaccination'' (86 FR 
61555), which allowed exclusions for religious exemptions.\64\ One of 
these commenters recommended that CMS develop an additional exclusion 
for this measure to account for sincerely held religious beliefs in 
order to align with Office of Civil Rights guidance.
---------------------------------------------------------------------------

    \64\ Conditions of Participation requirements from the interim 
final rule ``Medicare and Medicaid Programs; Omnibus COVID-19 Health 
Care Staff Vaccination'' (86 FR 61555) are no longer in effect due 
to the ``Medicare and Medicaid Programs; Policy and Regulatory 
Changes to the Omnibus COVID-19 Health Care Staff Vaccination 
Requirements; Additional Policy and Regulatory Changes to the 
Requirements for Long-Term Care (LTC) Facilities and Intermediate 
Care Facilities for Individuals With Intellectual Disabilities 
(ICFs-IID) To Provide COVID-19 Vaccine Education and Offer 
Vaccinations to Residents, Clients, and Staff; Policy and Regulatory 
Changes to the Long Term Care Facility COVID-19 Testing 
Requirements'' final rule (88 FR 36485).
---------------------------------------------------------------------------

    Additionally, one commenter noted that even though the current 
version of the HCP COVID-19 Vaccine measure excludes persons with 
medical contraindications from the measure's denominator, they believe 
that the exclusion may be inconsistently applied among IRFs and other 
healthcare settings.
    Response: We acknowledge that individual HCP may have sincerely 
held religious beliefs, observances, or practices that would prevent 
them from receiving a vaccine. However, we want to reiterate that 
neither the current version nor the proposed modified version of the 
measure mandate that HCP be up to date on their COVID-19 vaccination. 
The HCP COVID-19 Vaccine measure only requires reporting of vaccination 
rates for successful IRF QRP participation.
    With respect to the comment about exclusions being inconsistently 
applied, CMS has multiple processes in place to ensure reported patient 
data are accurate. State agencies conduct standard certification 
surveys for IRFs, and accuracy and completeness of the IRF-PAI are 
among the regulatory requirements that surveyors evaluate during 
surveys.\65\ Additionally, the IRF-PAI process has multiple regulatory 
requirements. Our regulations at Sec.  412.606(b) require that (1) the 
assessment accurately reflects the patient's status, (2) a clinician 
appropriately trained to perform a patient assessment using the IRF-PAI 
conducts or coordinates each assessment with the appropriate 
participation of health professionals, and (3) the assessment process 
includes direct observation, as well as communication with the 
patient.\66\ We take the accuracy of IRF-PAI assessment data very 
seriously, and routinely monitor the IRF QRP measures' performance, and 
will take appropriate steps to address any such issues, if identified, 
in future rulemaking.
---------------------------------------------------------------------------

    \65\ Center for Medicare and Medicaid Services. September 6, 
2022. Hospitals. https:/www.cms.gov/medicare/provider-enrollment-
and-certification/certificationandcomplianc/hospitals.
    \66\ 42 CFR 412.606 https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-B/part-412/subpart-P/section-412.606.
---------------------------------------------------------------------------

    Comment: One commenter suggested the measure needs to be 
restructured given the variation among States as to what information 
can be requested of staff and can be conditions of employment. These 
variations would make the ability to create any national average 
invalid. Another commenter suggested that without a regular cadence of 
boosters or a defined COVID-19 ``season,'' similar to influenza, 
modifying the definition of up to date is premature.
    Response: We acknowledge the commenter's concern regarding how 
State laws may impact an IRF's ability to collect data regarding HCP 
COVID-19 vaccination status in order to report on this measure, and 
note that these Federal requirements would remain regardless of 
fluctuating State requirements. We believe, however, that IRFs 
obtaining information on HCP COVID-19 vaccination status is important 
for determining reasonable measures to protect the health and safety of 
not only the patients whom the IRF serves, but other staff working 
within the facility. We clarify that the HCP COVID-19 Vaccine measure 
does not require providers to adopt mandatory vaccination policies. In 
addition, we recognize that the up to date COVID-19 vaccination 
definition may evolve due to the changing nature of the virus. Since 
the adoption of the current version of the measure, the public health 
response to COVID-19 has necessarily adapted to respond to the changing 
nature of the virus's transmission and community spread. As mentioned 
in the FY 2022 IRF PPS final rule (86 FR 42362), we received several 
public comments during the measure's pre-rulemaking process encouraging 
us to continue to update the measure as new evidence on COVID-19 
continues to arise and we stated our intention to continue to work with 
partners including FDA and CDC to consider any updates to the measure 
in future rulemaking as appropriate. We believe that the proposed 
measure modification aligns with the Administration's responsive 
approach to COVID-19 and will continue to support vaccination as the 
most effective means to prevent the worst consequences of COVID-19, 
including severe illness, hospitalization, and death.
    Comment: One commenter suggested CMS would be able to obtain the 
same information by examining community levels of COVID-19 vaccination.
    Response: This measure reports the vaccination rate among the HCPs 
eligible to work in the facility for at least one day during the 
reporting period, excluding persons with contraindications to COVID-19 
vaccination that are described by the CDC. We disagree that facility-
level HCP vaccination information can be obtained by examining 
community levels of COVID-19 vaccinations since facility-level rates 
could vary within the same community.
    Comment: A number of commenters raised concerns about the frequency 
and manner of data submission. Commenters noted that if the CDC revises 
the up to date definition in the middle of a reporting period, the data 
reported by providers will no longer be an accurate reflection of the 
facility. One commenter recommended CMS should adopt a ``fixed 
definition of vaccine coverage'' for calculating measure performance. 
Commenters noted that, without a single consistent resource for 
reporting instructions when the definition of up to date is revised, 
the risk of inaccurate reporting increases.
    Response: In response to the commenters' concerns that the up to 
date numerator definition may change during the reporting period, we 
refer commenters to section IX.C.1.a.4. of this final rule where we 
discuss how providers should refer to the definition of up to date as 
of the first day of the quarterly reporting period, which can be found 
at the following CDC NHSN web page: https://www.cdc.gov/nhsn/pdfs/hps/covidvax/UpToDateGuidance-508.pdf. The CDC notes that this 
aforementioned document will be updated quarterly to reflect any 
changes as COVID-19 guidance evolves, and notes that providers should 
use the definitions for the reporting period associated with the 
reporting weeks included in data submission. As such, the up to date 
vaccination definition that would be applicable during a particular 
reporting period should not change, which addresses the commenter's 
concern that there be a single consistent resource for reporting 
instructions when the definition of up to date is revised. If the 
requirements do change from one quarter to the next,

[[Page 51008]]

IRFs would have the up to date definition at the beginning of the 
quarter (using the aforementioned CDC NHSN web page), and have a 
minimum of 3 weeks to assess whether their HCP meet the definition of 
up to date before submitting HCP COVID-19 Vaccine measure data during 
the self-selected week of a corresponding month. IRFs would determine 
the up to date definition at the beginning of the quarter (using the 
aforementioned CDC NHSN web page) and would have a minimum of 3 weeks 
to determine whether their staff are up to date on vaccinations before 
submitting HCP COVID-19 Vaccine measure data during the self-selected 
week of a corresponding month.
    We interpret the commenter's recommendation to adopt a ``fixed 
definition of vaccine coverage'' as maintaining only one version of an 
up to date definition indefinitely. We thank the commenter for the 
suggestion. However, we note that in section IX.C.1.a.1.a of this final 
rule that as SARS-CoV-2 evolves, our COVID-19 vaccination strategy must 
remain responsive. When we adopted the HCP COVID-19 Vaccine measure in 
the FY 2022 IRF PPS final rule, we stated that the need for additional/
booster doses of COVID-19 vaccines had not been established and no 
additional doses had been recommended (86 FR 42390). To address the new 
variants of COVID-19, vaccine manufacturers have developed bivalent 
vaccines, which have been shown to increase immune responses to SARS-
CoV-2 variants. We continue to believe that vaccination remains the 
most effective means to prevent severe consequences of COVID-19 and 
feel it is important to update the specifications of the HCP COVID-19 
Vaccine measure to reflect most recent guidance that explicitly 
specifies for HCP to receive primary series and additional/booster 
doses in a timely manner.
    Comment: One commenter questioned if retroactive assessment of data 
will be required if the up to date definition were to change during the 
reporting period.
    Response: If the definition of up to date changes from one quarter 
to the next, IRFs would not have to submit data retroactively.
    Comment: One commenter suggested that if the measure continues to 
be included in the IRF QRP, CMS should reduce the burden of gathering 
data from all personnel captured within the measure's denominator 
population.
    Response: We did not propose changes to the measure denominator and 
disagree that the denominator criteria should be loosened. We emphasize 
that any HCP working in the facility for at least one working day 
during the reporting period, meeting denominator eligibility criteria, 
may come into contact with IRF patients, increasing the risk for HCP to 
patient transmission of infection. Therefore, we believe the measure as 
proposed has the potential to generate actionable data on up to date 
HCP COVID-19 vaccination rates that can be used to target quality 
improvement among IRF providers, including increasing up to date HCP 
COVID-19 vaccination coverage in IRFs, while also promoting patient 
safety and increasing the transparency of quality of care in the IRF 
setting.
    Comment: Two commenters recommended that the HCP COVID-19 Vaccine 
measure's reporting requirements should align more closely to those of 
the HCP Influenza Vaccine measure. One commenter notes that the HCP 
Influenza Vaccine measure does not require providers to track and 
report whether HCP receive up to date vaccinations. A few commenters 
suggested CMS consider limiting the reporting requirement to at least 
one week for each quarter and to work with the CDC to move toward a 
version of the measure that may be reported annually. One of the 
commenters who suggested annual reporting was generally supportive of 
the modification to the measure. Another commenter questioned if HCP 
without booster vaccinations will be mandated to receive boosters, and 
if booster vaccinations will be required annually or seasonally like 
the influenza vaccine.
    Response: As we stated in the FY 2024 IRF PPS proposed rule (88 FR 
20950), the measure developer (the CDC) noted that the model used for 
this measure is based on the Influenza Vaccination Coverage among HCP 
measure (CBE #0431), and it intends to utilize a similar approach for 
the HCP COVID-19 Vaccine measure if vaccination strategy becomes 
seasonal. Neither the current nor proposed modified versions of the HCP 
COVID-19 Vaccine measure mandate that HCPs receive an up to date COVID-
19 vaccine.
    Comment: Six commenters expressed concerns with the delay between 
data submission via the NHSN and public reporting on Care Compare, 
emphasizing that the up to date numerator definition may change between 
the time when data are submitted and when data are publicly reported. 
One commenter points out that it may mean that HCP who counted as up to 
date in a given quarter may no longer be up to date in the next quarter 
and CMS needs to clearly communicate what publicly reported data 
reflect.
    Response: We thank the commenters for expressing their concerns 
about the data lag between data submission and public reporting. We 
clarify that, as mentioned in the FY 2022 IRF PPS final rule (86 FR 
42496 through 42497), we revised our public reporting policy for this 
measure to use quarterly reporting, which allows the most recent 
quarter of data to be displayed, as opposed to an average of four 
rolling quarters. Additionally, the public display schedule of the HCP 
COVID-19 Vaccine measure aligns with IRF QRP public display policies 
finalized in the FY 2017 IRF PPS final rule (81 FR 52055), which allows 
IRFs to submit their IRF QRP data up to 4.5 months after the end of the 
reporting quarter. A number of administrative tasks must then occur in 
sequential order between the time IRF QRP data are submitted and 
reported in Care Compare to ensure the validity of data and to allow 
IRFs sufficient time to appeal any determinations of non-compliance 
with our requirements for the IRF QRP. We believe this reporting 
schedule, outlined in section IX.C.1.a.4. of this final rule is 
reasonable, and expediting this schedule may establish undue burden on 
providers and jeopardize the integrity of the data.
    Additionally, CMS does communicate the time periods that publicly 
reported data reflect. This information can be retrieved through the 
Care Compare site (https://www.medicare.gov/care-compare/) through 
``View Quality Measures,'' and then clicking on ``Get current data 
collection period.''
    Comment: One commenter believed the delay between when the 
information is collected and when it is actually publicly reported 
could cause confusion and damage the public's trust and confidence in 
the quality of care delivered in their community if the rate of up to 
date healthcare personnel vaccination is ``low'' due to the data lag. 
Another commenter noted that changing CDC definitions is challenging 
for health care professionals, and they do not believe that this 
information can be articulated in a manner for patients to fully digest 
in order to make meaningful health care decisions.

[[Page 51009]]

    Response: While we acknowledge that an IRF's percentage of HCP who 
are up to date with their COVID-19 vaccination could change if the CDC 
modifies it guidance, each provider will be measured against the same 
criteria within the same quarter, and the guideline for each quarter 
will be shared through the CDC website ahead of each quarter at the 
following NHSN web page: https://www.cdc.gov/nhsn/pdfs/hps/covidvax/UpToDateGuidance-508.pdf. If the requirements do change from one 
quarter to the next, IRFs would have the up to date definition at the 
beginning of the quarter and have a minimum of 3 weeks to assess 
whether their HCP meet the definition of up to date before submitting 
HCP COVID-19 Vaccine measure data during the self-selected week of a 
corresponding month.
    We also believe patients will be able to understand what changes to 
the up to date definition mean on Care Compare. We note that the public 
has been using the information displayed on Care Compare for the 
current HCP COVID-19 Vaccine measure since it was first publicly 
reported in 2022. CMS works closely with its Office of Communications 
and consumer groups when onboarding measures to the Care Compare 
websites, and we will do the same with the modified HCP COVID-19 
Vaccine measure to ensure that the measure description on Care Compare 
is clear and understandable for the general public.
    Comment: One commenter requested that CMS account for how CMS will 
publicly report the HCP COVID-19 Vaccine measure when the up to date 
definition in the numerator changes. They provide as example using CDC 
data where in the population greater than or equal to 65 years old, 
94.3 percent have completed the primary series (the current measure 
numerator definition), while only 42.6 percent have received a booster 
dose (the proposed measure numerator definition). This commenter does 
not believe that the two numbers should be trended and compared over 
time given that they are different definitions of vaccination.
    Response: We thank the commenter for the question, and we clarify 
that only one FY quarter of data is publicly reported at a time and the 
provider's performance is compared with its peers using data collected 
from the same FY quarter, and thus subject to the same definitions as 
set forth in the measure's guidelines. While the measure is only 
publicly reported one FY quarter at a time, we review measure trends as 
part of our routine monitoring activities and will exercise caution 
when monitoring measure trends especially during time periods when the 
CDC guidelines may change.
    Comment: One commenter inquired about if and where the HCP COVID-19 
Vaccine measure will be reported. This commenter also inquired about if 
facilities with more up to date vaccinations will get higher star-
ratings. Additionally, this commenter questioned if there will be 
additional reimbursement for collecting up to date vaccination rates of 
HCP. Lastly, the commenter inquired about how information about HCP 
vaccine percentages will be aggregated.
    Response: We thank the commenter for their questions. As mentioned 
in section IX.C.1.a.4. of this final rule, the HCP COVID-19 Vaccine 
measure will be publicly reported on Care Compare beginning with the 
September 2024 Care Compare refresh. Additionally, we will make 
available to IRFs a preview of their performance on the HCP COVID-19 
Vaccine measure on the IRF Provider Preview Report, which will be 
issued approximately 3 months prior to displaying the measure on Care 
Compare. In terms of star-ratings, the IRF QRP is not a part of the 
Hospital Quality Star Rating program. Furthermore, we reiterate that 
the IRF QRP is a pay-for-reporting program. Therefore, IRFs will only 
be financially penalized under the IRF QRP if they fail to comply with 
measure data submission requirements. There will not be additional 
reimbursement for collecting up to date vaccination rates of HCP or 
reimbursement based on HCP COVID-19 Vaccine measure performance. In 
response to the commenter's question about how percentages of HCP who 
are up to date with their COVID-19 vaccination will be aggregated, each 
quarter the CDC will calculate a single quarterly HCP COVID-19 
vaccination coverage rate for each facility, by taking the average of 
the data from the three weekly rates submitted by the facility for that 
quarter. If more than 1 week of data are submitted for the month, the 
most recent submitted week of the month will be used. We refer readers 
to the following CDC NHSN web page for additional information: https://www.cdc.gov/nhsn/pdfs/hps/covidvax/protocol-hcp-508.pdf.
    After careful consideration of the public comments we received, we 
are finalizing our proposal to modify the HCP COVID-19 Vaccine measure 
beginning with the FY 2025 IRF QRP as proposed.
b. Discharge Function Score Measure Beginning With the FY 2025 IRF QRP
(1) Background
    IRFs provide rehabilitation therapy in a resource-intensive 
inpatient hospital environment to patients with complex nursing, 
medical management, and rehabilitation needs, who require and can 
reasonably be expected to benefit from the multidisciplinary care 
provided in an IRF. Patients tend to have neurological conditions such 
as stroke, spinal cord injury, and brain injury; degenerative 
conditions including multiple sclerosis; congenital deformities; 
amputations; burns; active inflammatory conditions; severe or advanced 
osteoarthritis; or knee and hip joint replacements.\67\ In 2019, the 
most common condition treated by IRFs was stroke, which accounted for 
about one-fifth of IRF cases.\68\ For stroke patients, rehabilitation 
has been shown to be the most effective way to reduce stroke-associated 
motor impairments. Addressing these impairments is crucial as 
functional deficits affect patients' mobility, their capabilities in 
daily life activities, and their participation in society, which can 
lead to a lower quality of life.\69\
---------------------------------------------------------------------------

    \67\ 42 CFR 412.29.
    \68\ Medicare Payment Advisory Commission. Report to the 
Congress: Medicare and the Health Care Delivery System. June 2021. 
https://www.medpac.gov/wp-content/uploads/import_data/scrape_files/docs/default-source/reports/jun21_medpac_report_to_congress_sec.pdf.
    \69\ Hatem SM, Saussez G, Della Faille M, Prist V, Zhang X, 
Dispa D, Bleyenheuft Y. Rehabilitation of Motor Function After 
Stroke: A Multiple Systematic Review Focused on Techniques to 
Stimulate Upper Extremity Recovery. Front Hum Neurosci. 2016 Sep 
13;10:442. doi: 10.3389/fnhum.2016.00442. PMID: 27679565; PMCID: 
PMC5020059.
---------------------------------------------------------------------------

    Section 1886(j)(7)(F)(ii) of the Act, cross-referencing subsections 
(b), (c), and (d) of section 1899B of the Act, requires CMS to develop 
and implement standardized quality measures from five quality measure 
domains, including the domain of functional status, cognitive function, 
and changes in function and cognitive function, across post-acute care 
(PAC) settings, including IRFs. To satisfy this requirement, we adopted 
the Application of Percent of Long-Term Care Hospital (LTCH) Patients 
with an Admission and Discharge Functional Assessment and a Care Plan 
That Addresses Function (Application of Functional Assessment/Care 
Plan) measure for the IRF QRP in the FY 2016 IRF PPS final rule (80 FR 
47100 through 47111). While this process measure allowed for the 
standardization of functional assessments across assessment instruments 
and facilitated cross-setting data collection, quality

[[Page 51010]]

measurement, and interoperable data exchange, we believe it is now 
topped out \70\ and proposed to remove it in section VIII.C.1.c. of the 
proposed rule. While there are other outcome measures addressing 
functional status \71\ that can reliably distinguish performance among 
providers in the IRF QRP, these outcome measures are not cross-setting 
in nature because they rely on functional status items not collected in 
all PAC settings. In contrast, a cross-setting functional outcome 
measure would align measure specifications across settings, including 
the use of a common set of standardized functional assessment data 
elements.
---------------------------------------------------------------------------

    \70\ Centers for Medicare & Medicaid Services. 2022 Annual Call 
for Quality Measures Fact Sheet, p. 10. https://www.cms.gov/files/document/mips-call-quality-measures-overview-fact-sheet-2022.pdf.
    \71\ The measures include: Change in Self-Care Score for Medical 
Rehabilitation Patients (Change in Mobility for Medical 
Rehabilitation Patients, Discharge Self-Care Score for Medical 
Rehabilitation Patients), Discharge Mobility Score for Medical 
Rehabilitation Patients.
---------------------------------------------------------------------------

(a) Measure Importance
    Maintenance or improvement of physical function among older adults 
is increasingly an important focus of health care. Adults age 65 years 
and older constitute the most rapidly growing population in the United 
States, and functional capacity in physical (non-psychological) domains 
has been shown to decline with age.\72\ Moreover, impaired functional 
capacity is associated with poorer quality of life and an increased 
risk of all-cause mortality, postoperative complications, and cognitive 
impairment, the latter of which can complicate the return of a patient 
to the community from post-acute care.73 74 75 Nonetheless, 
evidence suggests that physical functional abilities, including 
mobility and self-care, are modifiable predictors of patient outcomes 
across PAC settings, including functional recovery or decline after 
post-acute care,76 77 78 79 rehospitalization 
rates,80 81 82 discharge to community,83 84 and 
falls.\85\
---------------------------------------------------------------------------

    \72\ High KP, Zieman S, Gurwitz J, Hill C, Lai J, Robinson T, 
Schonberg M, Whitson H. Use of Functional Assessment to Define 
Therapeutic Goals and Treatment. J Am Geriatr Soc. 2019 
Sep;67(9):1782-1790. doi: 10.1111/jgs.15975. Epub 2019 May 13. PMID: 
31081938; PMCID: PMC6955596.
    \73\ Clouston SA, Brewster P, Kuh D, Richards M, Cooper R, Hardy 
R, Rubin MS, Hofer SM. The Dynamic Relationship between Physical 
Function and Cognition in Longitudinal Aging Cohorts. Epidemiol Rev. 
2013;35(1):33-50. doi: 10.1093/epirev/mxs004. Epub 2013 Jan 24. 
PMID: 23349427; PMCID: PMC3578448.
    \74\ Michael YL, Colditz GA, Coakley E, Kawachi I. Health 
Behaviors, Social Networks, and Healthy Aging: Cross-Sectional 
Evidence from the Nurses' Health Study. Qual Life Res. 1999 
Dec;8(8):711-22. doi: 10.1023/a:1008949428041. PMID: 10855345.
    \75\ High KP, Zieman S, Gurwitz J, Hill C, Lai J, Robinson T, 
Schonberg M, Whitson H. Use of Functional Assessment to Define 
Therapeutic Goals and Treatment. J Am Geriatr Soc. 2019 
Sep;67(9):1782-1790. doi: 10.1111/jgs.15975. Epub 2019 May 13. PMID: 
31081938; PMCID: PMC6955596.
    \76\ Deutsch A, Palmer L, Vaughan M, Schwartz C, McMullen T. 
Inpatient Rehabilitation Facility Patients' Functional Abilities and 
Validity Evaluation of the Standardized Self-Care and Mobility Data 
Elements. Arch Phys Med Rehabil. 2022 Feb 11:S0003-9993(22)00205-2. 
doi: 10.1016/j.apmr.2022.01.147. Epub ahead of print. PMID: 
35157893.
    \77\ Hong I, Goodwin JS, Reistetter TA, Kuo YF, Mallinson T, 
Karmarkar A, Lin YL, Ottenbacher KJ. Comparison of Functional Status 
Improvements Among Patients With Stroke Receiving Postacute Care in 
Inpatient Rehabilitation vs Skilled Nursing Facilities. JAMA Netw 
Open. 2019 Dec 2;2(12):e1916646. doi: 10.1001/
jamanetworkopen.2019.16646. PMID: 31800069; PMCID: PMC6902754.
    \78\ Alcusky M, Ulbricht CM, Lapane KL. Postacute Care Setting, 
Facility Characteristics, and Poststroke Outcomes: A Systematic 
Review. Arch Phys Med Rehabil. 2018;99(6):1124-1140.e9. doi: 
10.1016/j.apmr.2017.09.005. PMID: 28965738; PMCID: PMC5874162.
    \79\ Chu CH, Quan AML, McGilton KS. Depression and Functional 
Mobility Decline in Long Term Care Home Residents with Dementia: a 
Prospective Cohort Study. Can Geriatr J. 2021;24(4):325-331. doi: 
10.5770/cgj.24.511. PMID: 34912487; PMCID: PMC8629506.
    \80\ Li CY, Haas A, Pritchard KT, Karmarkar A, Kuo YF, Hreha K, 
Ottenbacher KJ. Functional Status Across Post-Acute Settings Is 
Associated With 30-Day and 90-Day Hospital Readmissions. J Am Med 
Dir Assoc. 2021 Dec;22(12):2447-2453.e5. doi: 10.1016/
j.jamda.2021.07.039. Epub 2021 Aug 30. PMID: 34473961; PMCID: 
PMC8627458.
    \81\ Middleton A, Graham JE, Lin YL, Goodwin JS, Bettger JP, 
Deutsch A, Ottenbacher KJ. Motor and Cognitive Functional Status Are 
Associated with 30-day Unplanned Rehospitalization Following Post-
Acute Care in Medicare Fee-for-Service Beneficiaries. J Gen Intern 
Med. 2016 Dec;31(12):1427-1434. doi: 10.1007/s11606-016-3704-4. Epub 
2016 Jul 20. PMID: 27439979; PMCID: PMC5130938.
    \82\ Gustavson AM, Malone DJ, Boxer RS, Forster JE, Stevens-
Lapsley JE. Application of High-Intensity Functional Resistance 
Training in a Skilled Nursing Facility: An Implementation Study. 
Phys Ther. 2020;100(10):1746-1758. doi: 10.1093/ptj/pzaa126. PMID: 
32750132; PMCID: PMC7530575.
    \83\ Minor M, Jaywant A, Toglia J, Campo M, O'Dell MW. Discharge 
Rehabilitation Measures Predict Activity Limitations in Patients 
with Stroke Six Months after Inpatient Rehabilitation. Am J Phys Med 
Rehabil. 2021 Oct 20. doi: 10.1097/PHM.0000000000001908. Epub ahead 
of print. PMID: 34686630.
    \84\ Dubin R, Veith JM, Grippi MA, McPeake J, Harhay MO, 
Mikkelsen ME. Functional Outcomes, Goals, and Goal Attainment among 
Chronically Critically Ill Long-Term Acute Care Hospital Patients. 
Ann Am Thorac Soc. 2021;18(12):2041-2048. doi: 10.1513/
AnnalsATS.202011-1412OC. PMID: 33984248; PMCID: PMC8641806.
    \85\ Hoffman GJ, Liu H, Alexander NB, Tinetti M, Braun TM, Min 
LC. Posthospital Fall Injuries and 30-Day Readmissions in Adults 65 
Years and Older. JAMA Netw Open. 2019 May 3;2(5):e194276. doi: 
10.1001/jamanetworkopen.2019.4276. PMID: 31125100; PMCID: 
PMC6632136.
---------------------------------------------------------------------------

    The implementation of interventions that improve patients' 
functional outcomes and reduce the risks of associated undesirable 
outcomes as a part of a patient-centered care plan is essential to 
maximizing functional improvement. For many people, the overall goals 
of IRF care may include optimizing functional improvement, returning to 
a previous level of independence, or avoiding institutionalization. 
Several studies have reported that IRF care can improve patients' motor 
function at discharge for patients with various diagnoses, including 
traumatic brain injury and stroke.86 87 88 89 While patients 
generally improve in all functional domains at IRF discharge, evidence 
has shown that a significant number of patients continue to exhibit 
deficits in the domains of fall risk, gait speed, and cognition, 
suggesting the need for ongoing treatment. Assessing functional status 
as a health outcome in IRFs can provide valuable information in 
determining treatment decisions throughout the care continuum, such as 
the need for rehabilitation services and discharge 
planning,90 91 92 93 as well as

[[Page 51011]]

provide information to consumers about the effectiveness of 
rehabilitation and other IRF services delivered. Because evidence shows 
that older adults experience aging heterogeneously and require 
individualized and comprehensive health care, functional status can 
serve as a vital component in informing the provision of health care 
and thus indicate an IRF's quality of care.94 95
---------------------------------------------------------------------------

    \86\ Evans E, Krebill C, Gutman R, Resnik L, Zonfrillo MR, 
Lueckel SN, Zhang W, Kumar RG, Dams-O'Connor K, Thomas KS. 
Functional Motor Improvement during Inpatient Rehabilitation among 
Older Adults with Traumatic Brain Injury. PM R. 2022 Apr;14(4):417-
427. doi: 10.1002/pmrj.12644. PMID: 34018693; PMCID: PMC8606011.
    \87\ Kowalski RG, Hammond FM, Weintraub AH, Nakase-Richardson R, 
Zafonte RD, Whyte J, Giacino JT. Recovery of Consciousness and 
Functional Outcome in Moderate and Severe Traumatic Brain Injury. 
JAMA Neurol. 2021;78(5):548-557. doi: 10.1001/jamaneurol.2021.0084. 
PMID: 33646273; PMCID: PMC7922241.
    \88\ Li CY, Karmarkar A, Kuo YF, Haas A, Ottenbacher KJ. Impact 
of Self-Care and Mobility on One or More Post-Acute Care 
Transitions. J Aging Health. 2020;32(10):1325-1334. doi: 10.1177/
0898264320925259. PMID: 32501126; PMCID: PMC7718286.
    \89\ O'Dell MW, Jaywant A, Frantz M, Patel R, Kwong E, Wen K, 
Taub M, Campo M, Toglia J. Changes in the Activity Measure for Post-
Acute Care Domains in Persons With Stroke During the First Year 
After Discharge From Inpatient Rehabilitation. Arch Phys Med 
Rehabil. 2021 Apr;102(4):645-655. doi: 10.1016/j.apmr.2020.11.020. 
PMID: 33440132.
    \90\ Harry M, Woehrle T, Renier C, Furcht M, Enockson M. 
Predictive Utility of the Activity Measure for Post-Acute Care `6-
Clicks' Short Forms on Discharge Disposition and Effect on 
Readmissions: A Retrospective Observational Cohort Study. BMJ Open. 
2021;11:e044278. doi: 10.1136/bmjopen-2020-044278. PMID: 33478966; 
PMCID: PMC7825271.
    \91\ Chang FH, Lin YN, Liou TH, Lin JC, Yang CH, Cheng HL. 
Predicting Admission to Post-Acute Inpatient Rehabilitation in 
Patients with Acute Stroke. J Rehabil Med. 2020 Sep 
28;52(9):jrm00105. doi: 10.2340/16501977-2739. PMID: 32924065.
    \92\ Warren M, Knecht J, Verheijde J, Tompkins J. Association of 
AM-PAC ``6-Clicks'' Basic Mobility and Daily Activity Scores With 
Discharge Destination. Phys Ther. 2021 Apr;101(4): pzab043. doi: 
10.1093/ptj/pzab043. PMID: 33517463.
    \93\ Covert S, Johnson JK, Stilphen M, Passek S, Thompson NR, 
Katzan I. Use of the Activity Measure for Post-Acute Care ``6 
Clicks'' Basic Mobility Inpatient Short Form and National Institutes 
of Health Stroke Scale to Predict Hospital Discharge Disposition 
After Stroke. Phys Ther. 2020 Aug 31;100(9):1423-1433. doi: 10.1093/
ptj/pzaa102. PMID: 32494809.
    \94\ Criss MG, Wingood M, Staples WH, Southard V, Miller KL, 
Norris TL, Avers D, Ciolek CH, Lewis CB, Strunk ER. APTA Geriatrics' 
Guiding Principles for Best Practices in Geriatric Physical Therapy: 
An Executive Summary. J Geriatr Phys Ther. 2022 Apr-June;45(2):70-
75. doi: 10.1519/JPT.0000000000000342. PMID: 35384940.
    \95\ Cogan AM, Weaver JA, McHarg M, Leland NE, Davidson L, 
Mallinson T. Association of Length of Stay, Recovery Rate, and 
Therapy Time per Day With Functional Outcomes After Hip Fracture 
Surgery. JAMA Netw Open. 2020 Jan 3;3(1):e1919672. doi: 10.1001/
jamanetworkopen.2019.19672. PMID: 31977059; PMCID: PMC6991278.
---------------------------------------------------------------------------

    We proposed to adopt the Discharge Function Score (DC Function) 
measure \96\ in the IRF QRP beginning with the FY 2025 IRF QRP. This 
assessment-based outcome measure evaluates functional status by 
calculating the percentage of IRF patients who meet or exceed an 
expected discharge function score. We also proposed that this measure 
would replace the topped-out Application of Functional Assessment/Care 
Plan cross-setting process measure. Like the Application of Functional 
Assessment/Care Plan cross-setting process measure, the proposed DC 
Function measure is calculated using standardized patient assessment 
data from the IRF Patient Assessment Instrument (IRF-PAI).
---------------------------------------------------------------------------

    \96\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    The DC Function measure supports our current priorities. 
Specifically, the measure aligns with the Streamline Quality 
Measurement domain in CMS's Meaningful Measures 2.0 Framework in two 
ways. First, the proposed outcome measure could further CMS's objective 
to prioritize outcome measures by replacing the current cross-setting 
process measure (see section VIII.C.1.c. of the proposed rule). This 
proposed DC Function measure uses a set of cross-setting assessment 
items which would facilitate data collection, quality measurement, 
outcome comparison, and interoperable data exchange among PAC settings; 
existing functional outcome measures do not use a set of cross-setting 
assessment items. Second, this measure would add no additional provider 
burden since it would be calculated using data from the IRF-PAI that 
IRFs are already required to collect.
    The proposed DC Function measure would also follow a calculation 
approach similar to the existing functional outcome measures, which are 
endorsed by the CBE, with some modifications.\97\ Specifically, the 
measure (1) considers two dimensions of function \98\ (self-care and 
mobility activities) and (2) accounts for missing data by using 
statistical imputation to improve the validity of measure performance. 
The statistical imputation approach recodes missing functional status 
data to the most likely value had the status been assessed, whereas the 
current imputation approach implemented in existing functional outcome 
measures recodes missing data to the lowest functional status. A 
benefit of statistical imputation is that it uses patient 
characteristics to produce an unbiased estimate of the score on each 
item with a missing value. In contrast, the current approach treats 
patients with missing values and patients who were coded to the lowest 
functional status similarly, despite evidence suggesting varying 
measure performance between the two groups, which can lead to less 
accurate measure performances.
---------------------------------------------------------------------------

    \97\ The existing measures are the IRF Functional Outcome 
Measure: Discharge Self-Care Score for Medical Rehabilitation 
Patients measure (Discharge Self-Care Score), and the Inpatient 
Rehabilitation Facility (IRF) Functional Outcome Measure: Discharge 
Mobility Score for Medical Rehabilitation Patients measures 
(Discharge Mobility Score).
    \98\ Post-Acute Care Payment Reform Demonstration Report to 
Congress Supplement--Interim Report. May 2011. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Downloads/GAGE_PACPRD_RTC_Supp_Materials_May_2011.pdf.
---------------------------------------------------------------------------

(b) Measure Testing
    The measure development contractor used FY 2019 data to conduct 
testing on the DC Function measure to assess validity, reliability, and 
reportability, all of which informed interested parties' feedback and 
Technical Expert Panel (TEP) input (see section VIII.C.1.b.(3) of the 
proposed rule). Validity was assessed for the measure performance, the 
risk adjustment model, face validity, and statistical imputation 
models. Validity testing of measure performance entailed determining 
Spearman's rank correlations between the proposed measure's performance 
for providers with 20 or more stays and the performance of other 
publicly reported IRF quality measures. Results indicated that the 
proposed DC Function measure captures the intended outcome based on the 
directionalities and strengths of correlation coefficients and are 
further detailed below in Table 18.

[[Page 51012]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.069

    Validity testing of the risk adjustment model showed good model 
discrimination as the measure model has the predictive ability to 
distinguish patients with low expected functional capabilities from 
those with high expected functional capabilities.\99\ The ratios of 
observed-to-predicted discharge function score across eligible stays, 
by deciles of expected functional capabilities, ranged from 0.99 to 
1.01. Both the Cross-Setting Discharge Function TEPs and patient-family 
feedback showed strong support for the face validity and importance of 
the proposed measure as an indicator of quality of care (see section 
VIII.C.1.b.(3) of the proposed rule). Lastly, validity testing of the 
measure's statistical imputation models indicated that the models 
demonstrate good discrimination and produce more precise and accurate 
estimates of function scores for items with missing scores when 
compared to the current imputation approach implemented in IRF QRP 
functional outcome measures, specifically the IRF Functional Outcome 
Measure: Change in Self-Care Score for Medical Rehabilitation Patients 
measure (Change in Self-Care Score), the IRF Functional Outcome 
Measure: Change in Mobility Score for Medical Rehabilitation Patients 
measure (Change in Mobility Score), the IRF Functional Outcome Measure: 
Discharge Self-Care Score for Medical Rehabilitation Patients measure 
(Discharge Self-Care Score), and the IRF Functional Outcome Measure: 
Discharge Mobility Score for Medical Rehabilitation Patients measure 
(Discharge Mobility Score).
---------------------------------------------------------------------------

    \99\ ``Expected functional capabilities'' is defined as the 
predicted discharge function score.
---------------------------------------------------------------------------

    Reliability and reportability testing also yielded results that 
support the proposed DC Function measure's scientific acceptability. 
Split-half testing revealed the proposed measure's excellent 
reliability, indicated by an intraclass correlation coefficient value 
of 0.95. Reportability testing indicated high reportability (98 
percent) of IRFs meeting the public reporting threshold of 20 eligible 
stays. For additional measure testing details, we refer readers to the 
document titled Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report.\100\
---------------------------------------------------------------------------

    \100\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

(2) Competing and Related Measures
    Section 1886(j)(7)(D)(i) of the Act and section 1899B(e)(2)(A) of 
the Act require that, absent an exception under section 
1886(j)(7)(D)(i) and 1899B(e)(2)(B) of the Act, measures specified 
under section 1886(j)(7)(D)(ii) of the Act and section 1899B of the Act 
must be endorsed by the CBE with a contract under section 1890(a). In 
the case of a specified area or medical topic determined appropriate by 
the Secretary for which a feasible and practical measure has not been 
endorsed, section 1886(j)(7)(D)(ii) of the Act and section 
1899B(e)(2)(B) of the Act permit the Secretary to specify a measure 
that is not so endorsed, as long as due consideration is given to 
measures that have been endorsed or adopted by a consensus organization 
identified by the Secretary.
    The proposed DC Function measure is not CBE endorsed, so we 
considered whether there are other available measures that: (1) assess 
both functional domains of self-care and mobility in IRFs and (2) 
satisfy the requirement of the Act to develop and implement 
standardized quality measures from the quality measure domain of 
functional status, cognitive function, and changes in function and 
cognitive function across the PAC settings. While the Application of 
Functional Assessment/Care Plan measure assesses both functional 
domains and satisfies the Act's requirement, this current cross-setting 
process measure is not endorsed by a consensus organization and the 
performance on the Application of Functional Assessment/Care Plan 
measure among IRFs is so high and unvarying that this current measure 
does not offer meaningful distinctions in performance. Additionally, 
after review of other measures, we were unable to identify any measures 
endorsed or adopted by a consensus organization for IRFs that meet the 
aforementioned requirements. While the IRF QRP includes CBE endorsed 
outcome measures addressing functional status,\101\ they each assess a 
single domain of function, and are not cross-setting in nature because 
they rely

[[Page 51013]]

on functional status items not collected in all PAC settings.
---------------------------------------------------------------------------

    \101\ The measures include: Change in Self-Care Score for 
Medical Rehabilitation Patients Change in Mobility Score for Medical 
Rehabilitation Patients, Discharge Self-Care Score for Medical 
Rehabilitation Patients, and Discharge Mobility Score for Medical 
Rehabilitation Patients.
---------------------------------------------------------------------------

    Therefore, after consideration of other available measures, we 
found that the exceptions under sections 1886(j)(7)(D)(ii) and 
1899B(e)(2)(B) of the Act apply and proposed to adopt the DC Function 
measure beginning with the FY 2025 IRF QRP. We intend to submit the 
proposed measure to the CBE for consideration of endorsement when 
feasible.
(3) Interested Parties and Technical Expert Panel (TEP) Input
    In our development and specification of this measure, we employed a 
transparent process in which we sought input from interested parties 
and national experts and engaged in a process that allowed for pre-
rulemaking input in accordance with section 1890A of the Act. To meet 
this requirement, we provided the following opportunities for input 
from interested parties: a patient and family/caregiver advocates (PFA) 
focus group, two TEPs, and public comments through a request for 
information (RFI).
    First, the measure development contractor convened a PFA focus 
group, during which patients and caregivers provided support for the 
proposed measure concept. Participants emphasized the importance of 
measuring functional outcomes and found self-care and mobility to be 
critical aspects of care. Additionally, they expressed a strong 
interest in metrics assessing the number of patients discharged from 
particular facilities with improvements in self-care and mobility, and 
their views of self-care and mobility aligned with the functional 
domains captured by the proposed measure. All feedback was used to 
inform measure development efforts.
    The measure development contractor for the DC Function measure 
subsequently convened TEPs on July 14-15, 2021 and January 26-27, 2022 
to obtain expert input on the development of a cross-setting function 
measure for use in the IRF QRP. The TEPs consisted of interested 
parties with a diverse range of expertise, including IRF and PAC 
subject matter knowledge, clinical expertise, patient and family 
perspectives, and measure development experience. The TEPs supported 
the proposed measure concept and provided substantive feedback 
regarding the measure's specifications and measure testing data.
    First, the TEP was asked whether they prefer a cross-setting 
measure that is modeled after the currently adopted Discharge Mobility 
Score and Discharge Self-Care Score measures, or one that is modeled 
after the currently adopted Change in Mobility Score and Change in 
Self-Care Score measures. With the Discharge Mobility Score and Change 
in Mobility Score measures and the Discharge Self-Care Score and Change 
in Self-Care Score measures being both highly correlated and not 
appearing to measure unique concepts, the TEP favored the Discharge 
Mobility Score and Discharge Self-Care Score measures over the Change 
in Mobility Score and Change in Self-Care Score measures and 
recommended moving forward with utilizing the Discharge Mobility Score 
and Discharge Self-Care Score measures' concepts for the development of 
the cross-setting measure.
    Second, in deciding the standardized functional assessment data 
elements to include in the cross-setting measure, the TEP recommended 
removing redundant data elements. Strong correlations between scores of 
functional items within the same functional domain suggested that 
certain items may be redundant in eliciting information about patient 
function and inclusion of these items could lead to overrepresentation 
of a particular functional area. Subsequently, our measure development 
contractor focused on the Discharge Mobility Score measure as a 
starting point for cross-setting development due to the greater number 
of cross-setting standardized functional assessment data elements for 
mobility while also identifying redundant functional items that could 
be removed from a cross-setting functional measure.
    Third, the TEP supported including the cross-setting self-care 
items such that the cross-setting function measure would capture both 
self-care and mobility. Panelists agreed that self-care items added 
value to the measure and are clinically important to function. Lastly, 
the TEP provided refinements to imputation strategies to more 
accurately represent function performance across all PAC settings, 
including the support of using statistical imputation over the current 
imputation approach implemented in existing functional outcome measures 
in the PAC QRPs. We considered all the TEP's recommendations for 
developing a cross-setting function measure, and we applied their 
recommendations where technically feasible and appropriate. Summaries 
of the TEP proceedings titled Technical Expert Panel (TEP) for the 
Refinement of Long-Term Care Hospital (LTCH), Inpatient Rehabilitation 
Facility (IRF), Skilled Nursing Facility (SNF)/Nursing Facility (NF), 
and Home Health (HH) Function Measures Summary Report (July 2021 TEP) 
\102\ and Technical Expert Panel (TEP) for Cross-Setting Function 
Measure Development Summary Report (January 2022 TEP) \103\ are 
available on the CMS Measures Management System (MMS) Hub.
---------------------------------------------------------------------------

    \102\ Technical Expert Panel (TEP) for the Refinement of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) Function Measures Summary Report (July 2021 TEP) is 
available at https://mmshub.cms.gov/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
    \103\ Technical Expert Panel (TEP) for Cross-Setting Function 
Measure Development Summary Report (January 2022 TEP) is available 
at https://mmshub.cms.gov/sites/default/files/PAC-Function-TEP-Summary-Report-Jan2022-508.pdf.
---------------------------------------------------------------------------

    Finally, we solicited feedback from interested parties on the 
importance, relevance, and applicability of a cross-setting functional 
outcome measure for IRFs through an RFI in the FY 2023 IRF PPS proposed 
rule (87 FR 20244). Commenters were supportive of a cross-setting 
functional outcome measure that is inclusive of both self-care and 
mobility items, but also provided information related to potential risk 
adjustment methodologies as well as other measures that could be used 
to capture functional outcomes across PAC settings (87 FR 47070).
(4) Measure Applications Partnership (MAP) Review
    Our pre-rulemaking process includes making publicly available a 
list of quality and efficiency measures, called the MUC List, that the 
Secretary is considering adopting for use in the Medicare program, 
including our quality reporting programs. This allows multi-interested 
parties to provide recommendations to the Secretary on the measures 
included on the list.
    We included the DC Function measure under the IRF QRP in the 
publicly available MUC List for December 1, 2022.\104\ After the MUC 
List was published, the CBE convened MAP received four comments from 
interested parties in the industry on the 2022 MUC List. Two commenters 
were supportive of the measure and two were not. Among the commenters 
in support of the measure, one commenter stated that function scores 
are the most meaningful outcome measure in the IRF setting, as they not 
only assess patient outcomes but also can be used for clinical 
improvement processes. Additionally, this commenter noted the measure's 
good reliability and validity and that the measure is feasible to

[[Page 51014]]

implement. The second commenter supported including the measure in the 
IRF QRP measures we propose through rulemaking.
---------------------------------------------------------------------------

    \104\ Centers for Medicare & Medicaid Services. Overview of the 
List of Measures Under Consideration for December 1, 2022. https://mmshub.cms.gov/sites/default/files/2022-MUC-List-Overview.pdf.
---------------------------------------------------------------------------

    Commenters not in support of the measure raised the following 
concerns: the need for more detailed measure specifications, the 
complexity of calculating the expected discharge score, the measure's 
validity and usability, and the differences in denominator populations 
across PAC settings. We were able to address these concerns during the 
MAP PAC/LTC workgroup meeting held on December 12, 2022. Specifically, 
we clarified that the technical reports include detailed measure 
specifications, and that expected discharge scores are calculated by 
risk-adjusting the observed discharge scores (see section 
VIII.C.1.b.(5) of the proposed rule). We also noted that the measure 
exhibits good validity (see section VIII.C.1.b(1)(b) of the proposed 
rule) and clarified that the wide range of expected scores does not 
indicate poor validity and is consistent with the range of observed 
scores. We also pointed out that the measure is highly usable since it 
is similar in design and complexity to existing function measures and 
its data elements are already in use. Lastly, we explained that the 
denominator population in each measure setting represents the assessed 
population within the setting and the measure satisfies the requirement 
of the Act for a cross-setting measure in the functional status domain.
    Shortly after, several CBE convened MAP workgroups met to provide 
input on the proposed DC Function measure. First, the MAP Health Equity 
Advisory Group convened on December 6-7, 2022. The MAP Health Equity 
Advisory Group did not share any health equity concerns related to the 
implementation of the DC Function measure, and only requested 
clarification regarding measure specifications from the measure 
steward. The MAP Rural Health Advisory Group met on December 8-9, 2022, 
during which two of its members provided support for the DC Function 
measure and other MAP Rural Health Advisory Group members did not 
express rural health concerns regarding the measure.
    The MAP PAC/LTC workgroup met on December 12, 2022 and provided 
input on the proposed DC Function measure. During this meeting, we were 
able to address several concerns raised by interested parties after the 
publication of the MUC List. Specifically, we clarified that the 
expected discharge scores are not calculated using self-reported 
functional goals and are simply calculated by risk-adjusting the 
observed discharge scores (see section VIII.C.1.b.(5) of the proposed 
rule). Therefore, we believe that these scores cannot be ``gamed'' by 
reporting less-ambitious functional goals. We also pointed out that the 
measure is highly usable as it is similar in design and complexity to 
existing function measures and that the data elements used in this 
measure are already in use on the IRF-PAI submitted by IRFs. Lastly, we 
clarified that the DC Function measure is intended to supplement, 
rather than replace, existing IRF QRP measures for self-care and 
mobility and implements improvements on the existing Discharge Self-
Care Score and Discharge Mobility Score measures that make the proposed 
measure more valid and harder to game.
    The MAP PAC/LTC workgroup went on to discuss several concerns with 
the DC Function measure, including (1) whether the measure is cross-
setting due to denominator populations that differ among settings, (2) 
whether the measure would adequately represent the full picture of 
function, especially for patients who may have a limited potential for 
functional gain, and (3) that the range of expected scores was too 
large to offer a valid facility-level score. We clarified that the 
denominator population in each measure-setting represents the assessed 
population within the setting and that the measure satisfies the 
requirement of section 1886(j)(7) of the Act for a cross-setting 
measure in the functional status domain specified under section 
1899B(c)(1) of the Act. Additionally, we noted that the TEP had 
reviewed the item set and determined that all the self-care and 
mobility items were suitable for all settings. Further, we clarified 
that, because the DC Function measure would assess whether a patient 
met or exceeded their expected discharge score, it accounts for 
patients who are not expected to improve. Lastly, we noted that the DC 
Function measure has a high degree of correlation with the existing 
function measures and that the measure exhibits good validity and 
clarified that the wide range of expected scores does not indicate poor 
validity and is consistent with the range of observed scores. The PAC/
LTC workgroup voted to support the staff recommendation of conditional 
support for rulemaking, with the condition that we seek CBE 
endorsement.
    In response to the MAP PAC/LTC workgroup's preliminary 
recommendation, the CBE received two comments in support of the MAP 
PAC/LTC workgroup's preliminary recommendation of conditional support 
for rulemaking. One commenter recommended the DC Function measure under 
the condition that the measure be reviewed and refined such that its 
implementation supports patient autonomy and results in care that 
aligns with patients' personal functional goals. The second commenter 
provided support for the DC Function measure under the condition that 
it produces statistically meaningful information that can inform 
improvements in care processes, while also expressing concern that the 
measure is not truly cross-setting because: (1) the measure utilizes 
different patient populations in each setting-specific denominator, (2) 
the risk-adjustment models use setting-specific covariates, and (3) 
using a single set of cross-setting Section GG self-care and mobility 
function items in our standardized patient assessment instruments is 
not appropriate since the items may not be relevant given the 
differences in each PAC resident/patient population.
    Finally, the MAP Coordinating Committee workgroup convened on 
January 24-25, 2023. At this meeting, one interested party indicated 
their lack of support for the PAC/LTC workgroup's preliminary 
recommendation. The commenter expressed concern that the proposed DC 
Function measure competes with existing self-care and mobility measures 
in the IRF QRP. We noted that we monitor measures to determine whether 
they meet any measure removal factors, set forth in 42 CFR 
413.360(b)(2), and when identified, we may remove such measures through 
the rulemaking process. We noted again that the TEP had reviewed the 
item set and determined that all the self-care and mobility items were 
suitable for all settings. The MAP Coordinating Committee members 
expressed support for our review of existing measures for potential 
removal, as well as for the proposed DC Function measure, favoring the 
implementation of a single, standardized function measure across PAC 
settings. The Coordinating Committee unanimously upheld the workgroup 
recommendation of conditional support for rulemaking. We refer readers 
to the final MAP recommendations titled 2022-2023 MAP Final 
Recommendations.\105\
---------------------------------------------------------------------------

    \105\ 2022-2023 MAP Final Recommendations. https://mmshub.cms.gov/sites/default/files/2022-2023-MAP-Final-Recommendations-508.xlsx.
---------------------------------------------------------------------------

(5) Quality Measure Calculation
    The proposed DC Function measure is an outcome measure that 
estimates the percentage of IRF patients who meet or exceed an expected 
discharge score during the reporting period. The

[[Page 51015]]

proposed measure's numerator is the number of IRF stays with an 
observed discharge function score that is equal to or greater than the 
calculated expected discharge function score. The observed discharge 
function score is the sum of individual function item values at 
discharge. The expected discharge function score is computed by risk-
adjusting the observed discharge function score for each IRF stay. Risk 
adjustment controls for patient characteristics such as admission 
function score, age, and clinical conditions. The denominator is the 
total number of IRF stays with an IRF-PAI record in the measure target 
period (four rolling quarters) that do not meet the measure exclusion 
criteria. For additional details regarding the numerator, denominator, 
risk adjustment, and exclusion criteria, refer to the Discharge 
Function Score for Inpatient Rehabilitation Facilities (IRFs) Technical 
Report.\106\
---------------------------------------------------------------------------

    \106\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    The proposed DC Function measure implements a statistical 
imputation approach for handling ``missing'' standardized functional 
assessment data elements. The coding guidance for standardized 
functional assessment data elements allows for using ``Activity Not 
Attempted'' (ANA) codes, resulting in ``missing'' information about a 
patient's functional ability on at least some items, at admission and/
or discharge, for a substantive portion of IRF patients. Currently, 
functional outcome measures in the IRF QRP use a simple imputation 
method whereby all ANA codes or otherwise missing scores, on both 
admission and discharge records, are recoded to ``1'' or ``most 
dependent.'' Statistical imputation, on the other hand, replaces these 
missing values with a variable based on the values of other, non-
missing variables in the assessment and on the values of other 
assessments which are otherwise similar to the assessment with a 
missing value. Specifically, this proposed DC Function measure's 
statistical imputation allows missing values (that is, the ANA codes) 
to be replaced with any value from 1 to 6, based on a patient's 
clinical characteristics and codes assigned on other standardized 
functional assessment data elements. The measure implements separate 
imputation models for each standardized functional assessment data 
element used in the construction of the discharge score and the 
admission score. Relative to the current simple imputation method, this 
statistical imputation approach increases precision and accuracy and 
reduces the bias in estimates of missing item values. We refer readers 
to the Discharge Function Score for Inpatient Rehabilitation Facilities 
(IRFs) Technical Report \107\ for measure specifications and additional 
details.
---------------------------------------------------------------------------

    \107\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    We invited public comment on our proposal to adopt the DC Function 
measure, beginning with the FY 2025 IRF QRP. The following is a summary 
of the public comments received on our proposal to adopt the DC 
Function measure, beginning with the FY 2025 IRF QRP, and our 
responses:
    Comment: Two commenters supported the addition of the DC Function 
measure to the IRF QRP. One of these commenters agreed that the measure 
is a significant improvement upon existing function measures and notes 
the measure's potential to demonstrate the value of maintenance 
therapy. While supportive of the measure, one commenter believes the 
data sources for certain risk adjustment covariates, such as the Brief 
Interview of Mental Status (BIMS) to assess cognitive function, can be 
improved upon and urges CMS to closely monitor the appropriateness of 
the risk model used to estimate expected discharge scores. Another 
commenter noted that the measure does not impose additional provider 
burden, is an outcome measure rather than a process measure, and 
implements an imputation approach that improves upon the method used in 
the currently adopted Discharge Self-Care Score, Discharge Mobility 
Score, Change in Self-Care Score, and Change in Mobility Score 
measures. Both commenters encouraged continual evaluation of the 
imputation methodology for validity and any unintended negative 
consequences.
    Response: We thank the commenters for their support of the proposed 
measure and agree that the measure improves upon existing function 
measures implemented in the IRF QRP. We reevaluate measures implemented 
in the IRF QRP on an ongoing basis to ensure they have strong 
scientific acceptability and appropriately capture the care provided by 
IRFs. This monitoring includes the appropriateness and performance of 
both the risk models and imputation models used to calculate the 
measure. We also agree that the accuracy of the expected discharge 
function score is vital to the measure's performance but disagree that 
the data sources for cognitive function are flawed. As described in the 
FY 2019 IRF PPS final rule (83 FR 38544) and the FY 2020 proposed rule 
(84 FR 17294-17295), the cognitive items including the expression of 
ideas and wants, understanding verbal and non-verbal content, and the 
Brief Interview of Mental Status (BIMS) items have been thoroughly 
tested and have been shown to be valid. The reliability of these 
cognitive items was tested in the IRF setting through kappa statistics. 
Results indicated that most kappa values were above 0.60, which 
indicates strong reliability.\108\
---------------------------------------------------------------------------

    \108\ The Development and Testing of the Continuity Assessment 
Record and Evaluation (CARE) Item Set: Final Report on Reliability 
Testing Volume 2 of 3 https://www.cms.gov/files/document/development-and-testing-continuity-assessment-record-and-evaluation-care-item-set-final-report.pdf.
---------------------------------------------------------------------------

    Comment: One commenter who supported the measure requested a 
simplified overview of the risk adjustment methodology, as this would 
enable clinicians to provide more meaningful feedback in future years 
and also serve to alleviate clinician fear associated with an unknown 
measurement of the quality of care they provide.
    Response: We agree that it is important for clinicians to 
understand the proposed quality measure, and thus provided detailed 
specifications to ensure transparency with respect to the measure's 
calculation, including the risk adjustment methodology. At a high 
level, the `expected' discharge score is calculated by risk-adjusting 
the observed discharge score (that is, the sum of individual function 
item values at discharge) for admission functional status, age, and 
clinical characteristics using an ordinary least squares linear 
regression model. The model intercept and risk adjustor coefficients 
are determined by running the risk adjustment model on all eligible IRF 
stays. For more detailed measure specifications, we direct readers to 
the document titled Discharge Function Score for Inpatient 
Rehabilitation Facilities (IRFs) Technical Report.\109\
---------------------------------------------------------------------------

    \109\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    Comment: One commenter supported the proposed adoption of the DC 
Function measure, noting its importance as a patient-centered measure. 
However, this commenter strongly encouraged CMS to submit the measure 
for CBE endorsement.
    Response: We thank the commenter for their support and agree it is 
an important patient-centered measure. We

[[Page 51016]]

intend to submit the proposed measure to the CBE for consideration of 
endorsement when feasible.
    Comment: One commenter supported the proposed measure as it 
captures both self-care and mobility items and encouraged the review 
and refinement of the measure as needed. However, this commenter 
preferred separate quality measures for self-care and mobility to 
ensure each setting is able to capture the items most relevant to its 
patient population needs and goals and use the measures to determine 
meaningful quality improvement activities.
    Response: We thank the commenter for their support and agree with 
the importance of capturing both self-care and mobility items in the 
proposed measure, and for this reason, the Discharge Self-Care Score 
and Discharge Mobility Score measures are not proposed for removal. As 
with all other measures, we will routinely monitor this measure to 
ensure the measure maintains strong scientific acceptability and 
utility to PAC settings.
    Comment: Several commenters did not support the adoption of this 
proposed measure because it lacks CBE endorsement or has not undergone 
the CBE endorsement process. Three of these commenters noted that the 
CBE endorsement process provides information on whether or not the 
measure provides valuable information that can be used to inform 
improvements in care. Two other commenters pointed out that the measure 
received a MAP recommendation of ``conditional support for rulemaking 
pending endorsement by a consensus-based entity'' and believe there 
should be a discussion about competing measures, since the Discharge 
Self-Care Score and Discharge Mobility Score measures in the IRF QRP 
are CBE endorsed.
    Response: We direct readers to section IX.C.1.b.(2) of this final 
rule, where we discuss this topic in detail. Despite the current 
absence of CBE endorsement for this measure, we still believe it is 
important to adopt the DC Function measure into the IRF QRP because, 
unlike the Discharge Self-Care Score and Discharge Mobility Score 
measures, the DC Function measure relies on functional status items 
collected on the IRF-PAI and in all PAC settings, satisfies requirement 
of a cross-setting quality measure set forth in sections 
1886(j)(7)(F)(ii) and 1899B(c)(1)(A) of the Act, and assesses both 
domains of function. We also direct readers to section IX.C.1.b.(2) of 
this final rule, where we discuss measurement gaps that the DC Function 
measure fills in relation to competing and related measures. We also 
acknowledge the importance of the CBE endorsement process and plan to 
submit the proposed measure for CBE endorsement in the future. We 
direct readers to section IX.C.1.b.(1)(b) of this final rule, and the 
technical report for detailed measure testing results demonstrating 
that the measure provides meaningful information which can be used to 
improve quality of care, and to the TEP report summaries 
110 111 which detail TEP support for the proposed measure 
concept.
---------------------------------------------------------------------------

    \110\ Technical Expert Panel (TEP) for the Refinement of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) Function Measures Summary Report (July 2021 TEP). 
https://mms-test.battelle.org/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
    \111\ Technical Expert Panel (TEP) for Cross-Setting Function 
Measure Development Summary Report (January 2022 TEP). https://mmshub.cms.gov/sites/default/files/PAC-Function-TEP-Summary-Report-Jan2022-508.pdf.
---------------------------------------------------------------------------

    Comment: A few commenters oppose the adoption of this proposed 
measure, claiming that it is duplicative of the Discharge Self-Care 
Score and Discharge Mobility Score currently in the IRF QRP. They 
believe the adoption of the proposed measure will create confusion 
among clinicians, patients, and payers who review publicly displayed 
quality measure information. Two of these commenters added that if the 
DC Function Score measure is adopted, then the Discharge Self-Care 
Score and Discharge Mobility Score measures should be removed.
    Response: We disagree that the proposed measure is duplicative of 
the Discharge Self-Care Score and Discharge Mobility Score measures and 
believe all three measures add value to the IRF QRP measure set. As 
discussed in section IX.C.1.b.(2) of this final rule, the Discharge 
Self-Care Score and Discharge Mobility Score measures are not cross-
setting because they rely on functional status items not collected in 
all PAC settings and thus do not satisfy requirement of a cross-setting 
quality measure set forth in sections 1886(j)(7)(F)(ii) and 
1899B(c)(1)(A) of the Act. In contrast, the DC Function measure does 
include functional status items collected in each of the four PAC 
settings. Moreover, the DC Function measure captures information that 
is distinct from the Discharge Self-Care Score and Discharge Mobility 
Score measures. Specifically, the DC Function measure considers both 
dimensions of function (utilizing a subset of self-care and mobility GG 
items in the IRF-PAI), while the Discharge Self-Care Score and 
Discharge Mobility Score measures each consider one dimension of 
function (utilizing all self-care or mobility GG items, respectively). 
We intend for IRFs to use information from the DC Function measure and 
the Discharge Self-Care Score and Discharge Mobility Score measures 
when assessing functional areas that may be opportunities for 
improvement.
    Comment: Several commenters opposed the proposed DC Function 
measure because it combines self-care and mobility items collected on 
the IRF-PAI. Five of these commenters expressed a preference toward the 
Discharge Self-Care Score and Discharge Mobility Score measures 
currently adopted in the IRF QRP because they reflect the two 
dimensions of function separately. These five commenters believe a 
composite measure may disadvantage certain patient populations. The 
same commenters suggested that patients with limited function in their 
lower extremities may have more difficulty improving mobility while a 
patient with limited function in their upper extremities may have more 
difficulty improving self-care.
    Response: The DC Function measure is intended to summarize several 
cross-setting functional assessment items while meeting the 
requirements of sections 1886(j)(7)(F) and 1899B(c)(1)(A) of the Act. 
We agree with the commenters that the individual Discharge Self-Care 
Score and Discharge Mobility Score measures will continue to be useful 
to assess care quality in these dimensions, and for this reason, these 
two measures are not proposed for removal. Providers will be able to 
use information from both the DC Function measure and the Discharge 
Self-Care Score and Discharge Mobility Score measures when determining 
which functional areas may be opportunities for improvement. Moreover, 
we disagree that patients with lower functional performance on either 
self-care or mobility items will be disadvantaged in the proposed 
measure calculations. For each stay included in measure calculations, 
the observed function score is compared to the expected discharge 
score, which is adjusted to account for clinical characteristics, 
admission functional status, and demographic characteristics of the 
patient. Risk adjustment creates an individualized expectation for 
discharge function score for each stay that controls for these factors 
and ensures that each stay is measured against an expectation that is 
calibrated to the patient's individual circumstances when determining 
the numerator for each IRF.
    Comment: Several commenters stated that the DC Function measure has 
not

[[Page 51017]]

been tested, such as testing for reliability, validity, or feasibility.
    Response: We direct readers to section IX.C.1.b.(1)(b) of this 
final rule, where we discuss extensively the testing of the proposed DC 
Function measure. Testing demonstrated good validity for the measure 
performance, the risk adjustment model, face validity, and statistical 
imputation models; excellent reliability; and high reportability. The 
proposed measure would be calculated using data from the IRF-PAI that 
are already reported to the Medicare program for payment and quality 
reporting purposes and are therefore feasible to implement and require 
no additional provider burden. Additionally, we direct readers to 
section IX.C.1.b.(1)(b) of this final rule and to the Discharge 
Function Score for IRFs Technical Report \112\ for detailed measures 
testing results that support that the measure provides meaningful 
information which can be used to improve quality of care, as well as 
the TEP report summaries 113 114 which detail TEP support 
for the proposed measure concept.
---------------------------------------------------------------------------

    \112\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
    \113\ Technical Expert Panel (TEP) for the Refinement of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) Function Measures Summary Report (July 2021 TEP) is 
available at https://mms-test.battelle.org/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
    \114\ Technical Expert Panel (TEP) for Cross-Setting Function 
Measure Development Summary Report (January 2022 TEP) is available 
at https://mmshub.cms.gov/sites/default/files/PAC-Function-TEP-Summary-Report-Jan2022-508.pdf.
---------------------------------------------------------------------------

    Comment: Several commenters oppose the adoption of the DC Function 
measure because they do not believe it is appropriate or accurate for 
CMS to override the clinical judgement of the clinicians who are 
treating the patient by using statistical imputation to impute a value 
to a data element when an ANA code is used. Two of these commenters 
noted that the ANA codes allow clinicians to use their professional 
judgement when certain activities should not or could not be safely 
attempted by the patient, which may be due to medical reasons. 
Additionally, two of these commenters stated that among some patients 
not able to attempt certain self-care and mobility tasks at the time of 
admission, the use of ANA codes decreases significantly at the time of 
discharge, which they believe reflect the functional outcomes achieved 
during their IRF stay. One of these commenters additionally noted that 
a patient who cannot attempt an activity due to medical or safety 
concerns is considered dependent for that activity at that time.
    Response: We acknowledge that the ANA codes allow clinicians to use 
their professional judgement when certain activities should not or 
could not be attempted safely by the patient and that there may be 
medical reasons that a patient cannot safely attempt a task. We note 
that we did not propose any changes to the coding guidance for using 
ANA codes, and we would not expect IRF coding practices to change. 
However, we want to clarify that utilizing statistical imputation to 
calculate a quality measure does not override the clinical judgement of 
clinicians who are expected to continue determining whether certain 
activities can be safely attempted by patients at the time of admission 
and discharge and utilize that information to determine appropriate 
goals and treatment interventions for their IRF patients. Rather, 
statistical imputation is a component in measure calculation of 
reported data and improves upon the imputation approach currently 
implemented in the Change in Mobility Score, Change in Self-Care Score, 
Discharge in Mobility Score, and Discharge in Self-Care Score measures. 
In these currently adopted measures, ANA codes are always imputed to 1 
(dependent) when calculating the measure scores, regardless of a 
patient's own clinical and functional information. However, the 
imputation approach implemented in the proposed DC Function measure 
uses each patient's available functional and clinical information to 
estimate each ANA value had the item been completed. Testing 
demonstrates that, relative to the current simple imputation method, 
the statistical imputation approach used in this DC Function measure 
increases precision and accuracy and reduces bias in estimates of 
missing item values.
    Comment: Two commenters stated that clinicians do not have the 
autonomy to choose whether walk items or wheelchair items are the most 
appropriate choice for the patient at discharge. To illustrate this 
point, these commenters provided an example to show how the measure 
logic may not be equitable for walk patients versus wheelchair 
patients. The example states that if a patient walks 10 feet 
dependently because a second helper assists with a wheelchair due to 
poor balance and will use a wheelchair full time after discharge, then 
the patient's risk-adjusted expected outcomes would be based on their 
ability to walk, since a score was coded for Walk 10 feet on admission 
or discharge.
    Response: We disagree that clinicians do not have the autonomy to 
choose whether walk or wheelchair items should be assessed for a 
patient at discharge. Clinicians are expected to use their clinical 
judgement when determining whether certain activities can be safely 
attempted by the patients when completing the IRF-PAI, reporting ANA 
codes in measure data, and utilizing the assessment data to determine 
appropriate goals for their IRF patients. With respect to the example 
provided, we would like to point out that the use of walk and 
wheelchair items in the calculation of measure outcomes is similar to 
that of the existing Discharge Mobility measure: namely, wheelchair 
items are used only if walk items were coded as ANA at both admission 
and discharge, in order to maximize the use of walk item scores 
whenever they are available, including for patients who are scored on 
both walk and wheelchair items. Both the DC Function and Discharge 
Mobility Score measures would use the information about the patient's 
dependent walking at admission. The Discharge Mobility measure would 
then impute the lowest score (``dependent'') to the ANA walking items 
at discharge, while the DC Function measure may impute a higher score 
to those items, based on the clinical and functional covariates for 
that patient.
    Comment: Some commenters expressed concerns regarding the 
bootstrapping samples used during the development of the DC Function 
measure imputation model because they believe these samples are not 
representative of the full IRF population. These commenters believe the 
validity testing of the proposed DC Function imputation model is not 
accurate because the models are built using only the functional 
abilities of patients who had no Section GG items on the IRF-PAI coded 
ANA, and they believe this comprises a small percentage of the IRF 
population and exhibits clinical, demographic, and functional 
characteristics that likely differ from those of the entire IRF 
population. As such, two of these commenters stated that these 
imputation models should not be imposed on patients who had ANA 
assessments, as doing so could lead to unfair penalization of IRF 
providers treating certain patient populations and performance scores 
that are not representative of true functional gains achieved by 
patients during an IRF stay. Another one of four commenters further 
suggested that the current model of

[[Page 51018]]

imputing ANA patients as dependent on that functional item is likely 
more representative of a patient's functional capabilities than the 
statistical imputation approach, as a patient who is unable to complete 
an activity would be viewed as ``dependent'' for purposes of that 
activity's assessment at that time. This same commenter recommended for 
CMS to release more demographic data of the patient population that the 
bootstrapping model utilizes to understand if this population is truly 
representative of IRF patients.
    Response: We would like to clarify that bootstrapping samples were 
used only to determine validity of the imputation models; to develop 
the imputation models themselves, all stays without ANAs for each 
single item were used. As an example, when estimating the imputation 
model for GG0130A admission scores, all stays without ANAs for GG0130A 
at admission (>95 percent of eligible stays) were used. In other words, 
rather than using the relatively small subset of stays without any ANAs 
across all GG items, we used much larger subsets without ANAs on a 
given item. In fact, measure calculations using FY 2021 data utilized 
89-100 percent of stays in each of the discharge imputation models and 
in each of the non-walk/wheelchair admission imputation models. The 
percentage of stays in the walk/wheelchair admission imputation models 
ranges from around 45 percent to 73 percent, which is expected as these 
items have higher rates of skips based on the CMS guidance for coding 
the IRF-PAI. Given that 89-100 percent of samples are utilized in 
almost all the imputation models, the imputation models are, in fact, 
built upon samples that are representative of the IRF population. 
Furthermore, the imputation methodology builds upon the risk-adjustment 
methodology which has been in place for multiple years for existing 
measures. Risk adjustment creates an individualized expectation for the 
discharge function score for each stay that controls for clinical, 
demographic, and function characteristics to ensure that each stay is 
measured against an expectation that is calibrated to the patient's 
individual circumstances. Similarly, imputation creates an 
individualized prediction for each GG item value for each stay based on 
clinical, demographic, and function characteristics to ensure that each 
imputed value is calibrated to the patient's individual circumstances. 
Lastly, testing has indicated that discharge functional abilities of 
patients with ANA codes at admission tended to be higher than those 
coded as dependent at admission. Treating ANAs and dependent scores 
equivalently, as is done in the Discharge Self-Care Score and Discharge 
Mobility Score measures, may disadvantage patients who were truly 
scored as dependent at admission. Statistical imputation allows the DC 
Function measure to address this bias.
    Comment: Two commenters advocated for the release of more data and 
methodology pertaining to the statistical imputation approach. One 
commenter stated that this is the first time CMS is implementing a 
quality measure score with imputed data and that the report is unclear 
in how walk versus wheelchair patients are accounted for in this 
measure when there is an ANA code. This commenter shared results of an 
analysis they conducted on their own data which indicated that the 
sample of patients without an ANA can range from over 60 percent to 
over 90 percent depending on how the model handles dashes and ANA codes 
for walk and wheel patients, and this wide discrepancy shows the 
complexity of developing this measure and in verifying its results. The 
other commenter noted that the statistical imputation approach may 
falsely elevate overall discharge scores, and thus encouraged oversight 
and reporting related to the frequency of use of ANA codes on 
discharge.
    Response: We remind commenters that the four functional outcome 
measures currently used in the IRF QRP are calculated using imputed 
data. The current imputation approach in these four measures is to 
recode all ANA codes to 1 (dependent) for purposes of calculating the 
measure scores, regardless of a patient's reason for receiving IRF 
care, their demographics, or their clinical and functional 
characteristics. In contrast, the imputation approach of the proposed 
DC Function measure uses each patient's available primary reason for 
IRF care, their demographics, and their functional and clinical 
information to estimate each ANA value had the item been completed. 
Testing demonstrates that, relative to the current simple imputation 
method, the statistical imputation approach increases precision and 
accuracy and reduces bias in estimates of missing item values. 
Additionally, we are unsure which report is being referenced and direct 
readers to the document titled Discharge Function Score for Inpatient 
Rehabilitation Facilities (IRFs) Technical Report for more detailed 
measure specifications.\115\
---------------------------------------------------------------------------

    \115\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    We cannot respond to the findings of the analyses performed by the 
commenter since we do not have sufficient information. However, our 
analyses of FY 2021 data have indicated that around 89-100 percent of 
stays are used in each of the discharge imputation models and in each 
of the non-walk/wheelchair admission imputation models. The percentage 
of stays in the walk/wheelchair admission imputation models range from 
around 45 percent to 73 percent, which is expected as these items have 
higher rates of skips based on the CMS guidance in the IRF-PAI.
    Lastly, we disagree that the statistical imputation approach may 
falsely elevate overall discharge scores. The statistical imputation 
approach will in fact reflect more accurate performance scores, as 
indicated by testing results presented pertaining to statistical 
imputation, compared to the current simple imputation method.
    Comment: A few commenters stated that under the statistical 
imputation methodology, a patient's functional status could be recoded 
at a higher level based on ``the most likely score'' of other, 
completely unrelated functional items (for example, oral hygiene and 
the ability to go up and down steps) and reliance on completely 
unrelated functional items to impute function scores is not clinically 
or statistically appropriate.
    Response: We disagree that using a full set of clinical 
characteristics and functional items is not appropriate. The imputation 
models for the proposed DC Function measure use a similar set of 
covariates as the risk adjustment model for the Discharge Self-Care 
Score and Discharge Mobility Score measures which IRFs have been 
reporting since FY 2016. In addition to these covariates, the proposed 
DC Function measure's model adds the available information from all 
available Section GG functional items on the IRF-PAI. While less-
related functional variables are generally less correlated with a given 
item's score, and thus carry less weight in terms of how much they 
influence the imputed value, they still contribute to the overall model 
performance by improving overall model fit and reducing estimation 
error.
    Comment: A few commenters suggested that CMS be more involved with 
clinicians in discussions surrounding the assessment and coding

[[Page 51019]]

of patients rather than using an imputation approach if there is 
concern that ANA codes are not truly reflective of patients' functional 
abilities. One of these commenters also urged CMS to provide additional 
coding guidance for ANA use for the GG items in order to better 
standardize and reduce the use of ANA codes.
    Response: We engaged with PAC providers on more than one occasion. 
As described in section IX.C.1.b.(3) of this final rule, our measure 
development contractor convened two TEPs to obtain expert clinician 
input on the development of the measure. The TEPs consisted of 
interested parties with a diverse range of expertise, including IRF and 
other PAC subject matter knowledge, clinical expertise, and measure 
development experience in PAC settings. As described in the PAC QRP 
Functions TEP Summary Report--March 2022,\116\ panelists agreed that 
the recode approach used in the currently implemented Discharge Self-
Care Score, Discharge Mobility Score, Change in Self-Care Score, and 
Change in Mobility Score measures could be improved upon and reiterated 
that not all ANAs reflect dependence on a function activity. Based on 
the extensive testing results presented to the TEP, a majority of 
panelists favored the statistical imputation over alternative 
methodologies and an imputation method that is more accurate over one 
that is simpler.
---------------------------------------------------------------------------

    \116\ Technical Expert Panel (TEP) for Cross-Setting Function 
Measure Development, January 26-27, 2022 Summary Report. https://mmshub.cms.gov/sites/default/files/PAC-Function-TEP-Summary-Report-Jan2022-508.pdf. Page 20.
---------------------------------------------------------------------------

    Additionally, CMS continually provides training resources to 
providers to give guidance about how to code functional items,\117\ 
including the use of ANA codes.
---------------------------------------------------------------------------

    \117\ Centers for Medicare & Medicaid Services. Inpatient 
Rehabilitation Facility (IRF) Quality Reporting Program (QRP) 
Training. https://www.cms.gov/medicare/quality-initiatives-patient-assessment-instruments/irf-quality-reporting/irf-quality-reporting-training.
---------------------------------------------------------------------------

    Comment: One commenter believed self-care and mobility items in the 
IRF-PAI can be reported as a zero, resulting in the proposed imputation 
approach producing errors or needing to be recoded to a different 
measure; while another commenter sought clarification on measure 
calculations and stated that the DC Function measure calculates a risk 
adjusted ratio of observed to expected scores at discharge for all 
patients over 18 years old that do not meet exclusion criteria. While 
they supported the risk adjustment method, this commenter warned that 
it may give different results than the ``alternative standardization 
risk-adjustment model.''
    Response: The DC Function measure's items are neither recoded to 0 
nor recoded in another measure but are recoded to a value between 1 and 
6. The imputation approach is similar in complexity to the DC Function 
measure's risk adjustment approach, which is modeled after the approach 
in the currently adopted Discharge Self-Care Score, Discharge Mobility 
Score, Change in Self-Care Score, and Change in Mobility Score 
measures. Please reference section IX.C.1.b.(5) of this final rule for 
more information on the proposed imputation approach.
    We agree that it is important for clinicians to understand the 
proposed quality measure, and thus provided detailed specifications to 
ensure transparency with respect to the measure's calculation, 
including the risk-adjustment methodology. To clarify, the DC Function 
measure score is not a ratio. The measure is constructed by calculating 
the number of IRF stays where the expected score is higher than the 
observed score out of total stays. At a high level, the ``expected'' 
discharge score is calculated by risk-adjusting the observed discharge 
score (that is, the sum of individual function item values at 
discharge) for admission functional status, age, and clinical 
characteristics using an ordinary least squares linear regression 
model. The model intercept and risk adjustor coefficients are 
determined by running the risk adjustment model on all eligible IRF 
stays. For more detailed measure specifications, we direct readers to 
the document titled Discharge Function Score for Inpatient 
Rehabilitation Facilities (IRFs) Technical Report.\118\
---------------------------------------------------------------------------

    \118\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    Also, we are unsure of the ``alternative standardization risk-
adjustment model'' this commenter references and would like to clarify 
that the proposed risk adjustment model has undergone validity testing, 
showing good model discrimination as the measure model has the 
predictive ability to distinguish patients with low expected functional 
capabilities from those with high expected functional 
capabilities.\119\
---------------------------------------------------------------------------

    \119\ ``Expected functional capabilities'' is defined as the 
predicted discharge function score.
---------------------------------------------------------------------------

    Comment: One commenter stated that there is no minimum number of 
eligible stays from which to base the imputation method, potentially 
invalidating results.
    Response: We would like to clarify that imputation models are 
estimated using the entire population of eligible stays, and thus 
sample size is not a concern. For additional measure testing details, 
we refer readers to the document titled Discharge Function Score for 
Inpatient Rehabilitation Facilities (IRFs) Technical Report.\120\
---------------------------------------------------------------------------

    \120\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    Comment: One commenter expressed concern with the proposed 
statistical imputation approach utilized in the DC Function measure and 
suggested it might lead to this measure score varying significantly 
from the Discharge Self-Care Score and Discharge Mobility Score 
measures' scores.
    Response: The DC Function measure captures information that is 
distinct from the Discharge Self-Care Score and Discharge Mobility 
Score measures. Specifically, the DC Function measure considers both 
dimensions of function (utilizing a subset of self-care and mobility GG 
items), while the Discharge Self-Care Score and Discharge Mobility 
Score measures each consider one dimension of function (utilizing all 
self-care and mobility GG items, respectively). For these same reasons, 
we expect to see differences in outcome percentages among these three 
measures for reasons unrelated to the imputation approach used.
    Comment: Two commenters believe the measure's imputed and risk-
adjusted expected values will complicate clinicians' ability to review 
and validate information used for public reporting. Another commenter 
stated that the statistical imputation approach is a very complex 
calculation and understanding how performance is impacted may be 
difficult for both IRFs and the public. This commenter urges CMS to 
continuously evaluate this method and its impact impacts across the PAC 
settings.
    Response: The proposed measure uses methods that are similar in 
complexity to CBE-endorsed functional outcome measures that have been 
adopted in the PAC QRP for several years and will be similarly 
specified. As such, understanding performance should be no more 
difficult than understanding the currently adopted Discharge Self-Care 
Score, Discharge Mobility Score, Change in Self-Care Score, and Change 
in Mobility Score measures. As with all other measures, we will 
routinely monitor this measure's performance, including the statistical 
imputation

[[Page 51020]]

approach, to ensure the measure remains valid and reliable.
    Comment: One commenter requested that CMS provide more clarity on 
its imputation approach to recoding, specifically contrasting it with a 
Rasch analysis used in the PAC PPS prototype, to ensure transparency 
and clinical meaningfulness.
    Response: The Rasch analysis in the PAC PPS prototype produces a 
single value to which every single ANA is recoded for a given item 
across all patients and settings. By contrast, under the imputation 
approach for the DC Function measure, we estimate a different recode 
value for each patient, based on their clinical comorbidities, codes on 
all other GG items, and setting. We believe our approach accounts for 
several likely effects: setting-specific coding guidance and practice 
differences; function scores being correlated with clinical 
comorbidities; and functional scores for a given GG item being 
correlated with functional codes on other GG items, particularly on 
``adjacent'' (similar) items. Therefore, we believe recoding ANAs based 
on patients' specific clinical risk and using all available GG item 
scores (codes) is a more valid approach. For more detailed measure 
specifications, we direct readers to the document titled Discharge 
Function Score for Inpatient Rehabilitation Facilities (IRFs) Technical 
Report.\121\
---------------------------------------------------------------------------

    \121\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    Comment: Two commenters expressed concern that the proposed measure 
numerator is not wholly attributed to a facility's quality of care and 
that the calculation of the ``expected'' discharge score is opaque, 
resulting in difficulty for providers to determine the score for which 
they are striving. These commenters further noted that functional goals 
are not based on statistical regression and are identified via 
individual-specific goals related to function, independence, and 
overall health.
    Response: We agree with the commenter that functional goals are 
identified for each patient as a result of an individual assessment and 
clinical decisions, rather than statistics. However, we want to remind 
commenters that the DC Function measure is not calculated using the 
goals identified in clinical process. The ``expected'' discharge score 
is calculated by risk-adjusting the observed discharge score (that is, 
the sum of individual function item values at discharge) for admission 
functional status, age, and clinical characteristics using an ordinary 
least squares linear regression model. The model intercept and risk 
adjustor coefficients are determined by running the risk adjustment 
model on all eligible IRF stays. For more detailed measure 
specifications, we direct readers to the document titled Discharge 
Function Score for Inpatient Rehabilitation Facilities (IRFs) Technical 
Report.\122\ The risk-adjustment model for this measure controls for 
clinical, demographic, and function characteristics to ensure that the 
score fully reflects a facility's quality of care.
---------------------------------------------------------------------------

    \122\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    Comment: One commenter opposed the adoption of the proposed measure 
because this commenter has significant concern with the current 
calculations of the ``expected'' discharge score for the proposed 
measure. This commenter stated that there are identified discrepancies 
in the way that CMS calculates an ``expected'' discharge score for the 
existing Discharge Self-Care Score and Discharge Mobility Score 
measures, calculations are complex, and calculations of the 
``expected'' discharge value for multiple separate function items is 
unclear. As a result, this commenter believed it is premature to 
implement an expanded discharge function score measure and doing so 
will result in serious implementation burdens and technical challenges.
    Response: This commenter noted discrepancies in the way 
``expected'' discharge scores for current functional outcome measures 
are calculated but did not provide additional information regarding the 
discrepancies to which they were referring. CMS is unaware of any 
discrepancies and would require further details in order to respond to 
these concerns. Nonetheless, we believe the proposed measure's 
calculations of the ``expected'' discharge score has strong scientific 
acceptability based on measure testing results, as previously 
discussed. As with all other measures, we will routinely monitor this 
measure's performance, including the issue raised about the calculation 
of ``expected'' discharge scores, to ensure the measure remains valid 
and reliable.
    We would also like to clarify that the ``expected'' discharge score 
is not calculated for each function item separately. Instead, the 
``expected'' discharge score is calculated by risk-adjusting the 
observed discharge score, which is the sum of individual function item 
(observed) values at discharge. For more detailed measure 
specifications, we direct readers to the document titled Discharge 
Function Score for Inpatient Rehabilitation Facilities (IRFs) Technical 
Report.\123\
---------------------------------------------------------------------------

    \123\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    Comment: Several commenters disagreed with language in the proposed 
rule that characterized items coded with an ANA code (codes 07, 09, 10, 
and 88), a dash (-), and a skip ([caret]) as ``missing'' data since CMS 
provides distinct guidance and specifications for each code's use. 
Specifically, these commenters stated that ANA codes represent clinical 
information that the patient was incapable of performing a task for 
reasons specified by CMS in the IRF-PAI manual and thus are not 
considered ``missing data''; because these ANA codes represent clinical 
information, three of these commenters stated that imputation of these 
ANA codes based on other function activities would not improve the 
precision of the score.
    Response: We agree that ANA codes, a dash, and a skip have 
different meanings when used on the IRF-PAI. To clarify, the use of the 
term ``missing'' data refers to codes that are not coded 01, 02, 03, 
04, 05, or 06 which represent the amount of (or lack of) helper 
assistance a patient needs to complete a functional activity. ANA 
codes, a dash, and a skip are considered ``missing'' in the context of 
the measure calculations since the observed discharge score is the sum 
of 01-06 values from functional assessment items included in the 
observed discharge score. Utilizing statistical imputation to calculate 
the observed discharge score does not disregard the clinical 
information represented by ANA codes. Rather, statistical imputation is 
a component in measure calculation of reported data and improves upon 
the imputation approach currently implemented in the Change in Mobility 
Score, Change in Self-Care Score, Discharge in Mobility Score, and 
Discharge in Self-Care Score measures. In these measures, ANA codes are 
always imputed to 1 (dependent) when calculating the measure scores, 
regardless of a patient's own clinical and functional information. The 
imputation approach implemented in the proposed DC Function measure 
uses each patient's available functional and clinical information, 
including ANA codes on other functional assessment items, to

[[Page 51021]]

estimate each ANA value had the item been completed. Testing 
demonstrates that, relative to the current simple imputation method, 
the statistical imputation approach in used this DC Function measure 
increases precision and accuracy, while reducing bias in estimates of 
missing item values.
    Comment: Several commenters raised concerns about the extent to 
which the measure can be considered a cross-setting measure, and its 
utility for comparing performance across settings. Some of these 
commenters believe that calculating a cross-setting function measure 
with different populations across PAC settings will not be meaningful 
in characterizing patients or comparing their outcomes across the 
different PAC settings, and may lead to inaccurate comparisons for 
patients, caregivers, Medicare Advantage plans, Medicaid managed care 
plans, and other interested parties. The same commenters also stated 
that CMS should work with interested parties to standardize data so 
that interested parties can differentiate patients' abilities and 
disabilities in a wide range of functional levels across the PAC 
spectrum.
    Response: We acknowledge that the measure denominators differ 
across PAC settings. However, as clarified during the MAP PAC/LTC 
workgroup discussed in section IX.C.1.b.(4) of this final rule, the 
denominator population in each measure setting is the population 
included in the respective setting's quality reporting program, as 
stated in the FY 2023 IRF PPS final rule (87 FR 47082 and 87 FR 47074) 
and the FY 2018 SNF PPS final rule (82 FR 36598). Moreover, we would 
like to clarify that cross-setting measures do not necessarily suggest 
that facilities can be compared across settings. Instead, these 
measures are intended to compare providers within a specific setting 
while standardizing measurement of function across settings. The 
proposed measure does just this, by aligning measure specifications 
across settings and including the use of a common set of standardized 
functional assessment data elements. This alignment satisfies the 
requirements of section 1886(j)(7)(F)(i) of the Act for a cross-setting 
measure in the functional status domain specified under section 
1899B(c)(1) of the Act.
    Comment: One commenter requested the rationale as to why confidence 
intervals were not calculated and reported for the expected function 
scores and utilized in determining meaningful differences between the 
observed and expected function score. This commenter also stated that 
the minimum clinical difference in discharge function scores that 
indicates a change is meaningful to patient progress has not been 
identified.
    Response: The proposed DC Function measure uses the same approach 
in determining whether an observed discharge score is different than 
its associated expected discharge score as the currently adopted 
Discharge Self-Care Score and Discharge Mobility Score measures that 
are CBE endorsed. Specifically, the DC Function measure reports the 
proportion of a given provider's stays where observed discharge 
function score matches or exceeds expected discharge function score. 
The measure score is a continuous variable with values between 0 and 
100, allowing for intuitive interpretation and comparisons. Our TEP 
supported that patients and families are more likely to understand a 
measure that expresses functional outcome as a simple proportion of 
patients who meet expectation for their discharge functional status, 
rather than units of change in a scoring system that is unfamiliar to 
most Care Compare website users (the primary audience for this 
measure). Measure scores based on statistical significance of 
differences between observed and expected values (based on confidence 
intervals) place providers in broad categories, such as `No different 
than national average,' which do not allow more granular provider 
comparisons for the public reviewing the measure's data on Care 
Compare. Given the excellent reliability of the DC Function measure, we 
believe that reporting provider scores as broad categories is not 
warranted.
    Comment: One commenter noted the variability in median scores and 
believed this range suggests the measure may not be valid, and that the 
variability may be problematic when making comparisons among providers.
    Response: First, we would like to clarify that median scores are 
not used in the calculation of this measure. While we would require 
additional information regarding the median scores referenced in this 
comment to provide a more complete response, we acknowledge that the 
measure has a large range of average expected discharge scores, as 
calculated for each provider. This range is consistent with the range 
of observed discharge scores, indicating that the measure is capturing 
the range of patient's functional abilities, and thus, in fact, 
supports the validity of the measure.
    Comment: One commenter noted that intrinsic to the discharge scores 
are the associated admission scores, and suggested an analysis of this 
measure to assess the variability in initial admission function scores 
between hospitals for similar types of patients as differences may 
account for the gaps in the observed discharge function scores.
    Response: We acknowledge that the observed gap in discharge 
function scores may be due to variability in the initial admission 
function scores; nevertheless, the admission function scores are 
included as covariates in the risk adjustment model and thus are 
accounted for in the calculations of the expected discharge function 
scores.
    Comment: One commenter questioned CMS' characterization of the 
adjusted R-squared value of 0.65 for the proposed DC Function measure's 
risk adjustment model. This commenter believed a value of 0.65 suggests 
moderate, rather than ``good'' model discrimination. This commenter 
suggested CMS should address the ability of the risk adjustment model 
to make predictions by comparing R-squared values of the ``training'' 
and ``validation'' sets and reporting ``predicted R-squared'' values.
    Response: We want to clarify that the adjusted R-squared for the DC 
Function measure, as reported in the Discharge Function Score for 
Inpatient Rehabilitation Facilities (IRFs) Technical Report,\124\ was 
0.51. We believe that this value indicates ``good'' model 
discrimination, and it is comparable to those of the Discharge Self-
Care Score and Discharge Mobility Score measures (0.48-0.50). 
Additionally, because the measure model uses all available data, the 
concepts of ``training'' and ``validation'' sets (and any related 
``predicted R-squared'') are not applicable. Rather, adjusted R-squared 
values capture model fit for the risk-adjustment model.
---------------------------------------------------------------------------

    \124\ Discharge Function Score for Inpatient Rehabilitation 
Facilities (IRFs) Technical Report. https://www.cms.gov/files/document/irf-discharge-function-score-technical-report-february-2023.pdf.
---------------------------------------------------------------------------

    Comment: Two commenters expressed concern that the measure 
performance may not adequately demonstrate the advancement in 
functional ability a patient has gained across the mobility and 
selfcare domains during their IRF stay. One of these commenters 
believed that upper body dressing and lower body dressing are better 
indicators of patient functional success at discharge than items 
currently included in the DC Function measure, and the rationale for 
selecting certain function items to be captured in this measure seem to 
be based solely on ensuring cross-setting applicability and less on the 
accuracy of an ``expected'' function score.
    Response: We acknowledge that the cross-setting applicability was a

[[Page 51022]]

motivating factor in determining function items captured in the 
proposed DC Function measure, and upper body dressing and lower body 
dressing function items were not available across settings. 
Nonetheless, the proposed DC Function measure does reflect the progress 
of patients across both the mobility and selfcare domains. As stated in 
section IX.C.1.b.(3) of this final rule, the TEP supported the 
inclusion of both functional domains as self-care items impact mobility 
items and are clinically relevant to function. Additionally, the 
proposed measure is meant to supplement, rather than replace, the 
Discharge Self-Care Score and Discharge Mobility Score measures which 
implement the remaining self-care and mobility function items not 
captured in the DC Function measure. High correlations between the 
proposed measure and the Discharge Self-Care Score and Discharge 
Mobility Score measures (0.85 and 0.88, respectively) demonstrate that 
these three measures capture related but distinct aspects of provider 
care in relation to patients' function. The TEP understood these 
aforementioned considerations and supported the inclusion of the 
function items included in the proposed measure.
    Comment: Two commenters (one in support of this proposed measure, 
and one opposed) raised concerns that the measure does not account for 
cognition and communication. One commenter urged CMS to consider 
alternative assessments that better incorporate cognition and 
communication into the measure calculation. The other commenter 
similarly raised concerns that Section GG items in the IRF-PAI 
insufficiently capture all elements of function and do not adequately 
capture the outcomes required for safety and independence.
    Response: We agree that cognition and communication are critically 
important and related to the safety and independence of patients. 
Although not directly assessed for the purpose of measure calculation, 
this measure does indirectly capture a facility's ability to impact a 
patient's cognition and communication to the extent that these factors 
are correlated to improvements in self-care and mobility. That said, we 
agree that communication and cognition are important to assess 
directly, and facilities currently do so through completion of the 
BIMS, CAM(copyright), and items BB0700-BB0800 in the IRF-PAI. 
Additionally, CMS regularly assesses the measures in the IRF QRP for 
measurement gaps, and as described in section IX.D of this final rule, 
specifically identified cognitive improvement as a possible gap and 
sought feedback about how to best assess this clinical dimension. CMS 
will use this feedback as well as discussion with technical experts and 
empirical analyses to determine how to measure communication and 
cognition.
    Comment: Two commenters expressed concern regarding the validity or 
completeness of reported functional assessment data. One of these 
commenters recommended that CMS improve providers' reporting of 
functional assessment data before adopting this measure, as the 
inconsistency of PAC providers' recording of this information raises 
concerns about publicly reporting this measure and using this measure 
for payment. This commenter provided the example that some providers 
code patient function in response to payment incentives. Although there 
are currently no payment implications for this measure, this commenter 
noted that differential coding practices and profitability by case type 
across IRFs may contribute to differential profitability. Additionally, 
this commenter stated that the current imputation approach used in 
existing measures in the IRF QRP recodes any ANA code to the most or 
second most dependent level which would lead to a lower motor score and 
raise Medicare payment for the stay.
    Response: We acknowledge that the coding of GG items may be 
affected by payment and quality reporting considerations and are 
actively monitoring IRF coding practices. The imputation approach 
implemented in the currently adopted Discharge Self-Care Score and 
Discharge Mobility Score measures, which recodes any ANA code to the 
most dependent level, can exacerbate these incentives, particularly 
with respect to function at admission. We would like to point out that 
statistical imputation used in the proposed DC Function measure reduces 
these incentives by using all available relevant information to assign 
the most likely score, ranging from most to least dependent, to each GG 
item. We acknowledge the importance of utilizing valid assessment data 
and will continue to monitor this potential data validity concern and 
will reconsider the measure's implementation in the quality reporting 
program, if needed.
    CMS has multiple processes in place to ensure reported patient data 
are accurate. State agencies conduct standard certification surveys for 
IRFs, and accuracy and completeness of the IRF-PAI are among the 
regulatory requirements that surveyors evaluate during surveys.\125\ 
Additionally, the IRF-PAI process has multiple regulatory requirements. 
Our regulations at Sec.  412.606(b) require that (1) the assessment 
accurately reflects the patient's status, (2) a clinician appropriately 
trained to perform a patient assessment using the IRF-PAI conducts or 
coordinates each assessment with the appropriate participation of 
health professionals, and (3) the assessment process includes direct 
observation, as well as communication with the patient.\126\ We take 
the accuracy of IRF-PAI assessment data very seriously, and routinely 
monitor the IRF QRP measures' performance, and will take appropriate 
steps to address any such issues, if identified, in future rulemaking.
---------------------------------------------------------------------------

    \125\ Center for Medicare and Medicaid Services. September 6, 
2022. Hospitals. https://www.cms.gov/medicare/provider-enrollment-and-certification/certificationandcomplianc/hospitals.
    \126\ 42 CFR 412.606 https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-B/part-412/subpart-P/section-412.606.
---------------------------------------------------------------------------

    We note that the potential consequences of submitting false data 
and information in the IRF-PAI, including the potential for civil 
liability under the False Claims Act (31 U.S.C. 3729 to 3733) for 
knowingly presenting a false or fraudulent claim to the government for 
payment, provide strong incentives for providers to ensure that the 
data submitted in the IRF-PAI are accurate.
    Comment: One commenter raised concerns about the measure, noting 
that IRFs are allowed to have 5 percent of the IRF-PAI data incomplete.
    Response: We interpret the comment as referring to the 95 percent 
completion threshold for the Annual Increase Factor (AIF) update. IRFs 
must submit 95 percent of their assessments with 100 percent of the 
required data elements to avoid the 2 percent penalty.\127\ As with all 
our IRF QRP measures, we will continue to monitor this measure to 
identify any concerning trends as part of our routine monitoring 
activities to regularly assess measure performance, reliability, and 
reportability for all data submitted for the IRF QRP.
---------------------------------------------------------------------------

    \127\ Sec.  412.634(f) Requirements under the Inpatient 
Rehabilitation Facility (IRF) Quality Reporting Program (QRP). 
https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-B/part-412/subpart-P/section-412.634.
---------------------------------------------------------------------------

    Comment: One commenter believes that self-care and mobility items 
are not tracked across PAC settings, creating inconsistent reporting 
and undue burden on IRFs, and stating that IRFs are held to different 
standards compared to other settings.
    Response: In addition to the IRF, the items in the DC Function 
measure are

[[Page 51023]]

collected and tracked across the SNF, LTCH and Home Health setting. 
Therefore, we do not believe IRFs are held to a higher standard as it 
relates to collecting this information.
    After careful consideration of the public comments we received, we 
are finalizing our proposal to adopt the DC Function measure as an 
assessment-based outcome measure beginning with the FY 2025 IRF QRP.
c. Removal of the Application of Percent of Long-Term Care Hospital 
Patients With an Admission and Discharge Functional Assessment and a 
Care Plan That Addresses Function Beginning With the FY 2025 IRF QRP
    We proposed to remove the Application of Percent of Long-Term Care 
Hospital Patients with an Admission and Discharge Functional Assessment 
and a Care Plan That Addresses Function (Application of Functional 
Assessment/Care Plan) measure from the IRF QRP beginning with the FY 
2025 IRF QRP. Section 412.634(b)(2) of our regulations specifies eight 
factors we consider for measure removal from the IRF QRP, and we 
believe this measure should be removed because it satisfies two of 
these factors.
    First, the Application of Functional Assessment/Care Plan measure 
meets the conditions for measure removal factor one: measure 
performance among IRFs is so high and unvarying that meaningful 
distinctions in improvements in performance can no longer be made.\128\ 
Second, this measure meets the conditions for measure removal factor 
six: there is an available measure that is more strongly associated 
with desired patient functional outcomes. We believe the proposed DC 
Function measure discussed in section IX.C.1.b. of the proposed rule 
better measures functional outcomes than the current Application of 
Functional Assessment/Care Plan measure. We discuss each of these 
reasons in more detail below.
---------------------------------------------------------------------------

    \128\ For more information on the factors CMS uses to base 
decisions for measure removal, we refer readers to Sec.  
412.634(b)(2) Subpart P--Requirements under the Inpatient 
Rehabilitation Facility (IRF) Quality Reporting Program (QRP). 
https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-B/part-412/subpart-P/section-412.634.
---------------------------------------------------------------------------

    In regard to removal factor one, the Application of Functional 
Assessment/Care Plan measure has become topped out, with average 
performance rates reaching nearly 100 percent over the past 3 years 
(ranging from 99.8 percent to 99.9 percent during CYs 2019-
2021).129 130 131 For the 12-month period of third quarter 
of CY 2020 through second quarter of CY 2021 (July 1, 2020 through June 
30, 2021), IRFs had an average score for this measure of 99.8 percent, 
with nearly 80 percent of IRFs scoring 100 percent,\132\ and for CY 
2021, IRFs had an average score of 99.9 percent, with nearly 78 percent 
of IRFs scoring 100 percent.\133\ The proximity of these mean rates to 
the maximum score of 100 percent suggests a ceiling effect and a lack 
of variation that restricts distinction among IRFs.
---------------------------------------------------------------------------

    \129\ Centers for Medicare & Medicaid Services. Inpatient 
Rehabilitation Facilities Data Archive, 2021, Annual Files National 
Data 07-21. https://data.cms.gov/provider-data/archived-data/inpatient-rehabilitation-facilities.
    \130\ Centers for Medicare & Medicaid Services. Inpatient 
Rehabilitation Facilities Data Archive, 2022, Annual Files National 
Data 04-22. https://data.cms.gov/provider-data/archived-data/inpatient-rehabilitation-facilities.
    \131\ Centers for Medicare & Medicaid Services. Inpatient 
Rehabilitation Facilities Data Archive, 2022, Annual Files National 
Data 09-22. https://data.cms.gov/provider-data/archived-data/inpatient-rehabilitation-facilities.
    \132\ Centers for Medicare & Medicaid Services. Inpatient 
Rehabilitation Facilities Data Archive, 2022, Annual Files Provider 
Data 04-22. https://data.cms.gov/provider-data/archived-data/inpatient-rehabilitation-facilities.
    \133\ Centers for Medicare & Medicaid Services. Inpatient 
Rehabilitation Facilities Data Archive, 2022, Annual Files Provider 
Data 09-22. https://data.cms.gov/provider-data/archived-data/inpatient-rehabilitation-facilities.
---------------------------------------------------------------------------

    In regard to measure removal factor six, the proposed DC Function 
measure is more strongly associated with desired patient functional 
outcomes than this current process measure, the Application of 
Functional Assessment/Care Plan measure. As described in section 
VIII.C.b.(1)(b) of the proposed rule, the DC Function measure has the 
predictive ability to distinguish patients with low expected functional 
capabilities from those with high expected functional 
capabilities.\134\ We have been collecting standardized functional 
assessment elements across PAC settings since 2016, which has allowed 
for the development of the proposed DC Function measure and meets the 
statutory requirements to submit standardized patient assessment data 
and other necessary data with respect to the domain of functional 
status, cognitive function, and changes in function and cognitive 
function. In light of this development, this process measure, the 
Application of Functional Assessment/Care Plan measure which measures 
only whether a functional assessment is completed, and a functional 
goal is included in the care plan, is no longer necessary, and can be 
replaced with a measure that evaluates the IRF's outcome of care on a 
patient's function.
---------------------------------------------------------------------------

    \134\ ``Expected functional capabilities'' is defined as the 
predicted discharge function score.
---------------------------------------------------------------------------

    Because the Application of Functional Assessment/Care Plan measure 
meets measure removal factors one and six under Sec.  412.634(b)(2), we 
proposed to remove it from the IRF QRP beginning with the FY 2025 IRF 
QRP. We also proposed that public reporting of the Application of 
Functional Assessment/Care Plan measure would end by the September 2024 
Care Compare refresh or as soon as technically feasible when public 
reporting of the proposed DC Function measure would begin (see section 
VIII.G.3. of the proposed rule).
    Under our proposal, IRFs would no longer be required to report a 
Self-Care Discharge Goal (that is, GG0130, Column 2) or a Mobility 
Discharge Goal (that is, GG0170, Column 2) on the IRF-PAI beginning 
with patients admitted on October 1, 2023. We would remove the items 
for Self-Care Discharge Goals (that is, GG0130, Column 2) and Mobility 
Discharge Goals (that is, GG0170, Column 2) with the next release of 
the IRF-PAI. Under our proposal, these items would not be required to 
meet IRF QRP requirements beginning with the FY 2025 IRF QRP.
    We invited public comment on our proposal to remove the Application 
of Functional Assessment/Care Plan measure from the IRF QRP beginning 
with the FY 2025 IRF QRP. The following is a summary of the public 
comments received on our proposal and our responses:
    Comment: Several commenters supported the removal of the 
Application of Functional Assessment/Care Plan measure, along with the 
requirement to submit the associated goal items (that is, the Self-Care 
Discharge Goals and Mobility Discharge Goals), stating that the measure 
lacks variation in performance and is no longer meaningful, and noted 
its removal will reduce burden. Three of these commenters noted that 
the measure's removal should not be tied to the adoption of the DC 
Function measure because the measure is topped out and is no longer 
representative of meaningful distinctions in improvements and 
performance.
    Response: We thank the commenters for their support to remove the 
Application of Functional Assessment/Care Plan measure and the removal 
of the GG items from the IRF-PAI and agree that the measure provides 
limited value given the lack of variation. With respect to the 
commenters' request that we not tie this measure removal proposal to 
the adoption of the DC Function measure, we would like to

[[Page 51024]]

clarify that a cross-setting measure of function is required to meet 
the requirements set forth in sections 1886(j)(7)(F)(i) and 
1899B(c)(1)(A) of the Act. Thus, the removal of this measure is 
inherently dependent on the adoption of a new measure that would also 
meet the requirements of sections 1886(j)(7)(F)(i) and 1899B(c)(1)(A) 
of the Act.
    Comment: One commenter supported the removal of the Application of 
Functional Assessment/Care Plan measure, but also noted that it is 
important and integral to set and track individual patient functional 
goals for a patient's care plan.
    Response: We thank the commenter for their support to remove the 
Application of Functional Assessment/Care Plan measure from the IRF 
QRP. Additionally, we agree that it is critically important that 
facilities continue to set and track patient functional goals, even 
after the measure is removed. While CMS will not require the assessment 
or reporting of, items associated with this measure, IRFs have the 
option to continue collection within their own health records to meet 
patient needs.
    After consideration of the public comments we received, we are 
finalizing our proposal to remove the Application of Functional 
Assessment/Care Plan measure from the IRF QRP beginning with the FY 
2025 IRF QRP as proposed.
d. Removal of the IRF Functional Outcome Measure: Change in Self-Care 
Score for Medical Rehabilitation Patients and Removal of the IRF 
Functional Outcome Measure: Change in Mobility Score for Medical 
Rehabilitation Patients Beginning With the FY 2025 IRF QRP
    We proposed to remove the IRF Functional Outcome Measure: Change in 
Self-Care Score for Medical Rehabilitation Patients (Change in Self-
Care Score) and the IRF Functional Outcome Measure: Change in Mobility 
Score for Medical Rehabilitation Patients (Change in Mobility Score) 
measures from the IRF QRP beginning with the FY 2025 IRF QRP. Section 
412.634(b)(2) of our regulations specifies eight factors we consider 
for measure removal from the IRF QRP. We proposed removal of these 
measures because they satisfy measure removal factor eight: the costs 
associated with a measure outweigh the benefits of its use in the IRF 
QRP.
    Measure costs are multifaceted and include costs associated with 
implementing and maintaining the measures. On this basis, we proposed 
to remove these measures for two reasons. First, the costs to IRFs 
associated with tracking similar or duplicative measures in the IRF QRP 
outweigh any benefit that might be associated with the measures. 
Second, the costs to CMS associated with program oversight of the 
measures, including measure maintenance and public display, outweigh 
the benefit of information obtained from the measures. We discuss each 
of these in more detail below.
    We adopted the Change in Self-Care Score and Change in Mobility 
Score measures in the FY 2016 IRF PPS final rule (80 FR 47112 through 
47118) under section 1886(j)(7)(D)(ii) of the Act because the measures 
meet the functional status, cognitive function, and changes in function 
and cognitive function domain under section 1899B(c)(1) of the Act. Two 
additional measures addressing the functional status, cognitive 
function, and changes in function and cognitive function domain were 
adopted in the same program year: the IRF Functional Outcome Measure: 
Discharge Self-Care Score for Medical Rehabilitation Patients 
(Discharge Self-Care Score) and the IRF Functional Outcome Measure: 
Discharge Mobility Score for Medical Rehabilitation Patients (Discharge 
Mobility Score) measures. Given that the primary goal of rehabilitation 
is improvement in functional status, IRF clinicians have traditionally 
assessed and documented individual patients' functional status at 
admission and discharge to evaluate the effectiveness of the 
rehabilitation care provided.
    We proposed to remove the Change in Self-Care Score and Change in 
Mobility Score measures because we believe the IRF costs associated 
with tracking duplicative measures outweigh any benefit that might be 
associated with the measures. Since the adoption of these measures in 
2016, we have been monitoring the data and found that the scores for 
the two self-care functional outcome measures, Change in Self-Care 
Score and Discharge Self-Care Score, are very highly correlated in IRF 
settings (0.97).\135\ Similarly, in the monitoring data, we have found 
that, the scores for the two mobility score measures, Change in 
Mobility Score and Discharge Mobility Score, are very highly correlated 
in IRF settings (0.98).\136\ The high correlation between these 
measures suggests that the Change in Self-Care Score and Discharge 
Self-Care Score and the Change in Mobility Score and the Discharge 
Mobility Score measures provide almost identical information about this 
dimension of quality to IRFs and are therefore duplicative.
---------------------------------------------------------------------------

    \135\ Acumen, LLC and Abt Associates. Technical Expert Panel 
(TEP) for the Refinement of Long-Term Care Hospital (LTCH), 
Inpatient Rehabilitation Facility (IRF), Skilled Nursing Facility 
(SNF)/Nursing Facility (NF), and Home Health (HH) Function Measures, 
July 14-15, 2021: Summary Report. February 2022. https://mmshub.cms.gov/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
    \136\ Acumen, LLC and Abt Associates. Technical Expert Panel 
(TEP) for the Refinement of Long-Term Care Hospital (LTCH), 
Inpatient Rehabilitation Facility (IRF), Skilled Nursing Facility 
(SNF)/Nursing Facility (NF), and Home Health (HH) Function Measures: 
July 14-15, 2021: Summary Report. February 2022. https://mmshub.cms.gov/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
---------------------------------------------------------------------------

    Our proposal to remove the Change in Self-Care Score and the Change 
in Mobility Score measures is supported by feedback received from the 
TEP convened for the Refinement of LTCH, IRF, SNF/NF, and HH Function 
Measures. As described in section VIII.C.1.b(3) of the proposed rule, 
the TEP panelists were presented with analyses that demonstrated the 
``Change in Score'' and ``Discharge Score'' measure sets are highly 
correlated and do not appear to measure unique concepts, and they 
subsequently articulated that it would be sensible to retire either the 
``Change in Score'' or ``Discharge Score'' measure sets for both self-
care and mobility. Based on responses to the post-TEP survey, the 
majority of panelists (nine out of 12 respondents) suggested that only 
one measure is necessary. Of those nine respondents, six preferred 
retaining the ``Discharge Score'' measures over the ``Change in Score'' 
measures.\137\
---------------------------------------------------------------------------

    \137\ Acumen, LLC and Abt Associates. Technical Expert Panel 
(TEP) for the Refinement of Long-Term Care Hospital (LTCH), 
Inpatient Rehabilitation Facility (IRF), Skilled Nursing Facility 
(SNF)/Nursing Facility (NF), and Home Health (HH) Function Measures, 
July 14-15, 2021: Summary Report. February 2022. https://mmshub.cms.gov/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
---------------------------------------------------------------------------

    Additionally, we proposed to remove the Change in Self-Care Score 
and Change in Mobility Score measures because the program oversight 
costs outweigh the benefit of information that CMS, IRFs, and the 
public obtain from the measures. We must engage in various activities 
when administering the QRPs, such as monitoring measure results, 
producing provider preview reports, and ensuring the accuracy of the 
publicly reported data. Because these measures essentially provide the 
same information to IRFs and consumers as the Discharge Self-Care Score 
and Discharge Mobility Score measures, the costs to CMS associated with 
measure maintenance and public display outweigh the benefit of

[[Page 51025]]

information obtained from the measures.
    Because these measures meet the criteria for measure removal factor 
eight, we proposed to remove the Change in Self-Care Score and Change 
in Mobility Score measures from the IRF QRP beginning with the FY 2025 
IRF QRP. We also proposed that public reporting of the Change in Self-
Care Score and the Change in Mobility Score measure would end by the 
September 2024 Care Compare refresh or as soon as technically feasible.
    We invited public comment on our proposal to remove the Change in 
Self-Care Score and Change in Mobility Score measures from the IRF QRP 
beginning with the FY 2025 IRF QRP.
    The following is a summary of the public comments received on our 
proposal to remove the Change in Self-Care Score and Change in Mobility 
Score measures from the IRF QRP beginning with the FY 2025 IRF QRP and 
our responses.
    Comment: Several commenters expressed their support for the removal 
of the Change in Self-Care Score and the Change in Mobility Score 
measures, noting that these measures are duplicative of other measures 
and that their removal will reduce costs to IRFs and to CMS.
    Response: We thank the commenters for their support of the removal 
of the measures and agree, based on the testing we presented in the 
proposed rule, that the Change in Self-Care Score and Change in 
Mobility Score measures are duplicative of the Discharge Self-Care 
Score and Discharge Mobility Score measures.
    Comment: Several commenters did not agree with the removal of the 
Change in Self-Care Score and Change in Mobility Score measures because 
they believe these measures provide more information than the Discharge 
Self-Care Score and the Discharge Mobility Score measures. 
Specifically, some commenters stated that capturing the amount of 
change patients experience is more valuable than capturing whether 
patients meet or exceed an expected amount of change during their stay. 
One commenter noted that the greater variability in performance of the 
Change in Self-Care Score and Change in Mobility Score measures offers 
significantly greater opportunity to differentiate IRF performance, 
compared to the analogous Discharge Self-Care Score and Discharge 
Mobility Score measures.
    Response: We appreciate the perspective of the commenters and 
understand that there are advantages and disadvantages to retiring the 
Change in Self-Care Score and Change in Mobility Score versus the 
Discharge Self-Care Score and Discharge in Mobility Score measures. We 
weighed the tradeoffs of these measures in consultation with a TEP, 
comprised of 15 panelists with diverse perspectives and areas of 
expertise, including IRF representation.\138\ The majority of the TEP 
favored the retirement of the Change in Self-Care Score and Change in 
Mobility Score measures because they believed the Discharge Self-Care 
Score and Discharge in Mobility Score measures better capture a 
patient's relevant functional abilities. We agree that it is important 
for facilities to track the amount of change that occurs over the 
course of a stay for is patients and would like to point out that the 
removal of the Change in Self-Care Score and Change in Mobility Score 
measures does not preclude IRFs' abilities in this regard. However, we 
also believe that the Change in Self-Care Score and Change in Mobility 
Score measures are not intuitive to interpret for the primary audience 
of Care Compare, as the unit of change, and what constitutes a 
meaningful change, are unfamiliar to the vast majority of users, 
particularly prospective or current patients and their caregivers. This 
is in contrast to the Discharge Self-Care Score and Discharge Mobility 
Score measures, which are presented as a simple proportion. 
Additionally, as noted in section VII.C.1.b.1.b of this final rule, the 
correlations between the Change in Self-Care Score and Discharge Self-
Care Score measures and Change in Mobility Score and Discharge Mobility 
Score measures are very high (Spearman correlation: 0.97-0.98), 
indicating the measures capture almost identical concepts and lead to 
very similar rankings.\139\ As such, the testing does not support the 
claim that the Change in Self-Care Score and Change in Mobility Score 
measures provide significantly more information on which to compare 
facilities, as the relative rankings of facilities are very similar 
between the Change in Self-Care Score and Discharge Self-Care Score 
measures and the Change in Mobility Score and Discharge Mobility Score 
measures. Given the TEP's recommendation, the more intuitive 
interpretation, and the very high correlations, we believe there is 
more value in retiring the Change in Self-Care Score and Change in 
Mobility Score measures and retaining the Discharge Self-Care Score and 
Discharge Mobility Score measures.
---------------------------------------------------------------------------

    \138\ Acumen, LLC and Abt Associates. Technical Expert Panel 
(TEP) for the Refinement of Long-Term Care Hospital (LTCH), 
Inpatient Rehabilitation Facility (IRF), Skilled Nursing Facility 
(SNF)/Nursing Facility (NF), and Home Health (HH) Function Measures, 
July 14-15, 2021: Summary Report. February 2022. https://mmshub.cms.gov/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
    \139\ Acumen, LLC and Abt Associates. Technical Expert Panel 
(TEP) for the Refinement of Long-Term Care Hospital (LTCH), 
Inpatient Rehabilitation Facility (IRF), Skilled Nursing Facility 
(SNF)/Nursing Facility (NF), and Home Health (HH) Function Measures: 
July 14-15, 2021: Summary Report. February 2022. https://mmshub.cms.gov/sites/default/files/TEP-Summary-Report-PAC-Function.pdf.
---------------------------------------------------------------------------

    Comment: Two commenters raised concerns that the methodology used 
to calculate the Discharge Self-Care Score and Discharge Mobility Score 
measures does not account for functional abilities at admission in the 
way that the Change in Self-Care Score and Change in Mobility Score 
measures being proposed for removal do. One of these commenters 
requested that CMS clarify the extent to which these remaining 
Discharge Self-Care Score and Discharge Mobility Score measures would 
account for change in patients' function over time, as well as patient 
heterogeneity. Relatedly, another commenter noted that patients with 
higher discharge scores at the end of their IRF stay may include many 
patients who were admitted with high scores initially, and therefore, 
the quality and value of the IRF's care can be potentially 
misunderstood. These commenters also raised concerns about unintended 
consequences that could be introduced through the removal of the Change 
in Self-Care Score and Change in Mobility Score measures, such as the 
cherry-picking of patients or creating limited access to services for 
those with lower functional status. One of these commenters urged CMS 
to carefully evaluate whether the removal of the Change in Self-Care 
Score and Change in Mobility Score measures could lead to such 
unintended consequences.
    Response: We appreciate that measures of functional outcomes must 
account for patient case-mix to ensure fair and meaningful comparisons 
across facilities. Accordingly, the Discharge Self-Care Score and 
Discharge Mobility Score measures that would remain in the IRF QRP do 
in fact account for functional abilities at admission, as well as other 
relevant demographic and clinical characteristics (see, for example, 
Inpatient Rehabilitation Facility Quality Reporting Program Measure 
Calculations and Reporting User's Manual v4.0).\140\ Specifically, the

[[Page 51026]]

expected discharge scores, which patients must meet or exceed to meet 
for the measures' numerators are predicted using the patients' observed 
admission function scores plus the same clinical comorbidities and 
demographic characteristics as the corresponding Change in Self-Care 
Score and Change in Mobility Score measures. Given that the Discharge 
Self-Care Score and Discharge Mobility Score measures do account for 
functional abilities at admission, among other relevant clinical 
characteristics that can impact functional improvement, we do not 
anticipate that the removal of the Change in Self-Care Score and Change 
in Mobility Score measures will increase any incentive to cherry-pick 
patients or block access to care. We take the appropriate access to 
care in IRFs very seriously, and routinely monitor the performance of 
measures in the IRF QRP, including performance gaps across IRFs. We 
will continue to monitor closely whether any proposed changes to the 
IRF QRP have unintended consequences on access to care for high-risk 
patients. Should we find any unintended consequences, we will take 
appropriate steps to address these issues in future rulemaking.
---------------------------------------------------------------------------

    \140\ Centers for Medicare & Medicaid Services. Inpatient 
Rehabilitation Facility Quality Reporting Program Measure 
Calculations and Reporting User's Manual Version 4.0. October 2022. 
https://www.cms.gov/files/document/irf-quality-measure-calculations-and-reporting-users-manual-v40.pdf.
---------------------------------------------------------------------------

    Comment: One commenter stated that they do not support the removal 
of the Change in Self-Care Score and Change in Mobility Score measures, 
stating that these measures assess patients who meet or exceed a 
specific risk-adjusted goal, and as such are representative of IRF care 
as a whole.
    Response: We agree that there is value in assessing the extent to 
which patients meet or exceed an expected level of function, where the 
expected level of function accounts for a patient's own case mix. 
However, we would like to point out that this is exactly what the 
Discharge Self-Care Score and Discharge Mobility Score measures assess 
(which would be retained in the IRF QRP), as opposed to the Change in 
Self-Care and Change in Mobility Measure, which measure the risk-
adjusted change in function between admission and discharge.
    After consideration of the public comments we received, we are 
finalizing our proposal to remove the Change in Self-Care Score and 
Change in Mobility Score measures from the IRF QRP beginning with the 
FY 2025 IRF QRP as proposed.
2. IRF QRP Quality Measure Beginning With the FY 2026 IRF QRP
a. COVID-19 Vaccine: Percent of Patients/Residents Who Are Up to Date 
Measure Beginning With the FY 2026 IRF QRP
(1) Background
    COVID-19 has been and continues to be a major challenge for PAC 
facilities, including IRFs. The Secretary first declared COVID-19 a PHE 
on January 31, 2020. As of March 23, 2023, the U.S. has reported 
103,957,053 cumulative cases of COVID-19, and 1,123,613 total deaths 
due to COVID-19.\141\ Although all age groups are at risk of 
contracting COVID-19, older persons are at a significantly higher risk 
of mortality and severe disease following infection, with those over 
age 80 dying at five times the average rate.\142\ Older adults, in 
general, are prone to both acute and chronic infections owing to 
reduced immunity, and are a high-risk population.\143\ Adults age 65 
and older comprise over 75 percent of total COVID-19 deaths despite 
representing 13.4 percent of reported cases.\144\ COVID-19 has impacted 
older adults' access to care, leading to poorer clinical outcomes, as 
well as taking a serious toll on their mental health and well-being due 
to social distancing.\145\
---------------------------------------------------------------------------

    \141\ Centers for Disease Control and Prevention. COVID Data 
Tracker. https://covid.cdc.gov/covid-data-tracker/#cases_totalcases.
    \142\ United Nations. Policy Brief: The impact of COVID-19 on 
older persons. May 2020. https://unsdg.un.org/sites/default/files/2020-05/Policy-Brief-The-Impact-of-COVID-19-on-Older-Persons.pdf.
    \143\ Lekamwasam R, Lekamwasam S. Effects of COVID-19 pandemic 
on health and wellbeing of older people: a comprehensive review. Ann 
Geriatr Med Res. 2020 Sep;24(3):166-172.doi: 10.4235/agmr.20.0027. 
PMID: 32752587; PMCID: PMC7533189.
    \144\ Centers for Disease Control and Prevention. Demographic 
trends of COVID-19 cases and deaths in the US reported to CDC. COVID 
Data Tracker. https://covid.cdc.gov/covid-data-tracker/#demographics.
    \145\ United Nations. Policy Brief: The impact of COVID-19 on 
older persons. May 2020. https://unsdg.un.org/sites/default/files/2020-05/Policy-Brief-The-Impact-of-COVID-19-on-Older-Persons.pdf.
---------------------------------------------------------------------------

    Since the development of the vaccines to combat COVID-19, studies 
have shown they continue to provide strong protection against severe 
disease, hospitalization, and death in adults, including during the 
predominance of Omicron BA.4 and BA.5 variants.\146\ Initial studies 
showed the efficacy of FDA-approved or authorized COVID-19 vaccines in 
preventing COVID-19. Prior to the emergence of the Delta variant of the 
virus, vaccine effectiveness against COVID-19-associated 
hospitalization among adults aged 65 and older was 91 percent for those 
who were fully vaccinated with an mRNA vaccine (Pfizer-BioNTech or 
Moderna), and 84 percent for those receiving a viral vector vaccine 
(Janssen). Adults aged 65 and older who were fully vaccinated with an 
mRNA COVID-19 vaccine had a 94 percent reduction in risk of COVID-19 
hospitalization while those who were partially vaccinated had a 64 
percent reduction in risk.\147\ Further, after the emergence of the 
Delta variant, vaccine effectiveness against COVID-19-associated 
hospitalization for adults who were fully vaccinated was 76 percent 
among adults age 75 and older.\148\
---------------------------------------------------------------------------

    \146\ Chalkias S, Harper C, Vrbicky K, et al. A Bivalent 
Omicron-Containing Booster Vaccine Against COVID-19. N Engl J Med. 
2022 Oct 6;387(14):1279-1291. doi: 10.1056/NEJMoa2208343. PMID: 
36112399; PMCID: PMC9511634.
    \147\ Centers for Disease Control and Prevention. Fully 
Vaccinated Adults 65 and Older Are 94% Less Likely to Be 
Hospitalized with COVID-19. April 28, 2021. https://www.cdc.gov/media/releases/2021/p0428-vaccinated-adults-less-hospitalized.html.
    \148\ Grannis SJ, Rowley EA, Ong TC, et al. Interim Estimates of 
COVID-19 Vaccine Effectiveness Against COVID-19-Associated Emergency 
Department or Urgent Care Clinic Encounters and Hospitalizations 
Among Adults During SARS-CoV-2 B.1.617.2 (Delta) Variant 
Predominance--Nine States, June-August 2021. (Grannis SJ, et al. 
MMWR Morb Mortal Wkly Rep. 2021;70(37):1291-1293. doi.org/10.15585/mmwr.mm7037e2.
---------------------------------------------------------------------------

    More recently, since the emergence of the Omicron variants and 
availability of booster doses, multiple studies have shown that while 
vaccine effectiveness has waned, protection is higher among those 
receiving booster doses than among those only receiving the primary 
series.149 150 151 CDC data show that, among people age 50 
and older, those who have received both a primary vaccination series 
and booster doses have a lower risk of hospitalization and dying from 
COVID-19 than their non-

[[Page 51027]]

vaccinated counterparts.\152\ Additionally, a second vaccine booster 
dose has been shown to reduce risk of severe outcomes related to COVID-
19, such as hospitalization or death.\153\ Early evidence also 
demonstrates that the bivalent boosters, specifically aimed to provide 
better protection against disease caused by Omicron subvariants, have 
been quite effective, and underscores the role of up to date 
vaccination protocols in effectively countering the spread of COVID-
19.154 155
---------------------------------------------------------------------------

    \149\ Surie D, Bonnell L, Adams K, et al. Effectiveness of 
monovalent mRNA vaccines against COVID-19-associated hospitalization 
among immunocompetent adults during BA.1/BA.2 and BA.4/BA.5 
predominant periods of SARS CoV-2 Omicron variant in the United 
States--IVY Network, 18 States, December 26, 2021-August 31, 2022. 
MMWR Morb Mortal Wkly Rep. 2022;71(42):1327-1334. doi: 10.15585/
mmwr.mm7142a3.
    \150\ Andrews N, Stowe J, Kirsebom F, et al. Covid-19 vaccine 
effectiveness against the Omicron (B.1.1.529) variant. N Engl J Med. 
2022 Apr 21;386(16):1532-1546. doi 10.1056/NEJMoa2119451. PMID: 
35249272; PMCID: PMC8908811.
    \151\ Buchan SA, Chung H, Brown KA, et al. Estimated 
effectiveness of COVID-19 vaccines against Omicron or Delta 
symptomatic infection and severe outcomes. JAMA Netw Open. 2022 Sep 
1;5(9):e2232760.doi: 10.1001/jamanetworkopen.2022.32760. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2796615. PMID: 
36136332; PMCID: PMC9500552.
    \152\ Centers for Disease Control and Prevention. Rates of 
laboratory-confirmed COVID-19 hospitalizations by vaccination 
status. COVID Data Tracker. February 9, 2023. https://covid.cdc.gov/covid-data-tracker/#covidnet-hospitalizations-vaccination.
    \153\ Centers for Disease Control and Prevention. COVID-19 
vaccine effectiveness monthly update. COVID Data Tracker. November 
10, 2022. https://covid.cdc.gov/covid-data-tracker/#vaccine-effectiveness.
    \154\ Chalkias S, Harper C, Vrbicky K, et al. A bivalent 
omicron-containing booster vaccine against COVID-19. N Engl J Med. 
2022 Oct 6;387(14):1279-1291. doi: 10.1056/NEJMoa2208343. PMID: 
36112399; PMCID: PMC9511634.
    \155\ Tan, S.T., Kwan, A.T., Rodr[iacute]guez-Barraquer, I. et 
al. Infectiousness of SARS-CoV-2 breakthrough infections and 
reinfections during the Omicron wave. Nat Med 29, 358-365 (2023). 
https://doi.org/10.1038/s41591-022-02138-x.
---------------------------------------------------------------------------

(a) Measure Importance
    Despite the availability and demonstrated effectiveness of COVID-19 
vaccinations, significant gaps continue to exist in vaccination 
rates.\156\ As of March 22, 2023, vaccination rates among people age 65 
and older are generally high for the primary vaccination series (94.3 
percent) but lower for the first booster (73.6 percent among those who 
received a primary series) and even lower for the second booster (59.9 
percent among those who received a first booster).\157\ Additionally, 
though the uptake in boosters among people age 65 and older has been 
much higher than among people of other ages, booster uptake still 
remains relatively low compared to primary vaccination among older 
adults.\158\ Variations are also present when examining vaccination 
rates by race, gender, and geographic location.\159\ For example, 66.2 
percent of the Asian, non-Hispanic population have completed the 
primary series and 21.2 percent have received a bivalent booster dose, 
whereas 44.9 percent of the Black, non-Hispanic population have 
completed the primary series and only 8.9 percent have received a 
bivalent booster dose. Among Hispanic populations, 57.1 percent of the 
population have completed the primary series, and 8.5 percent have 
received a bivalent booster dose, while in White, non-Hispanic 
populations, 51.9 percent have completed the primary series and 16.2 
percent have received a bivalent booster dose.\160\ Disparities have 
been found in vaccination rates between rural and urban areas, with 
lower vaccination rates found in rural areas.161 162 Data 
show that 55.2 percent of the eligible population in rural areas have 
completed the primary vaccination series, as compared to 66.5 percent 
of the eligible population in urban areas.\163\ Receipt of bivalent 
booster doses among those eligible has been lower, with 18 percent of 
urban population having received a booster dose, and 11.5 percent of 
the rural population having received a booster dose.\164\
---------------------------------------------------------------------------

    \156\ Centers for Disease Control and Prevention. COVID Data 
Tracker: COVID-19 vaccinations in the United States. https://covid.cdc.gov/covid-data-tracker/#vaccinations_vacc-people-booster-percent-pop5.
    \157\ Centers for Disease Control and Prevention. COVID-19 
vaccination age and sex trends in the United States, national and 
jurisdictional. https://data.cdc.gov/Vaccinations/COVID-19-Vaccination-Age-and-Sex-Trends-in-the-Uni/5i5k-6cmh.
    \158\ Freed M, Neuman T, Kates J, Cubanski J. Deaths among older 
adults due to COVID-19 jumped during the summer of 2022 before 
falling somewhat in September. Kaiser Family Foundation. October 6, 
2022. https://www.kff.org/coronavirus-covid-19/issue-brief/deaths-among-older-adults-due-to-covid-19-jumped-during-the-summer-of-2022-before-falling-somewhat-in-september/.
    \159\ Saelee R, Zell E, Murthy BP, et al. Disparities in COVID-
19 Vaccination Coverage Between Urban and Rural Counties--United 
States, December 14, 2020-January 31, 2022. MMWR Morb Mortal Wkly 
Rep. 2022 Mar 4;71:335-340. doi: 10.15585/mmwr.mm7109a2. PMID: 
35239636; PMCID: PMC8893338.
    \160\ Centers for Disease Control and Prevention. COVID Data 
Tracker: Trends in demographic characteristics of people receiving 
COVID-19 vaccinations in the United States. https://covid.cdc.gov/covid-data-tracker/#vaccination-demographics-trends.
    \161\ Saelee R, Zell E, Murthy BP, et al. Disparities in COVID-
19 Vaccination Coverage Between Urban and Rural Counties--United 
States, December 14, 2020-January 31, 2022. MMWR Morb Mortal Wkly 
Rep. 2022 Mar 4;71:335-340. doi: 10.15585/mmwr.mm7109a2. PMID: 
35239636; PMCID: PMC8893338.
    \162\ Sun Y, Monnat SM. Rural-urban and within-rural differences 
in COVID-19 vaccination rates. J Rural Health. 2022 Sep;38(4):916-
922. doi: 10.1111/jrh.12625. PMID: 34555222; PMCID: PMC8661570
    \163\ Centers for Disease Control and Prevention. COVID Data 
Tracker. COVID-19 Vaccination Equity. https://covid.cdc.gov/covid-data-tracker/#vaccination-equity.
    \164\ Centers for Disease Control and Prevention. COVID-19 
Vaccination Equity. COVID Data Tracker. https://covid.cdc.gov/covid-data-tracker/#vaccination-equity.
---------------------------------------------------------------------------

    We proposed to adopt the COVID-19 Vaccine: Percent of Patients/
Residents Who Are Up to Date (Patient/Resident COVID-19 Vaccine) 
measure for the IRF QRP beginning with the FY 2026 IRF QRP. The 
proposed measure has the potential to increase COVID-19 vaccination 
coverage of patients in IRFs, as well as prevent the spread of COVID-19 
within the IRF patient population. The proposed Patient/Resident COVID-
19 Vaccine measure would also support the goal of CMS's Meaningful 
Measure Initiative 2.0 to ``Empower consumers to make good health care 
choices through patient-directed quality measures and public 
transparency objectives.'' The proposed Patient/Resident COVID-19 
Vaccine measure would be publicly reported on Care Compare and would 
provide patients, including those who are at high risk for developing 
serious complications from COVID-19, and their caregivers, with 
valuable information they can consider when choosing an IRF. The 
proposed Patient/Resident COVID-19 Vaccine measure would also 
facilitate patient care and care coordination during the hospital 
discharge planning process. For example, a discharging hospital, in 
collaboration with the patient and family, could use this proposed 
measure's publicly reported information on Care Compare to coordinate 
care and ensure patient preferences are considered in the discharge 
plan. Additionally, the proposed Patient/Resident COVID-19 Vaccine 
measure would be an indirect measure of IRF action. Since the patient's 
COVID-19 vaccination status would be reported at discharge from the 
IRF, if a patient is not up to date with their COVID-19 vaccination per 
applicable CDC guidance at the time they are admitted, the IRF has the 
opportunity to educate the patient and provide information on why they 
should become up to date with their COVID-19 vaccination. IRFs may also 
choose to administer the vaccine to the patient prior to their 
discharge from the IRF or coordinate a follow-up visit for the patient 
to obtain the vaccine at their physician's office or local pharmacy.
(b) Item Testing
    The measure development contractor conducted testing of the 
proposed standardized patient/resident COVID-19 vaccination coverage 
assessment item for the proposed Patient/Resident COVID-19 Vaccine 
measure using patient scenarios, draft guidance manual coding 
instructions, and cognitive interviews to assess IRFs' comprehension of 
the item and the associated guidance. A team of clinical experts 
assembled by our measure development contractor developed these patient 
scenarios to represent the most common scenarios that IRFs would 
encounter. The results of the item testing demonstrated that IRFs that 
used the draft guidance manual coding

[[Page 51028]]

instructions had strong agreement (that is, 84 percent) with the 
correct responses, supporting its reliability. The testing also 
provided information to improve both the item itself and the 
accompanying guidance.
(2) Competing and Related Measures
    Sections 1886(j)(7)(D)(i) and 1899B(e)(2)(A) of the Act require 
that, absent an exception under sections 1886(j)(7)(D)(ii) and 
1899B(e)(2)(B) of the Act, measures specified under section 
1886(j)(7)(D)(i) of the Act and section 1899B of the Act must be 
endorsed by a CBE with a contract under section 1890(a) of the Act. In 
the case of a specified area or medical topic determined appropriate by 
the Secretary for which a feasible and practical measure has not been 
endorsed, sections 1886(j)(7)(D)(ii) and 1899B(e)(2)(B) of the Act 
permit the Secretary to specify a measure that is not so endorsed, as 
long as due consideration is given to the measures that have been 
endorsed or adopted by a consensus organization identified by the 
Secretary. The proposed Patient/Resident COVID-19 Vaccine measure is 
not CBE endorsed, and after review of other endorsed and adopted 
measures, we were unable to identify any measures endorsed or adopted 
by a consensus organization for IRFs focused on capturing COVID-19 
vaccination coverage of IRF patients. We found only one related measure 
addressing COVID-19 vaccination, the COVID-19 Vaccination Coverage 
among Healthcare Personnel measure, adopted for the FY 2023 IRF QRP (86 
FR 42385 through 42396), which captures the percentage of HCP who 
receive a complete COVID-19 primary vaccination course.
    Therefore, after consideration of other available measures that 
assess COVID-19 vaccination rates among IRF patients, we believe the 
exceptions under sections 1886(j)(7)(D)(ii) and 1899B(e)(2)(B) of the 
Act apply. We intend to submit the proposed measure for consideration 
of endorsement by the CBE when feasible.
(3) Interested Parties and Technical Expert Panel (TEP) Input
    First, the measure development contractor convened a focus group of 
patient and family/caregiver advocates (PFAs) to solicit input. The 
PFAs felt a measure capturing raw vaccination rate, irrespective of IRF 
action, would be most helpful in patient and family/caregiver decision-
making. Next, TEP meetings were held on November 19, 2021 and December 
15, 2021 to solicit feedback on the development of patient/resident 
COVID-19 vaccination measures and assessment items for the PAC 
settings. The TEP panelists voiced their support for PAC patient/
resident COVID-19 vaccination measures and agreed that developing a 
measure to report the rate of vaccination in an IRF setting without 
denominator exclusions was an important goal. We considered the TEP's 
recommendations, and we applied the recommendations where technically 
feasible and appropriate. A summary of the TEP proceedings titled 
Technical Expert Panel (TEP) for the Development of Long-Term Care 
Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), Skilled 
Nursing Facility (SNF)/Nursing Facility (NF), and Home Health (HH) 
COVID-19 Vaccination-Related Items and Measures Summary Report is 
available on the CMS MMS Hub.\165\
---------------------------------------------------------------------------

    \165\ Technical Expert Panel (TEP) for the Development of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) COVID-19 Vaccination-Related Items and Measures Summary 
Report. https://mmshub.cms.gov/sites/default/files/COVID19-Patient-Level-Vaccination-TEP-Summary-Report-NovDec2021.pdf.
---------------------------------------------------------------------------

    To seek input on the importance, relevance, and applicability of a 
patient/resident COVID-19 vaccination coverage measure, we also 
solicited public comments in an RFI for publication in the FY 2023 IRF 
PPS proposed rule (87 FR 47038).\166\ Comments were generally positive 
on the concept of a measure addressing COVID-19 vaccination coverage 
among IRF patients. Some commenters included caveats with their support 
and requested further details regarding measure specifications and CBE 
endorsement. In addition, commenters voiced concerns regarding the 
evolving recommendations related to boosters and the definition of ``up 
to date,'' as well as whether an IRF length of stay would allow for 
meaningful distinctions among IRFs (87 FR 47071).
---------------------------------------------------------------------------

    \166\ 87 FR 20218.
---------------------------------------------------------------------------

(4) Measure Applications Partnership (MAP) Review
    The pre-rulemaking process includes making publicly available a 
list of quality and efficiency measures, called the Measures Under 
Consideration (MUC) List, that the Secretary is considering adopting 
for use in Medicare programs. This allows interested parties to provide 
recommendations to the Secretary on the measures included on the list. 
The Patient/Resident COVID-19 Vaccine measure was included on the 
publicly available 2022 MUC List for the IRF QRP.\167\
---------------------------------------------------------------------------

    \167\ Centers for Medicare & Medicaid Services. (2022). Overview 
of the List of Measures Under Consideration for December 1, 2022. 
https://mmshub.cms.gov/sites/default/files/2022-MUC-List-Overview.pdf.
---------------------------------------------------------------------------

    After the MUC List was published, the MAP received five comments 
from interested parties. Commenters were mostly supportive of the 
measure and recognized the importance of patients' COVID-19 
vaccination, and that measurement and reporting is one important method 
to help healthcare organizations assess their performance in achieving 
high rates of up to date vaccination. One commenter noted that patient 
engagement is critical at this stage of the pandemic, while another 
noted the criteria for inclusion in the numerator and denominator 
provide flexibility for the measure to remain relevant to current 
circumstances. Another commenter anticipated minimal implementation 
challenges since healthcare providers are already asking for patients' 
COVID-19 vaccination status at intake. Commenters who were not 
supportive of the measure raised several issues, including that the 
measure does not capture quality of care, concern about the evolving 
definition of the term ``up to date,'' that data collection would be 
burdensome, that administering the vaccine could impact the IRF 
treatment plan, and that a measure only covering one quarter may not be 
meaningful.
    Subsequently, several MAP workgroups met to provide input on the 
proposed measure. First, the MAP Health Equity Advisory Group convened 
on December 6, 2022. One MAP Health Equity Advisory Group member noted 
that the percentage of true contraindications for the COVID-19 vaccine 
is low, and the lack of exclusions on the measure is reasonable in 
order to minimize variation in what constitutes a 
contraindication.\168\ Similarly, the MAP Rural Health Advisory Group 
met on December 8, 2022, and requested clarification of the term ``up 
to date'' and noted concerns with the perceived level of burden for 
collection of data .\169\
---------------------------------------------------------------------------

    \168\ CMS Measures Management System (MMS). Measure 
Implementation: Pre-rulemaking MUC Lists and MAP reports. https://mmshub.cms.gov/measure-lifecycle/measure-implementation/pre-rulemaking/lists-and-reports.
    \169\ CMS Measures Management System (MMS). Measure 
Implementation: Pre-rulemaking MUC Lists and MAP reports. https://mmshub.cms.gov/measure-lifecycle/measure-implementation/pre-rulemaking/lists-and-reports.
---------------------------------------------------------------------------

    Next, the MAP PAC/LTC workgroup met on December 12, 2022. The MAP 
PAC/LTC workgroup's voting members

[[Page 51029]]

raised concerns brought up in public comments, such as provider 
actionability, lack of denominator exclusions, requirements for 
assessing patient vaccination status, evolving COVID-19 vaccination 
recommendations, and data reporting frequency for this measure. 
Additionally, MAP PAC/LTC workgroup members noted the potential 
inability of IRFs to administer the vaccine due to the shorter average 
length of stay as compared to other PAC settings. In response to 
workgroup member feedback, we noted that the intent of the Patient/
Resident COVID-19 Vaccine measure would be to promote transparency of 
data for patients to make informed decisions regarding care and is not 
intended to be a measure of IRF action. We also explained that this 
measure does not have exclusions for patient refusal since this measure 
was intended to report raw rates of vaccination, and this information 
is important for consumer choice. Additionally, we believe that PAC 
providers, including IRFs, are in a unique position to leverage their 
care processes to increase vaccination coverage in their settings to 
protect patients and prevent negative outcomes. We also noted that 
collection of these data will not require additional documentation or 
proof of vaccination. We clarified that the Patient/Resident COVID-19 
Vaccine measure would include the definition of up to date, so the 
measure would consider future changes in the CDC guidance regarding 
COVID-19 vaccination. We also clarified that the measure would continue 
to be a quarterly measure similar to the existing HCP COVID-19 Vaccine 
measure, as CDC has not determined whether COVID-19 is, or will be, a 
seasonal disease like influenza. Finally, we noted that the average 12-
day length of stay at IRFs is generally longer than patient stays at 
acute care hospitals. Given that health care is a continuum and every 
contact along the continuum provides an opportunity to encourage 
vaccination, IRFs have sufficient time to act on the patient's 
vaccination status. However, the MAP PAC/LTC workgroup reached a 60 
percent consensus on the vote of ``Do not support for rulemaking'' for 
this measure.\170\
---------------------------------------------------------------------------

    \170\ CMS Measures Management System (MMS). Measure 
Implementation: Pre-rulemaking MUC Lists and MAP reports. https://mmshub.cms.gov/measure-lifecycle/measure-implementation/pre-rulemaking/lists-and-reports.
---------------------------------------------------------------------------

    The MAP received four comments from industry commenters in response 
to the MAP PAC/LTC workgroup's recommendations. Interested parties 
generally understood the importance of COVID-19 vaccinations in 
preventing the spread of COVID-19, although a majority of commenters 
did not recommend the inclusion of the proposed Patient/Resident COVID-
19 Vaccine measure for the IRF QRP and raised several concerns. 
Specifically, commenters were concerned about vaccine hesitancy and 
providers' inability to influence results based on factors outside of 
their control. Commenters also noted that the measure has not been 
fully tested and encouraged CMS to monitor the measure for unintended 
consequences and ensure that the measure has meaningful results. One 
commenter raised concerns on whether patients' vaccination information 
would be easily available to IRFs as well as potential limitations with 
patients recounting vaccination status. One commenter was in support of 
the measure and provided recommendations for CMS to consider adding an 
exclusion for medical contraindications and submitting the measure for 
CBE endorsement.
    Finally, the MAP Coordinating Committee convened on January 24, 
2023, and noted concerns which were previously discussed in the MAP 
PAC/LTC workgroup, such as potential disruption to patient therapy due 
to vaccination and acuity of patients in the IRF setting. However, a 
MAP Coordinating Committee member noted that a patient's potential 
inability to complete rehabilitation was not a valid reason to withhold 
support of this measure, and that, because these patients have a high 
acuity, they are more vulnerable to COVID-19, further emphasizing the 
need to vaccinate them. MAP Coordinating Committee members also raised 
concerns discussed previously during the MAP PAC/LTC workgroup, 
including the shorter IRF length of stay and excluding medical 
contraindications from the denominator.
    The MAP Coordinating Committee recommended three mitigation 
strategies for the Patient/Resident COVID-19 Vaccine measure: (i) 
reconsider exclusions for medical contraindications, (ii) complete 
reliability and validity measure testing, and (iii) seek CBE 
endorsement. The MAP Coordinating Committee ultimately reached 81 
percent consensus on its voted recommendation of ``Do not support with 
potential for mitigation.'' Despite the MAP Coordinating Committee's 
vote, we believe it is still important to propose the Patient/Resident 
COVID-19 Vaccine measure for the IRF QRP. As we stated in section 
VIII.C.2.a.(3) of the proposed rule, we did not include exclusions for 
medical contraindications because the PFAs we met with told us that a 
measure capturing raw vaccination rate, irrespective of any medical 
contraindications, would be most helpful in patient and family/
caregiver decision-making. We do plan to conduct reliability and 
validity measure testing once we have collected enough data, and we 
intend to submit the proposed measure to the CBE for consideration of 
endorsement when feasible. We refer readers to the final MAP 
recommendations, titled 2022-2023 MAP Final Recommendations.\171\
---------------------------------------------------------------------------

    \171\ 2022-2023 MAP Final Recommendations. https://mmshub.cms.gov/sites/default/files/2022-2023-MAP-Final-Recommendations-508.xlsx
---------------------------------------------------------------------------

(5) Quality Measure Calculation
    The proposed Patient/Resident COVID-19 Vaccine measure is an 
assessment-based process measure that reports the percent of stays in 
which patients in an IRF are up to date on their COVID-19 vaccinations 
per the CDC's latest guidance.\172\ This measure has no exclusions and 
is not risk adjusted.
---------------------------------------------------------------------------

    \172\ The definition of ``up to date'' may change based on CDC's 
latest guidelines and is available on the CDC web page, ``Stay Up to 
Date with COVID-19 Vaccines Including Boosters,'' at https://www.cdc.gov/coronavirus/2019-ncov/vaccines/stay-up-to-date.html 
(updated March 2, 2023).
---------------------------------------------------------------------------

    The numerator for the proposed measure would be the total number of 
IRF stays in the denominator in which patients are up to date with 
their COVID-19 vaccination per CDC's latest guidance. The denominator 
for the proposed measure would be the total number of IRF stays 
discharged during the reporting period.
    The data source for the proposed Patient/Resident COVID-19 Vaccine 
measure is the IRF-PAI for IRF patients. For more information about the 
proposed data submission requirements, we refer readers to section 
VIII.F.3. of the proposed rule. For additional technical information 
about this proposed measure, we refer readers to the draft measure 
specifications document titled COVID-19 Vaccine: Percent of Patients/
Residents Who Are Up to Date Draft Measure Specifications.\173\ 
available on the IRF QRP Measures and Technical Information web page.
---------------------------------------------------------------------------

    \173\ COVID-19 Vaccine: Percent of Patients/Residents Who Are Up 
to Date Draft Measure Specifications. https://www.cms.gov/files/document/patient-resident-covid-vaccine-draft-specs.pdf.
---------------------------------------------------------------------------

    We invited public comments on the proposal to adopt the Patient/
Resident COVID-19 Vaccine measure beginning with the FY 2026 IRF QRP. 
The following is a summary of the public

[[Page 51030]]

comments received on our proposal and our responses.
    Comment: One commenter supported the measure noting it does not add 
significant burden.
    Response: We thank the commenter for their support.
    A number of commenters did not support the proposal to adopt the 
Patient/Resident COVID-19 Vaccine measure to the IRF QRP for various 
reasons. The following is a summary of these public comments received 
on our proposal and our responses.
    Comment: One commenter agreed with CMS's proposed justification 
that the measure has the potential to drive COVID-19 vaccination uptake 
among IRF patients and prevent the spread of COVID-19 in the IRF 
population and agreed that the measure could help empower consumers in 
making decisions about their care. Despite this, they still urged CMS 
to ensure that measures are appropriately specified and adequately 
tested and validated prior to implementation. This commenter also noted 
that, unlike the proposed HCP COVID-19 Vaccine measure, the 
specifications for this Patient/Resident COVID-19 Vaccine measure 
solely reference the definition of up to date as described on CDC's 
``Stay Up to Date'' website. Even though this definition more 
accurately reflects the most current Advisory Committee on Immunization 
Practices (ACIP) recommendation, the commenter urged CMS to ensure that 
this approach to specifying measures is valid and will not serve to 
cause confusion or reporting challenges in the future.
    However, several commenters did not support the proposal due to the 
measure not being fully tested for reliability and validity, and one 
commenter noted that even CMS stated that the measure would need to be 
tested for reliability and validity once enough data were collected. 
One commenter said it was unclear whether it is feasible for PAC 
facilities to collect and report information for the proposed measure. 
Another one of these commenters suggested CMS ``rushed through'' the 
validation process to add the measure to the IRF QRP as soon as 
possible because there is no support showing the measure is practical 
or feasible. Some commenters also encouraged CMS to delay 
implementation of the measure in the IRF QRP until the measure had been 
fully tested.
    Response: We are pleased that the commenter agrees with CMS's 
proposed rationale that the measure has the potential to drive COVID-19 
vaccination uptake among IRF patients, prevent the spread of COVID-19 
in the IRF population, and empower consumers in making decisions about 
their care.
    We also acknowledge the concerns brought up regarding the measure 
not being tested yet and commenters' reasons for not supporting the 
measure. However, we have tested the item proposed for the IRF-PAI to 
capture data for this measure and its feasibility and appropriateness. 
Since a COVID-19 vaccination item does not yet exist within the IRF-
PAI, we developed clinical vignettes to test item-level reliability of 
a draft Patient/Resident COVID-19 Vaccine item for the IRF-PAI. The 
clinical vignettes were a proxy for patient records with the most 
common and challenging cases providers would encounter, similar to the 
approach that CMS uses to train providers on all new assessment items, 
and the results demonstrated strong agreement (that is, 84 percent).
    Validity testing has not yet been completed, since the COVID-19 
vaccination item does not yet exist on the IRF-PAI. However, the 
Patient/Resident COVID-19 Vaccine measure was constructed based on 
prior use of similar items, such as the Percent of Residents or 
Patients Who Were Assessed and Appropriately Given the Seasonal 
Influenza Vaccine (Short Stay) for the IRF QRP and LTCH QRP.\174\ We 
have used these types of patient/resident vaccination assessment items 
in the calculation of vaccination quality measures in our PAC QRPs and 
intend to conduct reliability and validity testing for this specific 
Patient/Resident COVID-19 Vaccine measure once the COVID-19 vaccination 
item has been added to the IRF-PAI and we have collected sufficient 
data.
---------------------------------------------------------------------------

    \174\ 78 FR 47859 and 77 FR 53257.
---------------------------------------------------------------------------

    Additionally, we solicited feedback from our TEP on the proposed 
assessment item and its feasibility. No concerns were raised by the TEP 
regarding obtaining information required to complete the new COVID-19 
vaccination item.\175\
---------------------------------------------------------------------------

    \175\ Technical Expert Panel (TEP) for the Development of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) COVID-19 Vaccination-Related Items and Measures Summary 
Report. https://mmshub.cms.gov/sites/default/files/COVID19-Patient-Level-Vaccination-TEP-Summary-Report-NovDec2021.pdf.
---------------------------------------------------------------------------

    Comment: Several commenters did not support the measure and cited 
the CBE's MAP 2022-2023 review cycle where the MAP failed to reach 
consensus, and ultimately did not recommend the measure for rulemaking. 
One commenter said they were deeply concerned about the proposal to add 
the Patient/Resident COVID-19 Vaccine measure because it appeared as 
though CMS disregarded the recommendations of the MAP. Several of the 
commenters noted that the MAP is a multi-stakeholder panel of experts 
representing providers, patients, and payers, and encouraged CMS to 
address the MAP's concerns about the measure, including adding 
exclusions in the measure, conducting measure testing, and submitting 
the measure for CBE endorsement prior to adopting it in the IRF QRP.
    Response: As part of the pre-rulemaking process, HHS takes into 
consideration the recommendations of the MAP in selecting candidate 
quality and efficiency measures. HHS selects candidate measures and 
publishes proposed rules in the Federal Register, which allows for 
public comment and further consideration before a final rule is issued. 
If the CBE under contract with CMS has not endorsed a candidate 
measure, then HHS must publish a rationale for the use of the measure 
described in section 1890(b)(7)(B) of the Act in the notice. The MAP 
Coordinating Committee recommended three mitigation strategies for the 
Patient/Resident COVID-19 Vaccine measure: (i) reconsider exclusions 
for medical contraindications, (ii) complete reliability and validity 
measure testing, and (iii) seek CBE endorsement. We would like to 
reiterate that this measure is intended to promote transparency of data 
for patients/caregivers to make informed decisions for selecting 
facilities, providing potential patients and their caregivers with an 
important piece of information regarding vaccination rates as part of 
their process of identifying providers they would want to seek care 
from. As we stated in section IX.C.2.a.(3) of this final rule, we did 
not include exclusions for medical contraindications because the PFAs 
we met with told us that a measure capturing raw vaccination rate, 
irrespective of any medical contraindications, would be most helpful in 
patient and family/caregiver decision-making. We intend to add a new 
item to the IRF-PAI assessment tool to collect this information. We 
will then conduct measure testing once sufficient data on the COVID-19 
vaccination item are collected through the IRF-PAI and plan to submit 
the measure for CBE endorsement when it is technically feasible to do 
so.
    Comment: A few commenters believe the adoption of a patient-level 
measure of COVID-19 vaccination status might quickly become topped out 
due to lack of meaningful improvement in the

[[Page 51031]]

vaccination rate, comparing it to the Percent of Residents of Patients 
Who Were Assessed and Appropriately Given the Seasonal Influenza 
Vaccine (CBE #0680) that was removed from the IRF QRP measure set in 
the FY 2019 IRF PPS final rule (83 FR 38514). One of these commenters 
also stated that IRF performance on this proposed measure will fail to 
show meaningful distinctions in improvements since 94.3 percent of the 
United States population at least 65 years of age had completed their 
primary series as of May 2023.
    Response: We do not believe this measure is at risk of being 
retired early. The Patient/Resident COVID-19 Vaccine measure reports 
the percentage of patients in an IRF who are up to date on their COVID-
19 vaccinations per the CDC's latest guidance, rather than capturing 
the rates of primary vaccination series only. Because the measure 
reflects an up to date vaccination status, it minimizes the potential 
for topping out. We believe that continued monitoring of up to date 
vaccination among patients will remain an important tool to minimize 
severe illness, hospitalization, and death in PAC facilities. 
Additionally, we believe there is substantial room for improvement in 
measure performance. As of May 2023, while the vaccination rates among 
people 65 and older were high for the primary vaccination series (94.3 
percent), the vaccination rates were lower for the first booster dose 
(73.9 percent among those who received a primary series) and even lower 
for the second booster dose (60.4 percent among those who received a 
first booster).\176\
---------------------------------------------------------------------------

    \176\ Centers for Disease Control and Prevention. COVID-19 
vaccination age and sex trends in the United States, national and 
jurisdictional. May 11, 2023. https://data.cdc.gov/Vaccinations/COVID-19-Vaccination-Age-and-Sex-Trends-in-the-Uni/5i5k-6cmh.
---------------------------------------------------------------------------

    Comment: A few commenters were concerned that the Yes/No response 
options for the COVID-19 vaccination item in the IRF-PAI may be 
unreliable and lead to inaccurate and inconsistent reporting of data. 
One of these commenters noted that they are also concerned that a self-
reported up to date answer might not be accurate, which could lead to 
incorrect timing for the next dosage or inaccurate reporting overall. 
Two of these commenters said that it is unlikely most patients would 
have an understanding of the CDC's specific definition of up to date 
when answering a yes/no question for the patient assessment, which 
could also lead to potentially inaccurate data.
    Response: We disagree with the commenters. The results of the item 
testing conducted to test the COVID-19 vaccination item supported the 
use of a Patient-level COVID-19 Vaccination Coverage measure item. When 
the item was tested as drafted in the measure specifications with Yes/
No response options, overall agreement for IRFs was 84 percent. Across 
all provider types, those who used the CDC website, or the guidance 
manual and the CDC website had the highest percent agreement (100 
percent and 88 percent, respectively). We also believe the provision of 
two response options helps alleviate provider burden of providing 
additional details and information regarding the patient's vaccination 
status. Our TEP panelists indicated that they generally prefer items 
with less information in order to reduce IRFs' burden and that the 
nuance provided by the ``more information'' options could add 
additional burden and potential confusion.\177\ Additionally, coding 
guidance for this item would allow providers to use all sources of 
information available to obtain the vaccination data, such as patient 
interviews, medical records, proxy response, and vaccination cards 
provided by the patient or their caregivers.\178\ As with any other 
assessment item on the IRF-PAI, we expect IRF providers to work closely 
with the patient to obtain the most accurate response to the assessment 
question.
---------------------------------------------------------------------------

    \177\ Technical Expert Panel (TEP) for the Development of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) COVID-19 Vaccination-Related Items and Measures Summary 
Report. https://mmshub.cms.gov/sites/default/files/COVID19-Patient-Level-Vaccination-TEP-Summary-Report-NovDec2021.pdf.
    \178\ COVID-19 Vaccine: Percent of Patients/Residents Who Are Up 
to Date Draft Measure Specifications. https://www.cms.gov/files/document/patient-resident-covid-vaccine-draft-specs.pdf.
---------------------------------------------------------------------------

    Comment: A few commenters were concerned that the measure does not 
provide response options for patients who refuse to answer, refuse the 
vaccination, or are excluded due to medical contraindications or 
closely held religious beliefs. Another commenter urged CMS to consider 
adding an exclusion for medical contraindications, while still another 
noted that CMS has failed to address the recommendations of the CBE to 
explore adding medical exemptions to the measure.
    Response: We understand and thank the commenters for their 
recommendations about adding exclusions to the measure. Our measure 
development contractor convened a focus group of PFAs as well as a TEP 
that included interested parties from every PAC setting, to solicit 
input on patient/resident COVID-19 vaccination measures and assessment 
items. The PFAs told us that a measure capturing raw vaccination rates 
would be most helpful in patient and family/caregiver decision-making. 
Our TEP agreed that developing a measure to report the rate of 
vaccination without denominator exclusions was an important goal.\179\ 
Based on this feedback, we believe excluding patients/residents with 
contraindications from the measure would distort the intent of the 
measure of providing raw COVID-19 patient vaccination rates, while 
making the information more difficult for patients/caregivers to 
interpret, and therefore we did not include any exclusions.
---------------------------------------------------------------------------

    \179\ Technical Expert Panel (TEP) for the Development of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) COVID-19 Vaccination-Related Items and Measures Summary 
Report. https://mmshub.cms.gov/sites/default/files/COVID19-Patient-Level-Vaccination-TEP-Summary-Report-NovDec2021.pdf.
---------------------------------------------------------------------------

    Comment: Several commenters were concerned regarding the lack of a 
well-defined definition of up to date, and the burden it poses on 
providers to collect this data. One commenter said the ``moving target 
definition'' contributes to concerns about the reliability of the data 
collected. One commenter believed that the current specifications are 
flawed since the current numerator specifications refers the end user 
to a website outlining when primary and additional/booster dose(s) are 
recommended and stated that this lack of a well-defined set of 
specifications could negatively impact the reliability and validity of 
the measure.
    Response: The up to date concept is not new to providers and is 
currently in use by Nursing Home facilities for the short-stay and 
long-stay Percent of Residents Assessed and Appropriately Given the 
Pneumococcal Vaccine and Percent of Residents Who Received the 
Pneumococcal Vaccine measures. Beyond the historical use of this 
concept, ensuring that standards of care are up to date according to 
the relevant authorities remains a widespread goal for all providers. 
We believe that IRF providers should be staying current on the latest 
care guidelines for COVID-19 vaccination as part of best practice. 
Further, the IRF-PAI Guidance Manual will indicate how to code the item 
and providers could access the CDC website at any time to find the 
definition of up to date. The CDC has published FAQs that clearly state 
the definition of up to

[[Page 51032]]

date.\180\ In fact, when we tested the COVID-19 vaccination item, there 
was strong agreement with the correct responses when facilities used 
the available guidance, and rates of correct responses increased when 
facilities accessed the CDC website. Across all provider types, those 
who used the CDC website, or the guidance manual and the CDC website, 
had the highest percent agreement (100 percent and 88 percent 
respectively).
---------------------------------------------------------------------------

    \180\ Centers for Disease Control and Prevention. Frequently 
Asked Questions. May 15, 2023. https://www.cdc.gov/coronavirus/2019-ncov/vaccines/faq.html.
---------------------------------------------------------------------------

    Comment: One commenter noted that some patient stays may overlap 
between the period when new additional/booster dose(s) become available 
and/or the definition of up to date changes and requested clarification 
on how providers should account for such ``bridge'' cases.
    Response: Given this assessment item is completed at discharge, 
providers would code the item using guidance in place at the time of 
the patient's discharge. As previously discussed, this measure does not 
mandate or require patients to be up to date with their COVID-19 
vaccination. IRFs are successfully able to report the measure, and 
comply with the IRF QRP requirements, irrespective of the number of 
patients who have been vaccinated.
    Comment: Another commenter was concerned regarding the uncertainty 
about the seasonality of COVID-19, future vaccination schedules, and 
how often new versions of a COVID-19 vaccine will be available.
    Response: Beyond the historical use of the concept of up to date, 
ensuring that standards of care are up to date according to the 
relevant authorities remains a widespread goal for all providers. As 
the SARS-CoV-2 virus mutates, this vaccination measure takes a forward-
thinking approach to ensure that PAC patients are protected in the 
event of COVID-19 infection. Given that CDC guidelines may change over 
time in response to the virus, we believe the use of up to date will 
actually be simpler for facilities since it ensures that the measure 
specifications, item responses, and accompanying item guidance would 
not have to continually change. Additionally, CMS regularly reviews its 
measures as part of the measure maintenance process, and will re-
specify the measure in the future, if needed, based on any changes to 
guidelines.
    A number of commenters were concerned about the burden this measure 
places on providers and listed several types of burden including 
difficulty with data collection and keeping up with the definition of 
up to date. The following is a summary of those comments and our 
responses.
    Comment: Two commenters believe the proposed measure will pose 
unique challenges due to patients' different comorbidities and 
preexisting conditions that may impact which vaccine recommendation 
applies to them, and they believe that complying with the CDC 
guidelines may be challenging and time consuming for IRFs, especially 
if CDC revises its guidance. One of the commenters also noted that 
given the potential that there could be audits related to the COVID-19 
vaccine measures, that increased time, personnel and financial 
resources would be required to collect and report the required data for 
these measures, and they believe those resources would be better 
utilized for direct patient care and other quality improvement 
activities that more closely align with the primary mission of IRFs.
    Response: We disagree that this measure, if finalized, would take 
time away from patient care. We believe PAC providers should be 
assessing whether patients are up to date with COVID-19 vaccination as 
a part of their care, and even if they do not administer the vaccine, 
they can coordinate follow-up care for the patient to obtain the 
vaccine elsewhere. During our item testing, we heard from providers 
that they are routinely inquiring about COVID-19 vaccination status 
when admitting patients. CMS is committed to providing Medicare 
beneficiaries with high quality health care and therefore, routinely 
performs audits and reviews to ensure the standard of IRF care is 
maintained. We believe providers need to exercise due diligence as they 
stay abreast of standards of care and new evidence, as it becomes 
available. We believe IRFs consider vaccination essential to patient 
safety and quality care.
    Gathering information about a patient's vaccination status is an 
important part of developing and administering a comprehensive plan of 
care. Rather than taking time away from patient care, providers will be 
documenting information they are likely already collecting through the 
course of providing care to the patients. We would remind providers 
that IRFs are currently required to meet the IRF QRP requirements as 
authorized by section 1886(j)(7) of the Act, and it applies to 
freestanding IRFs, as well as inpatient rehabilitation units of 
hospitals or Critical Access Hospitals (CAHs) paid by Medicare under 
the IRF PPS.
    Comment: Two commenters believe that, as the CDC updates 
eligibility requirements for the latest versions of the COVID-19 
vaccine, keeping track of eligibility and what is considered up to date 
will be difficult for IRFs. One of these commenters stated that data 
infrastructure would be needed to capture the non-static definition of 
up to date to reassess vaccine status with each new revision of the 
reporting definition, and this would result in a heavy burden on data 
collection, analysis, and reporting programs.
    Response: We recognize that the up to date COVID-19 vaccination 
definition may evolve due to the changing nature of the virus, but we 
are also confident in IRFs' ability to understand these changes as they 
have been at the front lines of managing COVID-19 since the beginning 
of the pandemic. The public health response to COVID-19 has necessarily 
adapted to respond to the changing nature of the virus's transmission 
and community spread. As mentioned in the FY 2022 IRF PPS final rule 
(86 FR 42386), we received several public comments during the HCP 
COVID-19 Vaccine measure's pre-rulemaking process encouraging us to 
continue to evaluate the new evidence on COVID-19 as it continues to 
arise and we stated our intention to continue to work with partners, 
including FDA and CDC. We believe that the proposed measure aligns with 
the Administration's responsive approach to COVID-19 and will continue 
to support vaccination as the most effective means to prevent the worst 
consequences of COVID-19, including severe illness, hospitalization, 
and death. However, IRFs can choose how they want to manage tracking 
CDC information.
    Comment: A few commenters noted that collecting this information 
would be especially burdensome in cases where patients are unable or 
unwilling to provide the necessary information. One of these commenters 
also stated that patients will have cognitive, communication, and 
memory deficits that will cause barriers to appropriate communication 
and understanding of their vaccination status.
    Response: As noted in the COVID-19 Vaccine: Percent of Patients/
Residents Who Are Up to Date Draft Measure Specifications,\181\ 
providers will be able to use multiple sources of information available 
to obtain the vaccination data, such as patient interviews, medical

[[Page 51033]]

records, proxy response, and vaccination cards provided by the patient 
or their caregivers. Therefore, coding of this item in the IRF-PAI 
would not be limited to a patient's oral response. As with any 
assessment item, we will also publish coding guidance and instructions 
to further assist providers in collection of these data.
---------------------------------------------------------------------------

    \181\ COVID-19 Vaccine: Percent of Patients/Residents Who Are Up 
to Date Draft Measure Specifications. https://www.cms.gov/files/document/patient-resident-covid-vaccine-draft-specs.pdf.
---------------------------------------------------------------------------

    Comment: Commenters did not support the measure stating that IRFs 
do not typically administer vaccines and it would be an undue burden 
for rehabilitation units to store, provide, and report the 
administration of the COVID-19 vaccine.
    Response: This measure does not require IRF providers to administer 
the vaccine to the patients. While we know of no current indications of 
shortages or delays for the COVID-19 vaccines in IRF facilities and 
believe that facilities should be able to administer the vaccine if a 
patient is agreeable to receiving the vaccination, IRFs do not have to 
administer the vaccine themselves. They can arrange for the patient to 
obtain the vaccine outside of their facility or can work with community 
pharmacies to obtain vaccines.
    Several commenters did not support the measure as they do not think 
it is a measure of quality of care due to a lack of correlation between 
the vaccine uptake of patients and the quality of care a patient can 
expect when being admitted for a stay at an IRF and the inability of 
IRFs to affect the results. Commenters disagreed with CMS's statement 
in the proposed rule (86 FR 21000) that ``PAC providers, including 
IRFs, are in a unique position to leverage their care processes to 
increase vaccination coverage in their setting to protect patients and 
prevent negative outcomes.'' One commenter expressed significant 
logistical and clinician concerns with the proposal and its ability to 
quantify quality of care. They gave several reasons, which we address 
below.
    Comment: Two commenters noted that IRFs do not have immediate or 
ongoing access to COVID-19 vaccines and/or booster dose(s)s and will 
have difficulty reporting and demonstrating improvement on this 
measure.
    Response: While we believe facilities should be able to administer 
the vaccine if a patient is agreeable to receiving the vaccination, 
this measure does not require IRFs to administer the vaccine 
themselves. There are no current indications that there are vaccine 
shortages or delays for the COVID-19 vaccines in PAC facilities. 
However, IRFs can arrange for the patient to obtain the vaccine outside 
of their facility or can work with community pharmacies to obtain 
vaccines. We would also like to point out that the number of patients 
who have been vaccinated by an IRF does not impact an IRF's ability to 
successfully report the measure to comply with the requirements of the 
IRF QRP.
    Comment: Several commenters believe it is often infeasible or 
inappropriate to offer vaccination for patients due to length of stay, 
ability to manage side effects and medical contraindications, or other 
logistical challenges to gathering information from a patient who may 
have received care from multiple proximal providers. One commenter said 
that administering the vaccine could cause a readmission back to acute 
care or delay the patient's course of rehabilitation and extend their 
length of stay beyond the average time frame for which they receive 
payment. Therefore, these things would make it difficult for IRFs to 
manage and potentially improve their performance on this measure.
    Response: We understand concerns about PAC length of stay or effect 
of the vaccine on patient care. We believe providers should use 
clinical judgement to determine if a patient is eligible to receive the 
vaccination and avoid harm to the patient. It is the responsibility of 
the IRFs to determine when a patient is ready for discharge, keeping in 
mind patient's health and safety, which may necessitate a longer length 
of stay.
    However, we also believe that vaccination for high-risk 
populations, such as those in IRFs, is of paramount importance, and 
regardless of length of stay, a provider has the opportunity to educate 
the patient and provide information on why they should become up to 
date with COVID-19 vaccination, if they are not up to date at the time 
they are admitted. We believe vaccines can be scheduled at times that 
prevent or minimize disruptions with the patient treatment plan. For 
example, the vaccine could be given on a weekend or prior to discharge 
if the patient chooses to receive it. We would also like to point out 
that this measure does not mandate patients to be up to date with their 
COVID-19 vaccine. The number of patients who have been vaccinated in an 
IRF does not impact an IRF's ability to successfully report the measure 
to comply with the requirements of the IRF QRP.
    Comment: Other commenters said that most patients who are 
interested in receiving a vaccine have already received it from the 
referring hospital, long-term care hospital, skilled nursing facility 
or other setting where the patient received care prior to admission to 
the IRF, and therefore they did not think this measure would have an 
impact on the vaccination rates.
    Response: This measure is intended to provide the percent of 
patients who are up to date with their COVID-19 vaccination in an IRF 
at the time of discharge. This measure promotes transparency of raw 
data regarding COVID-19 vaccination rates for patients/caregivers to 
make informed decisions for selecting facilities. Irrespective of the 
patient's vaccination status, this measure will provide potential 
patients and their caregivers with an important piece of information 
regarding vaccination rates as part of their process of identifying 
providers they would want to seek care from, alongside other measures 
available on Care Compare, to make an informed, comprehensive decision. 
Additionally, we believe IRF providers would benefit in such situations 
where patients have already been vaccinated prior to admission, given 
this would mean the patient is up to date and reduce IRF burden to 
educate or vaccinate the patient.
    Comment: Several commenters list other factors affecting patient 
vaccination status outside of the IRF's control such as patient 
refusals and other cultural or religious reasons for a patient not 
receiving vaccination. One commenter believes COVID-19 vaccinations are 
still highly influenced by the political environment and political 
beliefs of patients/residents and their families. Therefore, they 
believe the percentage of patients who are vaccinated within an IRF 
will reflect the political leanings of the region in which the facility 
is located, and IRFs will not be able to influence this. Commenters 
noted that patients/residents may choose to forgo vaccination despite a 
provider's best efforts to encourage vaccination among their patients/
residents. One commenter stated that patients retain their right to 
decline a vaccine when they are admitted to an IRF and they believe 
patient acceptance of a vaccine does not measure an IRF's quality of 
care.
    Response: We appreciate providers' commitment to ensuring that 
patients are educated and encouraged to receive vaccinations, and we 
acknowledge that individual patients have a choice regarding whether to 
receive a COVID-19 vaccine or additional/booster dose(s), despite 
provider efforts. However, it is also true that patients and family/
caregivers have choices about selecting PAC providers, and it is our 
intention to empower them with the information they need to make an 
informed decision by publicly reporting the data we receive from IRFs 
on this measure. We

[[Page 51034]]

understand that despite provider efforts, there may be instances where 
a patient chooses not to be vaccinated, and we want to remind IRFs that 
this measure does not mandate that patients be up to date with their 
COVID-19 vaccine. The number of patients who have been vaccinated in an 
IRF does not impact an IRF's ability to successfully report the measure 
to comply with the requirements of the IRF QRP.
    Comment: One commenter said that even if the measure is intended to 
give patients and families information to make decisions about care, 
the lack of IRF access in many areas may reduce the impact of having 
IRFs collect this information. Several commenters believe the IRF's 
rate of vaccination will generally mirror the current COVID-19 
vaccination rate in an IRF's local community, which they do not believe 
is a reflection of an IRF's quality as a provider nor would it provide 
relevant or useful information through public reporting.
    Response: As described in section IX.C.2.a.(3) of this final rule, 
the measure development contractor convened TEP meetings to solicit 
feedback on the development of patient/resident COVID-19 vaccination 
measures. Analyses showed considerable variation in COVID-19 
vaccination rates among nursing homes by State and within State. 
Further, States with the lowest complete vaccination rates also show 
wider within-State variations in vaccination rates among nursing 
homes.\182\ The TEP panelists indicated that the presence of 
disparities in vaccination rates makes the patient-level vaccination 
measure meaningful to develop, and they broadly agreed that the 
vaccination gaps identified for nursing homes were also likely present 
within other PAC settings, including IRFs.\183\ Therefore, we believe 
that the information this measure will provide will still be valuable 
to potential IRF patients and their caregivers who have geographic 
limitations while seeking care. Additionally, this measure will provide 
potential patients and their caregivers with an important piece of 
information regarding vaccination rates as part of their process of 
identifying IRF providers they would want to seek care from, alongside 
other measures available on Care Compare to make a comprehensive 
decision.
---------------------------------------------------------------------------

    \182\ Technical Expert Panel (TEP) for the Development of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) COVID-19 Vaccination-Related Items and Measures Summary 
Report. https://mmshub.cms.gov/sites/default/files/COVID19-Patient-Level-Vaccination-TEP-Summary-Report-NovDec2021.pdf.
    \183\ Technical Expert Panel (TEP) for the Development of Long-
Term Care Hospital (LTCH), Inpatient Rehabilitation Facility (IRF), 
Skilled Nursing Facility (SNF)/Nursing Facility (NF), and Home 
Health (HH) COVID-19 Vaccination-Related Items and Measures Summary 
Report. https://mmshub.cms.gov/sites/default/files/COVID19-Patient-Level-Vaccination-TEP-Summary-Report-NovDec2021.pdf.
---------------------------------------------------------------------------

    Comment: Several commenters raised concerns about unintended 
consequences of receiving the vaccine during an IRF stay and believe 
they would interfere with a patient's therapy. They believe that 
scheduling a COVID-19 vaccine during a patient's relatively short 
length of stay, 12-13 days on average, could mean they have to forego 
several days of therapy they would otherwise need and be entitled to. 
One commenter noted that providers may have concerns that the side 
effects of a vaccine can interfere with or cause confusion while a 
patient is being diagnosed or treated during their hospitalization, and 
that the side effects of a vaccine like COVID-19 could delay needed 
intense therapy treatment. One commenter noted that the known side-
effects of the COVID-19 vaccine per the CDC, ``pain, redness, swelling 
at the injection site, tiredness, headache, muscle pain, chills, fever, 
and nausea,'' are contradictory to participating in intensive therapy, 
at least 3 hours a day, 5 days a week.
    Response: We understand and acknowledge commenters' concerns about 
potential side effects of COVID-19 vaccination on patient participation 
in IRF care and activities. However, vaccines can be scheduled at times 
that prevent or minimize disruptions to the patient treatment plan. For 
example, if an IRF is concerned about a patient's ability to perform in 
3 hours of therapy a day, the vaccine could be given on a weekend or 
prior to discharge. We support an IRF's use of clinical judgement to 
determine if a patient is eligible to receive the vaccination and if a 
patient chooses to receive one, to work with the patient to schedule 
the appropriate time to administer the vaccine. We also want to remind 
IRFs that they do not have to administer the COVID-19 vaccine. The 
number of patients who have been vaccinated in an IRF does not impact 
an IRF's ability to successfully report the measure to comply with the 
requirements of the IRF QRP
    Comment: One commenter pointed to the concerns raised by the MAP 
and other interested parties and believes CMS should consider the 
potential impacts of its approach on vaccination efforts. They caution 
that as providers are endeavoring to follow the vaccine guidelines and 
gain patient trust, this measure--as constructed--has the potential to 
adversely impact patient-provider relationships, trust, and provider 
performance.
    Response: We disagree with the commenter. We believe the proposed 
measure will support the goal of the CMS Meaningful Measure Initiative 
2.0 to ``Empower consumers to make good health care choices through 
patient-directed quality measures and public transparency objectives,'' 
and the PFAs we met with agreed that a measure capturing raw 
vaccination rates would be most helpful in patient and family/caregiver 
decision-making. Additionally, we take the appropriate access to care 
in IRFs very seriously, and routinely monitor the QRP measures' 
performance, including performance gaps across IRFs. We intend to 
monitor closely whether any proposed change to the IRF QRP has 
unintended consequences on access to care for high risk patients. 
Should we find any unintended consequences, we will take appropriate 
steps to address these issues in future rulemaking.
    Comment: Several commenters did not support adoption of this 
measure in light of the Administration's announcement of the end of the 
COVID-19 PHE on May 11. 2023. One of these commenters commended CMS for 
recognizing the burden of such a requirement included in the Hospital 
Conditions of Participation and working to remove it, but now questions 
the ``juxtaposition'' of proposing a vaccine uptake measure as a metric 
for quality of care. Another one of these commenters said that the end 
of the PHE will make it more challenging for patients to stay informed 
on the most recent guidance from the CDC. Finally, one of these 
commenters also brought up concerns about CDC's recent recommendations 
that individuals aged 65 and over ``may'' receive an additional dose of 
the updated vaccines.
    Response: Despite the announcement of the end of the COVID-19 PHE, 
many people continue to be affected by COVID-19, particularly seniors, 
people who are immunocompromised, and people with disabilities. As 
mentioned in the End of COVID-19 Public Health Emergency Fact 
Sheet,\184\ our response to the spread of SARS-CoV-2, the virus that 
causes COVID-19, remains a public health priority. Even beyond the end 
of the COVID-19 PHE, we will continue to work to protect Americans from 
the

[[Page 51035]]

virus and its worst impacts by supporting access to COVID-19 vaccines, 
treatments, and tests, including for people without health insurance. 
Given the continued impacts of COVID-19, we believe it is important to 
promote patient vaccination and education, which this measure aims to 
achieve. Accordingly, we are aligning our approach with those for other 
infectious diseases, such as influenza by encouraging ongoing COVID-19 
vaccination.\185\ Further, published coding guidance will indicate how 
to code the item taking into account CDC guidelines, and providers 
could access the CDC website at any time to find the definition of up 
to date. Lastly, this measure as proposed for the IRF QRP is not 
associated with the PHE declaration, or the Conditions of 
Participation. This measure is being proposed to address CMS's priority 
to empower consumers to make informed health care choices through 
patient-directed quality measures and public transparency, as with 
previous vaccination measures.
---------------------------------------------------------------------------

    \184\ U.S. Department of Health and Human Services. Fact Sheet: 
End of the COVID-19 Public Health Emergency. May 9, 2023. https://www.hhs.gov/about/news/2023/05/09/fact-sheet-end-of-the-covid-19-public-health-emergency.html.
    \185\ Medicare and Medicaid Programs; Policy and Regulatory 
Changes to the Omnibus COVID-19 Health Care Staff Vaccination 
Requirements; Additional Policy and Regulatory Changes to the 
Requirements for Long-Term Care (LTC) Facilities and Intermediate 
Care Facilities for Individuals With Intellectual Disabilities 
(ICFs-IID) To Provide COVID-19 Vaccine Education and Offer 
Vaccinations to Residents, Clients, and Staff; Policy and Regulatory 
Changes to the Long Term Care Facility COVID-19 Testing 
Requirements. (88 FR 36487).
---------------------------------------------------------------------------

    Comment: Two commenters noted that the draft item does not provide 
response options for patients who refuse to answer, refuse the 
vaccination, or are excluded due to medical contraindications or 
closely held religious beliefs. One commenter said that if CMS does add 
the measure to the IRF QRP, they must allow IRFs to report that they 
could not determine the patient's vaccination status. This commenter 
also noted that the CBE's MAP Health Equity Advisory Group ``expressed 
concerns about vaccine hesitancy due to cultural norms,'' and that if 
CMS adopts the proposed Patient/Resident COVID-19 Vaccine measure, IRFs 
should be able to report that they were unable to determine if a 
patient was vaccinated. Another commenter suggested that having a 
single yes or no item on the IRF-PAI without any requirements for 
documentation or validation of vaccination status would amount to a 
mere checkmark in a box with no evidence that it leads to improved 
quality of care.
    Response: We thank commenters for their recommendations about 
adding additional response options to the item for exclusions. However, 
as we have stated previously, the PFAs convened for our TEP told us 
that a measure capturing raw vaccination rates would be most helpful in 
patient and family/caregiver decision-making. The TEP agreed that 
developing a measure to report the rate of vaccination without 
denominator exclusions was an important goal. Based on this feedback, 
we believe excluding patients/residents with contraindications from the 
measure would distort the intent of the measure of providing raw COVID-
19 patient vaccination rates, while making the information more 
difficult for patients/caregivers to interpret, and hence did not 
include any exclusions.
    CMS has multiple processes in place to ensure reported patient data 
are accurate. State agencies conduct standard certification surveys for 
IRFs, and accuracy and completeness of the IRF-PAI are among the 
regulatory requirements that surveyors evaluate during surveys.\186\ 
Additionally, the IRF-PAI process has multiple regulatory requirements. 
Our regulations at Sec.  412.606(b) require that (1) the assessment 
accurately reflects the patient's status, (2) a clinician appropriately 
trained to perform a patient assessment using the IRF-PAI conducts or 
coordinates each assessment with the appropriate participation of 
health professionals, and (3) the assessment process includes direct 
observation, as well as communication with the patient.\187\ We take 
the accuracy of IRF-PAI assessment data very seriously, and routinely 
monitor the IRF QRP measures' performance, and will take appropriate 
steps to address any such issues, if identified, in future rulemaking.
---------------------------------------------------------------------------

    \186\ Centers for Medicare & Medicaid Services. Hospitals. 
September 6, 2022. https://www.cms.gov/medicare/provider-enrollment-and-certification/certificationandcomplianc/hospitals.
    \187\ 42 CFR 412.606 https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-B/part-412/subpart-P/section-412.606.
---------------------------------------------------------------------------

    We note that the potential consequences of submitting false data 
and information in the IRF-PAI, including the potential for civil 
liability under the False Claims Act (31 U.S.C. 3729 to 3733) for 
knowingly presenting a false or fraudulent claim to the government for 
payment, provide strong incentives for providers to ensure that the 
data submitted in the IRF-PAI are accurate.
    Comment: One commenter noted that the intent of the measure as 
proposed was unclear. This commenter referred to CMS' comment in the FY 
2024 IRF PPS proposed rule that the ``intent of the Patient/Resident 
COVID-19 Vaccine measures would be to promote transparency of data for 
patients to make informed decisions regarding care and is not intended 
to be a measure of IRF action.'' However, the commenter disagreed with 
this rationale, referencing the RFI in section VIII.D. of the proposed 
rule, Principles for Selecting and Prioritizing IRF QRP Quality 
Measures and Concepts under Consideration for Future Years. The 
commenter believes the proposed measure fails to qualify for the first 
proposed principle for selecting and prioritizing IRF QRP quality 
measure concepts under consideration for future years, 
``actionability.''
    Response: As stated in section VIII.D.2. of the proposed rule, to 
address actionability, IRF QRP measures should focus on structural 
elements, healthcare processes, and outcomes of care that have been 
demonstrated, such as through clinical evidence or other best 
practices, to be amenable to improvement and feasible for IRFs to 
implement. As stated previously, we believe this Patient/Resident 
COVID-19 Vaccine measure is an indirect measure of provider action. 
Providers have the opportunity to engage and educate patients on the 
benefits and importance of COVID-19 vaccination, especially in the IRF 
setting where patients are at higher risk of contracting COVID-19. 
Additionally, once collected these data will be available on the 
patient-level reports for IRF providers, which will further help 
providers decide on actions such as patient education and steps they 
can take to increase vaccination in their facility.
    After consideration of the public comments we received, we are 
finalizing our proposal to adopt the Patient/Resident COVID-19 Vaccine 
measure as an assessment-based measure beginning with the FY 2026 IRF 
QRP as proposed.

D. Principles for Selecting and Prioritizing IRF QRP Quality Measures 
and Concepts Under Consideration for Future Years--Request for 
Information (RFI)

1. Solicitation of Comments
    In the proposed rule, we invited general comments on the principles 
for identifying IRF QRP measures, as well as additional comments about 
measurement gaps, and suitable measures for filling these gaps. 
Specifically, we solicited comment on the following questions:
     Principles for Selecting and Prioritizing QRP Measures

[[Page 51036]]

    ++ To what extent do you agree with the principles for selecting 
and prioritizing measures?
    ++ Are there principles that you believe CMS should eliminate from 
the measure selection criteria?
    ++ Are there principles that you believe CMS should add to the 
measure selection criteria?
     IRF QRP Measurement Gaps
    ++ CMS requests input on the identified measurement gaps, including 
in the areas of cognitive function, behavioral and mental health, 
patient experience and patient satisfaction, and chronic conditions and 
pain management.
    ++ Are there gaps in the IRF QRP measures that have not been 
identified in this RFI?
     Measures and Measure Concepts Recommended for Use in the 
IRF QRP
    ++ Are there measures that you believe are either currently 
available for use, or that could be adapted or developed for use in the 
IRF QRP program to assess performance in the areas of (1) cognitive 
functioning, (2) behavioral and mental health, (3) patient experience 
and patient satisfaction, (4) chronic conditions, (5) pain management, 
or (6) other areas not mentioned in this RFI?
    CMS also sought input on data available to develop measures, 
approaches for data collection, perceived challenges or barriers, and 
approaches for addressing challenges. We received several comments in 
response to this RFI, which are summarized below.
Comments on Principles for Selecting and Prioritizing QRP Measure
    A few commenters expressed support for the measure selection and 
prioritization criteria identified by CMS in the RFI in the proposed 
rule, as well as those espoused through the National Quality Strategy 
and the ``Universal Foundation'' of quality measures. One commenter 
indicated that principles for measure selection and prioritization 
identified by CMS in the RFI are consistent with the principles 
inherent in the CMS Measure Management System and recommended that MMS 
measure development principles be integrated into the IRF QRP 
principles. The same commenter suggested that clearly delineated 
processes are required in order to guide the application of these 
principles.
    One commenter recommended that CMS consider the extent to which 
measures offer a well-rounded assessment of performance, are 
complementary, and demonstrate the patient's journey.
    Several commenters expressed concern about the addition of measures 
to the QRP and specifically requested that CMS consider the 
administrative burden associated with measure reporting. To reduce 
administrative burden, commenters suggested that CMS consider 
opportunity costs, and remove measures that are not tied to strategic 
quality improvement aims.
    In addition to administrative burden, other criteria that 
commenters suggested be considered as part of CMS' guiding principles, 
included: whether the measure is endorsed by a CBE; the extent to which 
the measure focuses on a salient healthcare issue; the measure's 
technical specifications, reliability and validity, implementation 
feasibility, and electronic availability of data.
    One commenter requested that CMS clearly explain how measures 
selected for development meet the set criteria used.
Comments on Principles for Selecting and Prioritizing QRP Measures and 
Measures and Measure Concepts Recommended for Use in the IRF QRP
    Although several commenters agreed with CMS on the presence of 
measurement gaps in the IRF QRP, particularly in the domain of 
cognitive functioning, one commenter stated that even if intended to 
fill a gap, additional measures to the IRF QRP could not be justified 
given the present administrative burden on IRFs. The commenter 
recommended that CMS continually evaluate whether measures are 
necessary and remove those that are deemed unnecessary. Another 
commenter indicated that CMS should neither add quality measures to the 
IRF QRP nor attempt to fill gaps until IRFs receive financial 
assistance for EHR systems.
Comments on Cognitive Function
    Several commenters supported the introduction of cognitive measures 
for future QRP measure sets, with one commenter indicating that 
cognitive function measures would provide additional context concerning 
IRF efficacy.
    Multiple commenters did not support the use of the CAM or BIMS as a 
source of data for use in measuring cognitive function. One commenter 
stated that neither the CAM nor BIMS provide clinical value to inform 
rehabilitation care planning or outcomes, including the change in 
cognitive functioning from admission to discharge. Commenters indicated 
that the BIMS was not developed as a tool to screen for the presence or 
absence of cognitive impairment and that it only captures selected 
elements of cognition, such as attention, short-term memory and verbal 
interaction, rather than executive functioning, judgement, reasoning, 
and higher-level cognitive functions. Commenters further stated that 
the BIMS scale shows low sensitivity identifying cognitive deficits 
that affect community placement.
    Other concerns about the BIMS for use in development of measures of 
cognitive functioning included the lack of physician buy-in for the 
BIMS, variation in the reliability of scoring, and limited utility of 
the BIMS for measuring and risk-adjusting patient cognition and 
communication.
    Although one commenter indicated that the proprietary nature of 
cognitive functioning instruments and administrative burden posed a 
challenge to adopting a cognitive assessment instrument, several 
commenters encouraged CMS to pursue alternative data sources and 
measures of cognitive functioning. Suggestions of ways to assess 
cognition included the Functional Independence MeasureTM 
(FIMTM) and patient-reported outcome measures. Another 
commenter encouraged CMS to select measures that are reliable, 
feasible, valid, and that are, or could be, endorsed by a consensus 
organization.
Comments on Behavioral and Mental Health
    Commenters voiced appreciation for CMS interest in addressing 
behavioral and mental health issues through the development of quality 
measures for the IRF QRP. Other commenters cited potential challenges 
to the adoption of behavioral and mental health measures. One commenter 
indicated that it would be difficult for IRFs to offer psychological 
services given the 3-hour therapy per day requirement.\188\ Another 
commenter indicated that such measures would not be relevant for the 
IRF setting, since patients with a severe behavioral or mental health 
impairment would be unlikely to participate in therapy, and inpatient 
rehabilitation would not be an appropriate setting. Should CMS still 
seek to develop behavioral and mental health quality measures, the 
commenter suggested consideration of the Patient Health Questionnaire 
(PHQ)-2 through PHQ-9, which are required for completion of the IRF-
PAI.
---------------------------------------------------------------------------

    \188\ Sec.  412.622(a)(3)(ii) Subpart P--Prospective Payment for 
Inpatient Rehabilitation Hospitals and Rehabilitation Units; Basis 
of payment.
---------------------------------------------------------------------------

    One commenter suggested that CMS consider adoption of measures that 
evaluate psychosocial functioning. One

[[Page 51037]]

commenter recommended that behavioral and mental health measures 
capture rehabilitative services, such as therapeutic recreation, that 
support activities that the patient is expected to enjoy post-
hospitalization.
Comments on Patient Experience and Patient Satisfaction
    A few commenters expressed support for the adoption of measures 
derived from patient experience surveys, including the IRF Experience 
of Care (EOC) survey. One commenter expressed preference for the use of 
the IRF EOC survey over the CoreQ Short Stay Discharge Survey (CoreQ 
survey) to measure patient experience, indicating that the IRF EOC 
survey addresses essential assessment areas (for example, goal setting, 
communications with staff, respect and privacy received, ability to 
obtain assistance when needed, cleanliness of the facility), whereas 
the CoreQ survey provides a more limited assessment and lacks the depth 
to drive quality improvement. Should CMS decide to use the CoreQ 
survey, the commenter recommended that CMS allow the fielding of 
supplemental questions, such as items from the IRF EOC survey. 
Regardless of which tool is used, the commenter urged CMS to ensure the 
reliability and validity of the measure and composites, subject the 
measure for review by a CBE, and to pursue the Consumer Assessment of 
Healthcare Providers and Services (CAHPS) trademark.
    One commenter, who did not support the inclusion of a patient 
experience or satisfaction measure in the IRF QRP, indicated that the 
administrative and financial costs associated with data collection, 
particularly for smaller, hospital-based IRFs, would be too high. The 
commenter further indicated that information gathered from these items 
would not be meaningful.
Comments on Chronic Condition and Pain Management
    One commenter indicated that, because pain is an inherent part of 
intensive rehabilitation therapy, rather than measuring whether pain 
exists or whether level of pain was assessed, a more meaningful pain 
measure would assess the extent to which IRF staff are responsive to 
and help manage patients' pain. The commenter suggested that the use of 
a patient-reported outcome measure would provide more meaningful 
information than a process measure of pain and would not increase 
burden to the IRF. Another commenter expressed concern about unintended 
consequences associated with measures related to pain management.
Comments on Other Measurement Gaps
    Some commenters believe measurement gaps to exist in areas not 
identified in the RFI. Other measures and measurement concepts 
identified by commenters included health equity; care for degenerative 
cognitive conditions; IRF workforce safety culture, engagement, and 
burnout; and measures of quality of life, such as the World Health 
Organization Quality of Life (WHOQOL) assessment and the Comprehensive 
Evaluation in Recreational Therapy for Physical Disabilities (CERT-Phys 
Dis).
    Response: We appreciate the input provided by commenters. While we 
will not be responding to specific comments submitted in response to 
this RFI in this final rule, we intend to use this input to inform our 
future measure development efforts.

E. Health Equity Update

1. Background
    In the FY 2023 IRF PPS proposed rule (87 FR 20247through 20254), we 
included an RFI entitled ``Overarching Principles for Measuring Equity 
and Healthcare Quality Disparities Across CMS Quality Programs.'' We 
define health equity as ``the attainment of the highest level of health 
for all people, where everyone has a fair and just opportunity to 
attain their optimal health regardless of race, ethnicity, disability, 
sexual orientation, gender identity, socioeconomic status, geography, 
preferred language, or other factors that affect access to care and 
health outcomes.'' \189\ We are working to advance health equity by 
designing, implementing, and operationalizing policies and programs 
that support health for all the people served by our programs and 
models, eliminating avoidable differences in health outcomes 
experienced by people who are disadvantaged or underserved, and 
providing the care and support that our enrollees need to thrive. Our 
goals outlined in the CMS Framework for Health Equity 2022-2023 \190\ 
are in line with Executive Order 13985, ``Advancing Racial Equity and 
Support for Underserved Communities Through the Federal Government.'' 
\191\ The goals included in the CMS Framework for Health Equity serve 
to further advance health equity, expand coverage, and improve health 
outcomes for the more than 170 million individuals supported by our 
programs, and set a foundation and priorities for our work, including: 
strengthening our infrastructure for assessment, creating synergies 
across the health care system to drive structural change, and 
identifying and working to eliminate barriers to CMS-supported 
benefits, services, and coverage. The CMS Framework for Health Equity 
outlines the approach CMS will use to promote health equity for 
enrollees, mitigate health disparities, and prioritize CMS's commitment 
to expanding the collection, reporting, and analysis of standardized 
data.\192\
---------------------------------------------------------------------------

    \189\ Centers for Medicare & Medicaid Services. Health Equity. 
October 3, 2022. https://www.cms.gov/pillar/health-equity.
    \190\ Centers for Medicare & Medicaid Services. CMS Framework 
for Health Equity 2022-2032. April 2022. https://www.cms.gov/files/document/cms-framework-health-equity-2022.pdf.
    \191\ The White House. Executive Order on Advancing Racial 
Equity and Support for Underserved Communities Through the Federal 
Government. Executive Order 13985, January 20, 2021. https://www.whitehouse.gov/briefing-room/presidential-actions/2021/01/20/executive-order-advancing-racial-equity-and-support-for-underserved-communities-through-the-federal-government/.
    \192\ Centers for Medicare and Medicaid Services. The Path 
Forward: Improving Data to Advance Health Equity Solutions. https://www.cms.gov/files/document/path-forwardhe-data-paper.pdf. July 11, 
2023.
---------------------------------------------------------------------------

    In addition to the CMS Framework for Health Equity, we seek to 
advance health equity and whole-person care as one of eight goals 
comprising the CMS National Quality Strategy (NQS).\193\ The NQS 
identifies a wide range of potential quality levers that can support 
our advancement of equity, including: (1) establishing a standardized 
approach for patient-reported data and stratification; (2) employing 
quality and value-based programs to address closing equity gaps; and 
(3) developing equity-focused data collections, regulations, oversight 
strategies, and quality improvement initiatives.
---------------------------------------------------------------------------

    \193\ Centers for Medicare & Medicaid Services. CMS National 
Quality Strategy? https://www.cms.gov/Medicare/Quality-Initiatives-
Patient-Assessment-Instruments/Value-Based-Programs/CMS-Quality-
Strategy.
---------------------------------------------------------------------------

    A goal of the NQS is to address persistent disparities that 
underlie our healthcare system. Racial disparities, in particular, are 
estimated to cost the U.S. $93 billion in excess medical costs and $42 
billion in lost productivity per year, in addition to economic losses 
due to premature deaths.\194\ At the same time, racial and ethnic 
diversity has increased in recent years with an increase in the 
percentage of people who identify as two or more races accounting for 
most of the change, rising from 2.9 percent to 10.2 percent between 
2010 and 2020.\195\

[[Page 51038]]

Therefore, we need to consider ways to reduce disparities, achieve 
equity, and support our diverse beneficiary population through the way 
we measure quality and display the data.
---------------------------------------------------------------------------

    \194\ Turner A. The Business Case for Racial Equity: A Strategy 
for Growth. April 24, 2018. W.K. Kellogg Foundation and Altarum. 
https://altarum.org/RacialEquity2018.
    \195\ Agency for Healthcare Research and Quality. 2022 National 
Healthcare Quality and Disparities Report. November 2022. https://www.ahrq.gov/research/findings/nhqrdr/nhqdr22/index.html.
---------------------------------------------------------------------------

    We solicited public comments via the aforementioned RFI on changes 
that we should consider in order to advance health equity. We refer 
readers to the FY 2023 IRF PPS final rule (87 FR 47072 through 47073) 
for a summary of the public comments and suggestions CMS received in 
response to the health equity RFI. In the proposed rule, we said we 
would take these comments into account as we continue to work to 
develop policies, quality measures, and measurement strategies on this 
important topic.
2. Anticipated Future State
    We are committed to developing approaches to meaningfully 
incorporate the advancement of health equity into the IRF QRP. One 
option we are considering is including social determinants of health 
(SDOH) as part of new quality measures.
    Social determinants of health are the conditions in the 
environments where people are born, live, learn, work, play, worship, 
and age that affect a wide range of health, functioning, and quality-
of-life outcomes and risks. They may have a stronger influence on the 
population's health and well-being than services delivered by 
practitioners and healthcare delivery organizations.\196\ Measure 
stratification by CMS is important for better understanding the 
differences in health outcomes from across different patient population 
groups according to specific demographic and SDOH variables. For 
example, when pediatric measures over the past two decades are 
stratified by race, ethnicity, and income, they show that outcomes for 
children in the lowest income households and for Black and Hispanic 
children have improved faster than outcomes for children in the highest 
income households or for White children, thus narrowing an important 
health disparity.\197\ This analysis and comparison of the SDOH items 
in the assessment instruments support our desire to understand the 
benefits of measure stratification. Hospital providers receive such 
information in their confidential feedback reports (CFRs) and we think 
this learning opportunity would benefit PAC providers. The goals of the 
CFR are to provide IRFs with their results so they can compare certain 
quality measures stratified by dual eligible status and race and 
ethnicity. The process is meant to increase providers' awareness of 
their data. We will solicit feedback from IRFs for future enhancements 
to the CFRs.
---------------------------------------------------------------------------

    \196\ Agency for Healthcare Research and Quality. 2022 National 
Healthcare Quality and Disparities Report. November 2022. https://www.ahrq.gov/research/findings/nhqrdr/nhqdr22/index.html.
    \197\ Agency for Healthcare Research and Quality. 2022 National 
Healthcare Quality and Disparities Report. November 2022. https://www.ahrq.gov/research/findings/nhqrdr/nhqdr22/index.html.
---------------------------------------------------------------------------

    In the proposed rule, we said that we are considering whether 
health equity measures we have adopted for other settings, such as 
hospitals,\198\ could be adopted in PAC settings. We said we were 
exploring ways to incorporate SDOH elements into the measure 
specifications. For example, we could consider a future health equity 
measure like screening for social needs and interventions using our 
current SDOH data items of preferred language, interpreter services, 
health literacy, transportation, and social isolation. With 30 percent 
to 55 percent of health outcomes attributed to SDOH,\199\ a measure 
capturing and addressing SDOH could encourage IRFs to identify 
patients' specific needs and connect them with the community resources 
necessary to overcome social barriers to their wellness. We could 
specify a health equity measure using the same SDOH data items that we 
currently collect as standardized patient assessment data elements 
under the IRF. These SDOH data items assess health literacy, social 
isolation, transportation problems, and preferred language (including 
need for or want of an interpreter). We also see value in aligning SDOH 
data items according to existing health information technology (IT) 
vocabulary and codes sets where applicable and appropriate such as 
those included in the Office of the National Coordinator for Health 
Information (ONC) United States Core Data for Interoperability (USCDI) 
\200\ across all care settings as we develop future health equity 
quality measures under our IRF QRP statutory authority. This would 
further the NQS' goal of aligning quality measures across our programs 
as part of the Universal Foundation.\201\
---------------------------------------------------------------------------

    \198\ Medicare Program; Hospital Inpatient Prospective Payment 
Systems for Acute Care Hospitals and the Long-Term Care Hospital 
Prospective Payment System and Policy Changes and Fiscal Year 2023 
Rates; Quality Programs and Medicare Promoting Interoperability 
Program Requirements for Eligible Hospitals and Critical Access 
Hospitals; Costs Incurred for Qualified and Non-Qualified Deferred 
Compensation Plans; and Changes to Hospital and Critical Access 
Hospital Conditions of Participation. 87 FR 49202 through 49215.
    \199\ World Health Organization. Social Determinants of Health. 
https://www.who.int/health-topics/social-determinants-of-health#tab=tab_1.
    \200\ United States Core Data for Interoperability (USCDI), 
https://www.healthit.gov/isa/united-states-core-data-interoperability-uscdi.
    \201\ Jacobs DB, Schreiber M, Seshamani M, Tsai D, Fowler E, 
Fleisher LA. Aligning Quality Measures across CMS--The Universal 
Foundation. N Engl J Med. 2023 Mar 2;338:776-779. doi: 10.1056/
NEJMp2215539. PMID: 36724323.
---------------------------------------------------------------------------

    Although we did not directly solicit feedback to our update, we did 
receive some public comments, which we summarize.
    Comment: Several commenters responded to our update on the 
continuing efforts to advance health equity. One commenter encouraged 
CMS to consider data collection reports as a starting point, and also a 
structural measure that is based on health equity priorities, similar 
to what has been adopted in other Medicare quality reporting programs.
    Two commenters supported the idea of measure stratification by 
certain SDOH, and one requested this information on all claims-based 
measures. Both commenters emphasized that any additional stratification 
of quality measures, including social risk factors and SDOH, would be 
of value to PAC providers, including IRFs.
    One commenter also noted that receiving patient-level data for 
claims-based measures on a more frequent basis would enable them to 
make better informed decisions. This commenter referenced the Hospital 
Inpatient Quality Reporting (IQR) Program which provides reports with 
patient-level data to hospitals and urged CMS to provide IRFs with the 
same level of detail in their quality data. They also noted that while 
having the measures stratified by SDOH would be helpful, they believe 
having it in a timely manner could have a more meaningful impact on 
equity and quality of care.
    We received some comments on other data points that may be useful 
in identifying and addressing health disparities. One commenter 
suggested focusing efforts on social risk factors that are of 
sufficient granularity to drive appropriate interventions at the 
individual level. Another commenter noted that while it is important to 
still try to understand differences by race and ethnicity to identify 
and address disparities that might stem from racism and social and 
economic inequities, they recommended against making generalizations 
about differences in health and health care simply based on race and 
ethnicity and to instead conduct more in-depth evaluations of 
underlying social and economic drivers of health. This commenter 
suggested

[[Page 51039]]

that CMS incentivize the collection and analysis of data on factors 
such as, but not limited to, disability status, veteran status, primary 
or preferred language, health literacy, food security, transportation 
access, housing stability, social support after discharge from an IRF, 
and a person's access to care. This same commenter, however, pointed 
out that any program must account for the fact that there are many 
contributors to health inequities, including personal factors, many of 
which are outside the control of IRFs. They encouraged CMS to have 
ongoing engagement with interested parties to best understand 
structural and socioeconomic barriers to health and to monitor for any 
unintended consequences. Finally, this commenter urged CMS to focus on 
improving care coordination as patients move between settings. However, 
another commenter requested CMS consider what is already being 
collected by providers prior to adding additional data collection 
requirements.
    One commenter encouraged CMS to thoughtfully consider the 
appropriate data collection of SDOH factors before attempting to report 
the data, given the resources required to implement new items in the 
electronic medical record. They pointed to the current work underway by 
the Office of Management and Budget (OMB) seeking feedback about 
combining race and ethnicity questions (88 FR 5375).
    One commenter recommended CMS consider including SDOH in new 
quality measures and in IRF payment and suggested it could be 
accomplished through the use of ICD-10 Z-codes as indicators of the 
additional resources required to care for patients.
    Response: We thank all the commenters for responding to our update 
on this important CMS priority. We will take your recommendations into 
consideration in our future work on health equity.

F. Form, Manner, and Timing of Data Submission Under the IRF QRP

1. Background
    We refer readers to the regulatory text at Sec.  412.634(b)(1) for 
information regarding the current policies for reporting IRF QRP data.
2. Reporting Schedule for the IRF-PAI Assessment Data for the Discharge 
Function Score Measure Beginning With the FY 2025 IRF.
    As discussed in section VIII.C.1.b. of the proposed rule, we 
proposed to adopt the Discharge Function Score (DC Function) measure 
beginning with the FY 2025 IRF QRP. We proposed that IRFs would be 
required to report these IRF-PAI assessment data related to the DC 
Function measure beginning with patients discharged on October 1, 2023, 
for purposes of the FY 2025 IRF QRP. Starting in CY 2024, IRFs would be 
required to submit data for the entire calendar year beginning with the 
FY 2026 IRF QRP. Because the DC Function measure is calculated based on 
data that are currently submitted to the Medicare program in the IRF-
PAI, there would be no new burden associated with data collection for 
this measure.
    We invited public comments on our proposal.
    We did not receive any comments on this proposed revision, and 
therefore, we are finalizing the revisions as proposed.
3. Reporting Schedule for the Data Submission of IRF-PAI Assessment 
Data for the COVID-19 Vaccine: Percent of Patients/Residents Who Are Up 
to Date Quality Measure Beginning With the FY 2026 IRF QRP
    As discussed in section VIII.C.2.a. of the proposed rule, we 
proposed to adopt the COVID-19 Vaccine: Percent of Patients/Residents 
Who Are Up to Date (Patient/Resident COVID-19 Vaccine) measure 
beginning with the FY 2026 IRF QRP. We proposed that IRFs would be 
required to report the IRF-PAI assessment data related to the Patient/
Resident COVID-19 Vaccine measure beginning with patients discharged on 
October 1, 2024, for purposes of the FY 2026 IRF QRP. Starting in CY 
2025, IRFs would be required to submit data for the entire CY beginning 
with the FY 2027 IRF QRP.
    We also proposed to add a new item to the IRF-PAI in order for IRFs 
to report this measure. Specifically, a new item would be added to the 
IRF-PAI discharge assessment to collect information on whether a 
patient is up to date with their COVID-19 vaccine at the time of 
discharge from the IRF. A draft of the new item is available in the 
COVID-19 Vaccine: Percent of Patients/Residents Who Are Up to Date 
Draft Measure Specifications.\202\
---------------------------------------------------------------------------

    \202\ COVID-19 Vaccine: Percent of Patients/Residents Who Are Up 
to Date. Draft Measure Specifications. https://www.cms.gov/files/document/patient-resident-covid-vaccine-draft-specs.pdf.
---------------------------------------------------------------------------

    We invited public comments on our proposal. The following is a 
summary of the public comments received on our proposal to require IRFs 
to report a new IRF-PAI assessment data item for the Patient/Resident 
COVID-19 Vaccine measure beginning with patients discharged on October 
1, 2024, and our responses.
    Comment: One commenter stated that this proposed measure has the 
potential to increase COVID-19 vaccination coverage of patients in 
IRFs, as well as prevent the spread of COVID-19 within the IRF patient 
population. However, given that the patient's COVID-19 vaccination 
status was proposed to be collected at discharge from the IRF rather 
than upon admission, they believe the opportunity is lost.
    Response: We believe that during a patient stay, IRFs have the 
opportunity to educate the patient and provide information on why they 
should become up to date, if a patient is not up to date with their 
vaccine at the time they are admitted. This is particularly important 
for patients in IRFs, who tend to be at higher risk for serious 
complications from COVID-19. If the patient is agreeable, the patient 
may receive the necessary vaccine to become up to date any time during 
their IRF stay prior to discharge.
    Comment: One commenter noted that IRFs have been reporting COVID-19 
vaccination and infection data to both State departments of health and 
the CDC's National Healthcare Safety Network (NHSN) and introducing a 
new IRF-PAI item would create the potential for duplicative reporting.
    Response: Currently, as part of the IRF QRP, we do not collect 
COVID-19 vaccination data for patients. CMS only collects COVID-19 
vaccination data for healthcare personnel via the NHSN. Therefore, 
addition of an IRF-PAI item for the purposes of collecting patient 
COVID-19 vaccination data would not lead to duplicative reporting at 
the Federal level.
    Comment: One commenter noted that the draft specifications for this 
measure do not specify what the preferred source would be, or how 
facilities should deal with conflicting information from different 
sources (for example, the patient responding that they are vaccinated, 
but the medical record suggesting they are not).
    Response: As described in the Draft Technical Specifications,\203\ 
providers will be able to use all sources of information available to 
obtain the vaccination data, such as patient interviews, medical 
records, proxy response, and vaccination cards provided by the patient 
or their caregivers. As with any assessment item in the IRF-PAI, we 
will also publish coding guidance and instructions to further aid 
providers in collection of

[[Page 51040]]

this data, including coding in situations with conflicting information.
---------------------------------------------------------------------------

    \203\ COVID-19 Vaccine: Percent of Patients/Residents Who Are Up 
to Date Draft Measure Specifications. https://www.cms.gov/files/document/patient-resident-covid-vaccine-draft-specs.pdf.
---------------------------------------------------------------------------

    After consideration of the public comments we received, we are 
finalizing our proposal to require IRFs to report a new IRF-PAI 
assessment data item for the Patient/Resident COVID-19 Vaccine measure 
beginning with patients discharged on October 1, 2024 for the FY 2026 
IRF QRP as proposed.

G. Policies Regarding Public Display of Measure Data for the IRF QRP

1. Background
    Section 1886(j)(7)(E) of the Act requires the Secretary to 
establish procedures for making the IRF QRP data available to the 
public after ensuring that IRFs have the opportunity to review their 
data prior to public display. For a more detailed discussion about our 
policies regarding public display of IRF QRP measure data and 
procedures for the IRF's opportunity to review and correct data and 
information, we refer readers to the FY 2017 IRF PPS final rule (81 FR 
52045 through 52048).
2. Public Reporting of the Transfer of Health (TOH) Information to the 
Provider--Post-Acute Care (PAC) Measure and TOH Information to the 
Patient--PAC Measure Beginning With the FY 2025 IRF QRP
    We proposed to begin publicly displaying data for the measures, TOH 
Information to the Provider--PAC Measure (TOH--Provider) and TOH 
Information to the--Patient PAC Measure (TOH--Patient) beginning with 
the September 2024 Care Compare refresh or as soon as technically 
feasible.
    We adopted these measures in the FY 2020 IRF PPS final rule (84 FR 
39099 through 39107). In response to the COVID-19 PHE, we issued an 
interim final rule (85 FR 27595 through 27596), which delayed the 
compliance date for the collection and reporting of the TOH--Provider 
and TOH--Patient measures to October 1st of the year that is at least 
one full FY after the end of the COVID-19 PHE. Subsequently, the CY 
2022 Home Health PPS Rate Update final rule (86 FR 62381 through 62386) 
revised the compliance date for the collection and reporting of the 
TOH--Provider and TOH--Patient measures under the IRF QRP to October 1, 
2022. Data collection for these two assessment-based measures in the 
IRF QRP began with patients discharged on or after October 1, 2022.
    We proposed to publicly display four rolling quarters of the data 
we receive for these two assessment-based measures, initially using 
data on discharges from January 1, 2023, through December 31, 2023 
(Quarter 1 2023 through Quarter 4 2023); and to begin publicly 
reporting data on these measures with the September 2024 refresh of 
Care Compare, or as soon as technically feasible. To ensure the 
statistical reliability of the data, we proposed that we would not 
publicly report an IRF's performance on a measure if the IRF had fewer 
than 20 eligible cases in any four consecutive rolling quarters for 
that measure. IRFs that have fewer than 20 eligible cases would be 
distinguished with a footnote that states, ``The number of cases/
patient stays is too small to publicly report.''
    We invited public comment on our proposal for the public display of 
the TOH--Provider and TOH--Patient assessment-based measures. The 
following is a summary of the public comments received on the proposal 
to publicly report these measures and our responses.
    Comment: Several commenters supported the proposal to publicly 
report the Transfer of Health Information to the Provider-PAC Measure 
and the Transfer of Health Information to the Patient-PAC Measure 
beginning with the September 2024 Care Compare refresh or as soon as 
technically feasible. One commenter believes the additional attention 
and focus on the transfer of health information would improve internal 
and external processes for patients and caregivers. Another commenter 
suggested stratification of the data would add value to consumers and 
providers.
    Response: We thank the commenters for their support and agree that 
the information will provide helpful information to consumers about an 
IRFs internal and external processes related to transfer of important 
health information. We also appreciate the suggestion for stratifying 
the data, and we will use this input to inform our future public 
reporting refinements.
    Comment: One commenter was not supportive of the proposal, saying 
that the reporting requirement would be duplicative of information IRFs 
are already required to collect and the measures would be redundant.
    Response: We want to clarify that the proposal would add no 
additional reporting requirements to the IRF QRP. IRFs began collecting 
the Transfer of Health information data elements for all patients 
discharged beginning October 1, 2022. In section IX.G.2 of this final 
rule, we proposed using data collected from January 1, 2023 through 
December 31, 2023 for the inaugural display of the measures on Care 
Compare beginning September 2024 or as soon as technically feasible.
    Comment: One commenter said they valued the public reporting of 
metrics that reflect the quality of care a patient received in an IRF 
but encouraged CMS to delay reporting of the TOH-Patient and TOH-
Provider measures until 2025, using discharges from January 1, 2024 
through December 31, 2024 (Quarter 1, 2024 through Quarter 4, 2024), 
given their recent adoption into the IRF QRP.
    Response: We disagree with the commenter. While the TOH-Patient and 
TOH-Provider measures original data collection start date was October 
1, 2020, we delayed the collection of the measures due to the COVID-19 
PHE. As the commenter noted, CMS revised the data collection to begin 
October 1, 2022, and while we have received some questions about the 
new data items on the IRF-PAI through our IRF QRP helpdesk, the number 
of questions have been minimal. Neither have there been any reported 
problems with the implementation of these items. The inaugural 
reporting period we proposed, January 1, 2023 through December 31, 2023 
(Quarter 1, 2023 through Quarter 4, 2023) is consistent with our public 
reporting proposals for other new IRF QRP measures. We do not agree 
that IRFs need more time to adjust for these measures.
    As a result of the public comments, we are finalizing our proposal 
to begin publicly displaying data for the measures: (1) Transfer of 
Health (TOH) Information to the Provider--Post-Acute Care (PAC) Measure 
(TOH-Provider); and (2) TOH Information to the Patient--PAC Measure 
(TOH-Patient) beginning with the September 2025 Care Compare refresh or 
as soon as technically feasible.
3. Public Reporting of the Discharge Function Score Measure Beginning 
With the FY 2025 IRF QRP
    We proposed to begin publicly displaying data for the Discharge 
Function Score (DC Function) measure beginning with the September 2024 
refresh of Care Compare, or as soon as technically feasible, using data 
collected from January 1, 2023 through December 31, 2023 (Quarter 1 
2023 through Quarter 4 2023). We proposed that an IRF's DC Function 
measure score would be displayed based on four quarters of data. 
Provider preview reports would be distributed to IRFs in June 2024, or 
as soon as technically feasible. Thereafter, an IRF's DC Function 
measure score would be publicly displayed based on four quarters of 
data and updated

[[Page 51041]]

quarterly. To ensure the statistical reliability of the data, we 
proposed that we would not publicly report an IRF's performance on the 
measure if the IRF had fewer than 20 eligible cases in any quarter. 
IRFs that have fewer than 20 eligible cases would be distinguished with 
a footnote that states: ``The number of cases/patient stays is too 
small to report.''
    We invited public comment on the proposal for the public display of 
the DC Function assessment-based measure beginning with the September 
2024 refresh of Care Compare, or as soon as technically feasible. The 
following is a summary of the public comments received on our proposal 
and our responses.
    Comment: One commenter provided support to publicly report the DC 
Function measure.
    Response: We thank the commenter for their support to publicly 
report the proposed measure.
    Comment: One commenter recommended that CMS specify when results 
will be provided to IRFs for their review, that CMS provide more 
patient-specific data, and clarify whether CMS uses results for 
``judgement or quality improvement or both.'' This commenter suggests 
CMS report ``comparative stratified functional status based on key risk 
factors at discharge'' to assist IRF improvements.
    Response: CMS plans to publicly display the DC Function measure 
score quarterly, based on four quarters of data. We refer readers to 
section IX.F.2 of this final rule for information about when the 
proposed DC Function measure will be publicly reported. Specifically, 
we proposed to begin publicly displaying data for the DC Function 
measure beginning with the September 2024 refresh of Care Compare, or 
as soon as technically feasible, using data collected from January 1, 
2023, through December 31, 2023 (Quarter 1 2023 through Quarter 4 
2023). Provider preview reports would be distributed to IRFs in June 
2024, or as soon as technically feasible. Thereafter, an IRF's DC 
Function measure score would be publicly displayed based on four 
quarters of data and updated quarterly.
    In regards to patient-specific data, IRFs can review key aspects of 
this measure, such as who did and did not meet the numerator criteria, 
in their own patent-level quality measure reports. In terms of the 
intended use of this measure, as with all QRPs, this measure will help 
inform Medicare beneficiaries and their caregivers when selecting IRF 
care and can be used by IRFs to monitor their own performance and 
improve care quality. Finally, we thank the commenter for their 
suggestion that CMS provide performance results stratified by key risk 
factors and will consider the feasibility of adding stratified 
performance scores to the provider preview report at a later date.
    Comment: One commenter expressed concern that IRFs with eligible 
stays requiring imputation during the first quarter of the measure 
period will not know the imputed values for their patients until the 
entire 12-month measure target period ends. Additionally, this 
commenter believes that after the first 12-month period ends and a new 
quarter begins, changes in imputed values from the first year will not 
be reflected in measure scores. The same commenter expressed concern 
for the inclusion of new IRFs in the proposed measure calculations, 
believing these IRFs will be excluded from the measure until they have 
a full 12 months of data.
    Response: New IRFs will not need 12 full months of data to receive 
scores but will receive scores with the following quarterly update. We 
propose to use data collected from January 1, 2023, through December 
31, 2023 (Quarter 1 2023 through Quarter 4 2023) for the first scores 
published. Therefore, IRFs will not need to wait 12 months for results. 
Also, because scores will be updated quarterly, results will consider 
new information provided that will impact scores from previous 
quarters.
    After consideration of the public comments we received, we are 
finalizing our proposal to begin publicly displaying data for the DC 
Function measure beginning with the September 2024 Care Compare refresh 
or as soon as technically feasible.
4. Public Reporting of the COVID-19 Vaccine: Percent of Patients/
Residents Who Are Up to Date Measure Beginning With the FY 2026 IRF QRP
    We proposed to begin publicly displaying data for the COVID-19 
Vaccine: Percent of Patients/Residents Who are Up to Date (Patient/
Resident COVID-19 Vaccine) measure beginning with the September 2025 
refresh of Care Compare, or as soon as technically feasible, using data 
collected for Q4 2024 (October 1, 2024 through December 31, 2024). We 
proposed that an IRF's percent of patients who are up to date, as 
reported under the Patient/Resident COVID-19 Vaccine measure, would be 
displayed based on one quarter of data. Provider preview reports would 
be distributed to IRFs in June 2025 for data collected in Q4 2024, or 
as soon as technically feasible. Thereafter, the percent of IRF 
patients who are up to date with their COVID-19 vaccinations would be 
publicly displayed based on one quarter of data updated quarterly. To 
ensure the statistical reliability of the data, we proposed that we 
would not publicly report an IRF's performance on the measure if the 
IRF had fewer than 20 eligible cases in any quarter. IRFs that have 
fewer than 20 eligible cases would be distinguished with a footnote 
that states: ``The number of cases/patient stays is too small to 
report.''
    We invited public comment on the proposal for the public display of 
the Patient/Resident COVID-19 Vaccine measure beginning with the 
September 2025 refresh of Care Compare, or as soon as technically 
feasible. The following is a summary of the public comments received 
and our responses.
    Comment: Several commenters questioned the value of reporting only 
one quarter of data, since community vaccination rates vary over time 
and as definitions update.
    Response: We believe it is important to make the most up to date 
data available to patients and their caregivers, which will support 
them in making essential decisions about their health care. We proposed 
the measure to be publicly reported on a rolling quarterly basis in 
order to align with the existing HCP COVID-19 Vaccine measure. This 
means the information would be updated quarterly with only the most 
recent data, such that the measure would be consumed as the most recent 
quarter of data refreshed. We believe averaging over 12 months would 
result in the dilution of the most recent and potentially more 
meaningful information, as opposed to the proposed method of reporting, 
which would result in publishing information that is more up to date 
and would not affect the data collection schedule established for 
submitting assessment data.
    Comment: We received comments on whether the public reporting of 
the measure would be meaningful or useful to consumers. One commenter 
said that as with most publicly reported data, there is a generous lag 
time from when the vaccine is administered, the data gathered and 
submitted, and their eventual display online.
    Response: The data will be posted on Care Compare as soon as 
technically feasible, and therefore having a one quarter reporting 
period reduces the lag following the data submission deadline. We 
believe this mitigates concerns that the data would not reflect 
`recent' information to consumers.
    Comment: Another commenter expressed concern about the impact of 
publicly reporting the data due to the

[[Page 51042]]

fact that potential patients may infer that a lower vaccination rate 
implies the facility has a certain political viewpoint on vaccinations, 
and that could influence their decision to choose the facility.
    Response: It is true that individual patients can make their own 
inference regarding the rates displayed publicly, and a provider may or 
may not be able to influence that. However, per 1899B(g) of the Act, 
CMS is statutorily obligated to publicly report IRF performance on the 
IRF QRP quality measures. This measure will provide potential patients 
and their caregivers with an important piece of information regarding 
vaccination rates as part of their process of identifying providers 
they would want to seek care from, alongside other measures available 
on Care Compare to make a comprehensive decision.
    After consideration of the public comments we received, we are 
finalizing our proposal to begin publicly displaying data for the 
Patient/Resident COVID-19 measure beginning with the September 2025 
Care Compare refresh or as soon as technically feasible.

X. Provisions of the Final Regulations

    In the final rule, we are adopting the provisions set forth in the 
FY 2024 IRF PPS proposed rule (88 FR 20950), specifically:
     We will update the CMG relative weights and average length 
of stay values for FY 2024, in a budget neutral manner, as discussed in 
section V. of this final rule.
     We will update the IRF PPS payment rates for FY 2024 by 
the market basket increase factor, based upon the most current data 
available, with a productivity adjustment required by section 
1886(j)(3)(C)(ii)(I) of the Act, as described in section VI. of this 
final rule.
     We will rebase and revise the IRF market basket to reflect 
a 2021 base year, as discussed in section VI. of this final rule.
     We will update the FY 2024 IRF PPS payment rates by the FY 
2024 wage index and the labor-related share in a budget-neutral manner, 
as discussed in section VI. of this final rule.
     We will calculate the IRF standard payment conversion 
factor for FY 2024, as discussed in section VI. of final rule.
     We will update the outlier threshold amount for FY 2024, 
as discussed in section VII. of this final rule.
     We will update the cost-to-charge ratio (CCR) ceiling and 
urban/rural average CCRs for FY 2024, as discussed in section VII. of 
this final rule.
     We will modify the regulation for IRF units to become 
excluded and paid under the IRF PPS as discussed in section VIII. of 
this final rule.
     We are also adopting updates to the IRF QRP in section IX. 
of this final rule as follows:
    ++ We are adopting the COVID-19 Vaccine: Percent of Patients/
Residents Who Are Up to Date (Patient/Resident COVID-19 Vaccine) 
measure.
    ++ We are adopting the Discharge Function Score (DC Function) 
measure.
    ++ We are modifying the COVID-19 Vaccination Coverage among 
Healthcare Personnel (HCP) (HCP COVID-19 Vaccine) measure.
    ++ We are removing the Application of Percent of Long-Term Care 
Hospital (LTCH) Patients with an Admission and Discharge Functional 
Assessment and a Care Plan That Addresses Function (Application of 
Functional Assessment/Care Plan) measure.
    ++ We are removing the IRF Functional Outcome Measure: Change in 
Self-Care Score for Medical Rehabilitation Patients (Change in Self-
Score) measure.
    ++ We are removing the IRF Functional Outcome Measure: Change in 
Mobility Score for Medical Rehabilitation Patients (Change in Mobility 
Score) measure.

XI. Collection of Information Requirements

    Under the Paperwork Reduction Act of 1995, we are required to 
provide 60-day notice in the Federal Register and solicit public 
comment before a collection of information requirement is submitted to 
the Office of Management and Budget (OMB) for review and approval. In 
order to fairly evaluate whether an information collection should be 
approved by OMB, section 3506(c)(2)(A) of the Paperwork Reduction Act 
of 1995 requires that we solicit comment on the following issues:
     The need for the information collection and its usefulness 
in carrying out the proper functions of our agency.
     The accuracy of our estimate of the information collection 
burden.
     The quality, utility, and clarity of the information to be 
collected.
     Recommendations to minimize the information collection 
burden on the affected public, including automated collection 
techniques.
    This final rule refers to associated information collections that 
are not discussed in the regulation text contained in this document.

A. Requirements for Updates Related to the IRF QRP Beginning With the 
FY 2025 IRF QRP

    An IRF that does not meet the requirements of the IRF QRP for a 
fiscal year will receive a 2-percentage point reduction to its 
otherwise applicable annual increase factor for that fiscal year.
    We believe that the burden associated with the IRF QRP is the time 
and effort associated with complying with the requirements of the IRF 
QRP. In section VIII.C. of the proposed rule, we proposed to modify one 
measure, adopt three new measures, and remove three measures from the 
IRF QRP.
    As stated in section VIII.C.1.a. of the proposed rule, we proposed 
that IRFs submit data on one modified quality measure, the COVID-19 
Vaccination Coverage among Healthcare Personnel (HCP) (HCP COVID-19 
Vaccine) measure beginning with the FY 2025 IRF QRP. The data is 
collected through the Centers for Disease Control and Prevention 
(CDC's) National Health Safety Network (NHSN). IRFs currently utilize 
the NHSN for purposes of meeting other IRF QRP requirements, including 
the current HCP COVID-19 Vaccine measure. IRFs will continue to submit 
the HCP COVID-19 Vaccine measure data to CMS through the NHSN. The 
burden associated with the HCP COVID-19 Vaccine measure is accounted 
for under the CDC's information collection request currently approved 
under OMB control number 0920-1317 (expiration date: January 31, 2024). 
Because we did not propose any updates to the form, manner, and timing 
of data submission for this HCP COVID-19 Vaccine measure, there will be 
no increase in burden associated with the proposal and refer readers to 
the FY 2022 IRF PPS final rule (86 FR 42399 through 42400) for these 
policies.
    In section VIII.C.1.b. of the proposed rule, we proposed to adopt 
the Discharge Function Score (DC Function) measure beginning with the 
FY 2025 IRF QRP. This assessment-based quality measure will be 
calculated using data from the IRF Patient Assessment Instrument (IRF-
PAI) that are already reported to CMS for payment and quality reporting 
purposes, and the burden is accounted for in the information collection 
request currently approved under OMB control number 0938-0842 
(expiration date: August 31, 2025). There will be no additional burden 
for IRFs associated with the DC Function measure since it does not 
require collection of new data elements.
    In section VIII.C.1.c. of the proposed rule, we also proposed to 
remove the

[[Page 51043]]

Application of Percent of Long-Term Care Hospital Patients with an 
Admission and Discharge Functional Assessment and a Care Plan That 
Addresses Function (Application of Functional Assessment/Care Plan) 
measure beginning with the FY 2025 IRF QRP. We believe that the removal 
of the Application of Functional Assessment/Care Plan measure will 
result in a decrease of 18 seconds (0.3 minutes or 0.005 hours) of 
clinical staff time at admission beginning with the FY 2025 IRF QRP. We 
believe the IRF-PAI item affected by the Application of Functional 
Assessment/Care Plan measure is completed by Occupational Therapists 
(OT), Physical Therapists (PT), Registered Nurses (RN), Licensed 
Practical and Licensed Vocational Nurses (LVN), and/or Speech-Language 
Pathologists (SLP) depending on the functional goal selected. We 
identified the staff type per item based on past IRF burden 
calculations in conjunction with expert opinion. Our assumptions for 
staff type were based on the categories generally necessary to perform 
an assessment. Individual providers determine the staffing resources 
necessary. Therefore, we averaged the national average for these labor 
types and established a composite cost estimate. This composite 
estimate was calculated by weighting each salary based on the following 
breakdown regarding provider types most likely to collect this data: OT 
45 percent; PT 45 percent; RN 5 percent; LVN 2.5 percent; SLP 2.5 
percent. For the purposes of calculating the costs associated with the 
collection of information requirements, we obtained mean hourly wages 
for these staff from the U.S. Bureau of Labor Statistics' (BLS) May 
2021 National Occupational Employment and Wage Estimates.\204\ To 
account for overhead and fringe benefits, we doubled the hourly wage. 
These amounts are detailed in Table 19.
---------------------------------------------------------------------------

    \204\ U.S. Bureau of Labor Statistics' (BLS) May 2021 National 
Occupational Employment and Wage Estimates. https://www.bls.gov/oes/current/oes_nat.htm.
[GRAPHIC] [TIFF OMITTED] TR02AU23.070

    We estimated that the burden and cost for IRFs for complying with 
requirements of the FY 2025 IRF QRP would decrease under our proposal. 
Specifically, we believe that there will be a 0.005 hour decrease in 
clinical staff time to report data for each IRF-PAI completed at 
admission. Using data from calendar year 2021, we estimated 511,938 
admission assessments from 1,133 IRFs annually. This equates to a 
decrease of 2,560 hours in burden at admission for all IRFs (0.005 hour 
x 511,938 admissions). Given 0.135 minutes of occupational therapist 
time at $86.04 per hour, 0.135 minutes of physical therapist time at 
$89.34 per hour, 0.015 minutes registered nurse time at $79.56 per 
hour, 0.0075 minutes of licensed vocational nurse time at $49.86 per 
hour, and 0.0075 minutes of speech language pathologist time at $82.52 
per hour to complete an average of 454 IRF-PAI admission assessments 
per IRF per year, we estimate the total cost will be decreased by 
$194.79 ($220,697.60 total reduction/1,133 IRFs) per IRF annually, or 
$220,697.60 for all IRFs annually based on the proposed removal of the 
Application of Functional Assessment/Care Plan measure.
    In section VIII.C.1.d. of the proposed rule, we proposed to remove 
the IRF Functional Outcome Measure: Change in Self-Care Score for 
Medical Rehabilitation Patients (Change in Self-Care Score) and the IRF 
Functional Outcome Measure: Change in Mobility Score for Medical 
Rehabilitation Patients (Change in Mobility Score) measures beginning 
with the FY 2025 IRF QRP. While these assessment-based quality measures 
were proposed for removal, the data elements used to calculate the 
measures will still be collected by IRFs for payment and quality 
reporting purposes, specifically for other quality measures under the 
IRF QRP. Therefore, we believe that the proposal to remove the Change 
in Self-Care Score and Change in Mobility Score measures will not 
decrease burden for IRFs.
    In section VIII.C.2.a. of the proposed rule, we proposed to adopt 
the COVID-19 Vaccine: Percent of Patients/Residents Who Are Up to Date 
(Patient/Resident COVID-19 Vaccine) measure beginning with the FY 2026 
IRF QRP. The proposed measure will be collected using the IRF-PAI. One 
data element will be added to the IRF-PAI at discharge in order to 
allow for collection of the Patient/Resident COVID-19 Vaccine measure, 
and we believe will result in an increase of 0.3 minutes of clinical 
staff time at discharge. We believe that the additional Patient/
Resident COVID-19 Vaccine measure's data element will be completed 
equally by registered nurses and licensed vocational nurses. Mean 
hourly wages for these staff are detailed in Table 19. However, 
individual IRFs determine the staffing resources necessary. Using data 
from CY 2021, we estimated a total of 779,274 discharges

[[Page 51044]]

on all patients regardless of payer from 1,133 IRFs annually. This 
equates to an increase of 3,896 hours in burden for all IRFs (0.005 
hour x 779,274 admissions). Given 0.15 minutes of registered nurse time 
at $79.56 per hour and 0.15 minutes of licensed vocational nurse time 
at $49.86 per hour to complete an average of 691 IRF-PAI discharge 
assessments per IRF per year, we estimate that the total cost of 
complying with the IRF QRP requirements will be increased by $222.52 
[($64.71/hr x 3,896 hours)/1,133 IRFs) per IRF annually, or $252,110.16 
($64.71/hr x 3,896 hours) for all IRFs annually based on the adoption 
of the Patient/Resident COVID-19 Vaccine measure. The information 
collection request approved under OMB control number 0938-0842 
(expiration date: August 31, 2025) will be revised and sent to OMB for 
approval.
    In summary, under OMB control number 0938-0842, the changes to the 
IRF QRP will result in a burden addition of $27.73 per IRF ($31,412.56/
1,133 IRFs). The total cost increase related to this information 
collection is approximately $31,412.56 and is summarized in Table 20.
[GRAPHIC] [TIFF OMITTED] TR02AU23.071

    We invited public comments on the proposed information collection 
requirements.
    The following is a summary of the public comments received on the 
proposed revisions and our responses:
    Comment: One commenter noted their disappointment that CMS 
continues to add and modify IRF QRP requirements while IRFs are still 
facing operational challenges related to the COVID-19 pandemic. They 
said the proposed modification to the HCP COVID-19 Vaccine measure 
beginning with the FY 2025 IRF QRP will add to their administrative 
burden and compliance costs. They also stated that the net effect of 
the removal of three current measures, the addition of two new 
measures, and the modification of one measure did not reduce any 
administrative burden associated with the IRF QRP.
    Response: We acknowledge that the net effect of our policies 
finalized in this final rule is an increase of $27.73 per IRF per year. 
However, despite the operational challenges imposed by the COVID-19 
pandemic, we must maintain our commitment to quality of care for all 
patients. In this final rule, we have sought to strike an appropriate 
balance between maintaining our commitment to quality of care with the 
impact on IRFs. The result is a reduction of the IRF QRP measure set 
from 18 to 17. We will continue to assess the IRF QRP measure set and 
use our Meaningful Measures Framework and measure removal criteria to 
guide decisions about future changes.
    Comment: Two commenters stated the estimate of 18 seconds or 0.3 
minutes of clinical staff time at discharge underestimates the burden 
of clinical staff to collect the Patient/Resident COVID-19 Vaccine 
measure. One of these commenters estimated the time required by a 
clinician to document a single item in the electronic medical record is 
around 7 seconds. This commenter also suggested the collection of the 
information from the patient to complete the data element will likely 
take far more than the remaining estimated 11 seconds, particularly due 
to the confusing nature and ongoing changes to the definition of ``up 
to date,'' as well as the time necessary to conduct a patient 
interview, reconcile information provided by the patient, review the 
medical records, or contact a proxy for the information. The commenter 
stated that CMS' estimate does not account for the time needed to 
modify their electronic medical record system or to train staff for 
this measure. The other commenter suggested that the clinician type 
included in the burden estimate for the Patient/Resident COVID-19 
Vaccine measure was not inclusive of the range of staff type that would 
need to receive an estimated hour of training. The commenter stated the 
training costs should be considered as a part of the burden estimate 
for completing the item.
    Response: The 18 seconds (0.3 minutes) estimated for this item is 
based on past IRF burden calculations and represents the time it takes 
to encode the IRF-PAI. As the commenter pointed out in their example, 
the patient must be assessed, and information gathered. After the 
patient assessment is completed, the IRF-PAI is coded with the 
information and submitted to the internet Quality Improvement and 
Evaluation System (iQIES), and it is these steps (after the patient 
assessment) that the estimated burden and cost captures. Finally, as we 
stated in section X.A. of this final rule, our assumptions

[[Page 51045]]

for staff type were based on the categories generally necessary to 
perform an assessment, and subsequently encode it, which is consistent 
with past collection of information estimates.\205\ While we 
acknowledge that some IRFs may train and utilize other personnel, our 
estimates are based on the categories of personnel necessary to 
complete the IRF-PAI.
---------------------------------------------------------------------------

    \205\ FY 2016 IRF PPS proposed rule (80 FR 23390).
---------------------------------------------------------------------------

    Comment: We received comments about the burden estimate for the DC 
Function Score measure. One commenter opposed the adoption of this 
measure given the growing burden of administering the IRF QRP, 
workforce shortages, and financial pressures. Two other commenters 
suggested that the measure's adoption will require software updates to 
implement and monitor the measure's complex calculations prior to CMS 
publishing results, as well as additional training and education for 
clinical and administrative personnel. One of these commenters 
recommended CMS should consider these costs because they impact the 
values presented in the FY 2024 IRF PPS proposed rule. Another 
commenter observed IRFs will still need to educate and train their 
clinicians on the new measure, incorporate discussion of this measure 
into their interdisciplinary team meetings, and create a solution that 
will calculate imputation values and the risk-adjusted expected 
discharge function score values in order to manage performance.
    Response: CMS continually looks for opportunities to minimize 
burden associated with collection of the IRF-PAI for information users 
through strategies that simplify collection and submission 
requirements. As discussed in sections IX.C.1.b. and X.A. of this final 
rule, this measure is modeled after the currently adopted Discharge 
Mobility Score and Discharge Self-Care Score measures, and we are not 
proposing changes to the number of items required or the reporting 
frequency of the items reported in the IRF-PAI for this DC Function 
measure. IRFs have been collecting the data elements used in the 
calculation of the DC Function measure since FY 2017. At that time, we 
standardized the collection instructions across all IRFs, ensuring that 
all instructions and notices are written in plain language, and by 
providing step-by-step examples for completing the IRF-PAI. CMS 
provides a dedicated help desk to support users and respond to 
questions about the data collection. Additionally, a dedicated IRF QRP 
web page houses multiple modes of tools, such as instructional videos, 
case studies, user manuals, and frequently asked questions which 
support understanding of the items collected for the DC Function 
measure and the IRF-PAI generally, and these can be used by current 
users and assist new users of the IRF-PAI. CMS utilizes a listserv to 
facilitate outreach to users, such as communicating timely and 
important new material(s), and we will use those outreach resources 
when providing training and information about the new DC Function 
measure. CMS creates data collection specifications for IRF electronic 
health record (EHR) software with `skip' patterns associated with the 
Quality Indicator items used for the DC Function measure to ensure the 
IRF-PAI is limited to the minimum data required to meet quality 
reporting requirements. These specifications are available free of 
charge to all IRFs and their technology partners. Further, these 
minimum requirements are standardized for all users of the IRF-PAI 
assessment forms. Finally, CMS calculates this measure for IRFs, and 
provides IRFs with various resources to review and monitor their own 
performance on this measure, including a free internet-based system 
through which users can access on-demand reports for feedback on the 
collection of the IRF-PAI associated with their facility.
    After considering the public comments received, and for the reasons 
outlined in this section of the final rule and our comment responses, 
we are finalizing the revisions as proposed.

XII. Regulatory Impact Analysis

A. Statement of Need

    This final rule updates the IRF prospective payment rates for FY 
2024 as required under section 1886(j)(3)(C) of the Act and in 
accordance with section 1886(j)(5) of the Act, which requires the 
Secretary to publish in the Federal Register on or before August 1 
before each FY, the classification and weighting factors for CMGs used 
under the IRF PPS for such FY and a description of the methodology and 
data used in computing the prospective payment rates under the IRF PPS 
for that FY. This final rule also implements section 1886(j)(3)(C) of 
the Act, which requires the Secretary to apply a productivity 
adjustment to the market basket increase factor for FY 2012 and 
subsequent years.
    Furthermore, this final rule adopts policy changes to the IRF QRP 
under the statutory discretion afforded to the Secretary under section 
1886(j)(7) of the Act. We are finalizing updates to the IRF QRP 
requirements beginning with the FY 2025 IRF QRP and FY 2026 IRF QRP. We 
are finalizing a modification to a current measure in the IRF QRP which 
we believe will encourage healthcare personnel to remain up to date 
with the COVID-19 vaccine, resulting in fewer cases, less 
hospitalizations, and lower mortality associated with the virus. We are 
finalizing the adoption of two new measures: one measure to maintain 
compliance with the requirements of section 1899B of the Act and 
replace the current cross-setting process measure with a measure that 
is more strongly associated with desired patient functional outcomes; 
and a second measure that supports the goals of CMS Meaningful Measures 
Initiative 2.0 to empower consumers with tools and information as they 
make healthcare choices as well as assist IRFs to leverage their care 
processes to increase vaccination coverage in their settings to protect 
residents and prevent negative outcomes. We are finalizing the removal 
of three measures from the IRF QRP as they meet the criteria specified 
at Sec.  412.634(b)(2) for measure removal.

B. Overall Impact

    We have examined the impacts of this rule as required by Executive 
Order 12866 on Regulatory Planning and Review (September 30, 1993), 
Executive Order 13563 on Improving Regulation and Regulatory Review 
(January 18, 2011), Executive Order 14094 entitled ``Modernizing 
Regulatory Review'' (April 6, 2023), the Regulatory Flexibility Act 
(RFA) (September 19, 1980, Pub. L. 96-354), section 1102(b) of the 
Social Security Act, section 202 of the Unfunded Mandates Reform Act of 
1995 (March 22, 1995; Pub. L. 104-4), Executive Order 13132 on 
Federalism (August 4, 1999) and the Congressional Review Act (5 U.S.C. 
804(2)).
    Executive Orders 12866 and 13563 direct agencies to assess all 
costs and benefits of available regulatory alternatives and, if 
regulation is necessary, to select regulatory approaches that maximize 
net benefits (including potential economic, environmental, public 
health and safety effects, distributive impacts, and equity). The 
Executive Order 14094 entitled ``Modernizing Regulatory Review'' 
(hereinafter, the Modernizing E.O.) amends section 3(f)(1) of Executive 
Order 12866 (Regulatory Planning and Review). The amended section 3(f) 
of Executive Order 12866 defines a ``significant regulatory action'' as 
an action that is likely to result in a rule: (1) having an annual 
effect on the economy of $200 million or more in any 1 year (adjusted 
every 3 years by the

[[Page 51046]]

Administrator of OIRA for changes in gross domestic product), or 
adversely affect in a material way the economy, a sector of the 
economy, productivity, competition, jobs, the environment, public 
health or safety, or State, local, territorial, or tribal governments 
or communities; (2) creating a serious inconsistency or otherwise 
interfering with an action taken or planned by another agency; (3) 
materially altering the budgetary impacts of entitlement grants, user 
fees, or loan programs or the rights and obligations of recipients 
thereof; or (4) raise legal or policy issues for which centralized 
review would meaningfully further the President's priorities or the 
principles set forth in this Executive order, as specifically 
authorized in a timely manner by the Administrator of OIRA in each 
case.
    A regulatory impact analysis (RIA) must be prepared for major rules 
with significant regulatory action/s and/or with significant effects as 
per section 3(f)(1) ($200 million or more in any 1 year). We estimate 
the total impact of the policy updates described in this final rule by 
comparing the estimated payments in FY 2024 with those in FY 2023. This 
analysis results in an estimated $355 million increase for FY 2024 IRF 
PPS payments. Additionally, we estimate that costs associated with 
updating the reporting requirements under the IRF QRP result in an 
estimated $31,783,532.15 additional cost in FY 2026 for IRFs. Based on 
our estimates, OMB's Office of Information and Regulatory Affairs has 
determined this rulemaking is significant per section 3(f)(1) as 
measured by the $200 million or more in any 1 year, and hence also a 
major rule under Subtitle E of the Small Business Regulatory 
Enforcement Fairness Act of 1996 (also known as the Congressional 
Review Act). Accordingly, we have prepared an RIA that, to the best of 
our ability, presents the costs and benefits of the rulemaking.

C. Anticipated Effects

1. Effects on IRFs
    The RFA requires agencies to analyze options for regulatory relief 
of small entities, if a rule has a significant impact on a substantial 
number of small entities. For purposes of the RFA, small entities 
include small businesses, nonprofit organizations, and small 
governmental jurisdictions. Most IRFs and most other providers and 
suppliers are small entities, either by having revenues of $8.0 million 
to $41.5 million or less in any 1 year depending on industry 
classification, or by being nonprofit organizations that are not 
dominant in their markets. (For details, see the Small Business 
Administration's final rule that set forth size standards for health 
care industries, at 65 FR 69432 at https://www.sba.gov/sites/default/files/2019-08/SBA%20Table%20of%20Size%20Standards_Effective%20Aug%2019%2C%202019_Rev.pdf, effective January 1, 2017 and updated on August 19, 2019.) Because 
we lack data on individual hospital receipts, we cannot determine the 
number of small proprietary IRFs or the proportion of IRFs' revenue 
that is derived from Medicare payments. Therefore, we assume that all 
IRFs (an approximate total of 1,133 IRFs, of which approximately 50 
percent are nonprofit facilities) are considered small entities and 
that Medicare payment constitutes the majority of their revenues. HHS 
generally uses a revenue impact of 3 to 5 percent as a significance 
threshold under the RFA. As shown in Table 21, we estimate that the net 
revenue impact of the final rule on all IRFs is to increase estimated 
payments by approximately 4.0 percent. The rates and policies set forth 
in this final rule will not have a significant impact (not greater than 
4 percent) on a substantial number of small entities. The estimated 
impact on small entities is shown in Table 21. MACs are not considered 
to be small entities. Individuals and States are not included in the 
definition of a small entity.
    In addition, section 1102(b) of the Act requires us to prepare an 
RIA if a rule may have a significant impact on the operations of a 
substantial number of small rural hospitals. This analysis must conform 
to the provisions of section 604 of the RFA. For purposes of section 
1102(b) of the Act, we define a small rural hospital as a hospital that 
is located outside of a Metropolitan Statistical Area and has fewer 
than 100 beds. As shown in Table 21, we estimate that the net revenue 
impact of this final rule on rural IRFs is to increase estimated 
payments by approximately 3.6 percent based on the data of the 135 
rural units and 12 rural hospitals in our database of 1,133 IRFs for 
which data were available. We estimate an overall impact for rural IRFs 
in all areas between 2.0 percent and 6.2 percent. As a result, we 
anticipate that this final rule will not have a significant impact on a 
substantial number of small entities.
    Section 202 of the Unfunded Mandates Reform Act of 1995 (Pub. L. 
104-04, enacted March 22, 1995) (UMRA) also requires that agencies 
assess anticipated costs and benefits before issuing any rule whose 
mandates require spending in any 1 year of $100 million in 1995 
dollars, updated annually for inflation. In 2023, that threshold is 
approximately $177 million. This final rule does not mandate any 
requirements for State, local, or tribal governments, or for the 
private sector.
    Executive Order 13132 establishes certain requirements that an 
agency must meet when it issues a proposed rule (and subsequent final 
rule) that imposes substantial direct requirement costs on State and 
local governments, preempts State law, or otherwise has federalism 
implications. As stated, this final rule will not have a substantial 
effect on State and local governments, preempt State law, or otherwise 
have a federalism implication.
2. Detailed Economic Analysis
    This final rule will update the IRF PPS rates contained in the FY 
2023 IRF PPS final rule (87 FR 47038). Specifically, this final rule 
will update the CMG relative weights and ALOS values, the wage index, 
and the outlier threshold for high-cost cases. This final rule will 
apply a productivity adjustment to the FY 2024 IRF market basket 
increase factor in accordance with section 1886(j)(3)(C)(ii)(I) of the 
Act. Further, this final rule rebases and revises the IRF market basket 
to reflect a 2021 base year. We are also modifying the regulation 
governing when IRF units can be excluded and paid under the IRF PPS.
    We estimate that the impact of the changes and updates described in 
this final rule would be a net estimated increase of $355 million in 
payments to IRFs. The impact analysis in Table 21 of this final rule 
represents the projected effects of the updates to IRF PPS payments for 
FY 2024 compared with the estimated IRF PPS payments in FY 2023. We 
determine the effects by estimating payments while holding all other 
payment variables constant. We use the best data available, but we do 
not attempt to predict behavioral responses to these changes, and we do 
not make adjustments for future changes in such variables as number of 
discharges or case-mix.
    We note that certain events may combine to limit the scope or 
accuracy of our impact analysis, because such an analysis is future-
oriented and, thus, susceptible to forecasting errors because of other 
changes in the forecasted impact time period. Some examples could be 
legislative changes made by the Congress to the Medicare program that 
would impact program funding, or changes specifically related to IRFs. 
Although some of these changes may not necessarily be specific to the 
IRF PPS, the nature of the Medicare program

[[Page 51047]]

is such that the changes may interact, and the complexity of the 
interaction of these changes could make it difficult to predict 
accurately the full scope of the impact upon IRFs.
    In updating the rates for FY 2024, we are implementing the standard 
annual revisions described in this final rule (for example, the update 
to the wage index and market basket increase factor used to adjust the 
Federal rates). We are also reducing the FY 2024 IRF market basket 
increase factor by a productivity adjustment in accordance with section 
1886(j)(3)(C)(ii)(I) of the Act. We estimate the total increase in 
payments to IRFs in FY 2024, relative to FY 2023, would be 
approximately $355 million.
    This estimate is derived from the application of the FY 2024 IRF 
market basket increase factor, as reduced by a productivity adjustment 
in accordance with section 1886(j)(3)(C)(ii)(I) of the Act, which 
yields an estimated increase in aggregate payments to IRFs of $305 
million. However, there is an estimated $50 million increase in 
aggregate payments to IRFs due to the update to the outlier threshold 
amount. Therefore, we estimate that these updates would result in a net 
increase in estimated payments of $355 million from FY 2023 to FY 2024.
    The effects of the updates that impact IRF PPS payment rates are 
shown in Table 21. The following updates that affect the IRF PPS 
payment rates are discussed separately below:
     The effects of the update to the outlier threshold amount, 
from approximately 2.5 percent to 3.0 percent of total estimated 
payments for FY 2024, consistent with section 1886(j)(4) of the Act.
     The effects of the annual market basket update (using the 
2021-based IRF market basket) to IRF PPS payment rates, as required by 
sections 1886(j)(3)(A)(i) and (j)(3)(C) of the Act, including a 
productivity adjustment in accordance with section 1886(j)(3)(C)(ii)(I) 
of the Act.
     The effects of applying the budget-neutral labor-related 
share and wage index adjustment, as required under section 1886(j)(6) 
of the Act, accounting for the permanent cap on wage index decreases 
when applicable.
     The effects of the budget-neutral changes to the CMG 
relative weights and ALOS values under the authority of section 
1886(j)(2)(C)(i) of the Act.
     The total change in estimated payments based on the FY 
2024 payment changes relative to the estimated FY 2023 payments.
3. Description of Table 21
    Table 21 shows the overall impact on the 1,133 IRFs included in the 
analysis.
    The next 12 rows of Table 21 contain IRFs categorized according to 
their geographic location, designation as either a freestanding 
hospital or a unit of a hospital, and by type of ownership; all urban, 
which is further divided into urban units of a hospital, urban 
freestanding hospitals, and by type of ownership; and all rural, which 
is further divided into rural units of a hospital, rural freestanding 
hospitals, and by type of ownership. There are 986 IRFs located in 
urban areas included in our analysis. Among these, there are 648 IRF 
units of hospitals located in urban areas and 338 freestanding IRF 
hospitals located in urban areas. There are 147 IRFs located in rural 
areas included in our analysis. Among these, there are 135 IRF units of 
hospitals located in rural areas and 12 freestanding IRF hospitals 
located in rural areas. There are 459 for-profit IRFs. Among these, 
there are 424 IRFs in urban areas and 35 IRFs in rural areas. There are 
571 non-profit IRFs. Among these, there are 480 urban IRFs and 91 rural 
IRFs. There are 103 government-owned IRFs. Among these, there are 82 
urban IRFs and 21 rural IRFs.
    The remaining four parts of Table 21 show IRFs grouped by their 
geographic location within a region, by teaching status, and by DSH 
patient percentage (PP). First, IRFs located in urban areas are 
categorized for their location within a particular one of the nine 
Census geographic regions. Second, IRFs located in rural areas are 
categorized for their location within a particular one of the nine 
Census geographic regions. In some cases, especially for rural IRFs 
located in the New England, Mountain, and Pacific regions, the number 
of IRFs represented is small. IRFs are then grouped by teaching status, 
including non-teaching IRFs, IRFs with an intern and resident to 
average daily census (ADC) ratio less than 10 percent, IRFs with an 
intern and resident to ADC ratio greater than or equal to 10 percent 
and less than or equal to 19 percent, and IRFs with an intern and 
resident to ADC ratio greater than 19 percent. Finally, IRFs are 
grouped by DSH PP, including IRFs with zero DSH PP, IRFs with a DSH PP 
less than 5 percent, IRFs with a DSH PP between 5 and less than 10 
percent, IRFs with a DSH PP between 10 and 20 percent, and IRFs with a 
DSH PP greater than 20 percent.
    The estimated impacts of each policy described in this rule to the 
facility categories listed are shown in the columns of Table 21. The 
description of each column is as follows:
     Column (1) shows the facility classification categories.
     Column (2) shows the number of IRFs in each category in 
our FY 2024 analysis file.
     Column (3) shows the number of cases in each category in 
our FY 2024 analysis file.
     Column (4) shows the estimated effect of the adjustment to 
the outlier threshold amount.
     Column (5) shows the estimated effect of the update to the 
IRF labor-related share and wage index, in a budget-neutral manner.
     Column (6) shows the estimated effect of the update to the 
CMG relative weights and ALOS values, in a budget-neutral manner.
     Column (7) compares our estimates of the payments per 
discharge, incorporating all of the policies reflected in this final 
rule for FY 2024 to our estimates of payments per discharge in FY 2023.
    The average estimated increase for all IRFs is approximately 4.0 
percent. This estimated net increase includes the effects of the IRF 
market basket update for FY 2024 of 3.4 percent, which is based on a 
IRF market basket increase factor of 3.6 percent, less a 0.2 percentage 
point productivity adjustment, as required by section 
1886(j)(3)(C)(ii)(I) of the Act. It also includes the approximate 0.6 
percent overall increase in estimated IRF outlier payments from the 
update to the outlier threshold amount. Since we are making the updates 
to the IRF wage index, labor-related share and the CMG relative weights 
in a budget-neutral manner, they will not be expected to affect total 
estimated IRF payments in the aggregate. However, as described in more 
detail in each section, they will be expected to affect the estimated 
distribution of payments among providers.

[[Page 51048]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.072


[[Page 51049]]


[GRAPHIC] [TIFF OMITTED] TR02AU23.073

4. Impact of the Update to the Outlier Threshold Amount
    The estimated effects of the update to the outlier threshold 
adjustment are presented in column 4 of Table 21.
    For the FY 2024 proposed rule, we used preliminary FY 2022 IRF 
claims data and based on that preliminary analysis, we estimated that 
IRF outlier payments as a percentage of total estimated IRF payments 
would be 2.3 percent in FY 2023. As we typically do between the 
proposed and final rules each year, we updated our FY 2022 IRF claims 
data to ensure that we are using the most recent available data in 
setting IRF payments. Therefore, based on an updated analysis of the 
most recent IRF claims data for this final rule, we estimate that IRF 
outlier payments as a percentage of total estimated IRF payments are 
2.5 percent in FY 2023. Thus, we are adjusting the outlier threshold 
amount in this final rule to maintain total estimated outlier payments 
equal to 3 percent of total estimated payments in FY 2024.
    The impact of this update to the outlier threshold amount (as shown 
in column 4 of Table 21) is to increase estimated overall payments to 
IRFs by 0.6 percentage point. We do not estimate that any group of IRFs 
would experience a decrease in payments from this proposed update.
5. Impact of the Wage Index, Labor-Related Share, and Wage Index Cap
    In column 5 of Table 21, we present the effects of the budget-
neutral update of the wage index and labor-related share, taking into 
account the permanent 5 percent cap on wage index decreases, when 
applicable. The changes to the wage index and the labor-related share 
are discussed together because the wage index is applied to the labor-
related share portion of payments, so the changes in the two have a 
combined effect on payments to providers. As discussed in section VI.E. 
of this final rule, we update the FY 2024 labor-related share from 72.9 
percent in FY 2023 to 74.1 percent in FY 2024. In aggregate, we do not 
estimate that these updates will affect overall estimated payments to 
IRFs. However, we do expect these updates to have small distributional 
effects. We estimate the largest decrease in payment from the update to 
the CBSA wage index and labor-related share to be a 2.3 percent 
decrease for IRFs in the Rural New England region and the largest 
increase in payment to be a 0.5 percent increase for IRFs in the Urban 
Middle Atlantic Region.
6. Impact of the Update to the CMG Relative Weights and ALOS Values
    In column 6 of Table 21, we present the effects of the budget-
neutral update of the CMG relative weights and ALOS values. In the 
aggregate, we do not estimate that these updates will affect overall 
estimated payments of IRFs. However, we do expect these updates to have 
small distributional effects, with the largest effect being an increase 
in payments of 0.2 percent to IRFs in the Rural New England region.
7. Effects of Modification of the Regulation for Excluded IRF Units 
Paid Under the IRF PPS
    As discussed in section VIII. of this final rule, we are amending 
the regulation text at Sec.  412.25(c)(1) in this final rule.
    We do not anticipate a financial impact associated with the 
modification of the regulation for excluded IRF units paid under the 
IRF PPS because an IRF unit would simply be opening on a different date 
(in the middle of a cost reporting period) than they otherwise would 
have (at the start of a cost reporting period). Although this 
modification to the regulatory requirements significantly reduces the 
burden of opening new IRF units and reduces IRF's construction costs, 
we do not believe that it will significantly affect IRF payments.
    In response to the need for availability of inpatient 
rehabilitation beds we are implementing changes to Sec.  412.25(c) to 
allow greater flexibility for hospitals to open excluded units, while 
minimizing the amount of effort that Medicare contractors would need to 
spend administering the regulatory requirements. We believe this change 
will provide IRFs greater flexibility when establishing an excluded 
unit at a time other than the start of a cost reporting period.
8. Effects of Requirements for the IRF QRP Beginning With FY 2025
    In accordance with section 1886(j)(7)(A) of the Act, the Secretary 
must reduce by 2 percentage points the annual market basket increase 
factor otherwise applicable to an IRF for a fiscal year if the IRF does 
not comply with the requirements of the IRF QRP for that fiscal year. 
In section IX.A. of the proposed rule, we discussed the method for 
applying the 2 percentage point reduction to IRFs that fail to meet the 
IRF QRP requirements.
    As discussed in section IX.C.1.a. of this final rule, we are 
finalizing the proposal to modify one measure in the IRF QRP beginning 
with the FY 2025 IRF QRP, the HCP COVID-19 Vaccine measure. We believe 
that the burden associated with the IRF QRP is the time and effort 
associated with complying with the non-claims-based measures 
requirements of the IRF QRP. The burden associated with the HCP COVID-
19 Vaccine measure is accounted for under the CDC PRA package currently 
approved under OMB control number 0920-1317 (expiration January 31, 
2024).
    As discussed in section IX.C.1.b. of this final rule, we are 
finalizing the proposal for IRFs to collect data on one new quality 
measure, the DC Function measure, beginning with assessments

[[Page 51050]]

completed on October 1, 2023. However, the measure utilizes data items 
that IRFs already report to CMS for payment and quality reporting 
purposes, and therefore the burden is accounted for in the PRA package 
approved under OMB control number 0938-0842 (expiration August 31, 
2025).
    As discussed in section IX.C.1.c. of this final rule, we are 
finalizing the proposal to remove the Application of Functional 
Assessment/Care Plan measure, from the IRF QRP, and this proposal would 
result in a decrease of 0.3 minutes of clinical staff time beginning 
with admission assessments completed on October 1, 2023. The proposed 
decrease in burden will be accounted for in a revised information 
collection request under OMB control number (0938-0842), and we 
provided impact information. We believe the data element for this 
quality measure is completed by occupational therapists (45 percent of 
the time or 0.135 minutes), physical therapists (45 percent of the time 
or 0.135 minutes), registered nurses (5 percent of the time or 0.015 
minutes), licensed practical and vocational nurses (2.5 percent of the 
time or 0.0075 minutes), or by speech-language pathologists (2.5 
percent of the time or 0.0075 minutes). For the purposes of calculating 
the costs associated with the collection of information requirements, 
we obtained mean hourly wages for these staff from the U.S. Bureau of 
Labor Statistics' (BLS) May 2021 National Occupational Employment and 
Wage Estimates.\206\ To account for overhead and fringe benefits, we 
doubled the hourly wage. These amounts are detailed in Table 22.
---------------------------------------------------------------------------

    \206\ U.S. Bureau of Labor Statistics' (BLS) May 2021 National 
Occupational Employment and Wage Estimates. https://www.bls.gov/oes/current/oes_nat.htm.
[GRAPHIC] [TIFF OMITTED] TR02AU23.074

    With 511,938 admissions from 1,133 IRFs annually, we estimated an 
annual burden decrease of 2,560 fewer hours (511,938 admissions x .005 
hours) and a decrease of $220,697.60 [2,560 hours x $86.21/hr)]. For 
each IRF we estimated an annual burden decrease of 2.26 hours (2,560 
hours/1,133 IRFs) at a savings of $194.79 ($220,697.60/1,133 IRFs).
    As discussed in section IX.C.1.d. of this final rule, we are 
finalizing the removal of two additional measures from the IRF QRP, the 
Change in Self-Care Score and Change in Mobility Score measures, 
beginning with assessments completed on October 1, 2023. However, the 
data items used in the calculation of this measure are used for other 
payment and quality reporting purposes, and therefore there is no 
change in burden associated with this proposal.
9. Effects of Requirements for the IRF QRP Beginning With FY 2026
    As discussed in section IX.C.2.a. of this final rule, we are 
finalizing the adoption of the Patient/Resident COVID-19 Vaccine 
measure, beginning with the FY 2026 IRF QRP. We estimated this measure 
would result in an increase of 0.3 minutes of clinical staff time 
beginning with discharge assessments completed on October 1, 2024. 
Although the increase in burden will be accounted for in a revised 
information collection request under OMB control number 0938-0842, we 
provided impact information. We estimated the data element for this 
quality measure would be completed by registered nurses (50 percent of 
the time or 0.15 minutes) or by licensed practical and vocational 
nurses (50 percent of the time or 0.15 minutes). For the purposes of 
calculating the costs associated with the collection of information 
requirements, we obtained mean hourly wages for these staff from the 
U.S. Bureau of Labor Statistics' (BLS) May 2021 National Occupational 
Employment and Wage Estimates.\207\ To account for overhead and fringe 
benefits, we doubled the hourly wage. These amounts are detailed in 
Table 22. With 779,274 discharges on all patients regardless of payer 
from 1,133 IRFs annually, we estimated an annual burden increase of 
3,896 hours (779,274 discharges x 0.005 hours) and an increase of 
$252,110.16 ($64.71/hr x 3,896 hours). For each IRF, we estimated an 
annual burden increase of 3.44 hours (3,896 hours/1,133 IRFs) at an 
additional cost of $222.52 ($252,110.16/1,133 IRFs).
---------------------------------------------------------------------------

    \207\ U.S. Bureau of Labor Statistics' (BLS) May 2021 National 
Occupational Employment and Wage Estimates. https://www.bls.gov/oes/current/oes_nat.htm.
---------------------------------------------------------------------------

    In summary, under OMB control number 0938-0842, the changes to the 
IRF QRP will result in an estimated increase in programmatic burden for 
1,133 IRFs. The total burden increase is approximately $31,412.56 for 
all IRFs and $27.73 per IRF and is summarized in Table 23.

[[Page 51051]]

[GRAPHIC] [TIFF OMITTED] TR02AU23.075

    We invited public comments on the overall impact of the IRF QRP 
proposals for FY 2025 and FY 2026.
    We did not receive any comments on the proposed revisions and 
therefore, we are finalizing the revisions as proposed.

D. Alternatives Considered

    The following is a discussion of the alternatives considered for 
the IRF PPS updates contained in this final rule.
    Section 1886(j)(3)(C) of the Act requires the Secretary to update 
the IRF PPS payment rates by an increase factor that reflects changes 
over time in the prices of an appropriate mix of goods and services 
included in the covered IRF services.
    We proposed to adopt a market basket increase factor for FY 2024 
that is based on a rebased and revised market basket reflecting a 2021 
base year. We considered the alternative of continuing to use the 2016-
based IRF market basket without rebasing to determine the market basket 
increase factor for FY 2024. However, we typically rebase and revise 
the market baskets for the various PPS every 4 to 5 years so that the 
cost weights and price proxies reflect more recent data. Therefore, we 
believe it is more technically appropriate to use a 2021-based IRF 
market basket since it allows for the FY 2024 market basket increase 
factor to reflect a more up-to-date cost structure experienced by IRFs.
    As noted previously in this final rule, section 1886(j)(3)(C) of 
the Act requires the Secretary to update the IRF PPS payment rates by 
an increase factor that reflects changes over time in the prices of an 
appropriate mix of goods and services included in the covered IRF 
services and section 1886(j)(3)(C)(ii)(I) of the Act requires the 
Secretary to apply a productivity adjustment to the market basket 
increase factor for FY 2024. Thus, in accordance with section 
1886(j)(3)(C) of the Act, we are updating the IRF prospective payments 
in this final rule by 3.4 percent (which equals the 3.6 percent 
estimated IRF market basket increase factor for FY 2024 reduced by a 
0.2 percentage point productivity adjustment as determined under 
section 1886(b)(3)(B)(xi)(II) of the Act (as required by section 
1886(j)(3)(C)(ii)(I) of the Act)).
    We considered maintaining the existing CMG relative weights and 
average length of stay values for FY 2024. However, in light of 
recently available data and our desire to ensure that the CMG relative 
weights and average length of stay values are as reflective as possible 
of recent changes in IRF utilization and case mix, we believe that it 
is appropriate to update the CMG relative weights and average length of 
stay values at this time to ensure that IRF PPS payments continue to 
reflect as accurately as possible the current costs of care in IRFs.
    We considered maintaining the existing outlier threshold amount for 
FY 2024. However, analysis of updated FY 2023 data indicates that 
estimated outlier payments would be less than 3 percent of total 
estimated payments for FY 2024, unless we updated the outlier threshold 
amount. Consequently, we are adjusting the outlier threshold amount in 
this final rule to maintain estimated outlier payments at 3 percent of 
estimated aggregate payments in FY 2024.
    We considered not modifying the regulation governing when IRF units 
can be excluded and paid under the IRF PPS. However, we believe that 
amending the regulation would provide hospitals greater flexibility 
when establishing an IRF.
    With regard to the proposal to modify the HCP COVID-19 Vaccine 
measure and to add the Patient/Resident COVID-19 Vaccine measure to the 
IRF QRP Program, the COVID-19 pandemic has exposed the importance of 
implementing infection prevention strategies, including the promotion 
of COVID-19 vaccination for HCP and patients/residents. We believe 
these measures would encourage healthcare personnel to get up to date 
with the COVID-19 vaccine and increase vaccine uptake in patients/
residents resulting in fewer cases, less hospitalizations, and lower 
mortality associated with the SARS-CoV-2 virus, but we were unable to 
identify any alternative methods for collecting the data. An 
overwhelming public need exists to target quality improvement among 
IRFs as well as provide data to patients and caregivers through 
transparency of data. Therefore, these measures have the potential to 
generate actionable data on COVID-19 vaccination rates.
    The proposal to replace the topped-out Application of Functional 
Assessment/Care Plan process measure with the proposed DC Function 
measure, which has strong scientific acceptability, satisfies the 
requirement that there be at least one cross-setting function measure 
in the PAC QRPs, including the IRF QRP, that uses standardized 
functional assessment data elements from standardized patient

[[Page 51052]]

assessment instruments. We considered the alternative of delaying the 
proposal of adopting the DC Function measure. However, given the 
proposed DC Function measure's strong scientific acceptability, the 
fact that it provides an opportunity to replace the current cross-
setting process measure (that is, the Application of Functional 
Assessment/Care Plan measure) with an outcome measure, and uses 
standardized functional assessment data elements that are already 
collected, we believe further delay of the DC Function measure is 
unwarranted. Further, the removal of the Application of Functional 
Assessment/Care Plan measure meets measure removal factors one and six, 
and no longer provides meaningful distinctions in improvements in 
performance. Finally, the removal of the Change in Self-Care Score and 
Change in Mobility Score measures meets measure removal factor eight, 
and the costs associated with these measures outweigh the benefits of 
their use in the program. Therefore, no alternatives were considered.

E. Regulatory Review Costs

    If regulations impose administrative costs on private entities, 
such as the time needed to read and interpret this final rule, we 
should estimate the cost associated with regulatory review. Due to the 
uncertainty involved with accurately quantifying the number of entities 
that will review the rule, we assume that the total number of unique 
commenters on the FY 2024 IRF PPS proposed rule will be the number of 
reviewers of this year's final rule. We acknowledge that this 
assumption may understate or overstate the costs of reviewing this 
final rule. It is possible that not all commenters reviewed the FY 2024 
IRF PPS proposed rule in detail, and it is also possible that some 
reviewers chose not to comment on the FY 2024 proposed rule. For these 
reasons, we thought that the number of commenters would be a fair 
estimate of the number of reviewers of this final rule.
    We also recognize that different types of entities are in many 
cases affected by mutually exclusive sections of this final rule, and 
therefore, for the purposes of our estimate we assume that each 
reviewer reads approximately 50 percent of the rule.
    Using the national mean hourly wage data from the May 2022 BLS for 
Occupational Employment Statistics (OES) for medical and health service 
managers (SOC 11-9111), we estimate that the cost of reviewing this 
rule is $123.06 per hour, including overhead and fringe benefits 
(https://www.bls.gov/oes/current/oes_nat.htm). Assuming an average 
reading speed, we estimate that it would take approximately 3 hours for 
the staff to review half of this final rule. For each reviewer of the 
rule, the estimated cost is $369.18 (3 hours x $123.06). Therefore, we 
estimate that the total cost of reviewing this regulation is $16,613.10 
($369.18 x 45 reviewers).

F. Accounting Statement and Table

    As required by OMB Circular A-4 (available at https://www.whitehouse.gov/wp-content/uploads/legacy_drupal_files/omb/circulars/A4/a-4.pdf), in Table 24 we have prepared an accounting 
statement showing the classification of the expenditures associated 
with the provisions of this final rule. Table 24 provides our best 
estimate of the increase in Medicare payments under the IRF PPS as a 
result of the updates presented in this final rule based on the data 
for 1,133 IRFs in our database.
[GRAPHIC] [TIFF OMITTED] TR02AU23.076

G. Conclusion

    Overall, the estimated payments per discharge for IRFs in FY 2024 
are projected to increase by 4.0 percent, compared with the estimated 
payments in FY 2023, as reflected in column 7 of Table 21.
    IRF payments per discharge are estimated to increase by 4.0 percent 
in urban areas and 3.6 percent in rural areas, compared with estimated 
FY 2023 payments. Payments per discharge to rehabilitation units are 
estimated to increase 4.5 percent in urban areas and 3.9 percent in 
rural areas. Payments per discharge to freestanding rehabilitation 
hospitals are estimated to increase 3.7 percent in urban areas and 2.8 
percent in rural areas.
    Overall, IRFs are estimated to experience a net increase in 
payments as a result of the policies in this final rule. The largest 
payment increase is estimated to be a 6.2 percent increase for IRFs 
located in the Rural Pacific region. The analysis above, together with 
the remainder of this preamble, provides an RIA.
    In accordance with the provisions of Executive Order 12866, this 
regulation was reviewed by OMB.
    Chiquita Brooks-LaSure, Administrator of the Centers for Medicare & 
Medicaid Services, approved this document on July 24, 2023.

Xavier Becerra,
Secretary, Department of Health and Human Services.
[FR Doc. 2023-16050 Filed 7-27-23; 4:15 pm]
BILLING CODE 4120-01-P