Inpatient Rehabilitation Facility - Patient Assessment Instrument

Inpatient Rehabilitation Assessment Instrument and Data Set for PPS for Inpatient Rehabilitation Facilities (CMS-10036)

Proposed-Specifications-for-IRF-QRP-Quality-Measures-and-SPADE

Inpatient Rehabilitation Facility - Patient Assessment Instrument

OMB: 0938-0842

⚠️ Notice: This form may be outdated. More recent filings and information on OMB 0938-0842 can be found here:

Document [pdf]

Download: pdf | pdf

April 2019

Proposed Specifications for IRF QRP
Quality Measures and Standardized
Patient Assessment Data Elements
(SPADEs)

Prepared for
Center for Clinical Standards and Quality
Centers for Medicare & Medicaid Services
Mail Stop C3-19-26
7500 Security Boulevard
Baltimore, MD 21244-1850
Prepared by
RTI International
3040 Cornwallis Road
Research Triangle Park, NC 27709
CMS Contract No. HHSM-500-2013-13015I
RAND Corporation
1776 Main Street
Santa Monica, California 90401-3208
CMS Contract No. HHSM-500-2013-13014I

Table of Contents

Table of Contents ........................................................................................................................................... i
Chapter 1 IMPACT ACT Measures Beginning with the FY 2022 IRF QRP ............................................... 1
Section 1. Cross-Setting Measures Development Work: An Introduction ............................................... 1
Section 2. Cross-Setting Proposed Measure: Transfer of Health Information to the Provider–Post-Acute
Care Measure ............................................................................................................................................ 2
Measure Description ............................................................................................................................. 2
Purpose/Rationale for the Quality Measure .......................................................................................... 3
Denominator ......................................................................................................................................... 6
Numerator ............................................................................................................................................. 6
Measure Time Window......................................................................................................................... 6
Items Included in the Quality Measure ................................................................................................. 7
Risk Adjustment.................................................................................................................................... 7
Quality Measure Calculation Steps ....................................................................................................... 7
Quality Measure Coding Steps ............................................................................................................. 8
Section 3. Cross-Setting Proposed Measure: Transfer of Health Information to the Patient–Post-Acute
Care Measure ............................................................................................................................................ 9
Measure Description ............................................................................................................................. 9
Purpose/Rationale for the Quality Measure .......................................................................................... 9
Denominator ....................................................................................................................................... 11
Numerator ........................................................................................................................................... 12
Measure Time Window....................................................................................................................... 12
Items Included in the Quality Measure ............................................................................................... 12
Risk Adjustment.................................................................................................................................. 13
Quality Measure Calculation Steps ..................................................................................................... 13
Quality Measure Coding Steps ........................................................................................................... 13
Section 4. Update to the Discharge to Community–Post Acute Care (PAC) Inpatient Rehabilitation
Facility (IRF) Quality Reporting Program (QRP) Measure .................................................................... 15
Measure Update .................................................................................................................................. 15
Measure Description ........................................................................................................................... 15
Purpose/Rationale for the Measure ..................................................................................................... 15
Denominator ....................................................................................................................................... 19
Numerator ........................................................................................................................................... 19
Target Population and Measure Exclusions ........................................................................................ 21
Data Sources ....................................................................................................................................... 24
Measure Time Window....................................................................................................................... 24
Statistical Risk Model and Risk Adjustment Covariates .................................................................... 24
i

Measure Calculation Algorithm .......................................................................................................... 26
Chapter 2 Standardized Patient Assessment Data Elements ....................................................................... 28
Section 1: Introduction ............................................................................................................................ 28
Background ......................................................................................................................................... 28
National Beta Test............................................................................................................................... 29
Section 2: Cognitive Function................................................................................................................. 33
Brief Interview for Mental Status (BIMS) .......................................................................................... 34
Confusion Assessment Method (CAM©) ........................................................................................... 37
Mental Status (Depressed Mood) ........................................................................................................ 39
Patient Health Questionnaire-2 to 9 (PHQ-2 to 9) .............................................................................. 40
Section 3: Special Services, Treatments, and Interventions (Including Nutritional Approaches) .......... 45
Chemotherapy (IV, Oral, Other) ......................................................................................................... 45
Radiation ............................................................................................................................................. 48
Oxygen Therapy (Intermittent, Continuous, High-concentration oxygen delivery system) ............... 50
Suctioning (Scheduled, As Needed) ................................................................................................... 52
Tracheostomy Care ............................................................................................................................. 54
Non-invasive Mechanical Ventilation (Bilevel Positive Airway Pressure [BiPAP], Continuous
Positive Airway Pressure [CPAP]) ..................................................................................................... 56
Invasive Mechanical Ventilator .......................................................................................................... 58
IV Medications (Antibiotics, Anticoagulation, Vasoactive Medications, Other) ............................... 60
Transfusions ........................................................................................................................................ 63
Dialysis (Hemodialysis, Peritoneal dialysis) ...................................................................................... 65
IV Access (Peripheral IV, Midline, Central line) ............................................................................... 67
Parenteral/IV Feeding ......................................................................................................................... 69
Feeding Tube ...................................................................................................................................... 71
Mechanically Altered Diet .................................................................................................................. 73
Therapeutic Diet.................................................................................................................................. 76
High-Risk Drug Classes: Use and Indication...................................................................................... 78
Section 4: Medical Conditions and Co-Morbidities................................................................................ 82
Pain Interference ................................................................................................................................. 82
Section 5: Impairments ........................................................................................................................... 85
Hearing and Vision Impairments ........................................................................................................ 85
Hearing................................................................................................................................................ 86
Vision .................................................................................................................................................. 89
Section 6: Proposed New Category: Social Determinants of Health ...................................................... 92
Standardized Data Elements to Assess for Social Determinants of Health ........................................ 92
Race and Ethnicity .............................................................................................................................. 92
ii

Preferred Language and Interpreter Services ...................................................................................... 94
Health Literacy.................................................................................................................................... 95
Transportation ..................................................................................................................................... 96
Social Isolation.................................................................................................................................... 97
APPENDIX A: Transfer of Health Information – Setting-Specific Language ....................................... 99
APPENDIX B: Discharge to Community–PAC IRF QRP Analyses ................................................... 101
APPENDIX C: National Beta Test Supplementary Tables .................................................................. 115

iii

Chapter 1 IMPACT ACT Measures Beginning with the FY 2022
IRF QRP
Section 1. Cross-Setting Measures Development Work: An Introduction
The Improving Medicare Post-Acute Care Transformation Act of 2014 (IMPACT Act), enacted
October 6, 2014, directs the Secretary of Health and Human Services to “specify quality measures on
which Post-Acute Care (PAC) providers are required under the applicable reporting provisions to submit
standardized patient assessment data” in several quality measure domains including incidence of major
falls, skin integrity and changes in skin integrity, medication reconciliation, functional status, transfer of
health information and care preferences when an individual transitions, and resource use and other
measures. The IMPACT Act requires the implementation of quality measures to address these measure
domains in Inpatient Rehabilitation Facilities (IRFs), Skilled Nursing Facilities (SNFs), Long-Term Care
Hospitals (LTCHs), and Home Health Agencies (HHAs).
The IMPACT Act also requires, to the extent possible, the submission of such quality measure
data through the use of a PAC assessment instrument and the modification of such instrument as
necessary to enable such use. This requirement refers to the collection of such data by means of the
Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI) for IRFs, LTCH Continuity
Assessment Record and Evaluation Data Set (LTCH CARE Data Set or LCDS) for LTCHs, the Minimum
Data Set (MDS) 3.0 for SNFs, the and the Outcome and Assessment Information Set (OASIS-D) for
HHAs.
For more information on the statutory history of the IRF, LTCH, or SNF QRP, please refer to the
FY 2015 final rules and for the HHA QRP, please refer to the CY 2016 final rules. More information on
the IMPACT Act is available at https://www.govtrack.us/congress/bills/113/hr4994.
In this document, we present specifications for the standardized patient assessment data elements
(SPADEs) and the following (2) measures proposed for adoption for the IRF QRP through the FY 2020
IRF PPS proposed rule:
The Transfer of Health Information measure concept consists of two companion measures:
1. Transfer of Health Information to the Provider–Post-Acute Care Measure
2. Transfer of Health Information to the Patient–Post-Acute Care Measure
We also provide updated specifications for the previously adopted Discharge to Community
measure.

Section 2. Cross-Setting Proposed Measure: Transfer of Health Information to the
Provider–Post-Acute Care Measure
Measure Description
The proposed measure, the Transfer of Health Information to the Provider, assesses for the timely
transfer of health information, specifically a reconciled medication list. This measure evaluates for the
transfer of information when a patient is transferred or discharged from their current setting to a
subsequent provider. For this proposed measure, the subsequent provider is defined as a short-term
general hospital, a SNF, intermediate care, home under care of an organized home health service
organization or hospice, hospice in an institutional facility, an IRF, an LTCH, a Medicaid nursing facility,
an inpatient psychiatric facility, or a critical access hospital.
This proposed measure, developed under the intent of the IMPACT Act, has been developed
conceptually as a standardized measure for the IRF, LTCH, SNF, and HHA settings. This proposed
measure is calculated by one standardized data element that asks at the time of discharge, did the facility
provide the patient’s current reconciled medication list to the subsequent provider. It also includes one
data element that asks the route of transmission of the reconciled medication list (Appendix A). In order
to track discharge to a subsequent provider, the IRF-PAI will be used to track discharge location status.
Guidance for what is considered a reconciled medication list is discussed in greater detail in the section
below. The measure is conceptualized uniformly across the PAC settings. The measure is calculated using
data from the IRF-PAI for IRF patients, the LCDS for LTCH patients, the MDS 3.0 assessment
instrument for SNF residents, and the OASIS-D for HHA patients. Data are collected and calculated
separately in each of the four settings using standardized data elements. The collection of this measure
and the components tied to the standardized data element used to calculate this measure are described in
Chapter 1, Section 1.
The Reconciled Medication List
This proposed measure evaluates if information was sent to the subsequent provider upon a PAC
discharge. Information, in this instance, is a reconciled medication list. To guide data collection efforts,
CMS outlines a general overview of what could be included in a reconciled medication list, but this is not
exhaustive of all information that could be transferred. We would like to stress that this information is for
guidance purposes only and is not a requirement for the types of information to be included in a PAC
provider’s reconciled medication list in order to meet the Transfer of Health Information to the Provider–
Post-Acute Care measure criteria. While the information for reconciled medication lists is guidance, we
anticipate that the timely transfer of medication information should drive safer care coordination.
For the purpose of providing guidance for this measure, a reconciled medication list is a list of the
current prescribed and over the counter (OTC) medications, nutritional supplements, vitamins, and
homeopathic and herbal products administered by any route to the patient/resident at the time of discharge
or transfer. Medications may also include but are not limited to total parenteral nutrition (TPN) and
oxygen. The current medications should include those that are: 1) active, including those that will be
discontinued after discharge; and 2) those held during the stay and planned to be continued/resumed after
discharge. If deemed relevant to the patient’s/resident’s care by the subsequent provider, medications
discontinued during the stay may be included.
A reconciled medication list often includes important information about: 1) the patient/resident including their name, date of birth, information, active diagnoses, known medication and other allergies,
and known drug sensitivities and reactions; and 2) each medication, including the name, strength, dose,
route of medication administration, frequency or timing, purpose/indication, any special instructions (e.g.,
crush medications), and, for any held medications, the reason for holding the medication and when
medication should resume. This information can improve medication safety. Additional information may
be applicable and important to include in the medication list such as the patient’s/resident’s weight and

date taken, height and date taken, patient’s preferred language, patient’s ability to self-administer
medication, when the last dose of the medication was administered by the discharging/transferring
provider, and when the final dose should be administered (e.g., end of treatment).
Documentation sources for reconciled medication list information include electronic and/or paper
records, including discharge summary records, a Medication Administration Record (MAR), Intravenous
Medication Administration Record (IVAR), home medication list, and physician orders.
The guidance given with respect to what is incorporated in a reconciled medication list is aligned
to the provisions in the proposed Discharge Planning for Hospitals, Critical Access Hospital, and Home
Health Agencies regulation, which outlines discharge planning and the documentation of medications
(please see: https://www.federalregister.gov/documents/2015/11/03/2015-27840/medicare-and-medicaidprograms-revisions-to-requirements-for-discharge-planning-for-hospitals). In addition, this guidance
follows the requirements finalized in the Reform of Requirements for Long-Term Care Facilities (please
see: https://www.federalregister.gov/documents/2016/10/04/2016-23503/medicare-and-medicaidprograms-reform-of-requirements-for-long-term-care-facilities).
Purpose/Rationale for the Quality Measure
In 2013, 22.3 percent of all acute hospital discharges were discharged to PAC settings, including
11 percent who were discharged to home under the care of a home health agency, and 9 percent who were
discharged to SNFs. 1 The proportion of patients being discharged from an acute care hospital to a PAC
setting was greater among beneficiaries enrolled in fee-for-service (FFS) Medicare. Among FFS patients
discharged from an acute hospital, 42 percent went directly to PAC settings. Of that percent, 20 percent
were discharged to a SNF, 18 percent were discharged to a Home Health Agency (HHA), three percent
were discharged to an IRF, and one percent were discharged to an LTCH. 2 Of the Medicare FFS
beneficiaries with an IRF stay in FYs 2016 and 2017, an estimated 10 percent were discharged or
transferred to an acute care hospital, 51 percent were discharged home with home health services, 16
percent were discharged or transferred to a SNF, and 1 percent were discharged or transferred to another
PAC setting (for example, another IRF, a hospice, or an LTCH). 3
The transfer and/or exchange of health information from one provider to another commonly takes
several forms including verbal (e.g. clinician to clinician communication by telephone or in-person),
paper-based (e.g. faxed or printed copies of records) and electronic communication (e.g. via health
information exchange network, using an electronic health/medical record, secure messaging). Health
information, such as medication information, that is incomplete or missing increases the likelihood of a

Tian, W. “An all-payer view of hospital discharge to postacute care,” May 2016. Available at: https://www.hcupus.ahrq.gov/reports/statbriefs/sb205-Hospital-Discharge-Postacute-Care.jsp.
2 Tian, W. “An all-payer view of hospital discharge to postacute care,” May 2016. Available at: https://www.hcupus.ahrq.gov/reports/statbriefs/sb205-Hospital-Discharge-Postacute-Care.jsp.
3 RTI International analysis of Medicare claims data for index stays in IRF 2016/2017. (RTI program reference: MM150).
1

patient/resident safety risk, often life-threatening. 4 5 6 7 8 9 Poor communication and coordination across
health care settings contributes to patient complications, hospital readmissions, ED visits, and medication
errors. 10 11 12 13 14 15 16 17 18 19 Communication has been cited as the third most frequent root cause in
sentinel events, which The Joint Commission defines 20 as a patient safety event that results in death,
permanent harm, or severe temporary harm. Failed or ineffective patient handoffs are estimated to play a
role in 20 percent of serious preventable adverse events. 21 When care transitions are enhanced through
care coordination activities, such as expedited patient information flow, these activities can reduce
duplication of care services and costs of care, resolve conflicting care plans and prevent medical errors. 22

Kwan, J. L., Lo, L., Sampson, M., & Shojania, K. G., “Medication reconciliation during transitions of care as a patient safety
strategy: a systematic review,” Annals of Internal Medicine, 2013, Vol. 158(5), pp. 397-403.
5 Boockvar, K. S., Blum, S., Kugler, A., Livote, E., Mergenhagen, K. A., Nebeker, J. R., & Yeh, J., “Effect of admission
medication reconciliation on adverse drug events from admission medication changes,” Archives of Internal Medicine, 2011,
Vol. 171(9), pp. 860-861.
6 Bell, C. M., Brener, S. S., Gunraj, N., Huo, C., Bierman, A. S., Scales, D. C., & Urbach, D. R., “Association of ICU or hospital
admission with unintentional discontinuation of medications for chronic diseases,” JAMA, 2011, Vol. 306(8), pp. 840-847.
7 Basey, A. J., Krska, J., Kennedy, T. D., & Mackridge, A. J., “Prescribing errors on admission to hospital and their potential
impact: a mixed-methods study,” BMJ Quality & Safety, 2014, Vol. 23(1), pp. 17-25.
8 Desai, R., Williams, C. E., Greene, S. B., Pierson, S., & Hansen, R. A., “Medication errors during patient transitions into
nursing homes: characteristics and association with patient harm,” The American Journal of Geriatric Pharmacotherapy,
2011, Vol. 9(6), pp. 413-422.
9 Boling, P. A., “Care transitions and home health care,” Clinical Geriatric Medicine, 2009, Vol.25(1), pp. 135-48.
10 Barnsteiner, J. H., “Medication Reconciliation: Transfer of medication information across settings—keeping it free from
error,” The American Journal of Nursing, 2005, Vol. 105(3), pp. 31-36.
11 Arbaje, A. I., Kansagara, D. L., Salanitro, A. H., Englander, H. L., Kripalani, S., Jencks, S. F., & Lindquist, L. A., “Regardless
of age: incorporating principles from geriatric medicine to improve care transitions for patients with complex needs,” Journal
of General Internal Medicine, 2014, Vol. 29(6), pp. 932-939.
12 Jencks, S. F., Williams, M. V., & Coleman, E. A., “Rehospitalizations among patients in the Medicare fee-for-service
program,” New England Journal of Medicine, 2009, Vol. 360(14), pp. 1418-1428.
13 Institute of Medicine. “Preventing medication errors: quality chasm series,” Washington, DC: The National Academies Press
2007. Available at: https://www.nap.edu/read/11623/chapter/1
14 Kitson, N. A., Price, M., Lau, F. Y., & Showler, G., “Developing a medication communication framework across continuums
of care using the Circle of Care Modeling approach,” BMC Health Services Research, 2013, Vol. 13(1), pp. 1-10.
15 Mor, V., Intrator, O., Feng, Z., & Grabowski, D. C., “The revolving door of rehospitalization from skilled nursing facilities,”
Health Affairs, 2010, Vol. 29(1), pp. 57-64.
16 Institute of Medicine. “Preventing medication errors: quality chasm series,” Washington, DC: The National Academies Press
2007. Available at: https://www.nap.edu/read/11623/chapter/1
17 Kitson, N. A., Price, M., Lau, F. Y., & Showler, G., “Developing a medication communication framework across continuums
of care using the Circle of Care Modeling approach,” BMC Health Services Research, 2013, Vol. 13(1), pp. 1-10.
18 Forster, A. J., Murff, H. J., Peterson, J. F., Gandhi, T. K., & Bates, D. W., “The incidence and severity of adverse events
affecting patients after discharge from the hospital.” Annals of Internal Medicine, 2003,138(3), pp. 161-167.
19 King, B. J., Gilmore‐Bykovskyi, A. L., Roiland, R. A., Polnaszek, B. E., Bowers, B. J., & Kind, A. J. “The
consequences of
poor communication during transitions from hospital to skilled nursing facility: a qualitative study,” Journal of the American
Geriatrics Society, 2013, Vol. 61(7), 1095-1102.
20 The Joint Commission, “Sentinel Event Policy” available at
https://www.jointcommission.org/sentinel_event_policy_and_procedures/
21 The Joint Commission. “Sentinel Event Data Root Causes by Event Type 2004 –2015.” 2016. Available at:
https://www.jointcommission.org/assets/1/23/jconline_Mar_2_2016.pdf.
22 Mor, V., Intrator, O., Feng, Z., & Grabowski, D. C., “The revolving door of rehospitalization from skilled nursing facilities,”
Health Affairs, 2010, Vol. 29(1), pp. 57-64.
4

23 24 25 26

The rising incidence of preventable adverse events, complications and hospital readmissions
have drawn national attention to the importance of the timely transfer of health information and care
preferences at transitions. However, there is limited information about the route or mode (for example,
paper-based, verbal, and electronic) of transmission used by PAC providers to transfer health information.
PAC provider health information exchange supports the goals of: high quality, personalized, and efficient
healthcare; care coordination and person-centered care; and supports real-time, data driven, clinical
decision making.
PAC patients often have complicated medication regimens and require efficient and effective
communication and coordination of care between settings, including transfer of detailed medication
information. 27 28 29 Individuals in PAC settings may be vulnerable to adverse health outcomes due to
insufficient medication information on the part of their health care providers, and their higher likelihood
for multiple comorbid chronic conditions, polypharmacy, and complicated transitions between care
settings. 30 31 Preventable adverse drug events (ADEs) occur after hospital discharge in a variety of settings
including PAC. 32
Patients in PAC settings are often taking multiple medications. Consequently, PAC providers
regularly are in the position of starting complex new medication regimens with little knowledge of the
patient or their medication history upon admission. Furthermore, inter-facility communication barriers
delay resolving medication discrepancies during transitions of care. 33 The transfer of a medication list
between providers is necessary for medication reconciliation interventions, which have been shown to be

Institute of Medicine, “Preventing medication errors: quality chasm series,” Washington, DC: The National Academies Press,
2007. Available at: https://www.nap.edu/read/11623/chapter/1.
24 Starmer, A. J., Sectish, T. C., Simon, D. W., Keohane, C., McSweeney, M. E., Chung, E. Y., Yoon, C.S., Lipsitz, S.R.,
Wassner, A.J., Harper, M. B., & Landrigan, C. P., “Rates of medical errors and preventable adverse events among
hospitalized children following implementation of a resident handoff bundle,” JAMA, 2013, Vol. 310(21), pp. 2262-2270.
25 Pronovost, P., M. M. E. Johns, S. Palmer, R. C. Bono, D. B. Fridsma, A. Gettinger, J.Goldman, W. Johnson, M. Karney, C.
Samitt, R. D. Sriram, A. Zenooz, and Y. C. Wang, Editors. Procuring Interoperability: Achieving High-Quality, Connected,
and Person-Centered Care. Washington, DC, 2018. National Academy of Medicine. Available at: https://nam.edu/wpcontent/uploads/2018/10/Procuring-Interoperability_web.pdf.
26
Balaban, R.B., Weissman, J.S., Samuel, P.A., & Woolhandler, S., “Redefining and redesigning hospital discharge to enhance
patient care: a randomized controlled study,” J Gen Intern Med, 2008, Vol. 23(8), pp. 1228-33.
27 Starmer A. J., Spector N. D., Srivastava R., West, D. C., Rosenbluth, G., Allen, A. D., Noble, E. L., … & Landrigen, C. P.,
“Changes in medical errors after implementation of a handoff program,” N Engl J Med, 2014, Vol. 37(1), pp. 1803-1812.
28 Kruse, C.S. Marquez, G., Nelson, D., & Polomares, O., “The use of health information exchange to augment patient handoff in
long-term care: a systematic review,” Applied Clinical Informatics, 2018, Vol. 9(4), pp. 752-771
29 Brody, A. A., Gibson, B., Tresner-Kirsch, D., Kramer, H., Thraen, I., Coarr, M. E., & Rupper, R., “High prevalence of
medication discrepancies between home health referrals and Centers for Medicare and Medicaid Services home health
certification and plan of care and their potential to affect safety of vulnerable elderly adults,” Journal of the American
Geriatrics Society, 2016, Vol. 64(11), pp. e166-e170.
30 Chhabra, P. T., Rattinger, G. B., Dutcher, S. K., Hare, M. E., Parsons, K., L., & Zuckerman, I. H., “Medication reconciliation
during the transition to and from long-term care settings: a systematic review,” Res Social Adm Pharm, 2012, Vol. 8(1), pp.
60-75.
31 Levinson, D. R., & General, I., “Adverse events in skilled nursing facilities: national incidence among Medicare beneficiaries.”
Washington, DC: U.S. Department of Health and Human Services, Office of the Inspector General, February 2014. Available
at: https://oig.hhs.gov/oei/reports/oei-06-11-00370.pdf.
32 Battles J., Azam I., Grady M., & Reback K., “Advances in patient safety and medical liability,” AHRQ Publication No. 170017-EF. Rockville, MD: Agency for Healthcare Research and Quality, August 2017. Available at:
https://www.ahrq.gov/sites/default/files/publications/files/advances-complete_3.pdf.
33
Patterson M., Foust J. B., Bollinger, S., Coleman, C., Nguyen, D., “Inter-facility communication barriers delay resolving
medication discrepancies during transitions of care,” Research in Social & Administrative Pharmacy (2018), doi:
10.1016/j.sapharm.2018.05.124.
23

a cost-effective way to avoid ADEs by reducing errors, 34 35 36 especially when medications are reviewed
by a pharmacist and when it is done in conjunction with the use of electronic medical records. 37
Denominator
IRF.

The target population is all patients discharged or transferred to a subsequent provider from an
IRF Denominator

The denominator is the total number of IRF Medicare Part A and Medicare Advantage (Part C)
patient stays ending in discharge to a short-term general hospital, a SNF, intermediate care, home under
care of an organized home health service organization or hospice, hospice in an institutional facility, a
swing bed, another IRF, an LTCH, a Medicaid nursing facility, an inpatient psychiatric facility, or a
critical access hospital. Discharge to one of these providers is determined based on response to the
discharge location item, 44D, of the IRF-PAI assessment, shown below:

44D. Patient’s discharge destination/living setting, using codes below: _________
(answer only if 44C = 1; if 44C = 0, skip to item 46)
(01. Home (e.g. private home/apt., board/care, assisted living, group home, transitional
living, other residential care arrangements); 02. Short-term General Hospital; 03. Skilled
Nursing Facility (SNF); 04: Intermediate care; 06. Home under care of organized home
health service organization; 50. Hospice (home); 51. Hospice (medical facility); 61. Swing
Bed; 62. Another Inpatient Rehabilitation Facility; 63. Long-Term Care Hospital (LTCH);
64. Medicaid Nursing Facility; 65. Inpatient Psychiatric Facility; 66. Critical Access
Hospital (CAH); 99. Not Listed
Numerator
IRF Numerator: The numerator is the number of stays for which the IRF-PAI indicated that the
following is true:
At the time of discharge, the facility provided a current reconciled medication list to the
subsequent provider (A2121= [1]).
Measure Time Window
The measure will be calculated quarterly. All IRF stays during the quarter will be included in the
denominator and are eligible for inclusion in the numerator. For patients with multiple stays during the
quarter, each stay is eligible for inclusion in the measure.

Boockvar, K. S., Blum, S., Kugler, A., Livote, E., Mergenhagen, K. A., Nebeker, J. R., & Yeh, J., “Effect of admission
medication reconciliation on adverse drug events from admission medication changes,” Archives of Internal Medicine, 2011,
Vol. 171(9), pp. 860-861.
35
Kwan, J. L., Lo, L., Sampson, M., & Shojania, K. G., “Medication reconciliation during transitions of care as a patient safety
strategy: a systematic review,” Annals of Internal Medicine, 2013, Vol. 158(5), pp. 397-403.
36
Chhabra, P. T., Rattinger, G. B., Dutcher, S. K., Hare, M. E., Parsons, K., L., & Zuckerman, I. H., “Medication reconciliation
during the transition to and from long-term care settings: a systematic review,” Res Social Adm Pharm, 2012, Vol. 8(1), pp.
60-75.
37
Agrawal A, Wu WY. “Reducing medication errors and improving systems reliability using an electronic medication
reconciliation system,” The Joint Commission Journal on Quality and Patient Safety, 2009, Vol. 35(2), pp. 106-114.

Items Included in the Quality Measure
One data element will be included to calculate the measure. One data element will be collected to
inform the internally consistency logic of the proposed measure.
Provision of Current Reconciled Medication List to Subsequent Provider at Discharge
A2121. Provision of Current Reconciled Medication List to Subsequent Provider at Discharge
At the time of discharge to another provider, did your facility provide the patient’s current reconciled
medication list to the subsequent provider?
Enter Code

0. No – Current reconciled medication list not provided to the subsequent provider
1. Yes – Current reconciled medication list provided to the subsequent provider
Route of Current Medication List Transmission

A2123. Route of Current Reconciled Medication List Transmission
Indicate the route(s) of transmission of the current reconciled medication list to the subsequent
provider and/or patient/family/caregiver.
1.
2.
Route of Transmission
To subsequent
To patient/family/
provider
caregiver
↓

Check all that apply ↓

A. Electronic Health Record
B. Health Information Exchange
Organization
C. Verbal (e.g., in-person, telephone, video
conferencing)
D. Paper-based (e.g., fax, copies, printouts)



E. Other Methods (e.g., texting, email, CDs)



A2121 = [1]
Step 3: Calculate the facility observed score
Divide the facility’s numerator count by its denominator count to obtain the observed score; in
other words, divide the results of Step 2 by the results of Step 1. Multiply by 100.
Quality Measure Coding Steps
The following steps are used to code the measure:
1. At discharge, code for the patient’s discharge location.
Identify discharge location with item 44D.
2. At discharge, code for if the facility provided the reconciled medication list to the
subsequent provider.
A valid response for item 44D would skip the coder into item A2121.
3. At discharge, code for the route of transmission.
A valid response for item A2121 [A2121 = 1] would skip the coder into item A2123. This
item is used for measure consistency logic.

Section 3. Cross-Setting Proposed Measure: Transfer of Health Information to the Patient–
Post-Acute Care Measure
Measure Description
This proposed measure assesses for and reports on the timely transfer of health information, i.e., a
current reconciled medication list, to the patient/resident when discharged from their current setting of
post-acute care to a private home/apartment, board and care home, assisted living, group home,
transitional living, or home under the care of an organized home health service organization or hospice.
This proposed measure, developed under the intent of the IMPACT Act, has been developed for
the IRF, LTCH, SNF, and HHA settings. This proposed measure is calculated by one standard data
element that asks at the time of discharge did the facility provide the patient’s/resident’s current
reconciled medication list to the patient, family, and/or caregiver. It also includes one data element that
asks the route of transmission of the reconciled medication list (Appendix A). In order to track discharge
to home, the IRF-PAI will be used to track discharge location status. The measure is conceptualized
uniformly across the PAC settings. The measure is calculated using data from the IRF-PAI for IRF
patients, the LCDS for LTCH patients, the MDS 3.0 assessment instrument for SNF residents, and the
OASIS-D for HHA patients. Data are collected and calculated separately in each of the four settings using
standardized data elements. The collection of this measure and the components tied to the standardized
data element used to calculate this measure are in Appendix A.
The Reconciled Medication List
The guidance related to a reconciled medication list for purposes of this measure, and the
information which may be included for the patient/resident and for each of the medications is provided in
Chapter 1, Section 2. It is recommended that a reconciled medication list that is transferred to the patient,
family, or caregiver use consumer-friendly terminology and plain language to ensure that the information
provided to patients and caregivers is clear and understandable, promoting transparent access to medical
record information. 38
Purpose/Rationale for the Quality Measure
In 2013, 22.3 percent of all acute hospital discharges were discharged to PAC settings, including
11 percent who were discharged to home under the care of a home health agency. 39 Of the Medicare FFS
beneficiaries with an IRF stay in fiscal years 2016 and 2017, an estimated 51 percent were discharged
home with home health services, 21 percent were discharged home with self-care, and .5 percent were
discharged with home hospice services 40
The communication of health information, such as a reconciled medication list, is critical to
ensuring safe and effective patient transitions from health care settings to home and/or other community
settings. Incomplete or missing health information, such as medication information, increases the

38
39

For an examples of plain language resources for healthcare information see: https://www.plainlanguage.gov/resources/contenttypes/healthcare/

Tian, W. “An all-payer view of hospital discharge to postacute care,” May 2016. Available at: https://www.hcupus.ahrq.gov/reports/statbriefs/sb205-Hospital-Discharge-Postacute-Care.jsp.

40 RTI International analysis of Medicare claims data for index stays in IRF 2016/2017. (RTI program reference: MM150).

likelihood of a patient safety risk, often life-threatening. 41 42 43 44 45 Individuals who use PAC settings are
particularly vulnerable to adverse health outcomes due to their higher likelihood of multiple comorbid
chronic conditions, polypharmacy, and complicated transitions between care settings. 46 47 Upon discharge
to home, individuals in PAC settings may be faced with numerous medication changes, new medication
regimes, and follow-up details. 48 49 50 The efficient and effective communication and coordination of
medication information may be critical to prevent potentially deadly adverse effects. When care
coordination activities enhance care transitions, these activities can reduce duplication of care services
and costs of care, resolve conflicting care plans, and prevent medical errors. 51 52
The transfer of a patient’s medication information to the patient, family, or caregiver is common
practice and supported by discharge planning requirements for participation in Medicare and Medicaid
programs. 53 54 However, there is limited information about the route or mode (for example, paper-based,
41

Kwan, J. L., Lo, L., Sampson, M., & Shojania, K. G., “Medication reconciliation during transitions of care as a patient safety
strategy: a systematic review,” Annals of Internal Medicine, 2013, Vol. 158(5), pp. 397-403.
42
Boockvar, K. S., Blum, S., Kugler, A., Livote, E., Mergenhagen, K. A., Nebeker, J. R., & Yeh, J., “Effect of admission
medication reconciliation on adverse drug events from admission medication changes,” Archives of Internal Medicine, 2011,
Vol. 171(9), pp. 860-861.
43
Bell, C. M., Brener, S. S., Gunraj, N., Huo, C., Bierman, A. S., Scales, D. C., & Urbach, D. R., “Association of ICU or
hospital admission with unintentional discontinuation of medications for chronic diseases,” JAMA, 2011, Vol. 306(8), pp.
840-847.
44
Basey, A. J., Krska, J., Kennedy, T. D., & Mackridge, A. J., “Prescribing errors on admission to hospital and their potential
impact: a mixed-methods study,” BMJ Quality & Safety, 2014, Vol. 23(1), pp. 17-25.
45
Desai, R., Williams, C. E., Greene, S. B., Pierson, S., & Hansen, R. A., “Medication errors during patient transitions into
nursing homes: characteristics and association with patient harm,” The American Journal of Geriatric Pharmacotherapy,
2011, Vol. 9(6), pp. 413-422.
46
Brody, A. A., Gibson, B., Tresner-Kirsch, D., Kramer, H., Thraen, I., Coarr, M. E., & Rupper, R. “High prevalence of
medication discrepancies between home health referrals and Centers for Medicare and Medicaid Services home health
certification and plan of care and their potential to affect safety of vulnerable elderly adults,” Journal of the American
Geriatrics Society, 2016, Vol. 64(11), pp. e166-e170.
47
Chhabra, P. T., Rattinger, G. B., Dutcher, S. K., Hare, M. E., Parsons, K., L., & Zuckerman, I. H., “Medication reconciliation
during the transition to and from long-term care settings: a systematic review,” Res Social Adm Pharm, 2012, Vol. 8(1), pp.
60-75.
48
Brody, A. A., Gibson, B., Tresner-Kirsch, D., Kramer, H., Thraen, I., Coarr, M. E., & Rupper, R. “High prevalence of
medication discrepancies between home health referrals and Centers for Medicare and Medicaid Services home health
certification and plan of care and their potential to affect safety of vulnerable elderly adults,” Journal of the American
Geriatrics Society, 2016, Vol. 64(11), pp. e166-e170.
49
Bell, C. M., Brener, S. S., Gunraj, N., Huo, C., Bierman, A. S., Scales, D. C., & Urbach, D. R., “Association of ICU or
hospital admission with unintentional discontinuation of medications for chronic diseases,” JAMA, 2011, Vol. 306(8), pp.
840-847.
50
Sheehan, O. C., Kharrazi, H., Carl, K. J., Leff, B., Wolff, J. L., Roth, D. L., Gabbard, J., & Boyd, C. M., “Helping older adults
improve their medication experience (HOME) by addressing medication regimen complexity in home healthcare,” Home
Healthcare Now. 2018, Vol. 36(1) pp. 10-19.
51 Mor, V., Intrator, O., Feng, Z., & Grabowski, D. C., “The revolving door of rehospitalization from skilled nursing facilities,”
Health Affairs, 2010, Vol. 29(1), pp. 57-64.
52 Starmer, A. J., Sectish, T. C., Simon, D. W., Keohane, C., McSweeney, M. E., Chung, E. Y., Yoon, C.S., Lipsitz, S.R.,
Wassner, A.J., Harper, M. B., & Landrigan, C. P., “Rates of medical errors and preventable adverse events among
hospitalized children following implementation of a resident handoff bundle,” JAMA, 2013, Vol. 310(21), pp. 2262-2270.
53 CMS, “Revision to state operations manual (SOM), Hospital Appendix A - Interpretive Guidelines for 42 CFR 482.43,
Discharge Planning” May 17, 2013. Available at: https://www.cms.gov/Medicare/Provider-Enrollment-andCertification/SurveyCertificationGenInfo/Downloads/Survey-and-Cert-Letter-13-32.pdf.
54 The State Operations Manual Guidance to Surveyors for Long Term Care Facilities (Guidance §483.21(c)(1) Rev. 11-22-17)
for discharge planning. CMS, “Revision to state operations manual (SOM), Hospital Appendix A - Interpretive Guidelines for
42 CFR 482.43, Discharge Planning” May 17, 2013. Available at: https://www.cms.gov/Medicare/Provider-Enrollment-andCertification/SurveyCertificationGenInfo/Downloads/Survey-and-Cert-Letter-13-32.pdf

verbal, and electronic) of transmission used by PAC providers to transfer health information. PAC
provider health information exchange with patients, families and caregivers supports the goals of: high
quality, personalized, and efficient healthcare; care coordination and person-centered care; and supports
real-time, data driven, clinical decision making.
Most PAC electronic health record (EHR) systems generate a discharge medication list. Further,
interventions to promote patient participation in medication management have been shown to be
acceptable and potentially useful for improving patient outcomes and reducing costs 55 56 and provision of
a reconciled medication list to patients/residents and their caregivers can improve transitional care. 57
Some clinical practice guidelines state the importance of medication safety and communicating
accurate medication information to the patient. For example, The Joint Commission’s National Patient
Safety Goals #4 and #5 for Home Care Accreditation (NPSG.03.06.01) include the following elements of
performance with respect to medication information that is important to provide to the patient or family
when discharging or transferring. 58
4. Provide the patient (or family as needed) with written information on the medications the patient
should be taking when he or she leaves the organization’s care (for example, name, dose, route,
frequency, purpose).
5. Explain the importance of managing medication information to the patient.
The AHRQ Project Re-Engineered Discharge (RED) Toolkit includes a number of medicationrelated strategies (e.g., active medication reconciliation, medication teaching for patients and caregivers,
development of medication list for patients and their health care providers). 59
Denominator
IRF Denominator
The denominator for this measure is the total number of IRF Medicare Part A and Medicare
Advantage (Part C) patient stays ending in discharge to a private home/ apartment (apt.), board/care,
assisted living, group home, transitional living or home under care of organized home health service
organization or hospice. Discharge to one of these locations is determined based on response to the
discharge location item, 44D, of the IRF-PAI assessment, shown below:

44D. Patient’s discharge destination/living setting, using codes below: ________
(answer only if 44C = 1; if 44C = 0, skip to item 46)
(01. Home (e.g. private home/apt., board/care, assisted living, group home, transitional
living, other residential care arrangements); 02. Short-term General Hospital; 03. Skilled
Nursing Facility (SNF); 04: Intermediate care; 06. Home under care of organized home
health service organization; 50. Hospice (home); 51. Hospice (medical facility); 61. Swing
Bed; 62. Another Inpatient Rehabilitation Facility; 63. Long-Term Care Hospital (LTCH);

Greene, J., & Hibbard, J. H. “Why does patient activation matter? An examination of the relationships between patient
activation and health-related outcomes,” Journal of General Internal Medicine, 2012, Vol. 27(5), pp. 520-526.
56 Phatak A, Prusi R, Ward, B., Hansen, L. O., Williams, M. V., Vetter, E., Chapman, N., Postelnick, M., “Impact of pharmacist
involvement in the transitional care of high-risk patients through medication reconciliation, medication education, and
postdischarge call-backs (IPITCH Study),” J Hosp Med., 2016, Vol. 11(1), pp. 39-44.
57 Toles, M., Colon-Emeric, C., Naylor, M. D., Asafu-Adjei, J., Hanson, L. C., “Connect-home: transitional care of skilled
nursing facility patients and their caregivers,” Am Geriatr Soc., 2017, Vol. 65(10), pp. 2322–2328.
58 The Joint Commission. National Patient Safety Goals, Effective January 2018, Home Care Accreditation Program. Available
at: https://www.jointcommission.org/assets/1/6/NPSG_Chapter_OME_Jan2018.pdf
59
Re-Engineered Discharge (RED) Toolkit. Agency for Healthcare Research and Quality, Rockville, MD. Available at:
http://www.ahrq.gov/professionals/systems/hospital/red/toolkit/index.html, Last accessed November, 28, 2018.
55

64. Medicaid Nursing Facility; 65. Inpatient Psychiatric Facility; 66. Critical Access
Hospital (CAH); 99. Not Listed
Numerator
IRF Numerator: The numerator is the number of stays for which the IRF-PAI indicated that the
following is true:
At the time of discharge, the facility provided a current reconciled medication list to the patient,
family and/or caregiver (A2122= [1]).
Measure Time Window
The measure will be calculated quarterly. All IRF stays during the quarter will be included in the
denominator and are eligible for inclusion in the numerator. For patients with multiple stay during the
quarter, each stay is eligible for inclusion in the measure.
Items Included in the Quality Measure
One data element will be included to calculate the measure. One data element will be collected to
inform the internally consistency logic of the proposed measure.
Provision of Current Reconciled Medication List to Patient at Discharge
A2122. Provision of Current Reconciled Medication List to Patient at Discharge
At the time of discharge, did your facility provide the patient’s current reconciled medication list to
the patient, family and/or caregiver?
Enter Code
0. No – Current reconciled medication list not provided to the patient, family and/or
caregiver
1. Yes – Current reconciled medication list provided to the patient, family and/or
caregiver

Route of Current Medication List Transmission
A2123. Route of Current Reconciled Medication List Transmission
Indicate the route(s) of transmission of the current reconciled medication list to the subsequent
provider and/or patient/family/caregiver.
1.
2.
Route of Transmission
To subsequent
To patient/family/
provider
caregiver
↓

Check all that apply ↓

A. Electronic Health Record
B. Health Information Exchange
Organization
C. Verbal (e.g., in-person, telephone, video
conferencing)
D. Paper-based (e.g., fax, copies, printouts)



E. Other Methods (e.g., texting, email, CDs)



Risk Adjustment
This measure is not risk-adjusted or stratified.
Quality Measure Calculation Steps
The following steps are used to calculate the measure:
1. Calculate the facility observed score (steps 1 through 3).
Step 1. Calculate the denominator count
Calculate the total number of patient stays with discharge to home based on the discharge
location item in Section A.
Step 2. Calculate the numerator count
Calculate the total number of stays where a reconciled medication list was transferred:
A2122 = [1]
Step 3: Calculate the facility observed score
Divide the facility’s numerator count by its denominator count to obtain the observed score;
in other words, divide the results of Step 2 by the results of Step 1. Multiply by 100.
Quality Measure Coding Steps
The following steps are used to code the measure:
1. At discharge, code for the patient’s discharge location.
Identify discharge location with item 44D.
2. At discharge, code for if the facility provided the reconciled medication list to the
patient, family and/or caregiver.
13

A valid response for item 44D would skip the coder into item A2122.
3. At discharge, code for the route of transmission.
A valid response for item A2122 [A2122 = 1] would skip the coder into item A2123. This
item is used for measure consistency logic.

Section 4. Update to the Discharge to Community–Post Acute Care (PAC) Inpatient
Rehabilitation Facility (IRF) Quality Reporting Program (QRP) Measure
Measure Update
The Discharge to Community–Post Acute Care (PAC) Inpatient Rehabilitation Facility (IRF)
Quality Reporting Program (QRP) measure was adopted for the IRF QRP in the FY 2017 IRF Prospective
Payment System (PPS) final rule (81 FR 52095 through 52103) to meet the requirement of the IMPACT
Act. Measure specifications were first published in July 2016. 60 These draft specifications include a new
proposed measure exclusion for baseline NF residents; there are no other changes to measure
specifications.
Measure Description
This measure assesses successful discharge to the community from a PAC setting, with successful
discharge to the community including no unplanned rehospitalizations and no death in the 31 days
following discharge. Specifically, this measure reports an IRF’s risk-standardized rate of Medicare feefor-service (FFS) patients who are discharged to the community following an IRF stay, and do not have
an unplanned readmission to an acute care hospital or LTCH in the 31 days following discharge to
community, and who remain alive during the 31 days following discharge to community. Community, for
this measure, is defined as home/self-care, with or without home health services, based on Patient
Discharge Status Codes 01, 06, 81, and 86 on the Medicare FFS claim. 61,62,63
We adopted four discharge to community measures for IRF, LTCH, SNF, and HH settings,
respectively. These measures are conceptualized uniformly across the PAC settings in terms of the
definition of the discharge to community outcome, the approach to risk adjustment, and the measure
calculation, with some differences where needed due to setting-specific considerations. It is important to
note that each measure is specific to the particular PAC setting (i.e., IRF, LTCH, SNF, or HH); we do not
pool PAC patients/residents across settings in the measure development and calculation.
Purpose/Rationale for the Measure
Discharge to a community setting is an important health care outcome for many patients/residents
for whom the overall goals of PAC include optimizing functional improvement, returning to a previous
level of independence, and avoiding institutionalization. Returning to the community is also an important
outcome for many patients/residents who are not expected to make functional improvement during their
PAC stay, and for patients/residents who may be expected to decline functionally due to their medical
condition. By assessing whether patients remain alive in the community without acute complications for
31 days following discharge, the DTC–PAC IRF QRP measure is a meaningful, patient- and familycentered measure of successful community discharge.

60 The original measure specifications are available at: https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/IRF-Quality-Reporting/Downloads/Measure-Specifications-for-FY17-IRF-QRP-FinalRule.pdf.
61 National Uniform Billing Committee Official UB-04 Data Specifications Manual 2018, Version 12, July 2017, Copyright
2017, American Hospital Association.

62 Patient discharge status codes 81 and 86 are intended for use on acute care claims only. However, since these codes have

sometimes been reported on PAC claims, we include them in our definition of community to credit the PAC provider for
discharging the patient to a community setting.

63 This definition is not intended to suggest that group homes, foster care, or other residential care settings included in the

definition of “community” for the purpose of this measure are the most integrated setting for any particular individual or
group of individuals under the Americans with Disabilities Act (ADA) and Section 504.

In addition to being an important outcome from a patient/resident and family perspective,
patients/residents discharged to community settings, on average, incur lower costs over the recovery
episode, compared with those discharged to institutional settings. 64,65 Given the high costs of care in
institutional settings, encouraging PACs to prepare patients for discharge to community, when clinically
appropriate, may have cost-saving implications for the Medicare program. 66 Also, providers have found
that successful discharge to community was a major driver of their ability to achieve savings, where
capitated payments for post-acute care were in place. 67 For patients/residents who require long-term care
due to persistent disability, discharge to community could result in lower long-term care costs for
Medicaid and for patients’/residents’ out-of-pocket expenditures. 68
Analyses conducted by the Medicare Payment Advisory Commission (MedPAC) using 2013
PAC data demonstrates the substantially higher costs of institutional PAC stays compared with HH
stays. 69 Average costs of HH stays ranged from $1,790 to $2,699 depending on the position of the HH
stay in a sequence of PAC care. Average costs of institutional PAC stays (including IRF, LTCH, or SNF
stays) ranged from $13,948 to $17,506, depending on the position of the institutional PAC stay in a
sequence of PAC care. 70
Analyses conducted for the Assistant Secretary for Planning and Evaluation (ASPE) on PAC
episodes, using a 5 percent sample of 2006 Medicare claims, revealed that relatively high average,
unadjusted Medicare payments are associated with discharge to institutional settings from IRFs, SNFs,
LTCHs or HHAs, as compared with payments associated with discharge to community settings. 71
Average, unadjusted Medicare payments associated with discharge to community settings ranged from $0
to $4,017 for IRF discharges, $0 to $3,544 for SNF discharges, $0 to $4,706 for LTCH discharges, and $0
to $992 for HHA discharges. In contrast, payments associated with discharge to non-community settings
were considerably higher, ranging from $11,847 to $25,364 for IRF discharges, $9,305 to $29,118 for
SNF discharges, $12,465 to $18,205 for LTCH discharges, and $7,981 to $35,192 for HHA discharges. 72
These expenditure estimates only include Medicare expenditures related to the immediate discharge
destination following SNF, LTCH, IRF or HH care, and not expenditures related to any subsequent
discharge destinations.
Measuring and comparing facility-level discharge to community rates is expected to help
differentiate among facilities with varying performance in this important domain, and to help avoid
disparities in care across patient/resident groups. Variation in discharge to community rates has been
reported within and across post-acute settings; across a variety of facility-level characteristics, such as
geographic location (for example, regional location, urban or rural location), ownership (for example, for64 Dobrez D, Heinemann AW, Deutsch A, Manheim L, Mallinson T. Impact of Medicare's prospective payment system for
65
66
67
68
69

inpatient rehabilitation facilities on stroke patient outcomes. American journal of physical medicine & rehabilitation /
Association of Academic Physiatrists. 2010;89(3):198-204.
Gage B, Morley M, Spain P, Ingber M. Examining Post Acute Care Relationships in an Integrated Hospital System. Final
Report. RTI International; 2009.
Ibid.
Doran JP, Zabinski SJ. Bundled payment initiatives for Medicare and non-Medicare total joint arthroplasty patients at a
community hospital: bundles in the real world. The Journal of Arthroplasty. 2015;30(3):353-355.
Newcomer RJ, Ko M, Kang T, Harrington C, Hulett D, Bindman AB. Health Care Expenditures After Initiating Long-term
Services and Supports in the Community Versus in a Nursing Facility. Medical Care. 2016 (54)3, 221-228.

Medicare Payment Advisory Commission. Paying for sequential stays in a unified prospective payment system
for post-acute care. In: Report to the Congress: Medicare and the Health Care Delivery System. June 2018.
Available at: http://www.medpac.gov/docs/default-source/reports/jun18_ch4_medpacreport_sec.pdf?sfvrsn=0.
70 Ibid.
71 Gage B, Morley M, Spain P, Ingber M. Examining Post Acute Care Relationships in an Integrated Hospital System. Final

Report. RTI International; 2009.

72 Ibid.

profit or nonprofit), and freestanding or hospital-based units; and across patient-level characteristics, such
as race and gender. 73,74,75,76,77,78 Discharge to community rates in the IRF setting have been reported to
range from about 60 to 80 percent. 79,80,81,82,83,84 Longer-term studies show that rates of discharge to
community from IRFs have decreased over time as IRF length of stay has decreased. 85,86 In the IRF
Medicare FFS population, using calendar year 2015-2016 national unadjusted data, we found that
approximately 64 percent of patients were discharged to the community; facility-level observed
discharges to community ranged from approximately 15 percent to 100 percent, with an interquartile
range of 9.3 percentage points. Greater variation in discharge to community rates is seen in the SNF
setting, with rates ranging from 31 to 65 percent. 87,88,89,90 A multi-center study of 23 LTCHs demonstrated
that 28.8 percent of 1,061 patients who were ventilator-dependent on admission were discharged to
73 Reistetter TA, Karmarkar AM, Graham JE, et al. Regional variation in stroke rehabilitation outcomes. Archives of physical

medicine and rehabilitation. 2014;95(1):29-38.

74 El-Solh AA, Saltzman SK, Ramadan FH, Naughton BJ. Validity of an artificial neural network in predicting discharge
75
76
77
78

80
81

82
83

88
89
90

destination from a postacute geriatric rehabilitation unit. Archives of physical medicine and rehabilitation. 2000;81(10):13881393.
Medicare Payment Advisory Commission. March 2018 Report to the Congress: Medicare Payment Policy. 2018.
Bhandari VK, Kushel M, Price L, Schillinger D. Racial disparities in outcomes of inpatient stroke rehabilitation. Archives of
physical medicine and rehabilitation. 2005;86(11):2081-2086.
Chang PF, Ostir GV, Kuo YF, Granger CV, Ottenbacher KJ. Ethnic differences in discharge destination among older patients
with traumatic brain injury. Archives of physical medicine and rehabilitation. 2008;89(2):231-236.
Berges IM, Kuo YF, Ostir GV, Granger CV, Graham JE, Ottenbacher KJ. Gender and ethnic differences in rehabilitation
outcomes after hip-replacement surgery. American journal of physical medicine & rehabilitation / Association of academic
physiatrists. 2008;87(7):567-572.
Galloway RV, Granger CV, Karmarkar AM, et al. The Uniform Data System for Medical Rehabilitation: report of patients
with debility discharged from inpatient rehabilitation programs in 2000-2010. American journal of physical medicine &
rehabilitation / Association of academic physiatrists. 2013;92(1):14-27.
Morley MA, Coots LA, Forgues AL, Gage BJ. Inpatient rehabilitation utilization for Medicare beneficiaries with multiple
sclerosis. Archives of physical medicine and rehabilitation. 2012;93(8):1377-1383.
Reistetter TA, Graham JE, Deutsch A, Granger CV, Markello S, Ottenbacher KJ. Utility of functional status for classifying
community versus institutional discharges after inpatient rehabilitation for stroke. Archives of physical medicine and
rehabilitation. 2010;91(3):345-350.
Gagnon D, Nadeau S, Tam V. Clinical and administrative outcomes during publicly-funded inpatient stroke rehabilitation
based on a case-mix group classification model. Journal of rehabilitation medicine. 2005;37(1):45-52.
DaVanzo J, El-Gamil A, Li J, Shimer M, Manolov N, Dobson A. Assessment of patient outcomes of rehabilitative care
provided in inpatient rehabilitation facilities (IRFs) and after discharge. Vienna, VA: Dobson DaVanzo & Associates, LLC;
2014.
Kushner DS, Peters KM, Johnson-Greene D. Evaluating Siebens Domain Management Model for Inpatient Rehabilitation to
Increase Functional Independence and Discharge Rate to Home in Geriatric Patients. Archives of physical medicine and
rehabilitation. 2015;96(7):1310-1318.
Galloway RV, Granger CV, Karmarkar AM, et al. The Uniform Data System for Medical Rehabilitation: report of patients
with debility discharged from inpatient rehabilitation programs in 2000-2010. American journal of physical medicine &
rehabilitation / Association of Academic Physiatrists. 2013;92(1):14-27.
Mallinson T, Deutsch A, Bateman J, et al. Comparison of discharge functional status after rehabilitation in skilled nursing,
home health, and medical rehabilitation settings for patients after hip fracture repair. Archives of physical medicine and
rehabilitation. 2014;95(2):209-217.
El-Solh AA, Saltzman SK, Ramadan FH, Naughton BJ. Validity of an artificial neural network in predicting discharge
destination from a postacute geriatric rehabilitation unit. Archives of physical medicine and rehabilitation. 2000;81(10):13881393.
Hall RK, Toles M, Massing M, et al. Utilization of acute care among patients with ESRD discharged home from skilled
nursing facilities. Clinical journal of the American Society of Nephrology: CJASN. 2015;10(3):428-434.
Stearns SC, Dalton K, Holmes GM, Seagrave SM. Using propensity stratification to compare patient outcomes in hospitalbased versus freestanding skilled-nursing facilities. Medical care research and review: MCRR. 2006;63(5):599-622.
Wodchis WP, Teare GF, Naglie G, et al. Skilled nursing facility rehabilitation and discharge to home after stroke. Archives of
physical medicine and rehabilitation. 2005;86(3):442-448.

home. 91 A single-center study found that 31 percent of LTCH hemodialysis patients were discharged to
home. 92 One study noted that 64 percent of beneficiaries who were discharged from the HH episode did
not use any other acute or post-acute services paid by Medicare in the 30 days after discharge. 93 However,
significant numbers of patients were admitted to hospitals (29 percent) and lesser numbers to SNF (7.6
percent), IRF (1.5 percent), HH (7.2 percent) or hospice (3.3 percent). 94
Discharge to community is an actionable health care outcome, as targeted interventions have been
shown to successfully increase discharge to community rates in a variety of post-acute settings. 95,96,97,98,99
Many of these interventions involve discharge planning, communication and care coordination, specific
rehabilitation strategies, such as addressing discharge barriers and improving medical and functional
status, or community-based transitional care services and supports.100,101,102,103,104,105,106, 107,108,109 The
91 Scheinhorn DJ, Hassenpflug MS, Votto JJ, et al. Post-ICU mechanical ventilation at 23 long-term care hospitals: a

multicenter outcomes study. Chest. 2007;131(1):85-93.

92 Thakar CV, Quate-Operacz M, Leonard AC, Eckman MH. Outcomes of hemodialysis patients in a long-term care hospital

setting: a single-center study. American journal of kidney diseases: the official journal of the National Kidney Foundation.
2010;55(2):300-306.
93 Wolff JL, Meadow A, Weiss CO, Boyd CM, Leff B. Medicare home health patients' transitions through acute and post-acute
care settings. Medical care. 2008;46(11):1188-1193.
94 Ibid.
95 Kushner DS, Peters KM, Johnson-Greene D. Evaluating Siebens Domain Management Model for Inpatient Rehabilitation to
Increase Functional Independence and Discharge Rate to Home in Geriatric Patients. Archives of physical medicine and
rehabilitation. 2015;96(7):1310-1318.
96 Wodchis WP, Teare GF, Naglie G, et al. Skilled nursing facility rehabilitation and discharge to home after stroke. Archives of
physical medicine and rehabilitation. 2005;86(3):442-448.
97 Berkowitz RE, Jones RN, Rieder R, et al. Improving disposition outcomes for patients in a geriatric skilled nursing facility.
Journal of the American Geriatrics Society. 2011;59(6):1130-1136.
98 Kushner DS, Peters KM, Johnson-Greene D. Evaluating use of the Siebens Domain Management Model during inpatient
rehabilitation to increase functional independence and discharge rate to home in stroke patients. PM&R: the journal of injury,
function, and rehabilitation. 2015;7(4):354-364.
99 O’Brien SR, Zhang N. Association between therapy intensity and discharge outcomes in aged Medicare skilled nursing
facilities admissions. Archives of Physical Medicine and Rehabilitation. 2018;99(1):107-115.
100 Kushner DS, Peters KM, Johnson-Greene D. Evaluating Siebens Domain Management Model for Inpatient Rehabilitation to
Increase Functional Independence and Discharge Rate to Home in Geriatric Patients. Archives of physical medicine and
rehabilitation. 2015;96(7):1310-1318.
101 Wodchis WP, Teare GF, Naglie G, et al. Skilled nursing facility rehabilitation and discharge to home after stroke. Archives of
physical medicine and rehabilitation. 2005;86(3):442-448.
102 Berkowitz RE, Jones RN, Rieder R, et al. Improving disposition outcomes for patients in a geriatric skilled nursing facility.
Journal of the American Geriatrics Society. 2011;59(6):1130-1136.
103 Kushner DS, Peters KM, Johnson-Greene D. Evaluating use of the Siebens Domain Management Model during inpatient
rehabilitation to increase functional independence and discharge rate to home in stroke patients. PM&R: the journal of injury,
function, and rehabilitation. 2015;7(4):354-364.
104 Jung HY, Trivedi AN, Grabowski DC, Mor V. Does more therapy in skilled nursing facilities lead to better outcomes in
patients with hip fracture? Physical therapy. 2016;96(1):81-89.
105 Camicia M, Wang H, DiVita M, Mix J, Niewczyk P. Length of stay at inpatient rehabilitation facility and stroke patient
outcomes. Rehabilitation Nursing: The Official Journal of the Association of Rehabilitation Nurses. 2016;41(2):78-90.
106 Buttke D, Cooke V, Abrahamson K, et al. A Statewide Model for assisting nursing home residents to transition successfully
to the community. Geriatrics. 2018;3(2):18.
107 Logue MD, Drago J. Evaluation of a modified community based care transitions model to reduce costs and improve
outcomes. BMC Geriatrics. 2013;13(1):94.
108 Carnahan JL, Slaven JE, Callahan CM, Tu W, Torke AM. Transitions from skilled nursing facility to home: the relationship
of early outpatient care to hospital readmission. Journal of the American Medical Directors Association. 2017;18(10):853859.
109 Rodakowski J, Rocco PB, Ortiz M, et al. Caregiver integration during discharge planning for older adults to reduce resource
use: a metaanalysis. J Am Geriatr Soc. 2017;65(8):1748-1755.

effectiveness of these interventions suggests that improvement in discharge to community rates among
post-acute care patients/residents is possible through modifying provider-led processes and interventions.
Denominator
The denominator for the discharge to community measure is the risk-adjusted expected number of
discharges to community. This estimate includes risk adjustment for patient characteristics with the
facility effect removed. The “expected” number of discharges to community is the predicted number of
risk-adjusted discharges to community if the same patients were treated at the average facility appropriate
to the measure.
The regression model used to calculate the denominator is developed using all non-excluded
facility stays in the national data. The denominator is computed in the same way as the numerator, but the
facility effect is set at the average. The descriptions of the discharge to community outcome, patient stays
included in the measure, and numerator calculation are provided below.
Numerator
The measure does not have a simple form for the numerator and denominator—that is, the risk
adjustment method does not make the observed number of community discharges the numerator, and a
predicted number the denominator. The measure numerator is the risk-adjusted estimate of the number of
patients who are discharged to the community, do not have an unplanned readmission to an acute care
hospital or LTCH in the 31-day post-discharge observation window, and who remain alive during the
post-discharge observation window. This estimate starts with the observed discharges to community and
is risk-adjusted for patient characteristics and a statistical estimate of the facility effect beyond case mix.
The numerator uses a model estimated on full national data specific to the IRF setting; it is
applied to the facility’s patient stays included in the measure and includes the estimated effect of that
facility. The prediction equation is based on a logistic statistical model with a two-level hierarchical
structure. The patient stays in the model have an indicator of the facility they are discharged from; the
effect of the facility is measured as a positive or negative shift in the intercept term of the equation. The
facility effects are modeled as belonging to a normal (Gaussian) distribution centered at 0 and are
estimated along with the effects of patient characteristics in the model. Numerator details are provided
below.
Numerator Details: Discharge to Community
Discharge to community is determined based on the “Patient Discharge Status Code” from the
IRF claim. Discharge to community is defined as discharge to home/self-care with or without home health
services. 110 Table 1 below lists the Patient Discharge Status Codes used to define community.

Table 1
Patient Discharge Status Codes Used to Determine Discharge to a Community Setting
Discharge Status Codes Indicating Discharge to a Community Setting
01

Discharged to home/self-care (routine discharge)

Discharged/transferred to home under care of organized home health service organization

Discharged to home or self-care with a planned acute care hospital readmission

Discharged/transferred to home under care of organized home health service organization with a
planned acute care hospital inpatient readmission

110 National Uniform Billing Committee Official UB-04 Data Specifications Manual 2018, Version 12, July 2017, Copyright

2017, American Hospital Association.

Patient discharge status codes 81 and 86 are intended for use on acute care claims only. However,
since these codes have sometimes been reported on PAC claims, we include them in our definition of
community to credit the PAC provider for discharging the patient to a community setting.
Numerator Details: Unplanned Readmissions in the 31-Day Post-Discharge Observation
Window
A patient who is discharged to the community is not considered to have a successful discharge to
community outcome for this measure if they have a subsequent unplanned readmission to an acute care
hospital or LTCH in the post-discharge observation window, which includes the day of discharge and the
31 days following day of discharge. We only assess the first readmission encountered in the postdischarge window. Our definition of acute care hospital includes hospitals paid under the Inpatient PPS
(IPPS), critical access hospitals (CAH), and psychiatric hospitals or units. Using acute care and LTCH
111
claims, we identify unplanned readmissions based on the CMS planned readmissions algorithm used in
the following PAC readmission measures, endorsed by the National Quality Forum (NQF) and used in
several CMS programs: (i) NQF #2510: Skilled Nursing Facility 30-Day All-Cause Readmission Measure
(SNFRM); (ii) NQF #2502: All-Cause Unplanned Readmission Measure for 30 Days Post Discharge
from Inpatient Rehabilitation Facilities; (iii) NQF #2512: All-Cause Unplanned Readmission Measure for
30 Days Post Discharge from Long-Term Care Hospitals; and (iv) NQF #2380: Rehospitalization During
the First 30 Days of Home Health. 112,113,114,115 These readmission measures are based on the HospitalWide All-Cause Readmission Measure (HWR) (CMS/Yale) (NQF #1789), 116 with some additions made
for the SNF, IRF, and LTCH setting measures. 117 The CMS planned readmission definition is based on
the claim from the readmission having a code for a diagnosis or procedure that is considered planned;
however, if a planned procedure is accompanied by a principal diagnosis in a specified list of acute
diagnoses, the readmission is reclassified as unplanned. Readmissions to psychiatric hospitals or units are
classified as planned readmissions. We use the most current available version of the CMS planned
readmission algorithm from the HWR measure specifications for measure calculation and make necessary
updates to the additions made for PAC settings to ensure the algorithm corresponds to our measurement
period.

111 Appendix E. Planned Readmission Algorithm Version 4.0 2018 (ICD-10). In: 2018 All-Cause Hospital Wide Measure

Updates and Specifications Report: Hospital-Level 30-Day Risk-Standardized Readmission Measure – Version 7.0. Available
at:
https://www.qualitynet.org/dcs/BlobServer?blobkey=id&blobnocache=true&blobwhere=1228890804653&blobheader=multi
part%2Foctet-stream&blobheadername1=ContentDisposition&blobheadervalue1=attachment%3Bfilename%3DHospWide_Readmission_AUS_Report_2018_328.pdf&blobcol=urldata&blobtable=MungoBlobs or
https://www.qualitynet.org/dcs/ContentServer?c=Page&pagename=QnetPublic%2FPage%2FQnetTier4&cid=121906985584
1.
112 NQF #2510: Skilled Nursing Facility 30-Day All-Cause Readmission Measure (SNFRM).
www.qualityforum.org/QPS/2510
113 NQF #2502: All-Cause Unplanned Readmission Measure for 30 Days Post Discharge from Inpatient Rehabilitation Facilities.
www.qualityforum.org/QPS/2502
114 NQF #2512: All-Cause Unplanned Readmission Measure for 30 Days Post Discharge from Long Term Care Hospitals.
www.qualityforum.org/QPS/2512
115 NQF #2380: Rehospitalization During the First 30 Days of Home Health. www.qualityforum.org/QPS/2380
116 NQF #1789: Hospital-Wide All-Cause Readmission Measure (HWR) (CMS/Yale). www.qualityforum.org/QPS/1789
117 Table 2-9. AHRQ CCS Single Level Procedure Codes and ICD-9 Procedure Codes Added to Yale’s Planned Readmission
Algorithm, for the Post-Acute Care Setting. In: Measure Specifications for Measures Adopted in the FY 2017 IRF QRP Final
Rule. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/IRF-QualityReporting/Downloads/Measure-Specifications-for-FY17-IRF-QRP-Final-Rule.pdf. Note: The ICD-9 codes listed in Table 2-9
were updated with ICD-10-CM codes for data starting October 1, 2015.

While this measure was developed with ICD-9-CM procedure and diagnosis codes, it has been
transitioned using the ICD-9-CM to ICD-10-CM crosswalk.
Numerator Details: Death in the 31-Day Post-Discharge Observation Window
Patients who are discharged to the community are not considered to have a successful discharge
to community outcome for this measure if they die in the post-discharge window, which includes the day
of discharge and the 31 days following day of discharge. Death in the post-discharge window is identified
based on date of death from Medicare eligibility files.
Target Population and Measure Exclusions
The target population for the measure is the group of Medicare FFS patients who are not excluded
for the reasons listed below.
Measure Exclusions
Exclusions for the discharge to community measure are listed below, along with the rationale and
data source for each exclusion. Baseline NF residence is based on data from the Minimum Data Set
(MDS). All other measure exclusion criteria are determined by processing Medicare claims and eligibility
data to determine whether the individual exclusion criteria are met. Only IRF stays that are preceded by a
short-term acute care stay in the 30 days prior to the IRF admission date are included in the measure.
Stays ending in transfers to the same level of care are excluded.
1) Age under 18 years
Rationale:
a. There is limited literature on discharge destination outcomes in this age group;
b. Patients in this age group represent a different cohort, likely living with their parents, and may be
expected to have higher discharge to community rates compared with the rest of the Medicare
population; and
c. Patients in this age group represent a small proportion of the post-acute Medicare FFS population.
Data source: Birth date and IRF admission date from Inpatient Standard Analytic File (SAF).
2) No short-term acute care stay within the 30 days preceding IRF admission
Rationale: Acute care claims from the 30 days prior to IRF admission provide the principal diagnosis
and other important patient data for risk adjustment. Patients without a short-term acute care
discharge within the 30 days prior to PAC admission are excluded from the measure, because
important risk adjustment data are missing.
Data source: Hospital discharge date in Inpatient SAF acute care claims in the 30 days before IRF
admission.
3) Discharges to psychiatric hospital
Rationale: Patients discharged to psychiatric hospital are excluded from the measure because
community living at the time of discharge may be potentially inappropriate or unsafe for them due to
their mental health or psychiatric condition.
Data source: Patient discharge status code from Inpatient SAF IRF claim.
4) Discharges against medical advice

Rationale: Patients who discharge themselves against medical advice are excluded because their care
plan may not have been fully implemented, and the discharge destination may not reflect the facility’s
discharge recommendation. Additionally, patients discharged against medical advice may potentially
be at higher risk of post-discharge readmissions or death, depending on their medical condition, or
due to potential non-adherence or non-compliance with care recommendations.
Data source: Patient discharge status code from Inpatient SAF IRF claim.
5) Discharges to disaster alternative care sites or federal hospitals
Rationale: Patients discharged to disaster alternative care sites are excluded because these discharges
are likely influenced by external emergency conditions and may not represent discretionary
discharges by the IRF provider. Discharges to federal hospitals are also excluded.
Data source: Patient discharge status code from Inpatient SAF IRF claim.
6) Discharges to court/law enforcement
Rationale: Patients who are discharged to court or law enforcement are likely ineligible for discharge
to the community due to legal restrictions.
Data source: Patient discharge status code from Inpatient SAF IRF claim.
7) Patients discharged to hospice or those with a hospice benefit in the 31-day post-discharge window
Rationale:
a. Patients discharged to hospice care and those with a hospice benefit in the post-discharge
observation window are terminally ill and have very different goals of care compared with nonhospice patients. For non-hospice patients, the primary goal of post-acute care is to return to
baseline, independent living in the community; death is an undesirable outcome in the nonhospice population. For patients on hospice, the goal is to provide them the opportunity to die
comfortably, at home or in a facility.
b. A large proportion of patients on hospice care die in the 31-day window following discharge from
the post-acute setting.
c. The hospice agency, not the post-acute care setting, makes the final decision of discharge to
hospice-home or hospice-facility.
Data source: Discharge to hospice is determined based on the Inpatient SAF IRF claim. Postdischarge hospice benefit is determined based on hospice enrollment dates (start and termination
dates) in the Enrollment Database (EDB).
8) Patients not continuously enrolled in Part A FFS Medicare for the 12 months prior to IRF admission
date, and at least 31 days after IRF discharge date
Rationale: Patients not continuously enrolled in Part A FFS Medicare for the 12 months prior to the
IRF admission date are excluded because risk adjustment for certain comorbidities requires
information on acute inpatient bills for one year prior to IRF admission. Patients not continuously
enrolled in Part A FFS Medicare for at least 31 days after IRF discharge are excluded because
readmissions and death must be observable in the 31-day post-discharge period. Patients without Part
A coverage or those who are enrolled in Medicare Advantage plans will not have complete inpatient
claims in the system.
Data source: EDB and Denominator Files.

9) Patients whose prior short-term acute care stay was for non-surgical treatment of cancer
Rationale: Patients whose prior short-term acute care stay was for non-surgical treatment of cancer
are excluded because they have a different trajectory for recovery after discharge, with a high
118
mortality rate. Exclusion of these patients is consistent with the HWR and PAC readmission
measures.
Data source: Diagnosis codes from the Inpatient SAF prior acute claim.
10) IRF stays that end in transfer to the same level of care
Rationale: IRF stays that end in transfer to the same level of care are excluded because their IRF
episode has not ended. For an IRF episode that involves transfer to the same level of care, only the
final IRF provider is included in the measure.
Data source: Patient discharge status code from Inpatient SAF IRF claim.
11) IRF stays with claims data that are problematic (e.g., anomalous records for stays that overlap
wholly or in part, or are otherwise erroneous or contradictory, stays not matched to the denominator
or EDB files, claims not paid)
Rationale: This measure requires accurate information from the IRF stay and prior short-term acute
care stay in the elements used for risk adjustment. No-pay IRF stays involving exhaustion of Part A
benefits are also excluded.
Data source: Inpatient SAF claims, EDB and denominator files.
12) Planned discharges to an acute or LTCH setting
Rationale: Planned discharges to an acute care hospital or LTCH are excluded because these patients
had a planned return to higher level of care and discharge to community is not appropriate for these
patients.
Data source: The planned readmission algorithm is applied to diagnosis and procedure codes found
on the first acute care or LTCH claim, if any, on the day of or day after index IRF discharge.
13) Medicare Part A benefits exhausted
Rationale: Patients who have exhausted their Medicare Part A coverage during the IRF stay are
excluded because the discharge destination decision may be related to exhaustion of benefits.
Data source: Inpatient SAF IRF claim
14) Patients who received care from a facility located outside of the United States, Puerto Rico or a U.S.
territory
Rationale: Patients who received care from foreign facilities may not have complete inpatient claims
in the system, and these facilities may not be subject to policy decisions related to this quality
measure.
Data source: CMS Certification Number from the Inpatient SAF IRF claim.

118

NQF #1789: Hospital-Wide All-Cause Readmission Measure (HWR) (CMS/Yale).
www.qualityforum.org/QPS/1789

15) New proposed exclusion: Patients who had a long-term NF stay in the 180 days preceding their
hospitalization and IRF stay, with no intervening community discharge between the long-term NF
stay and qualifying hospitalization for measure inclusion (i.e., baseline NF residents)
Rationale: Baseline NF residents did not live in the community prior to their IRF stay and discharge
to a community setting may not be a safe or expected outcome for these residents.
Data source: We examine historical MDS data in the 180 days preceding the qualifying prior acute
care admission and index IRF stay. Presence of an OBRA-only 119 assessment (not a SNF PPS
assessment) with no intervening community discharge between the OBRA assessment and acute care
admission date flags the index IRF stay as baseline NF resident.
Data Sources
This measure is based on Medicare FFS administrative claims and uses data in the Medicare
eligibility files, inpatient claims, and MDS. The eligibility files provide information such as date of birth,
date of death, sex, reasons for Medicare eligibility, periods of Part A coverage, and periods in the
Medicare FFS program. The data elements from the Medicare FFS claims are those basic to the operation
of the Medicare payment systems and include data such as date of admission, date of discharge,
diagnoses, procedures, indicators for use of dialysis services, and indicators of whether the Part A benefit
was exhausted. The inpatient claims data files contain patient-level PAC and other hospital records.
Historical MDS data are used to identify baseline NF residents. No data beyond those submitted in the
normal course of business are required from IRF providers for the calculation of this measure.
The following are the specific files used for measure calculation with links to their
documentation:
•

Medicare Inpatient claims (SAF), Index PAC claims - Documentation for the Medicare claims
data is provided online by Research Data Assistance Center (ResDAC). The following web page
includes data dictionaries for the Inpatient SAF (Inpatient Research Identifiable File (RIF)):
http://www.resdac.org/cms-data/files/ip-rif/data-documentation

•

Medicare Enrollment Database - Information about the EDB may be found at:
http://aspe.hhs.gov/datacncl/datadir/cms.htm

•

Medicare Denominator file - Information and documentation are available at:
https://aspe.hhs.gov/report/data-health-and-well-being-american-indians-alaska-natives-andother-native-americans-data-catalog/medicare-denominator-file and
ftp://ftp.cdc.gov/pub/health_statistics/nchs/datalinkage/Denominator%20(edited).pdf.

•

Minimum Data Set (MDS) - Documentation available at: https://www.resdac.org/cmsdata/files/mds-3.0

Measure Time Window
The measure is calculated using two years of data. All IRF stays during the two-year time
window, except those that meet the exclusion criteria, are included in the measure. For patients with
multiple stays during the two-year time window, each stay is eligible for inclusion in the measure. Data
from CY 2012-2013 were used to develop this measure. The analyses in this document are based on CY
2015-2016 data.
Statistical Risk Model and Risk Adjustment Covariates
We used a hierarchical logistic regression method to predict the probability of discharge to
community. Patient characteristics related to discharge and a marker for the specific discharging facility
119 OBRA = Omnibus Budget Reconciliation Act

are included in the equation. The equation is hierarchical in that both individual patient characteristics are
accounted for, as well as the clustering of patient characteristics by facility. The statistical model
estimates both the average predictive effect of the patient characteristics across all facilities, and the
degree to which each facility has an effect on discharge to community that differs from that of the average
facility. The facility effects are assumed to be randomly distributed around the average (according to a
normal distribution). When computing the facility effect, hierarchical modeling accounts for the known
predictors of discharge to community, on average, such as patient characteristics, the observed facility
rate, and the number of facility stays eligible for inclusion in the measure. The estimated facility effect is
determined mostly by the facility’s own data if the number of patient discharges is relatively large (as the
estimate would be relatively precise) but is adjusted toward the average if the number of patient
discharges is small (as that would yield a less precise estimate).
We used the following model:
Let Yij, denote the outcome (equal to 1 if patient i is discharged to community, 0 otherwise) for a
patient i at facility j; Zij denotes a set of risk adjustment variables. We assume the outcome is related to
the risk adjusters via a logit function with dispersion:
(1)
logit(Prob(Yij =1)) = αj + β*Zij + εij
2
αj = µ + ωj ; ωj ~ N(0, τ )
where Z ij = (Z1, Z2, ... Zk) is a set of k patient-level risk adjustment variables; αj represents the facility-specific
intercept; µ is the adjusted average outcome across all facilities; τ2 is the between-facility variance component;
and ε ~N(0,σ2) is the error term. The hierarchical logistic regression model is estimated using SAS software
(PROC GLIMMIX: SAS/STAT User’s Guide, SAS Institute Inc.).
The estimated equation is used twice in the measure. The sum of the probabilities of discharge to
community of all patients in the facility measure, including both the effects of patient characteristics and
the facility, is the “predicted number” of discharges to community after adjusting for the facility’s case
mix. The same equation is used without the facility effect to compute the “expected number” of
discharges to community for the same patients at the average facility. The ratio of the predicted-toexpected number of discharges to community is a measure of the degree to which discharges to
community are higher or lower than what would otherwise be expected. This standardized risk ratio
(SRR) is then multiplied by the mean discharge to community rate for all facility stays for the measure,
yielding the risk-standardized discharge to community rate for each facility. Please note that the
estimation procedure is recalculated for each measurement period. Re-estimating the models for each
measurement period allows the estimated effects of the patient characteristics to vary over time as patient
case-mix and medical treatment patterns change.
Risk adjustment variable descriptions are provided below. See Appendix B, Table B-1 for the full
list of variables in the risk adjustment models.
1. Age and sex groups.
2. End stage renal disease (ESRD) or disability as original reason for entitlement.
3. Principal diagnosis (Clinical Classifications Software (CCS) groups) from the prior acute stay in
the past 30 days. The principal diagnosis codes from the prior acute claim are grouped clinically
using the CCS groupings developed by the Agency for Healthcare Research and Quality
(AHRQ).120,121

120 AHRQ CCS groupings of ICD-9 codes - Documentation available at: http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp
121 AHRQ CCS groupings of ICD-10 codes - Documentation available at: https://www.hcup-

us.ahrq.gov/toolssoftware/ccs10/ccs10.jsp

4. IRF case-mix groups.
5. Surgical procedure categories (if present) based on the prior acute stay in the past 30 days. The
procedures are grouped using the CCS groupings of procedures developed by AHRQ.122
6. Indicator for ESRD status.
7. Dialysis in prior acute stay where ESRD not indicated.
8. Length of prior acute hospital stay in days for patients whose prior acute stay was in a nonpsychiatric hospital (categorical variables are used to account for nonlinearity); indicator of prior
psychiatric hospital stay for patients whose prior acute stay was in a psychiatric hospital.
9. Comorbidities based on prior acute stay in the past 30 days or based on a one year look back,
depending on the specific comorbidity. Comorbidities are clustered using the Hierarchical
Condition Categories [HCC] groups used by CMS. 123
10. Number of prior acute hospital discharges in the past year, not including the hospitalization in the
30 days prior to the IRF stay.
Measure Calculation Algorithm
The following steps describe the calculation algorithm/measure logic for the discharge to
community measures:
Step 1:

Identify patients meeting the criteria for the target population, after applying
measure exclusions.

Step 2:

Identify patients meeting the numerator criteria, i.e., discharge to community, no
unplanned readmissions on the day of discharge or in the 31 days following
discharge, and no death on the day of discharge or in the 31 days following
discharge.

Step 3:

Identify presence or absence of risk adjustment variables for each patient.

Step 4:

Calculate the predicted and expected number of discharges to community for
each facility using the hierarchical logistic regression model.
The predicted number of discharges to community for each facility is calculated
as the sum of the predicted probability of discharge to community for each
patient discharged from the facility and included in the measure, including the
facility-specific effect.
To calculate the predicted number of discharges to community, predj, for index
facility stays at facilityj, we used the following equation:
predj = Σlogit-1(µ + ωi + β*Zij)

(2)

where the sum is over all stays in facilityj, and ωi is the random intercept.
To calculate the expected number expj, we used the following equation:

122 Ibid
123 CMS-HCC Mappings of ICD-9 and ICD-10 Codes: Mappings are included in the software at the following website:

http://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Risk-Adjustors.html

expj = Σlogit-1 (µ + β*Zij)
Step 5:

(3)

Calculate the SRR for each facility, as the ratio of the predicted to expected
number of discharges to community.
To calculate the facility-wide SRR, SRRj, we used the following equation:
SRRj = predj/expj

Step 6:

(4)

Calculate the risk-standardized discharge to community rate for each facility.
To aid interpretation, the facility-wide SRRj, obtained from equation (4) is then
multiplied by the overall national raw discharge to community rate for all facility
stays, Ῡ, to produce the facility-wide risk-standardized discharge to community
rate (RSRj).
To calculate the risk-standardized discharge to community rate for each facility,
we used the following equation:
RSR j = SRR j*Ῡ

(5)

NOTE: Because the statistic described in Step 6 is a complex function of parameter estimates, resampling using bootstrapping may be necessary to derive a confidence interval estimate for the final riskstandardized rate, to characterize the uncertainty of the estimate.
See Appendix B for risk adjustment model results and providers’ observed and risk-standardized
score distributions.

Chapter 2 Standardized Patient Assessment Data Elements
Section 1: Introduction
The Improving Medicare Post-Acute Care Transformation Act of 2014 (IMPACT Act) requires
CMS to develop, implement, and maintain standardized patient assessment data elements (SPADEs) for
post-acute care (PAC) settings. The four PAC settings specified in the IMPACT Act are home health
agencies (HHAs), inpatient rehabilitation facilities (IRFs), long term care hospitals (LTCHs), and skilled
nursing facilities (SNFs). The goals of implementing cross-setting SPADEs are to facilitate care
coordination, interoperability, and improve Medicare beneficiary outcomes.
Existing PAC assessment instruments (i.e., OASIS for HHAs, IRF-PAI for IRFs, LCDS for
LTCHs, and the MDS for SNFs) often collect data elements pertaining to similar concepts, but the
individual data elements -- questions and response options -- vary by assessment instrument. With a few
exceptions, the data elements collected in these assessment instruments are not currently standardized or
interoperable, therefore, patient responses across the assessment instruments cannot be compared easily.
The IMPACT Act further requires that the assessment instruments described above be modified
to include core data elements on health assessment categories and that such data be standardized and
interoperable. Implementation of a core set of standardized assessment items across PAC settings has
important implications for Medicare beneficiaries, families, providers, and policymakers. CMS is
proposing standardized patient assessment data elements for five categories specified in the IMPACT Act.
These categories are:
1.

Cognitive function (e.g., able to express ideas and to understand normal speech) and mental
status (e.g., depression and dementia)

2. Special services, treatments, and interventions (e.g., need for ventilator, dialysis, chemotherapy,
and total parenteral nutrition)
3. Medical conditions and co-morbidities (e.g., diabetes, heart failure, and pressure ulcers)
4. Impairments (e.g., incontinence; impaired ability to hear, see, or swallow)
5. Other categories as deemed necessary by the Secretary
Background
In the following sections, we present additional information on the SPADEs proposed in the FY
2020 IRF PPS proposed rule. We include detailed specifications of the proposed data elements along with
a mockup of how the SPADE is proposed to appear in the IRF-PAI. We outline how each SPADE is
relevant to the care of patients in the IRF, review its current use in existing PAC assessment item sets, and
summarize any prior testing of the data elements. For SPADEs that were included in the National Beta
Test, which was conducted by RAND between November 2017 and August 2018, we present detailed
information on data element performance.
Evidence supporting the SPADEs proposed in the IRF PPS proposed rule comes from several
sources including the Post-Acute Care Payment Reform Demonstration (PAC PRD), Minimum Data Set
3.0 (MDS 3.0) testing, and the National Beta Test. The most relevant metrics for evaluation of SPADE
performance (i.e., feasibility and reliability) include the amount of missing data, time to administer the
data element, and interrater reliability (IRR). Interrater reliability is the level of agreement between two
raters; that is, the extent to which two different individuals would code the same response when presented
with the same information. Typically, percent agreement and the kappa statistic – or weighted kappas, for
ordinal data – are used to represent IRR. The kappa statistic is preferred in most cases because there are
agreed-upon conventions for its interpretation and it corrects for chance agreement between raters.

However, kappa is sensitive to prevalence rates; when prevalence rates are extremely high or low, the
resulting kappa statistic does accurately convey the level of agreement. 124 125 126 127 In those cases,
percent agreement is preferred. The evidence offered for the SPADEs in the sections below follow
standard conventions in reporting both percent agreement and kappas or weighted kappas to describe
IRR.
Post-Acute Care Payment Reform Demonstration (PAC PRD)
Some prior evidence for the proposed SPADEs comes from the PAC PRD. The PAC PRD was
mandated by the Deficit Reduction Act of 2005 to examine the relative costliness and outcomes of similar
types of Medicare beneficiaries discharged to different PAC settings (i.e., HHAs, IRFs, LTCHs, and
SNFs). To meet these aims, the study collected standardized assessment data, using the Continuity
Assessment Record and Evaluation (CARE) across PAC settings to measure patient severity and case-mix
across settings in over 200 providers in 11 geographically diverse markets. The standardized assessment
data allowed cross-setting comparisons of the factors associated with costs and outcomes, as well as
service substitution among post-acute providers, all else being equal about the patient. Further
information on the design and methods of the PAC PRD can be found at https://www.cms.gov/ResearchStatistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.
Testing of the Minimum Data Set 3.0 (MDS 3.0)
Additional testing information comes from the national testing of the MDS 3.0. 128 During a sixyear period starting in 2003, CMS engaged in a national project to create an improved version of the
MDS 2.0. A joint RAND/Harvard team employed an iterative development process that culminated in the
national testing of the MDS 3.0 in 2006-2007. The national validation and evaluation testing of the MDS
3.0 included 71 community nursing homes (3,822 residents) and 19 Veterans Health Administration
nursing homes (764 residents), distributed throughout the regions of the United States. The evaluation
was designed to test and analyze interrater reliability, validity of key items, response rates for interview
items, feedback on changes from participating nurses, and time to complete the MDS assessment. In
addition, the national test design allowed comparison of item distributions between MDS 3.0 and MDS
2.0. Further information on the design and methods of MDS 3.0 testing can be found at
https://www.cms.gov/Medicare/Quality-Initiatives-Patient-AssessmentInstruments/NursingHomeQualityInits/downloads/MDS30FinalReport.pdf.
National Beta Test
Purpose and Goals
The National Beta Test was conducted to evaluate the reliability and validity of candidate
SPADEs and to support the identification of data elements for standardization across PAC settings, in
accordance with the mandates of the IMPACT Act. To test SPADE performance within each setting,
sufficient numbers of patients/residents needed to be included in each of the four settings to enable
setting-specific performance estimates. Further, the participating patients/residents needed to represent
124

Cicchetti, D. V., Feinstein, A. R. (1990). High agreement but low kappa: II. Resolving the paradoxes. Journal of
Clinical Epidemiology, 43(6):551-558.
125 Xu, S., Lorber, M. F. (2014). Interrater agreement statistics with skewed data: Evaluation of alternatives to
Cohen’s kappa. Journal of Consulting and Clinical Psychology, 82(6):1219-1227.
126 Byrt, T., Bishop, J., Carlin, J. B. (1993). Bias, prevalence and kappa. Journal of Clinical Epidemiology,
46(5):423-429.
127 McHugh, M. L. (2002). Interrater reliability: the kappa statistic. Biochemia Medica, 22(3):276-282.
128 Saliba, D., & Buchanan, J. (2008). Development and validation of a revised nursing home assessment tool: MDS
3.0. Santa Monica, CA: RAND Corporation. 2008. Available at
http://www.cms.hhs.gov/NursingHomeQualityInits/Downloads/MDS30FinalReport.pdf

adequate coverage of the clinical range of patients/residents receiving care nationally in each of the four
PAC settings. To evaluate the suitability of the SPADEs for cross-setting use, sufficient numbers of
facilities/agencies of each setting type needed to be included in the test. These facilities/agencies needed
to reflect a reasonable range of geographic diversity relative to PAC settings nationally.
Many large national studies of patients and health conditions are designed to generate estimates
and make comparisons of rates of conditions or severity of patients on one or more clinical characteristics
(e.g., cognitive status). To do this, these studies seek to recruit a proportionally balanced representative
sample, and employ casemix models and/or sampling weights to the data. In contrast, the National Beta
Test was designed to generate valid and robust national SPADE performance estimates (i.e., time to
complete and IRR), which fundamentally requires acceptable geographic diversity, sufficient sample size,
and reasonable coverage of the range of clinical characteristics. To meet these requirements, the National
Beta Test was carefully designed so that data could be collected from a wide range of environments,
allowing for thorough evaluation of candidate SPADE performance in all PAC settings. These analyses
included extensive checks on the sampling design (e.g., generating results by market and by urbanicity) to
identify possible limitations to the generalizability of results. Results of these sensitivity analyses are not
included in this document, but will be described in detail in the forthcoming volumes of the National Beta
Test Final Report (see https://www.cms.gov/Medicare/Quality-Initiatives-Patient-AssessmentInstruments/Post-Acute-Care-Quality-Initiatives/IMPACT-Act-of-2014/IMPACT-Act-Downloads-andVideos.html).
In order to help readers interpret evidence from the National Beta Test that is included for some
SPADEs, we include an abridged description of the National Beta Test design and methods below. An indepth technical discussion of the design and methods of the National Beta Test can be found in the
document titled “Development and Evaluation of Candidate SPADEs_National Beta Test Background
and Methods,” available at https://www.cms.gov/Medicare/Quality-Initiatives-Patient-AssessmentInstruments/Post-Acute-Care-Quality-Initiatives/IMPACT-Act-of-2014/IMPACT-Act-Downloads-andVideos.html.
Design and Sampling
The National Beta Test included PAC providers in a total of 14 markets across the country. This
number of markets was chosen to be similar to the design used for the PAC PRD. A multi-stage stratified
random sampling plan was used to obtain the sample of 14 geographic/metropolitan areas, or “markets,”
in the United States, and then a sample of eligible PAC facilities was compiled from those markets. To be
eligible for selection, markets had to meet the following criteria:
•

Sampled markets would yield a predefined number of PAC facilities/agencies of each type for the
sample (12 SNFs, 10 HHAs, at least four LTCHs or IRFs, and at least one LTCH)

•

The predefined number of facilities/agencies within the markets were expected to have flow rates
large enough to obtain the targeted number of assessments per facility

•

The predefined number of facilities/agencies had to be located within two hours of one another to
facilitate completion of assessments in a timely manner

Of 306 markets in the U.S., 64 were deemed eligible. The random sampling of the 14 markets
was stratified by U.S. Census division to enhance geographic representation yielding the following 14
markets: Boston, MA, Chicago, IL, Dallas, TX, Durham, NC, Fort Lauderdale, FL, Harrisburg, VA,
Houston, TX, Kansas City, MO, Los Angeles, CA, Nashville, TN, Philadelphia, PA, Phoenix, AZ, St.
Louis, MO, and San Diego, CA. Given these markets are a random sample, they are expected to be
representative of the set of 64 eligible facilities and findings generalizable to the set of eligible facilities.
The target numbers of providers by setting within these 14 markets were 28 IRFs, 28 LTCHs, 84
SNFs, and 70 HHAs, totaling 210 PAC providers. The number of settings was determined based on

standard sample size calculations which included the numbers of facilities and patients rather than the
proportions of the populations they represented. The power calculations indicated that 28 providers per
setting type (two in each market) would yield sufficient numbers of admissions during the field period to
obtain robust estimates of candidate SPADE performance. This minimum number was adopted as the
recruitment target for IRFs and LTCHs; additional SNFs and HHAs were targeted so as to enhance
sample diversity in light of the larger proportion of these setting types nationally. A total of 143 PAC
facilities were successfully recruited across 14 U.S. markets (35 HHAs, 22 IRFs, 26 LTCHs, 60 SNFs) to
participate in the National Beta Test. Although this number falls short of targets both overall and by
setting, this shortfall was offset by extending the field period, allowing for the accrual of more eligible
patient/resident admissions and discharges.
Eligibility
The National Beta Test SPADEs included in this proposed rule were evaluated for performance
among a sample of communicative patients/residents (who could make themselves understood through
any means). All communicative patients/residents who were admitted to a participating provider site
during the field period and were Medicare beneficiaries covered under one of the PAC prospective
payment systems were eligible for the admission assessment, and all those who completed an admission
assessment and were discharged during the field period were eligible for the discharge assessment.
National Beta Test enrollment of non-communicative patients/residents was not tied to an admission date
so as to ensure availability of sufficient numbers within the field period for evaluation of three data
elements developed specifically for non-communicative patients/residents (observational assessments of
cognitive status, mood, and pain). Although this ensured availability of sufficient numbers of noncommunicative patients/residents for testing of the non-communicative data elements, it precluded
assessing these patients/residents with non-interview SPADEs at admission. The three data elements
developed specifically for non-communicative patients/residents are not included in this proposed rule,
thus the non-communicative sample from the National Beta Test is not described further here.
Section 1557 of the Patient Protection and Affordable Care Act 129 states that facilities that deliver
PAC services under Medicare are required to provide qualified interpreters to their patients/residents with
limited English proficiency. Facilities have discretion in how they furnish qualified interpreters, including
the use of remote interpreters (i.e., high-quality telephone or video services). As described above, the
focus of the National Beta Test was to establish the feasibility and validity of the data elements within and
across PAC settings. Including limited English proficiency patients/residents in the sample would have
required the Beta test facilities to engage or involve translators during the test assessments. In planning
the National Beta Test, we anticipated that this would have added undue complexity to what
facilities/agencies were being asked to do, and would have undermined the ability of facility/agency staff
to complete the requested number of assessments within the assessment window (e.g., Admission Days 37) and within the study field period. In light of the strong existing evidence for the feasibility of all
patient/resident interview SPADEs included in this proposed rule (BIMS, Pain Interference, PHQ) when
administered in other languages, either through standard PAC workflow (e.g., as tested and currently
collected in the MDS 3.0) and/or through rigorous translation and testing (e.g., PHQ), the performance of
translated versions of these patient/resident interview SPADEs did not need to be further evaluated. In
addition, because their exclusion did not threaten our ability to achieve acceptable geographic diversity,
sufficient sample size, and reasonable coverage of the range of PAC patient/resident clinical
characteristics, the exclusion of limited English proficiency patients/residents was not considered a
limitation to interpretation of the National Beta Test results.

129

https://www.hhs.gov/civil-rights/for-individuals/section-1557/index.html

Data Collection
Admission assessments were completed between admission days 3-7; discharge assessments
could be completed from two days prior to discharge through the discharge date. Trained research nurses
and/or staff at participating PAC facilities/agencies administered all assessments. A subset of the
admission assessments was completed by research nurse/facility staff assessor pairs to allow for
evaluation of interrater reliability. Power analyses indicated that reliability estimates required a minimum
of 194 paired assessments, time to complete estimates could be compared across settings for detection of
small effect sizes with a minimum of 274 assessments per setting, and as few as 460 assessments would
be sufficient to evaluate aspects of validity (e.g., group differences, associations with other clinical
variables, etc.) with small to moderate effect sizes. Therefore, average assessment contributions per
participating facility/agency were calculated for each of these goals (i.e., paired assessments, assessments
completed by facility/agency staff, total admission assessments) and communicated throughout the study
period to guide the data collection and track progress. These minimums were more easily attainable in
SNFs and HHAs due to the larger number of participating facilities/agencies. However, participating
LTCHs and IRFs were able to collectively meet these targets by the end of the field period. The total
number of admission assessments is shown in Appendix C, Table 1.1. This table also shows the number
of assessments from which completion times were estimated, and the number of assessments that were
conducted by paired raters and contributed to evaluation of interrater reliability (IRR). In addition to
meeting the minimum sample size requirements, the data collection yielded very small rates of missing
data, speaking to the overall feasibility of the proposed SPADEs. Table 1.2 in Appendix C shows
completion rates by Beta protocol module. Module completion rates ranged from 93.8 to 98.2 percent,
and nearly 90 percent of the communicative admission sample completed all assessment modules. More
information on the design and methods of the National Beta Test can be found in the document titled
“Development and Evaluation of Candidate SPADEs_National Beta Test Background and Methods,”
available at https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/PostAcute-Care-Quality-Initiatives/IMPACT-Act-of-2014/IMPACT-Act-Downloads-and-Videos.html.

Section 2: Cognitive Function
Impairments in cognitive function can result from many underlying conditions, including
dementia, Alzheimer’s Disease, stroke, brain injury, side effects of medication, metabolic and/or
endocrine imbalances, and delirium. 130 Cognitive impairments may affect a patient or resident’s ability to
recover from illness or injury, or they may be a sign of an acute condition (e.g., hypoxia) that requires
immediate intervention. Cognitive impairment that manifests with behavioral symptoms—or that impairs
a patient’s ability to communicate, prompting behavioral disturbances—may put the patient or resident or
others in the care setting at risk for injury or assault, or may signal unmet patient or resident needs (e.g.,
pain management). Screening for the presence of impairment can help ensure appropriate and timely
intervention.
A substantial proportion of PAC patients and residents experience cognitive impairment,
delirium, communication impairment, or behavioral distress. Testing from the PAC PRD found that about
one-third of patients and residents in PAC settings were classified as having moderately or severely
impaired cognitive function. 131 132 About one-third exhibited disorganized thinking and altered level of
consciousness, and about one-half exhibited inattention. Fewer than 7 percent of patients and residents
exhibited signs and symptoms of behavioral distress in the PAC PRD.
Therapeutic interventions can improve patient outcomes, and evidence suggests that treatment
(e.g., drugs, physical activity) can stabilize or delay symptom progression in some patients, thereby
improving quality of life. 133 134 135 In addition, assessments help PAC providers to better understand the
needs of their patients by establishing a baseline for identifying changes in cognitive function and mental
status (e.g., delirium), elucidating the patient’s ability to understand and participate in treatments during
their stay, highlighting safety needs (e.g., to prevent falls), and identifying appropriate support needs at
the time of discharge. The standardized assessment of patient or resident cognition supports clinical
decision-making, early clinical intervention, person-centered care, and improved care continuity and
coordination. The use of valid and reliable standardized assessments can aid in the communication of
information within and across providers, enabling the transfer of accurate health information.
CMS has identified several data elements as applicable for cross-setting use in standardized
assessment of cognitive impairment. The proposed data elements comprise:
1. The Brief Interview for Mental Status (BIMS);
2. The Confusion Assessment Method (CAM)

130

National Institute on Aging. (2013). “Assessing Cognitive Impairment in Older Patients: A Quick Guide for
Primary Care Physicians.” Available at https://www.nia.nih.gov/alzheimers/publication/assessing-cognitiveimpairment-older-patients
131 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment
reform demonstration: Final report. Research Triangle Park, NC: RTI International. 2012. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/ResearchReports-Items/PAC_Payment_Reform_Demo_Final.html.
132 This estimate is based on responses to the Brief Interview for Mental Status (BIMS) in a study of
patient/residents in the Post-Acute Care Payment Reform Demonstration (Gage et al., 2012).
133 Casey, D. A., Antimisiaris, D., & O’Brien, J. (2010). “Drugs for Alzheimer’s disease: are they effective?”
Pharmacy and Therapeutics 35(4): 208.
134 Bherer, L., Erickson, K. I., & Liu-Ambrose, T. (2013). “A review of the effects of physical activity and exercise
on cognitive and brain functions in older adults.” Journal of Aging Research. 2013.
135 Langa, K. M., & Levine, D. A. (2014). “The diagnosis and management of mild cognitive impairment: a clinical
review.” JAMA 312(23): 2551-2561.

The data elements proposed involve different aspects of cognition (e.g., short term memory,
comprehension), types of data (e.g., interview, performance-based), and are collected by various modes
(e.g., clinician assessed, patient reported).
Brief Interview for Mental Status (BIMS)
The Brief Interview for Mental Status (BIMS) is a performance-based cognitive assessment
developed to be a brief cognition screener with a focus on learning and memory. The BIMS evaluates
repetition, recall with and without prompting, and temporal orientation.
Relevance to IRFs
The BIMS is currently included in the IRF-PAI assessment on admission. Assessing cognitive
functioning is critical in IRF settings, as cognitive impairments are common among IRF patients.
Although more comprehensive cognitive assessment is commonplace in IRFs (e.g., instruments
incorporated in speech therapy or administered by neuropsychologists), standardized assessment tools can
provide comparable baseline information if uniformly administered to all patients and standardized across
provider types. An estimated 22.2 percent of IRF patients are moderately impaired, and 11.6 percent are
severely impaired, as assessed by the BIMS in the PAC PRD. 136 Patients with brain injury and stroke are
commonly transferred to IRFs for intensive post-acute care: approximately 21 percent of IRF patients
have a primary diagnosis of stroke, and approximately 8 percent have a primary diagnosis of brain
injury. 137 In addition, cognitive impairments are associated with engagement in rehabilitation
therapies, 138 and individuals with severe cognitive impairment as measured by BIMS at IRF admission
are more likely to be readmitted after discharge. 139 Cognitive impairment has significant implications for
patient resource utilization, ability to participate in rehabilitation therapies, and care planning in IRFs.
The standardized assessment of cognitive function using the BIMS would provide important information
for care planning, care transitions, patient safety, and resource use in IRFs.
Proposed Data Elements for the Assessment of Cognitive Function: The BIMS
C0100. Should Brief Interview for Mental Status (C0200-C0500) be Conducted?
Attempt to conduct interview with all patients.
Enter Code

0.No (patient is rarely/never understood)  Skip to XXXX
1. Yes  Continue to C0200, Repetition of Three Words

Brief Interview for Mental Status (BIMS)
C0200. Repetition of Three Words

136

Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. (2012). Post-acute care
payment reform demonstration: Final report. Research Triangle Park, NC: RTI International. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/ResearchReports-Items/PAC_Payment_Reform_Demo_Final.html.
137 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9
Inpatient rehabilitation facility services. Retrieved from http://www.medpac.gov/-documents-/reports
138 E.J. Lenze, M.C. Munin, M.A. Dew, et al. (2004). Adverse effects of depression and cognitive impairment on
rehabilitation participation and recovery from hip fracture. Int J Geriatr Psychiatry 19: 472–478
139 Gage B., Morley M., Smith L., et al. (2012). Post-Acute Care Payment Reform Demonstration (Final report,
Volume 4 of 4). Research Triangle Park, NC: RTI International.

Enter Code

Ask patient: “I am going to say three words for you to remember. Please repeat the words
after I have said all three. The words are: sock, blue and bed. Now tell me the three
words.”
Number of words repeated after first attempt
0. None
1. One
2. Two
3. Three
After the patient’s first attempt, repeat the words using cues (“sock, something to
wear; blue, a color; bed, a piece of furniture”). You may repeat the words up to two
more times.
C0300. Temporal Orientation (orientation to year, month, and day)
Enter Code

Enter Code

Ask patient: “Please tell me what year it is right now.”

A. Able to report correct year
0. Missed by > 5 years or no answer
1. Missed by 2-5 years
2. Missed by 1 year
3. Correct
Ask patient: “What month are we in right now?”

B. Able to report correct month
0. Missed by > 1 month or no answer
1. Missed by 6 days to 1 month
2. Accurate within 5 days
Ask patient: “What day of the week is today?”

C. Able to report correct day of the week
0. Incorrect or no answer
1. Correct

C0400. Recall
Enter Code

Enter Code

Ask patient: “Let's go back to an earlier question. What were those three words that I asked
you to repeat?” If unable to remember a word, give cue (something to wear; a color; a
piece of furniture) for that word.

A. Able to recall “sock”
0. No - could not recall
1. Yes, after cueing ("something to wear")
2. Yes, no cue required
B. Able to recall “blue”
0. No - could not recall
1. Yes, after cueing ("a color")
2. Yes, no cue required
C. Able to recall “bed”
0. No - could not recall
1. Yes, after cueing ("a piece of furniture")
2. Yes, no cue required

C0500. BIMS Summary Score
Enter Score

Add scores for questions C0200-C0400 and fill in total score (00-15)
Enter 99 if the patient was unable to complete the interview

Current use
The BIMS data elements are currently used in the MDS and the IRF-PAI.
Prior evidence supporting use of the BIMS
The BIMS data elements were tested in the PAC PRD, where they showed substantial to almost
perfect reliability of 0.71 to 0.91 (weighted kappas) when used across all four PAC settings. The lowest
agreement was on the “repetition of three words” memory data element, with a kappa of 0.71, which still
falls within the range of substantial agreement. PAC PRD testing also demonstrated feasibility of the
BIMS for use in IRFs and found evidence of strong reliability of the BIMS data elements in the IRF
setting. In addition, the BIMS data elements were found to be predictive of higher patient cost. 140 The
BIMS data elements were also included in the national MDS 3.0 test in nursing homes and showed almost
perfect reliability. 141 Agreement ranged from 0.86 to 0.99 (standard kappa). The BIMS data elements
were found to be highly correlated (0.906) with a gold-standard measure of cognitive function, the
Modified Mini-Mental Status (3MS) exam. 142
Evidence supporting use of the BIMS from the National Beta Test
Assessing impairment: In the Beta testing, the BIMS was administered at admission to 646
patients/residents in the HHA setting, 786 in IRF, 496 in LTCH, and 1,134 in SNF (n = 3,062 overall).
Overall, 5 percent of patients/residents met criteria for being severely impaired, 18 percent moderately
impaired, and 76 percent intact. In the IRF setting, 3 percent were severely impaired, 15 percent
moderately impaired, and 82 percent intact. Patients in the IRF setting showed similar impairment levels
to those in an HHA and somewhat lower impairment than those in an LTCH or SNF. Setting-specific
admission frequencies for BIMS data elements and the overall impairment category at admission are
shown in Appendix C, Table 2.1.1.
Missing data: In general, there were low rates of missing data for BIMS items. Overall, itemlevel missing data ranged from 0.4 to 1.7 percent and ranged from 0.3 to 0.9 percent in the IRF setting.
For all settings, missing data rates were slightly higher for recall of current day of the week. In general,
the low rate of missing data indicates feasibility of administration.
Time to complete: To assess feasibility of administration, the length of time to administer the
BIMS was assessed among 445 patients/residents in HHA, 537 in IRF, 332 in LTCH, and 494 in SNF (n
= 1,808 overall). Overall mean time to complete the BIMS was 2.2 minutes (SD = 1.2 minutes). Time to
complete in the IRF setting was 1.8 minutes (SD = 0.9 minutes).
Interrater reliability: The interrater reliability (IRR) was excellent for the BIMS as measured by
kappa and percent agreement of paired raters (n = 966 paired assessments across settings; n = 259 paired
assessments in IRF). Across all settings, the kappa for the BIMS Impairment Category classification
(based on the BIMS total score) was 0.91; in the IRF setting, the kappa was 0.85. The kappas for
individual items within the BIMS ranged from 0.83 to 0.93 across all settings and ranged from 0.81 to
140

Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment
reform demonstration: Final report. Research Triangle Park, NC: RTI International. 2012. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/ResearchReports-Items/PAC_Payment_Reform_Demo_Final.html.
141 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0.
Appendices. Santa Monica, CA: RAND Corporation. 2008. Available at :
https://www.cms.gov/Medicare/Quality-Initiatives-Patient-AssessmentInstruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.
142 Saliba, D., Buchanan, J., Edelen, M. O., Streim, J., Ouslander, J., Berlowitz, D., & Chodosh, J. (2012). “MDS
3.0: Brief interview for mental status.” Journal of the American Medical Directors Association 13(7): 611-617.

0.91 in the IRF setting. Overall kappa values were not estimated for two items within the BIMS because
the proportion of patients across settings with correct responses was out of range for a stable kappa
estimate. Similarly, in the IRF setting, kappa was not estimated for three BIMS items. Percent agreement
for the BIMS Impairment Category classification was 96 percent across all settings and 95 percent in the
IRF setting. Percent agreement for the individual items ranged from 94 to 98 percent across settings and
from 94 to 99 percent in IRFs. Please refer to Table 2.1.2 in Appendix C for kappa and percent agreement
statistics for all BIMS items.
Confusion Assessment Method (CAM©)
The Confusion Assessment Method (CAM) is a widely used delirium screening tool. 143 Delirium,
when undetected or untreated, can increase the likelihood of complications, rehospitalization, and death
compared to patients/residents without delirium. 144
Although multiple versions of the CAM have been developed, CMS is proposing that the Short
version be adopted for standardized patient assessment data elements. The Short CAM contains only four
items (i.e., items 1 to 4) from the original Confusion Assessment Method (Long CAM). These items
focus on an acute change in mental status, inattention, disorganized thinking, and altered level of
consciousness.
Relevance to IRFs
The IRF-PAI does not include items to assess signs and symptoms of delirium, although delirium
is common among IRF populations. In PAC PRD testing using the CAM, high proportions of IRF
patients demonstrated signs and symptoms of delirium: 57.3 percent showed inattention, 44.1 percent
showed disorganized thinking, and 21.4 percent showed altered level of consciousness. 145 Delirium may
also interfere with functional recovery and a patient’s ability to actively participate in intensive
rehabilitation therapies, 146 147 which is required by IRFs. In addition, presence of delirium has
implications for administering and interpreting cognitive assessments, 148 149 which has implications for
assessing recovery and anticipated benefits of cognitive rehabilitation for IRF patients. As such, assessing
IRF patients for signs and symptoms of delirium is critical for care planning and decision making in IRF
settings, and for ensuring that IRF patients can maximally benefit from rehabilitation therapies.
The standardized assessment of delirium and reversible confusion using the Short CAM would
provide important information for care planning, care transitions, patient safety, and resource use in IRFs.

143

De, J., & Wand, A. P. (2015). Delirium screening: a systematic review of delirium screening tools in
hospitalized patients. The Gerontologist, 55(6), 1079-1099.
144 Marcantonio, E. R., Kiely, D. K., Simon, S. E., John Orav, E., Jones, R. N., Murphy, K. M., & Bergmann, M. A.
(2005). Outcomes of older people admitted to postacute facilities with delirium. Journal of the American
Geriatrics Society, 53(6): 963-969.
145 Unpublished data from the PAC PRD Public Comments sample, 2008-2010.
146 Marcantonio, E. R., Simon, S. E., Bergmann, M. A., et al. (2003). Delirium symptoms in postacute care:
Prevalent, persistent, and associated with poor functional recovery. Journal of the American Geriatrics Society,
51:4e9.
147 Kiely, D. K., Jones, R. N., Bergmann, M. A., et al. (2006). Association between delirium resolution and
functional recovery among newly admitted postacute facility patients. Journal of Gerontol A Biol Sci Med Sci.
61:204e208.
148 Landi, F., Liperoti, R,, Bernabei, R. (2011). Postacute rehabilitation in cognitively impaired patients:
comprehensive assessment and tailored interventions. Journal of the American Medical Association
149

12:395e397.

McCusker, J., Cole, M., Dendukuri, N., et al. (2001). Delirium in older medical inpatients and subsequent
cognitive and functional status: A prospective study. Canadian Medical Association Journal 165:575e583.

Proposed Data Elements for the Assessment of Cognitive Function: CAM
C1310. Signs and Symptoms of Delirium (from CAM©)
Code after completing Brief Interview for Mental Status or Staff Assessment and reviewing medical
record.
A. Acute Onset Mental Status Change
Enter Code

Is there evidence of an acute change in mental status from the patient's baseline?
0. No
1. Yes

Coding:
0. Behavior not
present
1. Behavior
continuously
present, does
not fluctuate
2. Behavior
present,
fluctuates
(comes and
goes, changes
in severity)

Enter Code in Boxes

B. Inattention - Did the patient have difficulty focusing attention, for
example, being easily distractible or having difficulty keeping track
of what was being said?
C. Disorganized thinking - Was the patient's thinking
disorganized or incoherent (rambling or irrelevant
conversation, unclear or illogical flow of ideas, or unpredictable
switching from subject to subject)?
D. Altered level of consciousness - Did the patient have altered
level of consciousness as indicated by any of the following
criteria?
■ vigilant - startled easily to any sound or touch
■ lethargic - repeatedly dozed off when being asked questions, but
responded to voice or touch
■ stuporous - very difficult to arouse and keep aroused for the
interview
■ comatose - could not be aroused
Confusion Assessment Method. © 1988, 2003, Hospital Elder Life Program. All rights reserved.
Adapted from: Inouye SK et al. Ann Intern Med. 1990; 113:941-8. Used with permission.
Current use

The Short CAM data elements are currently collected in the MDS and the LCDS, and the scoring
is based on staff observations of signs and symptoms of delirium. While the Short CAM data elements are
used in both assessment tools, the response options currently differ. The current version of the LCDS
includes two response options (yes/no, indicating that the behavior is present or not present), whereas the
MDS offers three response options (behavior continuously present, does not fluctuate; behavior present,
fluctuates; behavior not present). The LCDS and MDS versions of the CAM also differ slightly in
wording and criteria for the “Altered Level of Consciousness” item.
Prior evidence supporting use of the CAM
A version of the CAM, with the addition of an item to assess psychomotor retardation, was tested
in the national MDS 3.0 test in nursing homes. Reliabilities were substantial or almost perfect. Overall

average kappa ranged from 0.85 to 0.89 and items ranged from 0.78 to 0.90 (standard kappa). 150 Based on
a meta-analysis of diagnostic accuracy in nine studies, the CAM demonstrated moderate sensitivity (82
percent, 95 percent confidence interval [CI]: 69–91 percent) and high specificity (99 percent, 95 percent
CI: 87–100 percent), respectively, using a delirium diagnosis (Diagnostic and Statistical Manual of
Mental Disorders IV) as the standard. 151
Evidence supporting use of the CAM from the National Beta Test
Assessing impairment: In the Beta testing, we administered the version of the CAM that is
currently collected in the MDS 3.0, that is, the version with three response options. The CAM was
administered at admission to 630 patients/residents in HHA, 771 in IRF, 471 in LTCH, and 1,101 in SNF
(n = 2,973 overall). Overall, 5 percent of patients/residents had evidence of mental status change from
baseline, 12 percent had difficultly focusing (3 percent continuously), 6 percent had disorganized thinking
(1 percent continuously), and 4 percent had altered consciousness (1 percent continuously). In the IRF
setting specifically, 6 percent of patients/residents had evidence of mental status change from baseline, 14
percent had difficultly focusing (3 percent continuously), 7 percent had disorganized thinking (2 percent
continuously), and 4 percent had altered consciousness (1 percent continuously). Setting-specific
frequencies for CAM data elements at admission are shown in Appendix C, Table 2.2.1.
Missing data: Overall, there were very low rates of missing data for the CAM. Across all settings,
item level missing data did not exceed 0.4 percent for any of the four CAM items. Similarly, in the IRF
setting item level missing data did not exceed 0.4 percent. For all settings, missing data rates were slightly
higher for the change in mental status from baseline item (0.4 percent missing). In general, the low rate of
missing data indicates feasibility of administration.
Time to complete: To assess feasibility of administration, time to complete was assessed for 375
patients/residents in HHA, 472 in IRF, 284 in LTCH, and 405 in SNF (n = 1,536 overall). Overall mean
time to complete the CAM was 1.4 minutes (SD = 0.7 minutes) across settings. In the IRF setting, mean
time to complete the CAM was 1.3 minutes (SD = 0.6 minutes).
Interrater reliability: The interrater reliability (IRR) was good for the CAM as measured by
kappa and percent agreement of paired raters (n = 914 paired assessments across settings; n = 245 paired
assessments in IRF). The kappa for the focusing attention item was good (0.66) across settings and
moderate in the IRF setting (0.55). Across all settings, kappa was not estimated for the other three items
within the CAM because the proportion of patients across settings with correct responses was out of range
for a stable kappa estimate. In IRFs, however, the evidence of change in mental status from baseline item
did yield a kappa of 0.60, reflecting moderate IRR. Percent agreement for the CAM across settings was
high for all four CAM items: evidence of change of mental status from baseline (96 percent), whether the
patient had difficulty focusing attention (91 percent), had disorganized thinking (94 percent), and had
altered consciousness (96 percent). Percent agreement in the IRF setting was similarly high for the four
CAM items (93 percent, 89 percent, 93 percent, and 97 percent, respectively). Please refer to Table 2.2.2
in Appendix C for kappa and percent agreement statistics for all CAM items.
Mental Status (Depressed Mood)
Depression is the most common mental health condition in older adults, yet under-recognized and
thus under-treated. Existing data show that depressed mood is relatively common in patients and residents
receiving PAC services. The PAC PRD found that about nine percent of individuals in PAC were
150 Saliba, D., & Buchanan, J. (2008). Development and validation of a revised nursing home assessment tool: MDS 3.0

151

Appendices. Santa Monica, CA: RAND Corporation. Available at: https://www.cms.gov/Medicare/QualityInitiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf..

Shi, Q., Warren, L., Saposnik, G., & MacDermid, J. C. (2013). Confusion assessment method: a systematic
review and meta-analysis of diagnostic accuracy. Neuropsychiatric Disease and Treatment, 9, 1359.

classified as having likely depression. 152 The prevalence varied from a low of seven percent of
beneficiaries in SNFs to a high of 11 percent in IRFs. 153 Almost half of nursing home residents in the
United States with an active diagnosis of depression at the time of admission are not receiving psychiatric
treatment (medication or psychological therapy) for the condition. 154
Older adults with depression may exhibit different symptoms than younger adults, including
fatigue, insomnia, irritable mood, confusion, and lack of focus. 155 Some medications and medical
conditions, such as heart disease, stroke, or cancer, may also cause depressive symptoms in older adults.26
Diagnosis and treatment of depression can lead to significant improvement of symptoms, as measured on
depression assessment scales. Depressive symptoms improve in 60 to 80 percent of elderly patients taking
an antidepressant medication. 156 Psychosocial treatments of depression in older adults have been shown
to be more effective than no treatment, based on self-rated and clinician-rated measures of depression. 157
158

Assessments of the signs and symptoms of depression help PAC providers to better understand
the needs of their patients and residents by prompting further evaluation (i.e., to establish a diagnosis of
depression); elucidating the patient’s or resident’s ability to participate in therapies for conditions other
than depression during their stay; and identifying appropriate ongoing treatment and support needs at the
time of discharge. The standardized assessment of depression among PAC patients and residents supports
clinical decision-making, early clinical intervention, person-centered care, and improved care continuity
and coordination. The use of valid and reliable standardized assessments can aid in the communication of
information within and across providers, further enabling the transfer of accurate health information.
Standardized Data Elements to Assess Depressed Mood
CMS has identified the Patient Health Questionnaire-2 to 9 (PHQ-2 to 9) data elements for
standardized assessment of depressed mood.
Patient Health Questionnaire-2 to 9 (PHQ-2 to 9)
The Patient Health Questionnaire-2 to 9 (PHQ-2 to 9) data elements use a summed item scoring
approach to first screen for signs and symptoms of depressed mood in patients and residents by assessing

152

This estimate is based on patient responses to a question about being sad in the two weeks prior to the
assessment interview in a study of patient/residents in the PAC PRD (Gage et al., 2012). If they responded
“often” or “always,” they were considered to have depression.
153 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment
reform demonstration: Final report. Research Triangle Park, NC: RTI International. 2012. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/ResearchReports-Items/PAC_Payment_Reform_Demo_Final.html.
154 Ulbricht, C. M., Rothschild, A. J., Hunnicutt, J. N., & Lapane, K. L. (2017). Depression and cognitive
impairment among newly admitted nursing home residents in the USA. International Journal of Geriatric
Psychiatry, 32(11), 1172-1181.
155 National Institute on Aging. (2011). Depression and Older Adults. Retrieved from
https://www.nia.nih.gov/health/depression-and-older-adults
156 Lebowitz, B. D., Pearson, J. L., Schneider, L. S., Reynolds, C. F., Alexopoulos, G. S., Bruce, M. L., ... &
Mossey, J. (1997). Diagnosis and treatment of depression in late life: Consensus statement update. Journal of
the American Medical Association, 278(14): 1186-1190.
157 Scogin, F., & McElreath, L. (1994). Efficacy of psychosocial treatments for geriatric depression: a quantitative
review. Journal of Consulting and Clinical Psychology, 62(1):69-74.
158 Wei, W., Sambamoorthi, U., Olfson, M., Walkup, J. T., & Crystal, S. (2005). Use of psychotherapy for
depression in older adults. American Journal of Psychiatry, 162(4), 711-717.

the two cardinal criteria for depression: depressed mood and anhedonia (inability to feel pleasure). 159 At
least one of the two must be present for a determination of probable depression, which signals the need
for continued assessment of the additional seven PHQ symptoms. The interview is concluded if a
respondent screens negative for the first two symptoms.
Relevance to IRFs
The PHQ-2 to 9 would provide valuable patient information for use in IRFs. The IRF-PAI does
not currently assess the signs and symptoms of depression, though depression is common among IRF
patients. In PAC PRD, 11.3 percent of IRF patients screened positive for depressive symptoms as
assessed by the PHQ-2, more than the other three PAC settings. 160 This highlights the importance of
screening for depressed mood among patients in IRF settings. Depressed mood may influence patient
participation in rehabilitation therapies and may affect the validity of cognitive assessments, 161 162 and
therefore has significant implications for monitoring and supporting progress toward rehabilitation goals
among IRF patients. The PHQ-2 demonstrated high reliability in IRF settings in PAC PRD testing. 163
The standardized assessment of the signs and symptoms of depression using the PHQ-2 to 9
would provide important information for care planning, care transitions, and resource use in IRFs.

159

American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. Third Edition.
Washington, DC: American Psychiatric Association; 1980.
160 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. (2012). Post-acute care
payment reform demonstration: Final report. Research Triangle Park, NC: RTI International. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/ResearchReports-Items/PAC_Payment_Reform_Demo_Final.html.
161 E.J. Lenze, M.C. Munin, M.A. Dew, et al. (2004). Adverse effects of depression and cognitive impairment on
rehabilitation participation and recovery from hip fracture. International Journal Geriatrics Psychiatry 19: 472–
478
162 Lequerica AH, Kortte K. (2010). Therapeutic engagement: a proposed model of engagement in medical
rehabilitation. American journal of physical medicine & rehabilitation 89(5):415-22.
163 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. (2012). Post-acute care
payment reform demonstration: Final report. Research Triangle Park, NC: RTI International. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/ResearchReports-Items/PAC_Payment_Reform_Demo_Final.html.

Proposed Data Elements for the Assessment of Cognitive Function: PHQ-2 to 9
D0150. Patient Mood Interview (PHQ-2 to 9)
Say to patient: “Over the last 2 weeks, have you been bothered by any of the following
problems?"
If symptom is present, enter 1 (yes) in column 1, Symptom Presence.
If yes in column 1, then ask the patient: "About how often have you been bothered by this?"
Read and show the patient a card with the symptom frequency choices. Indicate response in column 2,
Symptom Frequency.
2.
1. Symptom Presence
2. Symptom Frequency
1.
Symptom
0. No (enter 0 in column 2)
0. Never or 1 day
Symptom
Frequency
1. Yes (enter 0-3 in column 2)
1. 2-6 days (several days)
Presence
9. No response (leave column 2
blank)

2. 7-11 days (half or more of
the days)

3. 12-14 days(nearly every day)

Enter Scores in Boxes

A. Little interest or pleasure in doing things
B. Feeling down, depressed, or hopeless
If either D0150A2 or D0150B2 is coded 2 or 3, CONTINUE asking the questions below. If
not, END the PHQ interview and SKIP to next section.
C. Trouble falling or staying asleep, or sleeping too much
D. Feeling tired or having little energy
E. Poor appetite or overeating
F. Feeling bad about yourself – or that you are a failure or have let
yourself or your family down
G. Trouble concentrating on things, such as reading the newspaper
or watching television
H. Moving or speaking so slowly that other people could have
noticed. Or the opposite – being so fidgety or restless that you have
been moving around a lot more than usual
I. Thoughts that you would be better off dead, or of hurting
yourself in some way
D0160. Total Severity Score
Enter Score Add scores for all frequency responses in Column 2, Symptom Frequency. Total score must
be between 00 and 27.
Enter 99 if unable to complete interview (i.e., Symptom Frequency is blank for 3 or more
required items)

Current use
The PHQ-2 data elements are currently in use in the OASIS. The PHQ-9 data elements, which
include the two questions used in the PHQ-2 plus additional items, are in use in MDS.

Prior evidence supporting use of PHQ-2 and PHQ-9
The PHQ-2 is a brief, reliable screening tool for assessing signs and symptoms of depression.
Among studies conducted in primary care centers with large samples of adults, the PHQ-2 has performed
well as both a screening tool for identifying depression and to assess depression severity. 164 165 It has
also been shown to be sensitive to changes in a patient’s mood. Across 15 studies that assessed the
diagnostic accuracy of the PHQ-2 against a recognized gold-standard instrument for the diagnosis of
major depression in adults, sensitivity estimates (based on the summed-item approach to scoring and a
cutoff score of 3) have varied, ranging between 39 percent and 97 percent (median value = 77 percent);
specificity estimates (based on the summed-item approach to scoring and a cutoff score of three) have
been higher and more stable (kappas ranged from 0.74 to 0.91). 166 It is thus a viable option for
standardization, with the benefits of the shorter assessment counterbalancing the limitation of the lower
sensitivity.
The PHQ-9 was also tested in the national MDS 3.0 test in nursing homes. For the two presence
items in the PHQ-2 (little interest in doing things; feeling down, depressed or hopeless), kappa statistics
were almost perfect and ranged from 0.98 to 0.99. 167 The PHQ-9 was also found to have agreement with
the Modified Schedule for Affective Disorders and Schizophrenia (m-SADS), a gold-standard measure
for mood disorder, in residents without severe cognitive impairment (weighted kappa=0.69) and with the
Cornell Depression Scale, a gold-standard measure for mood disorder, in residents with severe cognitive
impairment (correlation = 0.63). 168 In addition, the Staff Time and Resource Intensity Verification
(STRIVE) study conducted in a national sample of nursing homes by CMS concluded the PHQ-9 used in
the MDS 3.0 was the “best measure” for identifying individuals with higher wage-weighted staff time,
defined as the time that nursing home staff spent caring for residents. 169
Evidence supporting use of PHQ-2 to 9 from the National Beta Test
Assessing depressed mood: The PHQ-2 to 9 was administered to assess depressed mood in Beta
testing. As a hybrid measure, the PHQ-2 to 9 uses the first two elements (PHQ-2) as a gateway item for
the longer PHQ-9. The assessor only administers the full PHQ-9 if the initial score on the PHQ-2 passes a
threshold indicating possible depression. A patient/resident who did not evidence possible depression
based on the PHQ-2 would not receive the seven additional elements contained in the PHQ-9. In Beta
testing, the PHQ-2 to 9 was administered to 646 patients/residents in the HHA setting, 786 in IRF, 496 in
LTCH, and 1,134 in SNF settings (n = 3,062 overall).
Across settings, 38 percent of patients/residents reported having little interest in doing things and
43 percent reported feeling down, depressed, or hopeless at some point in the last 14 days. Among IRF
patients, 39 percent reported having little interest in doing things, and 43 percent reported feeling down,
depressed, or hopeless. About 1 in 10 IRF patients experienced little interest in doing things and 1 in 12

165 Löwe, B., Kroenke, K., & Gräfe, K. (2005). Detecting and monitoring depression with a two-item questionnaire (PHQ-2).

Journal of Psychosomatic Research, 58(2): 163-171.

166 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform

IRF patients experienced feeling down, depressed or hopeless nearly every day over the past two weeks.
Similarly, about 1 in 10 IRF patients experienced these symptoms, on half or more of the days.
Over one in four patients/residents (28 percent) across settings passed the PHQ-2 to 9 threshold
based on one or both of these symptoms, and continued to complete the remaining seven data elements.
This positive screen rate was similar in the IRF setting (27 percent). Detailed symptom endorsement and
frequency for the PHQ – 2 to 9 is shown in Appendix C Table 3.1.1. The average PHQ-2 only score
across settings was 2.4 (SD = 1.7), and 2.3 (SD = 1.7) in the IRF setting. The average full PHQ-9 score
across settings was 11.9 (SD = 5.3), and 11.8 (SD = 5.3) in the IRF setting. The PHQ-9 has thresholds to
indicate the severity of probable depression. 170 Both across settings and in IRFs, the largest group of
patients/residents screening positive on the PHQ-2 and continuing on to complete the full PHQ-9 was
classified in the mild (31percent and 36 percent in IRF) or moderate (32 percent and 32 percent in IRF)
severity group. The mean scores and severity threshold proportions are shown in Table 3.1.1 of Appendix
C.
Missing data: Overall, there were low rates of missing data for the PHQ-2 to 9. Across all
settings, item level missing data did not exceed 5.2 percent for any of the items. Similarly, in the IRF
setting, item level missing data did not exceed 4.8 percent for any of the items. Missing data rates, overall
and in IRFs, were greatest for the moving and speaking slowly item. In general, the low rate of missing
data indicates feasibility of administration.
Time to complete: Time to complete was examined among 428 assessments in HHA, 515 in IRF,
305 in LTCH, and 479 in SNF (n = 1,727 overall). Among patients/residents who only received the PHQ2, time to complete was an average of 1.7 minutes (SD = 1.1) overall. Average time to complete the PHQ2 in the IRF setting was 1.5 minutes (SD = 0.9). Among patients receiving the full PHQ-9, time to
complete was an average of 4.0 minutes (SD = 1.2). In the IRF setting, time to complete the PHQ-9 was
3.7 minutes on average (SD = 1.2). Without regard for PHQ-2 versus PHQ-9 stratification, the mood data
elements took an average of 2.3 minutes (SD = 1.5) to complete across settings, and 2.0 minutes (SD =
1.3) in the IRF setting.
Interrater reliability: Interrater reliability was assessed for 196 patients/residents in HHA, 254 in
IRF, 231 in LTCH, and 267 in SNF (n = 948 overall). IRR for all symptom presence and frequency items
was excellent: kappas ranged from 0.95 – 1.00 for the four settings combined, and 0.87-1.00 in IRF. IRR
regarding eligibility for the full PHQ-9 based on PHQ-2 responses was nearly perfect: kappa for whether
to continue from the PHQ-2 to the full PHQ-9 was 0.98 across settings, and 0.98 in IRF. Finally, for
patients/residents who received the full PHQ-9, the IRR for sum of symptom frequencies was nearly
perfect (0.96 overall, and 0.95 in IRF).
Percent agreement was also nearly perfect, ranging from 97 percent to 100 percent overall, and 93
percent to 100 percent in IRF. For eligibility to complete the full PHQ-9, percent agreement was 99
percent across settings, and 99 percent in IRF. For the sum of symptom frequencies, percent agreement
was 95 percent across settings and 94 percent in IRF. Please refer to Table 3.1.2 in Appendix C for kappa
and percent agreement statistics for all PHQ items.

170 Kroenke, K., Spitzer, R., & Williams, J. (2001). The PHQ-9 validity of a brief depression severity measure. Journal of

General Internal Medicine, Available from. 16, 606–13.

Section 3: Special Services, Treatments, and Interventions (Including Nutritional
Approaches)
Some medical conditions require complex clinical care, consisting of special services, treatments,
and interventions. The implementation of these interventions typically indicates conditions of a more
serious nature and can be life-sustaining. Patients and residents who need them may have few clinical
alternatives. Conditions requiring the use of special services, treatments, and interventions can have a
profound effect on an individual’s health status, self-image, and quality of life. Providers should be aware
of the patient or resident’s clinical needs in order to plan the provision of these important therapies and to
ensure the continued appropriateness of care and support care transitions. The assessment of special
services, treatments, and interventions may also help to identify resource use intensity by capturing the
medical complexity of patients/residents.
Standardized Data Elements to Assess for Special Services, Treatments, and Interventions
CMS has identified data elements for cross-setting standardization of assessment for special
services, treatments, and interventions in the areas of cancer, respiratory, and other treatments, as well as
nutritional approaches and high-risk medications. The proposed data elements are:
1. Chemotherapy (IV, Oral, Other);
2. Radiation;
3. Oxygen therapy (Intermittent, Continuous, High-concentration oxygen delivery system);
4. Suctioning (Scheduled, As needed);
5. Tracheostomy Care;
6. Non-invasive Mechanical Ventilator (Bilevel Positive Airway Pressure [BiPAP]; Continuous
Positive Airway Pressure [CPAP]);
7. Invasive Mechanical Ventilator;
8. Intravenous (IV) Medications (Antibiotics, Anticoagulation, Vasoactive Medications, Other);
9. Transfusions;
10. Dialysis (Hemodialysis, Peritoneal dialysis);
11. Intravenous (IV) Access (Peripheral IV, Midline, Central line);
12. Parenteral/IV Feeding;
13. Feeding Tube;
14. Mechanically Altered Diet;
15. Therapeutic Diet; and
16. High-Risk Drug Classes: Use and Indication.
Chemotherapy (IV, Oral, Other)
Chemotherapy is a type of cancer treatment that uses medications to destroy cancer cells. Receipt
of this treatment indicates that a patient has a malignancy (cancer) and therefore has a serious, often lifethreatening or life-limiting condition. Both intravenous (IV) and oral chemotherapy have serious side
effects, including nausea/vomiting, extreme fatigue, risk of infection (due to a suppressed immune
system), anemia, and an increased risk of bleeding (due to low platelet counts). Oral chemotherapy can be
as potent as chemotherapy given by IV but can be significantly more convenient and less resourceintensive to administer. Because of the toxicity of these agents, special care must be exercised in handling
and transporting chemotherapy drugs. IV chemotherapy may be given by peripheral IV but is more
45

commonly given via an indwelling central line, which raises the risk of bloodstream infections. The need
for chemotherapy predicts resource intensity, both because of the complexity of administering these
potent, toxic drug combinations following specific protocols and because of what the need for
chemotherapy signals about the patient’s underlying medical condition. Furthermore, the resource
intensity of IV chemotherapy is higher than for oral chemotherapy, as the protocols for administration and
the care of the central line (if present) require significant resources.
Relevance to IRFs
Chemotherapy (either in general or specific routes of administration) is not assessed at present in
the IRF-PAI. Patients in the rehabilitation setting with cancer and who are receiving chemotherapy may
be different than other patients in terms of their rehabilitation stay requirements, their potential for
rehabilitation functional gains, and their risk of return to the acute care setting. In addition, these patients
may require more intensive medical care and monitoring than some other populations of patients (e.g., lab
work, nursing care). Individuals impaired by cancer or chemotherapy treatments have been shown to
make functional gains in the IRF setting. 171 172 Some cancer patients can benefit from 3 hours of therapy
per day and benefit from multi-modal types of therapy to address heterogeneous needs that can include
neurologic issues, orthopedic problems, general conditioning, pain management, and lymphedema
management. 173 174 However, cancer patients in an inpatient rehabilitation unit are at risk of transfer back
to the acute care setting at rates ranging from 17 percent to of 35 percent. 175 176 Receipt of chemotherapy
has implications for care planning, assessing functional gains, and estimating patient length of stay and
resource utilization in IRF setting.
Given the resource intensity of administering chemotherapy and the side effects and potential
complications of these highly-toxic medications, assessing whether the patient is receiving chemotherapy
would provide important information for care planning, clinical decision making, and resource use in
IRFs.

171 Marciniak, C. M., Sliwa, J. A., Spill, G., Heinemann, A. W., & Semik, P. E. (1996). Functional outcome following

rehabilitation of the cancer patient. Archives of physical medicine and rehabilitation, 77(1): 54-57. Available at
http://www.sciencedirect.com/science/article/pii/S0003999396902208
172 McKinley, W. O., Huang, M. E., & Tewksbury, M. A. (2000). Neoplastic vs. traumatic spinal cord injury: an inpatient
rehabilitation comparison. American journal of physical medicine & rehabilitation, 79(2): 138-144. Available at
http://journals.lww.com/ajpmr/Abstract/2000/03000/Neoplastic_vs__Traumatic_Spinal_Cord_Injury__An.5.aspx
173 Fialka-Moser, V., Crevenna, R., Korpan, M., & Quittan, M. (2003). CANCER REHABILITATION. Journal of rehabilitation
medicine, 35(4): 153-162. Available at https://www.medicaljournals.se/jrm/content/abstract/10.1080/16501970306129.
174 Hewitt, M., Maxwell, S., & Vargo, M. M. (2007). Policy issues related to the rehabilitation of the surgical cancer patient.
Journal of surgical oncology, 95(5): 370-385.Available at http://onlinelibrary.wiley.com/doi/10.1002/jso.20777/pdf.
175 Guo, Y., Persyn, L., Palmer, J. L., & Bruera, E. (2008). Incidence of and risk factors for transferring cancer patients from
rehabilitation to acute care units. American Journal of Physical Medicine & Rehabilitation, 87(8): 647-653.Available at
http://journals.lww.com/ajpmr/Abstract/2008/08000/Incidence_of_and_Risk_Factors_for_Transferring.6.aspx.
176 Asher, A., Roberts, P. S., Bresee, C., Zabel, G., Riggs, R. V., & Rogatko, A. (2014). Transferring inpatient rehabilitation
facility cancer patients back to acute care (TRIPBAC). PM&R, 6(9): 808-813.Available at
http://www.sciencedirect.com/science/article/pii/S1934148214000240.

Proposed Data Elements for the Assessment of Special Services, Treatments, and Interventions:
Chemotherapy
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Cancer Treatments
A1. Chemotherapy
A2. IV
A3. Oral
A10. Other

Current use
Chemotherapy is currently assessed in the MDS. It first assesses whether the resident received
chemotherapy while not a resident of the assessing facility and within the last 14 days, and then whether
the resident has received chemotherapy while a resident and within the last 14 days while a resident. The
MDS data element does not assess the route of chemotherapy.
Prior evidence supporting use of Chemotherapy (IV, Oral, Other)
An IV Chemotherapy data element was found to be feasible for cross-setting use in the PAC
PRD. 177 In nursing homes, a checkbox for chemotherapy during the last 5 days was shown to have perfect
agreement (100 percent) among rater pairs in the national MDS 3.0 test. 178
Evidence supporting use of Chemotherapy (IV, Oral, Other) from the National Beta Test
Assessing Chemotherapy: One item assessed whether Chemotherapy was performed during the
assessment period. If indicated, three follow-up items assessed whether the Chemotherapy was
administered via intravenous (IV), oral, or other route. In Beta testing, the data elements were
administered to 629 patients/residents in the HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF
settings (n = 2,926 overall). Across settings, the overwhelming majority of patients/residents (99 percent)
did not receive Chemotherapy. In the IRF setting, specifically, only three percent of patients had
Chemotherapy treatment noted. More detailed rates of Chemotherapy implementation across settings is
shown in Appendix C, Table 4.1.1.
Missing data: Overall, there were very low rates of missing responses for the Chemotherapy
items. Across all settings, missingness did not exceed 0.7 percent for each of the four items. In the IRF
setting missingness was 0.5 percent for each of the four items. The low rate of missing data indicates
feasibility of administration.

177 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform

Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554 overall). Average time to complete the Chemotherapy items
was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: The IRR was excellent for the Chemotherapy data element as measured by
percent agreement of paired raters (n = 882 paired assessments across settings; n = 236 paired
assessments in IRF). Kappas were not estimated for the Chemotherapy sub-elements because the
proportion of patients and residents receiving chemotherapy was out of range for stable kappa estimates.
Percent agreement was perfect (100 percent) for all four Chemotherapy items across settings as well as in
the IRF setting. Please refer to Table 4.1.2 in Appendix C for percent agreement statistics for the
Chemotherapy items.
Radiation
Radiation is a type of cancer treatment that uses high-energy radiation to shrink tumors and kill
cancer cells by damaging their DNA. However, it can also damage normal cells, leading to side effects
such as fatigue, skin irritation or damage, hair loss, nausea, and delayed side effects such as fibrosis (scar
tissue formation), damage to the bowels if radiation was delivered to the abdominal region, memory loss,
and infrequently, a second cancer due to radiation exposure. Radiation is a mainstay of cancer treatment;
about half to two-thirds of all patients with cancer receive radiation therapy at some point in their
treatment course. 179 180 The indications range from early-stage cancer treated with curative intent to
palliative radiation therapy, such as to treat metastatic cancer; tumors that are pressing on the spine or
growing within bones, causing severe pain; or shrinking a tumor near the esophagus, which can inhibit
swallowing. There are many types of radiation, such as external-beam radiation therapy and internal
radiation therapy (brachytherapy that is delivered from sources placed inside or on the body), and
systemic radiation therapy (in which the patient swallows or receives an injection of a radioactive
substance).
Relevance to IRFs
As noted above, individuals impaired by cancer, or its treatments, including chemotherapy or
radiation, have been shown to make functional gains in the IRF setting, and cancer patients can benefit
from intensive rehabilitation therapies. In particular, patients with brain tumors who are receiving
concurrent radiation during an IRF stay make greater functional gains compared to those who are not. 181
182 However, cancer patients in an inpatient rehabilitation unit are at risk of transfer back to the acute care
setting, at rates ranging from 17 percent to of 35 percent. 183 184 Receipt of radiation therapy has

179 Yamada, Y. Principles of Radiotherapy. In Cancer rehabilitation: principles and practice. Eds. Stubblefield Michael D. and

O’Dell Michael W. 2009.

180 National Cancer Institute. (2010). Radiation Therapy for Cancer. Available at https://www.cancer.gov/about-

cancer/treatment/types/radiation-therapy/radiation-fact-sheet

181 Marciniak, C. M., Sliwa, J. A., Heinemann, A. W., & Semik, P. E. (2001). Functional outcomes of persons with brain tumors

after inpatient rehabilitation. Archives of physical medicine and rehabilitation, 82(4): 457-463. Available at
http://www.sciencedirect.com/science/article/pii/S0003999301948294.
182 McKinley, W. O., Huang, M. E., & Tewksbury, M. A. (2000). Neoplastic vs. traumatic spinal cord injury: an inpatient
rehabilitation comparison. American Journal of Physical Medicine & Rehabilitation, 79(2): 138-144. Available at
http://journals.lww.com/ajpmr/Abstract/2000/03000/Neoplastic_vs__Traumatic_Spinal_Cord_Injury__An.5.aspx
183 Guo, Y., Persyn, L., Palmer, J. L., & Bruera, E. (2008). Incidence of and risk factors for transferring cancer patients from
rehabilitation to acute care units. American Journal of Physical Medicine & Rehabilitation, 87(8): 647-653.Available at
http://journals.lww.com/ajpmr/Abstract/2008/08000/Incidence_of_and_Risk_Factors_for_Transferring.6.aspx.
184 Asher, A., Roberts, P. S., Bresee, C., Zabel, G., Riggs, R. V., & Rogatko, A. (2014). Transferring inpatient rehabilitation
facility cancer patients back to acute care (TRIPBAC). PM&R 6(9): 808-813.Available at
http://www.sciencedirect.com/science/article/pii/S1934148214000240

implications for care planning, assessing functional gains, and estimating patient length of stay and
resource utilization in IRF setting.
Therefore, assessing whether the patient is receiving Radiation would provide important
information for care planning, clinical decision making, and resource use in IRFs.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Radiation
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Cancer Treatments
B1. Radiation

Current use
Radiation is currently assessed in the MDS. It first assesses whether the resident received
radiation while not a resident of the assessing facility and within the last 14 days, and then whether the
resident received radiation while a resident and within the last 14 days.
Prior evidence supporting use of Radiation
In nursing homes, a checkbox for radiation during the last 5 days was shown to have perfect
agreement (100 percent) among rater pairs in the national MDS 3.0 test. 185
Evidence supporting use of Radiation from the National Beta Test
Assessing Radiation: One item assessed whether Radiation was performed during the assessment
period. In Beta testing, the data element was administered to 629 patients/residents in the HHA setting,
762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926 overall). Across settings, only 3
patients/residents (1 in SNF, 2 in HHA; zero percent after rounding) received radiation. Detailed
Radiation data are shown in Appendix C, Table 4.2.1.
Missing data: Overall, there were very low rates of missing responses for the Radiation item.
Across all settings, missingness was 0.7 percent. Similarly, in the IRF setting missingness was 0.5
percent. The low rate of missing data indicates feasibility of administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the Radiation item was 0.22
minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD = 0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). Kappas are not reported for the Radiation data
element because its proportion was out of range for a stable kappa estimate. Percent agreement for the
Radiation data element was perfect, both across settings and in the IRF specifically. Please refer to Table
4.2.2 in Appendix C for percent agreement statistics for the Radiation items.
185 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.

Oxygen Therapy (Intermittent, Continuous, High-concentration oxygen delivery system)
Oxygen therapy provides a patient/resident with supplemental oxygen when medical conditions
(e.g., chronic obstructive pulmonary disease [COPD], pneumonia, severe asthma) prevent the patient or
resident from adequately oxygenating their bloodstream. Oxygen administration is a resource-intensive
intervention, as it requires specialized equipment: a reliable source of oxygen, various delivery systems
(e.g., oxygen concentrator, liquid oxygen containers, and high-pressure systems), and the patient interface
(e.g., nasal cannula, various types of masks). Accessories are also required (e.g., regulators, filters, tubing,
etc.). While the equipment is generally the same for both sub-elements of this data element (continuous
vs. intermittent), the main differences between delivering oxygen intermittently versus continuously are
the severity of the underlying illness (which often requires more hours per day of oxygen therapy), and
the bedside nursing care to set up the oxygen delivery system if the patient is unable (whether physically
or cognitively) to do so independently.
Relevance to IRFs
There are currently no items in IRF-PAI addressing oxygen use in the IRF setting. Use of oxygen
is a marker of clinical complexity and medical risk, potential for functional gains, and resource use in the
IRF setting. Stroke, spinal cord injury, brain injury, and other neurologic conditions are commonly
addressed conditions in the IRFs; a subset of patients with these conditions are at risk of dysphagia and
inability to handle oral secretions which could result in aspiration pneumonia and may require
supplemental oxygen use. When pneumonia is present as a comorbidity among IRF patients, it can be
associated with longer length of stay, lower discharge functional status ratings, and lower odds of home
discharge. 186 In addition, patients with cardiac conditions (some of whom may require oxygen therapy)
represent approximately 5 percent of IRF cases. 187 Patients’ use of oxygen therapy has important
implications for ability to participate in intensive rehabilitation therapies (3 hours per day, 5 days per
week), and ability to make functional gains over the course of rehabilitation, which may affect length of
stay. Assessing whether a patient is receiving Oxygen Therapy would provide important information for
care planning, clinical decision making, care transitions, and resource use in IRFs.

186 Ahmed, I., Graham, J. E., Karmarkar, A. M., Granger, C. V., & Ottenbacher, K. J. (2013). In-patient rehabilitation outcomes

following lower extremity fracture in patients with pneumonia. Respiratory Care 58(4): 601-606. Available at
http://rc.rcjournal.com/content/58/4/601.short
187 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9 Inpatient
rehabilitation facility services. Available at http://www.medpac.gov/-documents-/reports

Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Oxygen Therapy
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Respiratory Therapies
C1. Oxygen Therapy
C2. Continuous
C3. Intermittent
C4. High-concentration

Current use
Oxygen Therapy is currently assessed in the MDS. It first assesses whether the resident received
oxygen therapy while not a resident of the assessing facility and within the last 14 days, and then whether
the resident has received oxygen therapy while a resident and within the last 14 days. The MDS data
element does not assess the type of oxygen therapy.
Prior evidence supporting use of Oxygen Therapy (Continuous, Intermittent, High-concentration
oxygen delivery system)
A related data element on high concentration oxygen use (FiO2>40 percent) was used and found
feasible for cross-setting use in the PAC PRD. 188 In nursing homes, a checkbox for oxygen therapy
during the last five days was shown to have reliability ranging from 0.93 to 0.96 (kappas) in the national
MDS 3.0 test. 189
Evidence supporting use of Oxygen Therapy from the National Beta Test
Assessing Oxygen Therapy: One item assessed whether Oxygen Therapy was performed during
the assessment period. If indicated, three follow-up items assessed therapy type: Intermittent, Continuous,
and use of a High-concentration Delivery System. In Beta testing, the data element Oxygen Therapy
(Intermittent, Continuous, High-concentration Delivery System) was administered to 629
patients/residents in the HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926
overall). Across settings, one in five patients/residents (20 percent) received Oxygen Therapy whereas in
the IRF setting, 17 percent received Oxygen Therapy.
Across settings, the most common type of Oxygen Therapy was Intermittent Therapy (14
percent). Only 6 percent of patients/residents had Continuous Therapy, and 1 percent of patients/residents
had High-concentration Delivery System. This pattern was similar in the IRF setting as well, where
Intermittent Therapy was the most common (11 percent). Continuous Therapy (8 percent) and High188 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform

concentration Delivery (1 percent) were less common. Detailed Oxygen Therapy implementation data are
shown in Appendix C, Table 4.3.1.
Missing data: Overall, there were very low rates of missing responses for the Oxygen Therapy
items. Across all settings, missingness was less than 0.9 percent. In the IRF setting specifically,
missingness was less than 0.4 percent. The low rate of missing data indicates feasibility of administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the Oxygen Therapy data
element was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25
minutes (SD = 0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). The kappa for implementation of Oxygen Therapy
was substantial/good both overall (0.82) and in the IRF setting (0.80). Kappa for the Intermittent Therapy
sub-data element was 0.81 overall and 0.76 in the IRF setting, and kappa for the Continuous Therapy subdata element was 0.55 overall and 0.68 in the IRF setting. Kappas are not reported for the High
Concentration Therapy sub-data element because its proportions were out of range for a stable kappa
estimate. Percent agreement for the data elements was excellent/almost perfect. Across settings, percent
agreement ranged from 93 to 99 percent. Percent agreement in the IRF setting was also excellent/almost
perfect ranging from (94 to 100 percent). Please refer to Table 4.3.2 in Appendix C for kappa and percent
agreement statistics for all Oxygen Therapy items.
Suctioning (Scheduled, As Needed)
Suctioning is used to clear secretions from the airway when a person cannot clear those secretions
on his or her own due to a variety of reasons, including excess production of secretions from a pulmonary
infectious process or neurological deficits that inhibit the ability to cough, swallow, etc. It is done by
aspirating secretions through a catheter connected to a suction source.
Types of suctioning include oropharyngeal and nasopharyngeal suctioning, nasotracheal
suctioning, and suctioning through an artificial airway such as a tracheostomy tube. Oropharyngeal and
nasopharyngeal suctioning are a key part of many patients’ care plans, both to prevent the accumulation
of secretions that can lead to aspiration pneumonias (a common condition in patients with inadequate gag
reflexes) and to relieve obstructions from mucus plugging during an acute or chronic respiratory
infection, which often lead to desaturations and increased respiratory effort. Suctioning can be done on a
scheduled basis if the patient is judged to clinically benefit from regular interventions; or can be done as
needed, such as when secretions become so prominent that gurgling or choking is noted, or a sudden
desaturation occurs from a mucus plug. As suctioning is generally performed by a care provider rather
than independently, this intervention can be quite resource-intensive if it occurs every hour, for example,
rather than once a shift. It also signifies an underlying medical condition that prevents patients from
clearing their secretions effectively, which also means they need increased nursing care more generally
(such as after a stroke or during an acute respiratory infection).
Relevance to IRFs
Pneumonia and dysphagia are two conditions that may occur in the IRF setting that may
necessitate the use of suctioning of secretions. Stroke, spinal cord injury, brain injury, and other
neurologic conditions are commonly addressed conditions and qualifying conditions for IRFs; a subset of
patients with these conditions are at risk of dysphagia and inability to handle oral secretions, which could
result in aspiration pneumonia and may require suctioning. As mentioned above, pneumonia, which in
many cases could require suctioning of respiratory secretions, and when present as a comorbidity among
IRF lower extremity fracture patients, is associated with longer length of stay, lower discharge functional

status ratings, and lower odds of home discharge. 190 Additionally, pneumonia (which in some cases could
require suctioning) is a common reason for interruptions in rehabilitation programs and for short stay
transfers to an acute care setting among several classes of IRF patients (e.g., bacterial pneumonia caused
26.4 percent of preventable short-stay transfers to an acute care setting among IRF patients with traumatic
brain injury and 66.7 percent of preventable short-stay transfers to an acute setting among IRF patients
with spinal cord injury). 191 The need for suctioning may affect patients’ ability to full participate in the
intensive rehabilitation program (3 hours per day, 5 days per week) in the IRF setting. Assessing whether
Suctioning is being performed for a patient would provide important information for care planning,
clinical decision making, care transitions, and resource use in IRFs.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Suctioning
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Respiratory Therapies
D1. Suctioning
D2. Scheduled
D3. As Needed
Current use
Suctioning is currently assessed in the MDS. It first assesses whether the resident received
suctioning while not a resident of the assessing facility and within the last 14 days, and then whether the
resident received suctioning while a resident and within the last 14 days. The MDS data element does not
assess whether the suctioning is scheduled or as needed.
Prior evidence supporting use of Suctioning (Scheduled, As Needed)
In the PAC PRD, suctioning was assessed as part of the Trach Tube with Suctioning data
element, which evaluated whether patients or residents had a tracheostomy tube or needed suctioning.
This related data element was found feasible for cross-setting use in the PAC PRD. 192 In nursing homes, a

190 Ahmed, I., Graham, J. E., Karmarkar, A. M., Granger, C. V., & Ottenbacher, K. J. (2013). In-patient rehabilitation outcomes

following lower extremity fracture in patients with pneumonia. Respiratory Care 58(4): 601-606.Available at
http://rc.rcjournal.com/content/58/4/601.short
191 Middleton, A., Graham, J. E., Krishnan, S., & Ottenbacher, K. J. (2016). Program Interruptions and Short-Stay Transfers
Represent Potential Targets for Inpatient Rehabilitation Care-Improvement Efforts. American Journal of Physical Medicine &
Rehabilitation 95(11): 850-861. Available at
http://journals.lww.com/ajpmr/Fulltext/2016/11000/Program_Interruptions_and_Short_Stay_Transfers.9.aspx
192 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform
demonstration: Final report. Research Triangle Park, NC: RTI International. 2012. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.

checkbox for suctioning during the last 5 days was shown to have perfect agreement (100 percent) among
rater pairs in the national MDS 3.0 test. 193
Evidence supporting use of Suctioning (Scheduled, As Needed) from the National Beta Test
Assessing Suctioning: One item assessed whether Suctioning was provided during the assessment
period. If indicated, two follow-up items assessed therapy type: Scheduled or As Needed. In Beta testing,
the data element Suctioning (Scheduled, As Needed) was administered to 629 patients/residents in the
HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926 overall).
Across settings, most patients/residents (99 percent) did not have Suctioning noted, and those that
did noted As Needed Suctioning (1 percent). In the IRF setting, only 1 percent of patients/residents had
Suctioning indicated, all of which were noted As Needed. Detailed Suctioning findings are shown in
Appendix C, Table 4.4.1.
Missing data: Overall, there were very low rates of missing responses for the Suctioning items.
Across all settings, missingness was less than 0.9 percent. In the IRF setting specifically, missingness for
any Suctioning item was less than 0.4 percent. The low rate of missing data indicates feasibility of
administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554 overall). Average time to complete the Suctioning items was
0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). The IRR was excellent for the Suctioning data
element as measured by percent agreement of paired raters. Kappas were not estimated for the Suctioning
data element because the proportion of patients and residents receiving suctioning was out of range for
stable kappa estimates. Percent agreement for the data elements ranged from 98 to 99 percent across
settings and 99 to 100 percent in the IRF setting. Please refer to Table 4.4.2 in Appendix C for kappa and
percent agreement statistics for all Suctioning items.
Tracheostomy Care
A tracheotomy is a surgical procedure that consists of making a direct airway opening
(tracheostomy) into the trachea (windpipe). Tracheostomies are created primarily for reasons such as to
bypass an obstructed upper airway; in chronic cases, to enable the removal of secretions from the airway;
and to deliver oxygen to the patient’s lungs. For example, patients with a need for long-term ventilation
(such as those in a persistent vegetative state or those who require long-term ventilator weaning but are
alert and oriented); patients with tumors of the upper airway; patients with severe neck, mouth, or chest
wall injuries; patients with degenerative neuromuscular diseases such as amyotrophic lateral sclerosis
(ALS); patients with spinal cord injuries; and patients with airway burns are some of the examples of the
indications for a tracheostomy. Generally, in all of these cases we note that suctioning is necessary to
ensure that the tracheostomy is clear of secretions, which can inhibit successful oxygenation of the
individual. Often, individuals with tracheostomies are also receiving supplemental oxygenation. The
presence of a tracheostomy, permanent or temporary, warrants careful monitoring and immediate
intervention should the tracheostomy become occluded, or in the case of a temporary tracheostomy, the
devices used become dislodged.

193 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.

For patients with a tracheostomy, tracheostomy care, which primarily consists of cleansing,
dressing changes, and replacement of the tracheostomy cannula (tube), is a critical part of their care plans.
Regular cleansing is important to prevent infection, such as pneumonia, and to prevent any occlusions
with which there are risks for inadequate oxygenation. While in rare cases the presence of a tracheostomy
is not associated with increased care demands (and in some of those instances, the care of the
tracheostomy is performed by the patient), in general the presence of such a device is associated with
increased patient risk, and clinical care services will necessarily include close monitoring to ensure that
no life-threatening events occur because of the tracheostomy.
Relevance to IRFs
Patients with deficits in respiratory drive or in respiratory muscle strength may require prolonged
mechanical ventilation that would require a tracheostomy; such deficits may be present in patients with
stroke, traumatic brain injury, spinal cord injury, or other neurologic conditions that serve as IRF
qualifying conditions. In addition, the presence of a tracheostomy tube itself may be a marker of resource
use and functional gains among key populations of IRF patients. For example, stroke patients admitted to
IRFs with medical tubes, including tracheostomies, have been found to have longer lengths of stay, lower
admission and discharge FIM scores, and more medical complications. 194 Tracheostomy care may also
affect a patient’s capacity to participate in intensive rehabilitation therapies. As such, it is important to
assess tracheostomy care in IRF settings for purposes of care planning and determining resource use.
Assessing whether Tracheostomy Care is being performed for a patient would provide important
information for care planning, clinical decision making, care transitions, and resource use in IRFs.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Tracheostomy Care
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
E1. Tracheostomy Care
Current use
Tracheostomy Care is currently assessed in the MDS. It first assesses whether the resident
received tracheostomy care while not a resident of the assessing facility and within the last 14 days, and
then whether the resident received tracheostomy care while a resident and within the last 14 days.
Prior evidence supporting use of Tracheostomy Care
In nursing homes, a checkbox for tracheostomy care during the last 5 days was shown to have
perfect agreement (100 percent) among rater pairs in the national MDS 3.0 test. 195
Evidence supporting use of Tracheostomy Care from the National Beta Test
Assessing Tracheostomy Care: One item assessed whether Tracheostomy Care was performed
during the assessment period. In Beta testing, the data element was administered to 629 patients/residents
in the HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926 overall). Across
194 Roth, E. J., Lovell, L., Harvey, R. L., Bode, R. K., & Heinemann, A. W. (2002). Stroke rehabilitation. Stroke 33(7): 1845-

1850. Available at http://stroke.ahajournals.org/content/33/7/1845.short

195 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.

settings, 1 percent of patients received Tracheostomy Care. In the IRF setting specifically, 1 percent had
Tracheostomy Care noted. Detailed Tracheostomy Care findings across settings are shown in Appendix
C, Table 4.5.1.
Missing data: Overall, there were very low rates of missing responses for the Tracheostomy Care
item. Across all settings, missingness was 1.2 percent. Similarly, in the IRF setting missingness was 0.5
percent. The low rate of missing data indicates feasibility of administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554 overall). Average time to complete the Tracheostomy Care item
was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). The IRR was excellent for the Tracheostomy Care
data element as measured by percent agreement of paired raters. Kappa was not estimated for the
Tracheostomy Care data element because the proportion of patients and residents receiving tracheostomy
care was out of range for a stable kappa estimate. Percent agreement for the data element was 100 percent
across settings and in the IRF setting. Please refer to Table 4.5.2 in Appendix C for percent agreement
statistics for the Tracheostomy Care item.
Non-invasive Mechanical Ventilation (Bilevel Positive Airway Pressure [BiPAP], Continuous
Positive Airway Pressure [CPAP])
BiPAP and CPAP are respiratory support devices that prevent the airways from closing by
delivering slightly pressurized air through a mask continuously or via electronic cycling throughout the
breathing cycle. A BiPAP/CPAP mask provides breathing support through the provision of positive
airway pressure that prevents airways from collapsing down during the respiratory cycle. Non-invasive
mechanical ventilation differs from invasive mechanical ventilation because the interface with the patient
is a mask rather than an endotracheal tube that is passed into the windpipe. BiPAP and CPAP have a
variety of clinical indications, from obstructive sleep apnea, to acute respiratory infections, to progressive
neuromuscular decline leading to respiratory failure. The key difference between BiPAP and CPAP is that
BiPAP, as the name implies, delivers two different pressure levels, a higher pressure to support inhalation
and a lower pressure to prevent the airways from collapsing during exhalation while CPAP delivers the
same amount of positive airway pressure throughout the breathing cycle. These interventions signify
underlying medical conditions in the patient who requires their use.
Relevance to IRFs
BiPAP and CPAP use are not currently assessed in IRF-PAI. Many populations of patients
admitted to IRFs are at increased risk of sleep-disordered breathing that could require use of CPAP or
BIPAP, including stroke patients (about 21 percent of IRF patients), individuals with neurological
conditions (about 20 percent of IRF patients), and cardiac patients (about 5 percent of IRF patients). 196
For example, sleep disordered breathing has been identified as common in stroke patients, and is a riskfactor for stroke itself and stroke recurrence; treatment of stroke patients with OSA with CPAP has been

196 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9 Inpatient

rehabilitation facility services. Retrieved from http://www.medpac.gov/-documents-/reports

associated with improved functional motor outcomes. 197 198 199 200 In addition, neurological conditions
and spinal cord injuries, which are qualifying conditions for admission to an IRF, can be associated with
respiratory muscle weakness, which could require non-invasive mechanical ventilation (i.e., CPAP,
BiPAP). Noninvasive mechanical ventilation may improve outcomes in patients admitted to IRFs for
cardiac or pulmonary rehabilitation, and may improve pulmonary rehabilitation outcomes in patients with
interstitial lung disease and COPD patients. 201 202 Use of noninvasive mechanical ventilation may also
have implications for daytime energy, and patient motivation to actively participate in intensive
rehabilitation therapies in the IRF setting, as well as being a marker of clinical complexity and resource
use. As such, use of noninvasive mechanical ventilation is important to assess in IRF settings for purposes
of care planning and resource use.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Non-invasive Mechanical Ventilation
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
G1. Non-invasive Mechanical Ventilator
G2. BiPAP
G3. CPAP
Current use
Non-Invasive Mechanical Ventilation is currently assessed in the LCDS and the MDS. The LCDS
uses a checklist format, including an item asking if the patient has non-invasive ventilator (BIPAP,
CPAP) treatment at admission. The MDS first assesses whether the resident received non-invasive
mechanical ventilation while not a resident of the assessing facility and within the last 14 days, and then
whether the resident received non-invasive mechanical ventilation while a resident and within the last 14

197 Brooks, D., Davis, L., Vujovic-Zotovic, N., Boulias, C., Ismail, F., Richardson, D., & Goldstein, R. S. (2010). Sleep-

disordered breathing in patients enrolled in an inpatient stroke rehabilitation program. Archives of Physical Medicine and
Rehabilitation 91(4): 659-662. Available at http://www.sciencedirect.com/science/article/pii/S0003999310000298
198 Brown, D. L. (2006). Sleep disorders and stroke. Copyright© 2006 by Thieme Medical Publishers, Inc., 333 Seventh
Avenue, New York, NY 10001, USA. Seminars in Neurology 26(1): 117-122 Available at https://www.thiemeconnect.com/products/ejournals/html/10.1055/s-2006-933315
199 Davis, A. P., Billings, M. E., Longstreth, W. T., & Khot, S. P. (2013). Early diagnosis and treatment of obstructive sleep
apnea after stroke Are we neglecting a modifiable stroke risk factor? Neurology: Clinical Practice 3(3): 192-201. Available at
http://cp.neurology.org/content/3/3/192.short
200 Ryan, C. M., Bayley, M., Green, R., Murray, B. J., & Bradley, T. D. (2011). Influence of continuous positive airway pressure
on outcomes of rehabilitation in stroke patients with obstructive sleep apnea. Stroke, STROKEAHA-110. Available at
http://stroke.ahajournals.org/content/42/4/1062.short
201 Köhnlein, T., Schönheit-Kenn, U., Winterkamp, S., Welte, T., & Kenn, K. (2009). Noninvasive ventilation in pulmonary
rehabilitation of COPD patients. Respiratory medicine 103(9): 1329-1336. Available at
http://www.sciencedirect.com/science/article/pii/S0954611109000997
202 Dreher, M., Ekkernkamp, E., Schmoor, C., Schoenheit-Kenn, U., Winterkamp, S., & Kenn, K. (2015). Pulmonary
rehabilitation and noninvasive ventilation in patients with hypercapnic interstitial lung disease. Respiration 89(3): 208-213.
Available at https://www.ncbi.nlm.nih.gov/pubmed/25677159

days. The LCDS and MDS data elements do not assess whether the non-invasive mechanical ventilation
is BiPAP or CPAP.
Prior evidence supporting use of Non-invasive Mechanical Ventilation (BiPAP, CPAP)
A checkbox item for Non-invasive Ventilation (CPAP) was tested in the PAC PRD and was
found to be feasible for cross-setting use. 203
Evidence supporting use of Non-invasive Mechanical Ventilation (BiPAP, CPAP) from the
National Beta Test
Assessing Non-invasive Mechanical Ventilation: One item assessed whether a Non-invasive
Mechanical Ventilator was noted during the assessment period. If indicated, two follow-up items assessed
whether this Non-invasive Mechanical Ventilator was Bilevel Positive Airway Pressure (BiPAP) or
Continuous Positive Airway Pressure (CPAP). In Beta testing, the data element was administered to 629
patients/residents in the HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926
overall). Across settings overall, 5 percent of assessments noted use of a Non-invasive Mechanical
Ventilator. In the IRF setting specifically, 6 percent noted a Non-invasive Mechanical Ventilator. With
regard to specific Non-invasive Mechanical Ventilator, 2 percent of assessments across settings noted
BiPAP and 3 percent noted CPAP. In IRF, CPAP was more common (6 percent) than BiPAP (1 percent).
Detailed findings regarding non-invasive mechanical ventilator are shown in Appendix C, Table 4.7.1.
Missing data: Overall, there were very low rates of missing responses for the Non-Invasive
Mechanical Ventilator items. Across all settings, missingness was less than 1.2 percent. In the IRF setting
specifically, missingness was 0.5 percent or less. The low rate of missing data indicates feasibility of
administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554 overall). Average time to complete the Non-invasive
Mechanical Ventilator items was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF
setting was 0.25 minutes (SD = 0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). Kappas for the Non-invasive Mechanical Ventilator
items are not reported because their proportions were out of range for stable kappa estimates. Percent
agreement for the data elements ranged from 97 to 98 percent across settings and from 98 to 100 percent
in the IRF setting. Please refer to Table 4.7.2 in Appendix C for percent agreement statistics for all Noninvasive Mechanical Ventilator items across settings.
Invasive Mechanical Ventilator
Invasive mechanical ventilator includes any type of electrically or pneumatically powered closedsystem mechanical support devices, to ensure adequate ventilation of the patient who is unable to support
his or her own respiration. Patients receiving closed-system ventilation include those receiving ventilation
via a tracheostomy, as well as those patients with an endotracheal tube (e.g., nasally or orally intubated).
Depending on the patient’s underlying diagnosis, clinical condition, and prognosis, he or she may or may
not be a candidate for weaning off the ventilator. For instance, certain medical conditions such as lung
infections are expected to improve or resolve to a point where the patient can support his or her own
respiration, whereas chronic neurodegenerative diseases are likely to progress over time and therefore
preclude the patient from weaning and eventually having the tube removed.

203 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform

Ventilation in this manner is a resource-intensive therapy associated with life threatening
conditions without which the patient would not survive. However, ventilator use has inherent risks
requiring close monitoring, and failure to adequately care for the patient who is ventilator dependent can
lead to iatrogenic events such as death, pneumonia, and sepsis. Mechanical ventilation further signifies
the complexity of the patient’s underlying medical and/or surgical condition.
Relevance to IRFs
Although the frequency of patients receiving invasive mechanical ventilation varies widely across
IRF settings, IRF patients who are ventilator dependent can participate and benefit from intensive
rehabilitation programs, 204 and early initiation of rehabilitation for such patients may be associated with
improved outcomes. Invasive mechanical ventilation is associated with high daily and aggregate costs. In
a national study of mechanical ventilation use in the United States, the estimated aggregated costs were
$27 billion, 12 percent of all hospital costs. 205 Assessment of whether the patient is on Invasive
Mechanical Ventilation would provide important information for care planning, clinical decision making,
care transitions, and resource use in IRFs.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Invasive Mechanical Ventilator
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Respiratory Therapies
F1. Invasive Mechanical Ventilator (ventilator or respirator)

Current use
Invasive mechanical ventilator is currently assessed in the LCDS and MDS. The MDS first
assesses whether the resident received invasive mechanical ventilation while not a resident of the
assessing facility and within the last 14 days, and then whether the resident received invasive mechanical
ventilation while a resident and within the last 14 days. The LCDS includes an item that assesses use and
type of invasive mechanical ventilator support (e.g., weaning or non-weaning).

204 Make B, Gilmartin M, Brody JS, Snider GL. (1984). Rehabilitation of ventilator-dependent subjects with lung disease. The

concept and initial experience. Chest 86:358-65.

205 Wunsch, H., Linde-Zwirble, W. T., Angus, D. C., Hartman, M. E., Milbrandt, E. B., & Kahn, J. M. (2010). The epidemiology

of mechanical ventilation use in the United States. Critical Care Med 38(10): 1947-1953.

Prior evidence supporting use of Invasive Mechanical Ventilator
Checkbox items for ventilator (weaning and non-weaning) were tested in the PAC PRD and were
found to be feasible for cross-setting use. 206 A version of the item was tested in the MDS 3.0 National
Evaluation Study and had perfect agreement (100 percent). 207
Evidence supporting use of Invasive Mechanical Ventilator from the National Beta Test
Assessing Invasive Mechanical Ventilator: One item assessed whether an Invasive Mechanical
Ventilator was noted during the assessment period. In Beta testing, the data element was administered to
629 patients/residents in the HHA setting, 762 in IRF, 448 in LTCH, and 1087 in SNF settings (n = 2,926
overall). Across settings overall, only 13 assessments (zero percent after rounding) noted use of an
Invasive Mechanical Ventilator. One of these 13 patients was in the IRF setting (twelve were in an
LTCH). Detailed Invasive Mechanical Ventilator findings across settings are shown in Appendix C, Table
4.6.1.
Missing data: Overall, there were very low rates of missing responses for the Invasive
Mechanical Ventilator item. Across all settings, missingness was 1.2 percent for the item. In the IRF
setting specifically, missingness was 0.5 percent. The low rate of missing data indicates feasibility of
administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the Invasive Mechanical
Ventilator item was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was
0.25 minutes (SD = 0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). The IRR was excellent for the Invasive Mechanical
Ventilator data element as measured by percent agreement of paired raters. Kappa was not estimated for
the Invasive Mechanical Ventilator data element because the proportion was out of range for a stable
kappa estimate. Percent agreement for the data element was 100 percent across settings and in the IRF
setting. Please refer to Table 4.6.2 in Appendix C for percent agreement statistics for the Invasive
Mechanical Ventilator item across all settings.
IV Medications (Antibiotics, Anticoagulation, Vasoactive Medications, Other)
Intravenous (IV) medications are drugs or biologics that are administered via intravenous push
(bolus), single, intermittent, or continuous infusion through a tube placed into the vein, including one that
allows the fluids to enter the circulation through one of the larger heart vessels or more peripherally
through a vein, e.g., commonly referred to as central midline, or peripheral ports.
This data element is important to collect, as IV medications are more resource intensive to
administer than oral medications and signify a higher patient complexity (and often higher severity of
illness). The clinical indications for each of the sub-types of IV medications proposed (antibiotics,
anticoagulants, vasoactive, and other) are very different. IV antibiotics are used for severe infections
when a) the bioavailability of the oral form of the medication would be inadequate to kill the pathogen; b)
an oral form of the medication does not exist; or c) the patient is unable to take the medication by mouth.
Due to growing concern about antimicrobial resistance, antibiotic stewardship initiatives are aimed at
206 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform

increasing evidence-based antibiotic prescribing and decreasing antibiotic overuse. While data on the
antibiotic(s) would not be collected, collecting data on the use of IV antibiotics overall in the four PAC
settings would assist with monitoring the implementation of evidence-based prescribing guidelines
moving forward.
IV anticoagulants refer to anti-clotting medications (“blood thinners”) often used for the
prevention and treatment of deep vein thrombosis and other thromboembolic complications. IV
anticoagulants are commonly used in patients with limited mobility (either chronically or acutely, in the
post-operative setting), who are therefore at risk of deep vein thrombosis, or patients with certain cardiac
arrhythmias such as atrial fibrillation. When a patient is on an IV anticoagulant, they require frequent
monitoring of laboratory values to ensure appropriate anticoagulation status.
Vasoactive medications affect blood pressure and/or heart rate by causing dilation or constricting
of the blood vessels. Vasoactive medications are used to treat septic shock, cardiac arrest, and other
cardiac function issues. Continuous infusions of vasoactive medications require close observation of the
patient, including constant monitoring of blood pressure and heart rate, in order to respond quickly to any
changes.
Relevance to IRFs
IRF-PAI does not currently assess delivery of IV Medications nor subtypes thereof. Several
classes of patients with IRF-Qualifying Conditions are at risk of infections that could require intravenous
antibiotics (e.g., post-operative infections in patients admitted after a lower extremity fracture or joint
replacement; urinary tract infections among catheterized patients or those with urinary retention, which is
common among those with neurological conditions, stroke, debility, brain injury or spinal cord injury;
aspiration pneumonia among the same population of patients with neurological or debility related
conditions that could impair ability to swallow). Several groups of patients with IRF qualifying conditions
are at increased risk of venous thromboembolism (i.e., deep venous thrombosis or pulmonary embolism)
that could require initiation of intravenous anticoagulation, including those admitted after lower extremity
fracture, lower extremity joint replacement, major multiple trauma, spinal cord injury, traumatic brain
injury patients, stroke patients, and other patients whose mobility has been limited due to other neurologic
conditions. For example, incidence of DVT varies from 16.4 percent to 100 percent among stroke, spinal
cord injury, or traumatic brain injury patients not receiving prophylaxis, and incidence remains high when
prophylactic measures (e.g., pneumatic compression, compression stockings, mobilization, medication)
are used. 208 In addition, use of IV antibiotics could represent a medical complication or comorbidity that
places key classes of IRF patients at risk of a program interruption or transfer to an acute care setting. Of
preventable program interruptions among IRF patients, among the most frequent included urinary tract
infections among patients with stroke and traumatic brain injury (28.2 percent and 42.9 percent of
preventable program interruptions, respectively). Infection is among the most common admitting
diagnosis for short-stay transfers from IRFs to acute care setting for patients with stroke, traumatic brain
injury, and spinal cord injury. Thus, given the increased risk for IV medication use among patients with
IRF-qualifying conditions and its association with the interruption of rehabilitation therapies, and the fact
that it is a marker of clinical complexity and resource use, it is important to assess IV medication use in
IRFs. The standardized assessment of IV Medications, including the type of medications, would provide
important information for care planning, clinical decision making, patient safety, care transitions, and
resource use in IRFs.

208 Akman, M. N., Cetin, N., Bayramoglu, M., Isiklar, I., & Kilinc, S. (2004). Value of the D-dimer test in diagnosing deep vein

thrombosis in rehabilitation inpatients. Archives of Physical Medicine and Rehabilitation 85(7): 1091-1094. Available at

http://www.sciencedirect.com/science/article/pii/S0003999304000255

Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions: IV
Medications
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
H1. IV Medications
H2. Vasoactive medications
H3. Antibiotics
H4. Anticoagulation
H10. Other

Current use
The item IV Medications is currently assessed in the LCDS and MDS. The LCDS uses a checklist
format, including an item asking if the patient is being administered any IV Medications at admission.
The MDS first assesses whether the resident received IV medications while not a resident of the assessing
facility and within the last 14 days, and then whether the resident received IV medications while a
resident and within the last 14 days. The MDS data element does not assess the type of IV medications.
Prior evidence supporting use of IV Medications
A similar but more focused data element, IV Vasoactive Medications, was tested in the PAC PRD
and found to be feasible across PAC settings. This data element was specific to the IV administration of
vasoactive drugs (e.g., pressors, dilators, continuous medication for pulmonary edema) that increase or
decrease blood pressure and/or heart rate.
In nursing homes, a checkbox for IV medications during the last five days was shown to have
reliability of 0.95 (kappa) in the national MDS 3.0 test. 209
Evidence supporting use of IV Medications from the National Beta Test
Assessing IV Medications: One item assessed whether IV Medications were noted during the
assessment period. If indicated, three follow-up items assessed specific types of IV Medications
(Antibiotics, Anticoagulation, or Other). In Beta testing, the data element was administered to 629
patients/residents in the HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926
overall).
Across settings, one in four assessments (25 percent) had IV Medications noted. For specific
types of IV Medication, 16 percent had Antibiotics noted, 8 percent had Anticoagulation noted, and 7
percent had other IV Medications noted. In IRF, 17 percent noted IV Medications. For the specific types
of IV Medication, 8 percent had antibiotics noted, 6 percent had anticoagulation noted, and 5 percent had
other IV Medications noted. Detailed IV Medications findings across settings are shown in Appendix C,
Table 4.8.1.

209 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.

Missing data: Overall, there were very low rates of missing responses for the IV Medications
items. Across all settings, missingness was less than 0.9 percent. In the IRF setting specifically,
missingness for the IV Medication items did not exceed 0.9 percent. The low rate of missing data
indicates feasibility of administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the IV Medications items
was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). With the exception of the Anticoagulation subelement, the IRRs were fair to good for the IV Medications data element as measured by kappa and
percent agreement of paired raters. The kappa for the overarching IV Medications data element was 0.70
across settings and 0.61 in the IRF setting. The kappa for the Antibiotics sub-element was 0.88 across
settings. The kappa for the Anticoagulation sub-element was 0.13 across settings, placing it in the
“slight/poor” range. Consultation with assessors suggested that this low kappa was likely due to
inconsistent interpretation of the coding instructions, which will be improved in the future with more
comprehensive guidance. The kappa for the Other sub-element was 0.46 across settings. In the IRF
setting, kappa was not estimated for the type of IV Medication data elements because the proportions
were out of range for stable kappa estimates. Percent agreement for the data element ranged from 88 to 96
percent across settings and from 91 to 98 percent in the IRF setting. Please refer to Table 4.8.2 in
Appendix C for IRR statistics for all IV Medications items.
Transfusions
Transfusions are the administration of blood or blood products (e.g., platelets, synthetic blood
products) into the bloodstream. Blood transfusions are highly protocolized, with multiple safety checks
and monitoring required during and after the infusion to avoid adverse events. Coordination with the
facility’s blood bank is necessary, as well as documentation by clinical staff to ensure compliance with
regulatory requirements. In addition, the need for transfusions signifies underlying patient complexity that
is likely to require additional nursing staff and care coordination, and impacts planning for transitions of
care, as transfusions are not performed in all PAC settings. Receipt of transfusions is also important to
assess for case mix adjustment due to the need for added resources and to the extent that receipt of
transfusions indicates a more medically complex patient.
Relevance to IRFs
Data regarding blood transfusions are not currently collected in the IRF-PAI. Key populations of
IRF patients may benefit from blood transfusions during their rehabilitation stay. For example, patients
with fractures of the lower extremity and major joint replacements of the lower extremity are IRF
qualifying conditions and represent approximately 12 percent and approximately 8 percent of IRF cases
annually, respectively. 210 As in other settings, blood transfusions are resource-intensive, requiring
laboratory testing, coordination with the blood bank, intensive bedside nursing care and monitoring, and
can be associated with adverse reactions. Because need for and receipt of a blood transfusion can be a
marker of clinical complexity and resource use, assessment of receipt of transfusions is warranted in the
IRF setting. The standardized assessment of patients’ receipt of transfusions would provide important
information for care planning, clinical decision making, patient safety, care transitions, and resource use
in IRFs.

210 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9 Inpatient

rehabilitation facility services. Retrieved from http://www.medpac.gov/-documents-/reports

Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Transfusions

O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Other

I1. Transfusions
Current use
Transfusions are currently assessed in the MDS. It first assesses whether the resident received
transfusions while not a resident of the assessing facility and within the last 14 days, and then whether the
resident received transfusions while a resident and within the last 14 days.
Prior evidence supporting use of Transfusions
In nursing homes, a checkbox for transfusions in the past five days was shown to have reliability
of 0.67 (kappa) in the national MDS 3.0 test. 211
Evidence supporting use of Transfusions from the National Beta Test
Assessing Transfusions: One item assessed whether Transfusions were performed during the
assessment period. In Beta testing, the data element was administered to 629 patients/residents in the
HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926 overall). Across settings,
only 14 patient/resident assessments (zero percent after rounding) noted Transfusions. Five of these 14
patients (1 percent) were in the IRF setting specifically. Detailed Transfusion findings across settings are
shown in Appendix C, Table 4.9.1.
Missing data: Overall, there were very low rates of missing responses for the Transfusions item.
Across all settings, missingness was 1.0 percent for the item. In the IRF setting specifically, missingness
was 0.9 percent. The low rate of missing data indicates feasibility of administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the Transfusion item was
0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). Kappas are not reported for the Transfusions data
element because the proportion was out of range for a stable kappa estimate. Percent agreement for the
Transfusions data element was perfect overall (100 percent), and nearly perfect in the IRF specifically (99
percent). Please refer to Table 4.9.2 in Appendix C for setting-specific percent agreement statistics for the
Transfusion item.

211 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.

Dialysis (Hemodialysis, Peritoneal dialysis)
Dialysis is used primarily in the case of end stage kidney failure. It is a process by which waste,
salt, and excess water are removed from the body and key electrolytes such as sodium, potassium, and
bicarbonate are maintained at a safe level. Hemodialysis is conducted using an artificial kidney, an
external hemodialyzer, which filters the blood. During peritoneal dialysis, the dialysate is injected into the
peritoneal (abdominal) cavity, excess fluid and waste products are drawn out of the blood and into the
dialysate, and the fluid is then drained. Hemodialysis sessions are typically performed three times a week
and last up to four hours each. Peritoneal dialysis can be performed continuously overnight or
intermittently during the day.
Both forms of dialysis (hemodialysis and peritoneal dialysis) are resource intensive, not only
during the actual dialysis process but before, during, and following. Patients who need and undergo
dialysis procedures are at high risk for physiologic and hemodynamic instability from fluid shifts and
electrolyte disturbances as well as infections that can lead to sepsis. Further, patients receiving
hemodialysis are often transported to a different facility, or, at a minimum, to a different part of the
facility if the IRF is adjacent to a dialysis center or provides dialysis services onsite. Close monitoring for
fluid shifts, blood pressure abnormalities, and other adverse effects is required prior to, during, and
following each dialysis session. Nursing staff typically perform peritoneal dialysis at the bedside, and, as
with hemodialysis, close monitoring is required.
Relevance to IRFs
IRF-PAI does not presently collect data regarding receipt of dialysis, or the type thereof. In PAC
PRD 2.1 percent of IRF patients received hemodialysis. 212 There is a paucity of information about the
impact of end-stage renal disease (ESRD) and receipt of dialysis in the IRF setting. However, some
studies have found dialysis patients in IRFs to have longer lengths of stay 213 and poorer function
performance outcomes. 214 ESRD and receipt of dialysis has been found to be related to functional
outcomes in geriatric patients. For example, routine dialysis can lead to fatigue on non-dialysis days,
which may result in decreased physical activity and impede participation in therapies for some patients. 215
Finally, ESRD patients are at increased risk of amputations, which is a qualifying condition among
Medicare IRF patients (3 to 4 percent of IRF cases). 216 Dialysis is a time intensive service that requires
coordination with specialists and close monitoring of vital signs and laboratory studies and carries with it
risks of complications and infections. Accordingly, dialysis may impact patients’ ability to participate in
an intensive rehabilitation program, resource use, and functional gains, and assessment of receipt of
dialysis services in the IRF setting is warranted for resource use and care planning purposes. Assessing
Dialysis (Hemodialysis, Peritoneal dialysis) would provide important information for care planning,
clinical decision making, patient safety, care transitions, and resource use in IRFs.

212 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. (2012). Post-acute care payment

reform demonstration: Final report. Research Triangle Park, NC: RTI International. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.
213 Forrest, G. P. (2004). Inpatient rehabilitation of patients requiring hemodialysis. Archives of Physical Medicine and
Rehabilitation 85(1): 51-53. Available at http://www.sciencedirect.com/science/article/pii/S0003999303003666
214 Cowen, T. D., Huang, C. T., Lebow, J., Devivo, M. J., & Hawkins, L. N. (1995). Functional outcomes after inpatient
rehabilitation of patients with end-stage renal disease. Archives of physical medicine and rehabilitation 76(4): 355-359.
Available at http://www.sciencedirect.com/science/article/pii/S000399939580661X
215 Farragher, J., & Jassal, S. V. (2012). Rehabilitation of the geriatric dialysis patient. Blackwell Publishing Ltd. In Seminars in
Dialysis 25(6): 649-656. Available at http://onlinelibrary.wiley.com/doi/10.1111/sdi.12014/full
216 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9 Inpatient
rehabilitation facility services. Retrieved from http://www.medpac.gov/-documents-/reports

Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Dialysis

O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Other

J1. Dialysis
J2. Hemodialysis
J3. Peritoneal dialysis
Current use
The data element Dialysis is currently assessed in the LCDS and MDS. The LCDS uses a
checklist format, including an item asking if the patient receives dialysis as part of the patient’s treatment
plan. The MDS first assesses whether the resident received dialysis while not a resident of the assessing
facility and within the last 14 days, and then whether the resident received dialysis while a resident and
within the last 14 days. The LCDS and MDS data elements do not assess the type of dialysis.
Prior evidence supporting use of Dialysis (Hemodialysis, Peritoneal dialysis)
In nursing homes, a data element assessing dialysis in the past five days was tested in the national
MDS 3.0 test and shown to have almost perfect reliability (kappas of 0.91 to 0.93). 217
Evidence supporting use of Dialysis (Hemodialysis, Peritoneal dialysis) from the National Beta
Test
Assessing Dialysis: One item assessed whether Dialysis was noted during the assessment period.
If indicated, two follow-up items assessed whether the Dialysis was Hemodialysis or Peritoneal Dialysis.
In Beta testing, the data element was administered to 629 patients/residents in the HHA setting, 762 in
IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926 overall). Across settings overall, 5 percent of
assessments noted use of Dialysis. In the IRF setting specifically, Dialysis was noted for 5 percent of
patients. With regard to specific forms of Dialysis, the vast majority of noted Dialysis was Hemodialysis.
Only seven assessments overall, and three in IRF (both zero percent after rounding) indicated Peritoneal
Dialysis. Detailed findings regarding Dialysis are shown in Appendix C, Table 4.10.1.
Missing data: Overall, there were very low rates of missing responses for the Dialysis items.
Across all settings, missingness was less than1 percent. In the IRF setting specifically, missingness did
not exceed 0.9 percent. The low rate of missing data indicates feasibility of administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the Dialysis item was 0.22
minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD = 0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). The majority of kappas are not reported for the
217 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.

Dialysis data element because the proportions both overall and for each setting were out of range for a
stable kappa estimate. Percent agreement for Dialysis was nearly perfect overall and in the IRF
specifically (98 percent). The same was true for the two types of dialysis across settings (98 percent and
100 percent, respectively), and also in the IRF (98 percent and 100 percent, respectively). Please refer to
Table 4.10.2 in Appendix C for percent agreement statistics for all Dialysis items.
IV Access (Peripheral IV, Midline, Central line)
Intravenous (IV) access refers to a catheter inserted into a vein for a variety of clinical reasons,
including long-term medication treatment, hemodialysis, large volumes of blood or fluid, frequent access
for blood samples, intravenous fluid administration, total parenteral nutrition (TPN), or in some instances
the measurement of central venous pressure.
The sub-elements associated with IV Access distinguish between peripheral access and central
access. Further, different types of central access are specified. The rationale for distinguishing between a
peripheral IV and central IV access is that central lines confer higher risks associated with life threatening
events such as pulmonary embolism, infection, and bleeding. Patients with central lines, including those
peripherally inserted or who have subcutaneous central line “port” access, always require vigilant nursing
care to ensure patency of the lines and importantly to ensure that such invasive lines are free from any
potentially life-threatening events such as infection, air embolism, as well as bleeding from an open
lumen.
Relevance to IRFs
The presence of intravenous access is not currently assessed in IRF-PAI, nor are specific subtypes
of intravenous access. The need for IV access in IRFs is common: in PAC PRD, 7.2 percent of IRF
patients received central line management. 218 Presence of IV access and type is a marker of clinical
complexity (i.e., need for a medication that can be administered through the IV route and nursing care
need), and accordingly represents a marker of resource use and an important consideration for care
planning. Assessing IV Access would provide important information for care planning, clinical decision
making, patient safety, care transitions, and resource use in IRFs.

218 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. (2012). Post-acute care payment

Proposed Data Elements for the Assessment of Special Services, Treatments, and Interventions: IV
Access
O0110. Special Treatments, Procedures, and Programs
Check all of the following treatments, procedures, and programs that apply on admission.
Check all that apply
Other
O1. IV Access
O2. Peripheral
O3. Midline
O4. Central (e.g., PICC, tunneled, port)

Current use
The IV Access data element is not currently included in any of the PAC assessments.
Prior evidence supporting use of IV Access
The IV Access data element was not tested in the PAC PRD but that study did test a related data
element, Central Line Management, which was found feasible for cross-setting use.
Evidence supporting use of IV Access from the National Beta Test
Assessing IV Access: One item assessed whether Intravenous (IV) Access was noted during the
assessment period. If indicated, four follow-up items assessed whether the IV was a Peripheral Line,
Midline Catheter, Central Line, or Other form of IV. In Beta testing, the data elements were administered
to 629 patients/residents in the HHA setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n =
2,926 overall). Across settings, 24 percent of assessments noted use of IV Access. The rate in the IRF
specifically was 22 percent. For the specific type of IV Access noted, a Central Line was most common
across settings (13 percent) followed closely with Peripheral IV (11 percent). Midline Catheter (2 percent)
and Other (1 percent) were less common. In the IRF setting, a Peripheral IV was most common (14
percent), followed by a Central Line (6 percent), Other IV (2 percent), and Midline Catheter (1 percent).
Detailed findings regarding IV Access are shown in Appendix C, Table 4.11.1.
Missing data: Overall, there were very low rates of missing responses for the IV Access items.
Across all settings, missingness was less than 1.4 percent. In the IRF setting specifically, missingness was
less than 0.8 percent. The low rates of missing data indicate feasibility of administration.
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the IV Access item was
0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). IRR was excellent across settings for the IV Access
item (kappa = 0.90) as well as for type of access Peripheral (kappa=0.81) and Central (kappa=0.85).
Similarly, IRR was substantial/good in the IRF specifically for the IV Access item (0.81) and type of
access Peripheral (kappa=0.81). Percent agreement for the data element was almost perfect. Across
settings, percent agreement was 96 percent for IV Access generally, as well as the subsequent types of IV
Access (96 to 98 percent). In the IRF specifically, percent agreement was 94 percent for the general IV
68

Access item, and the subsequent types were also excellent or almost perfect (96 to 99 percent). Please
refer to Table 4.11.2 in Appendix C for kappa and percent agreement statistics for all IV Access items.
Parenteral/IV Feeding
Patients can be fed parenterally (i.e., intravenously) to bypass the usual process of eating and
digestion. The person receives nutritional formulas containing salts, glucose, amino acids, lipids, and
added vitamins. Parenteral/IV feeding is often used following surgery, when feeding by mouth or
digestive system is not possible, when a patient's digestive system cannot absorb nutrients due to chronic
disease, or if a patient's nutritional requirement cannot be met by tube feeding and supplementation. The
need for parenteral/IV feeding indicates a clinical complexity that prevents the patient from meeting
his/her nutritional needs enterally. Overall, parenteral/IV feeding is a form of nutritional support that can
be used to prevent or address malnutrition. 219 Without treatment, malnutrition can lead to a host of
negative consequences, including a decline in health, poorer physical and cognitive function, increased
use of health care services, earlier institutionalization, and increased risk of death. 220
Malnutrition is prevalent among older adults, a population commonly served in PAC settings. A
study showed that 58.3 percent of hospitalized patients diagnosed with malnutrition in the U.S. in 2010
were over age 65. 221 Additionally, as mentioned above, parenteral/IV feeding is often used to provide
nutrition for patients with specific diseases. For example, parenteral/IV feeding can be utilized for
individuals with inflammatory bowel disease, a condition which is common in older adults. 222 223 224
Parenteral/IV feeding is more resource intensive than other forms of nutrition, as it often involves
monitoring of blood chemistries and maintenance of a central line. Therefore, assessing a patient’s need
for parenteral feeding is important for care planning and case mix adjustment. In addition to the risks
associated with central and peripheral intravenous access, parenteral/IV feeding is associated with
significant risks such as embolism and sepsis.
Relevance to IRFs
Parenteral feeding is jointly assessed with tube feeding at present in the IRF-PAI and is also
assessed separately. As in other settings, parenteral nutrition indicates clinical complexity and resource
use requiring frequent blood work, central venous access, risk of infection, and more intensive nursing
care; these are important in the IRF setting for resource use and care planning. Need for parenteral or IV
feeding also indicates the nutritional status of the patient, and accordingly could be an important marker
for potential resource use and functional gains, particularly among key classes of IRF patients. For
example, patients with severe malnutrition are at higher risk for a variety of complications. 225 Among IRF
patients with stroke, an IRF qualifying condition, malnutrition (which may or may not require
219 National Collaborating Centre for Acute Care (UK). (2006). Nutrition support for adults: oral nutrition support, enteral tube

feeding and parenteral nutrition. Methods, Evidence & Guidance. Retrieved from
https://www.nice.org.uk/guidance/cg32/evidence/full-guideline-194889853
220 Evans, C. (2005). Malnutrition in the elderly: a multifactorial failure to thrive. The Permanente Journal, 9(3), 38.
221 Corkins, M. R., Guenter, P., DiMaria‐Ghalili, R. A., Jensen, G. L., Malone, A., Miller, S., ... & American Society for
Parenteral and Enteral Nutrition. (2014). Malnutrition diagnoses in hospitalized patients: United States, 2010. Journal of
Parenteral and Enteral Nutrition, 38(2), 186-195.
222 Semrad, C. E. (2012). Use of parenteral nutrition in patients with inflammatory bowel disease. Gastroenterology &
hepatology, 8(6), 393.
223 Mullady, D. K., & O'Keefe, S. J. (2006). Treatment of intestinal failure: home parenteral nutrition. Nature Reviews
Gastroenterology and Hepatology, 3(9), 492.
224 Taleban, S., Colombel, J. F., Mohler, M. J., & Fain, M. J. (2015). Inflammatory bowel disease and the elderly: a review.
Journal of Crohn's and Colitis, 9(6), 507-515.
225 Dempsey, D. T., Mullen, J. L., & Buzby, G. P. (1988). “The link between nutritional status and clinical outcome: can
nutritional intervention modify it?” American Journal of Clinical Nutrition 47(2): 352-356.

parenteral/IV feeding) has been associated with poorer rehabilitation outcomes and has been found to be
associated with length of stay and functional outcomes among stroke patients in some IRFs. 226 As
Parenteral/IV Feeding and nutritional state can be indicative of clinical complexity, resource use,
potential ability to participate in an intensive rehabilitation program, and potential for functional gains,
the standardized assessment Parenteral/IV Feeding would provide important information for IRFs.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Parenteral/IV Feeding
K0520. Nutritional Approaches
Check all of the following nutritional approaches that apply on admission.
Check all that apply
A. Parenteral/IV feeding
Current use
Different versions of the Parenteral/IV Feeding data element are currently collected in the
OASIS, IRF-PAI, LCDS, and the MDS. The OASIS data element assesses whether the patient is
receiving parenteral nutrition at home. The IRF-PAI includes a check box data element to assess total
parenteral nutrition (TPN) with a 3-day look-back period. The LCDS includes a checklist to assess
whether the patient receives TPN at admission. The MDS first assesses whether the patient received
parental/IV feeding while not a resident of the assessing facility and within the last 7 days, and then
whether the patient received parental/IV feeding while a resident and within the last 7 days.
Prior evidence supporting use of Parenteral/IV Feeding
A similar data element, the Total Parenteral Nutrition, was tested in the PAC PRD and found to
be feasible across PAC settings. Parental/IV feeding in the last five days was shown to have almost
perfect reliability (kappa of 0.95) in the national MDS 3.0 test in nursing homes. 227
Evidence supporting use of Parenteral/IV Feeding from the National Beta Test
Assessing Parenteral/IV Feeding: The Parenteral/IV Feeding data element was included in the
National Beta Test. This data element was administered to 629 patients/residents in the HHA setting, 762
in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926 overall). Across settings, only 1 percent of
assessments indicated parenteral/IV feeding. In the IRF setting, 1 percent of assessments noted
parenteral/IV feeding. Detailed Parenteral/IV Feeding implementation is shown in Appendix C, Table
5.1.1 for all four settings.
Missing data: Overall, there were very low rates of missing responses for the Parenteral/IV
Feeding data element. Across all settings, missingness was 1.3 percent. In the IRF setting specifically,
missingness was 0.8 percent. The low rates of missing data indicate feasibility of administering this data
element across PAC provider settings.

226 Finestone, H. M., Greene-Finestone, L. S., Wilson, E. S., & Teasell, R. W. (1996). Prolonged length of stay and reduced

functional improvement rate in malnourished stroke rehabilitation patients. Archives of Physical Medicine and Rehabilitation
77(4): 340-345. Available at https://www.ncbi.nlm.nih.gov/pubmed/8607756
227 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.
Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf

Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the Parenteral/IV Feeding
item was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes
(SD = 0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). Kappas are not reported for the parenteral/IV
feeding data element because its proportion was too low for a stable kappa estimate. Percent agreement
was perfect at 100 percent for the Parenteral/IV feeding data element in the four settings combined as
well as in the IRF setting specifically. Please refer to Table 5.1.2 in Appendix C for setting-specific
percent agreement statistics for the Parenteral/IV Feeding item.
Feeding Tube
The Feeding Tube data element refers to enteral nutrition, which is the delivery of a nutritionally
complete diet containing protein, carbohydrate, fat, water, minerals, and vitamins directly into the
stomach, duodenum, or jejunum. It is typically used for patients/residents who have a functional
gastrointestinal tract but are unable to maintain an adequate or safe oral intake. This data element assesses
if the patient/resident received enteral nutrition during the assessment period.
Enteral nutrition is a form of nutritional support that can be used to prevent or address
malnutrition. 228 Without treatment, malnutrition can lead to a host of negative consequences, including a
decline in health, poorer physical and cognitive function, increased use of health care services, earlier
institutionalization, and increased risk of death. 229
Malnutrition is prevalent among older adults, a population commonly served in PAC settings. A
study showed that 58.3 percent of hospitalized patients diagnosed with malnutrition in the U.S. in 2010
were over age 65. 230 Additionally, enteral nutrition can be used to provide nutrition for patients with
specific diseases. For example, tube feeding can be utilized for individuals with stroke 231 and those with
head and neck cancer, 232 conditions which are common in older adults. 233 234
Assessing use of a feeding tube can inform resource utilization, care planning, and care
transitions.
Relevance to IRFs
At present, tube feeding is jointly assessed in a single item with parenteral nutrition in IRF-PAI.
Administration of tube feeding implies nutritional needs that cannot be met by standard oral feeds, either
due to poor oral intake and inability to meet nutritional goals or due to aspiration risk. For IRF patients,
228 National Alliance for Infusion Therapy and the American Society for Parenteral and Enteral Nutrition Public Policy

Committee and Board of Directors. (2010). Disease‐related malnutrition and enteral nutrition therapy: a significant problem
with a cost‐effective solution. Nutrition in Clinical Practice, 25(5), 548-554.
229 Evans, C. (2005). Malnutrition in the elderly: a multifactorial failure to thrive. The Permanente Journal, 9(3), 38.
230 Corkins, M. R., Guenter, P., DiMaria‐Ghalili, R. A., Jensen, G. L., Malone, A., Miller, S., ... & American Society for
Parenteral and Enteral Nutrition. (2014). Malnutrition diagnoses in hospitalized patients: United States, 2010. Journal of
Parenteral and Enteral Nutrition, 38(2), 186-195.
231 Corrigan, M. L., Escuro, A. A., Celestin, J., & Kirby, D. F. (2011). Nutrition in the stroke patient. Nutrition in Clinical
Practice, 26(3), 242-252.
232 Raykher, A., Russo, L., Schattner, M., Schwartz, L., Scott, B., & Shike, M. (2007). Enteral nutrition support of head and neck
cancer patients. Nutrition in Clinical Practice, 22(1), 68-73.
233 Centers for Disease Control and Prevention (CDC. (2012). Prevalence of stroke--United States, 2006-2010. MMWR.
Morbidity and Mortality Weekly Report, 61(20), 379.
234 VanderWalde, N. A., Fleming, M., Weiss, J., & Chera, B. S. (2013). Treatment of older patients with head and neck cancer: a
review. Oncologist, 18(5):568-78.

tube feeding can imply risk of aspiration and aspiration-related complications such as pneumonia, as well
as additional equipment and nursing resources. There are specific groups of IRF patient for whom tube
feeding can serve as a proxy for risk of dysphagia, ability to fully participate in an intensive rehabilitation
program, clinical complexity, and nutritional status. As mentioned above, malnutrition, which may or
may not require tube feeding, has been associated with poorer rehabilitation outcomes among geriatric
stroke patients and has been found to be associated with length of stay and functional outcomes among
stroke patients in some IRF Settings. 235 236 Feeding tubes themselves also appear to have important
implications: Stroke patients admitted to IRFs with medical tubes, including feeding tubes, have been
found to have longer lengths of stay, lower admission and discharge FIM scores, and more medical
complications and feeding tubes have been associated with greater functional improvements over the
course of IRF stays for severe stroke patients. 237 238 Because it can be indicative of clinical complexity,
resource use, and potential functional gains, assessment of tube feeding in the IRF setting would provide
important information for care planning, care transitions, and resource use in IRFs. 239
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Feeding Tube
K0520. Nutritional Approaches
Check all of the following nutritional approaches that apply on admission.
Check all that apply
B. Feeding tube (e.g., nasogastric or abdominal (PEG))

Current use
A version of the Feeding Tube data element is currently assessed in three existing PAC
assessments. The data element Enteral Nutrition is currently collected in the OASIS, with a question
asking if the patient is receiving enteral nutrition at home. In the IRF-PAI, a Swallowing Status data
element captures some information related to enteral nutrition through the response option
“Tube/Parenteral Feeding.” The MDS data element, Feeding tube – nasogastric or abdominal (PEG), first
assesses whether a resident used a feeding tube while not a resident of the assessing facility and within the
last 7 days and then whether the resident used a feeding tube while a resident and within the last 7 days.

235.Finestone, H. M., Greene-Finestone, L. S., Wilson, E. S., & Teasell, R. W. (1996). Prolonged length of stay and reduced

functional improvement rate in malnourished stroke rehabilitation patients. Archives of Physical Medicine and Rehabilitation
77(4): 340-345. Available at https://www.ncbi.nlm.nih.gov/pubmed/8607756
236 Aptaker, R. L., Roth, E. J., Reichhardt, G., Duerden, M. E., & Levy, C. E. (1994). Serum albumin level as a predictor of
geriatric stroke rehabilitation outcome. Archives of Physical Medicine and Rehabilitation 75(1): 80-84. Available at
https://www.ncbi.nlm.nih.gov/pubmed/8291969
237 James, R., Gines, D., Menlove, A., Horn, S. D., Gassaway, J., & Smout, R. J. (2005). Nutrition support (tube feeding) as a
rehabilitation intervention. Archives of Physical Medicine and Rehabilitation 86(12): 82-92. Available at
http://www.sciencedirect.com/science/article/pii/S0003999305011421
238 Roth, E. J., Lovell, L., Harvey, R. L., Bode, R. K., & Heinemann, A. W. (2002). Stroke rehabilitation. Stroke 33(7): 18451850. Available at http://stroke.ahajournals.org/content/33/7/1845.short
239 Dempsey, D. T., Mullen, J. L., & Buzby, G. P. (1988). “The link between nutritional status and clinical outcome: can
nutritional intervention modify it?” American Journal of Clinical Nutrition 47(2): 352-356.

Prior evidence supporting use of Feeding Tube
In the national MDS 3.0 test in nursing homes, the Feeding Tube data element, collected for the
last five days, was shown to have almost perfect reliability (kappa of 0.89). 240
Evidence supporting use of Feeding Tube from the National Beta Test
Assessing Feeding Tube: The Feeding Tube data element was included in the National Beta Test.
This data element was administered to 629 patients/residents in the HHA setting, 762 in IRF, 448 in
LTCH, and 1,087 in SNF settings (n = 2,926 overall). Across settings, 3 percent of assessments indicated
use of a Feeding Tube. In the IRF setting, 3 percent of assessments noted use of a Feeding Tube. Detailed
Feeding Tube implementation is shown in Appendix C, Table 5.2.1 for all four settings.
Missing data: There were very low rates of missing data for the Feeding Tube data element both
overall (1.3 percent) and in the IRF setting (0.8 percent).
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554 overall). Average time to complete the Feeding Tube item was
0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). Kappas are not reported for the Feeding Tube data
element because its proportion was too low for a stable kappa estimate. Percent agreement was 100
percent across settings and in the IRF setting. Please refer to Table 5.2.2 in Appendix C for settingspecific percent agreement statistics for the Feeding Tube item.
Mechanically Altered Diet
A mechanically altered diet is one that is specifically prepared to alter the texture or consistency
of food to facilitate oral intake. Examples include soft solids, puréed foods, ground meat, and thickened
liquids. A mechanically altered diet should not automatically be considered a therapeutic diet.
The provision of a mechanically altered diet is resource intensive, as it signifies difficulty
swallowing/eating safely (dysphagia). Often, nurses are required to slowly feed patients meals consisting
of a mechanically altered diet rather than having them eat independently. Dysphagia is frequently
associated with various health conditions, including: nervous system-related diseases (e.g., cerebral palsy
and Parkinson’s disease); stroke; head injury; head, neck, and esophagus cancers; head, neck, and chest
injuries; and dementia. 241 In the absence of treatment, swallowing disorders can lead to malnutrition,
dehydration, aspiration pneumonia, poor overall health, chronic lung disease, choking, and death. 242 In
addition, other consequences can include lack of interest and enjoyment related to eating or drinking, and
embarrassment or isolation tied to social situations involving eating. 243
The prevalence of dysphagia is high in older adults, a population commonly served in PAC
settings. A study of a geriatric population living independently found that the lifetime prevalence of a

240 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf
241 National Institute on Deafness and Other Communication Disorders. (2017). Dysphagia. Retrieved from
https://www.nidcd.nih.gov/health/dysphagia
242 American Speech-Language-Hearing Association. (Undated). Adult Dysphagia. Retrieved from
https://www.asha.org/PRPSpecificTopic.aspx?folderid=8589942550§ion=Overview
243 American Speech-Language-Hearing Association. (Undated). Adult Dysphagia. Retrieved from
https://www.asha.org/PRPSpecificTopic.aspx?folderid=8589942550§ion=Overview

swallowing disorder was 38 percent, and current prevalence of a swallowing disorder was 33 percent. 244
Additionally, increasing age has been shown to be associated with a higher likelihood of swallowing
problems in the previous year. 245 Beyond general aging effects on swallowing physiology, age-related
disease is the main risk factor for dysphagia in older adults. 246 Stroke and dementia are examples of two
conditions that are common among the elderly that may contribute to issues with swallowing. 247
Furthermore, discharge to a PAC setting is more likely among those with dysphagia. A study
examining burden among inpatients diagnosed with dysphagia found that individuals with dysphagia had
a 33.2 percent higher likelihood of being discharged to a PAC facility compared to patients without
dysphagia. 248
Assessing whether a patient requires a mechanically altered diet is important in ensuring patient
safety and can inform care planning, care transitions, and resource utilization.
Relevance to IRFs
Patients with severe malnutrition are at higher risk for a variety of complications. 249 Use of a
mechanically altered diet or supervision is currently assessed in the IRF-PAI. Mechanically altered diets
are particularly relevant for many common populations of IRF patients including those with strokes,
neurologic conditions, and brain injuries, which account for 19.5 percent, 13.1 percent and 8.7 percent of
patients in IRFs in 2014 and are IRF qualifying conditions. 250 Owing to neurological changes, these
patients may be at risk of aspiration and related complications, and as such many benefit from the use of a
mechanically altered diet with thickened liquids or pureed solids. As mechanically altered diets are a
marker of dysphagia, they are a marker both of clinical complexity, complication risk, and resource use
among key groups of IRF patients. Dysphagia commonly affects stroke patients, with rates varying
widely in the literature from 37 percent to 78 percent depending upon the setting and screening instrument
used. 251 This is a risk for malnutrition, which has been found to be common among stroke patients and
associated with worse functional outcomes and more complications. 252 Dysphagia is also common among
patients with traumatic brain injury, with an incidence as high as 93 percent among traumatic brain injury
patients admitted to rehabilitation. 253 Many other neurologic disorders, for which patients may be
admitted to an IRF, may feature dysphagia that may benefit from a mechanically altered diet. 254 Because
244 Roy, N., Stemple, J., Merrill, R. M., & Thomas, L. (2007). Dysphagia in the elderly: preliminary evidence of prevalence, risk

factors, and socioemotional effects. Annals of Otology, Rhinology & Laryngology, 116(11), 858-865.

245 Bhattacharyya, N. (2014). The prevalence of dysphagia among adults in the United States. Otolaryngology--Head and Neck

Surgery, 151(5), 765-769.

246 Sura, L., Madhavan, A., Carnaby, G., & Crary, M. A. (2012). Dysphagia in the elderly: management and nutritional

considerations. Clinical Interventions in Aging, 7, 287.

247 Ibid.

248 Patel, D. A., Krishnaswami, S., Steger, E., Conover, E., Vaezi, M. F., Ciucci, M. R., & Francis, D. O. (2017). Economic and

survival burden of dysphagia among inpatients in the United States. Diseases of the Esophagus, 31(1), dox131.

249 Dempsey, D. T., Mullen, J. L., & Buzby, G. P. (1988). “The link between nutritional status and clinical outcome: can

nutritional intervention modify it?” American Journal of Clinical Nutrition 47(2): 352-356.

250 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9 Inpatient

rehabilitation facility services. Retrieved from http://www.medpac.gov/-documents-/reports

251 Martino, R., Foley, N., Bhogal, S., Diamant, N., Speechley, M., & Teasell, R. (2005). Dysphagia after stroke. stroke 36(12):

2756-2763. Available at http://stroke.ahajournals.org/content/36/12/2756.

252 Finestone, H. M., & Greene-Finestone, L. S. (2003). Rehabilitation medicine: 2. Diagnosis of dysphagia and its nutritional

management for stroke patients. Canadian Medical Association Journal 169(10): 1041-1044. Available at
http://www.cmaj.ca/content/169/10/1041.full
253 Hansen, T. S., Engberg, A. W., & Larsen, K. (2008). Functional oral intake and time to reach unrestricted dieting for patients
with traumatic brain injury. Archives of Physical Medicine and Rehabilitation 89(8): 1556-1562. Available at
http://www.sciencedirect.com/science/article/pii/S0003999308003237
254 Buchholz, D. W. (1993). Dysphagia associated with neurological disorders. Acta oto-rhino-laryngologica belgica 48(2): 143155. Available at https://www.ncbi.nlm.nih.gov/pubmed/8209677

it can be a marker of clinical complexity, resource use, and can be related to potential for functional
rehabilitation gains, assessing whether an IRF patient requires a mechanically altered diet would provide
important information for care planning, care transitions, patient safety, and resource use in IRF.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Mechanically Altered Diet
K0520. Nutritional Approaches
Check all of the following nutritional approaches that apply on admission.
Check all that apply
C. Mechanically altered diet - require change in texture of food or liquids (e.g., pureed
food, thickened liquids)
Current use
Mechanically Altered Diet is currently assessed in the MDS. It first assesses whether the resident
received a mechanically altered diet while not a resident and within the last 7 days, and then whether the
resident received a mechanically altered diet while a resident and within the last 7 days.
Prior evidence supporting use of Mechanically Altered Diet
In the national MDS 3.0 test in nursing homes, the Mechanically Altered Diet data element was
shown to have almost perfect reliability (kappas from 0.90 to 0.96). 255
Evidence supporting use of Mechanically Altered Diet from the National Beta Test
Assessing Mechanically Altered Diet: The Mechanically Altered Diet data element was included
in the National Beta Test. The data element was administered to 629 patients/residents in the HHA
setting, 762 in IRF, 448 in LTCH, and 1,087 in SNF settings (n = 2,926 overall). Across settings, 10
percent of assessments indicated Mechanically Altered Diet. In the IRF setting, 15 percent of assessments
noted Mechanically Altered Diet. Detailed Mechanically Altered Diet implementation is shown in
Appendix C, Table 5.3.1 for all four settings.
Missing data: There were very low rates of missing data for the Mechanically Altered Diet data
element both overall (1.2 percent) and in the IRF setting (0.7 percent).
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554, overall). Average time to complete the Mechanically Altered
Diet item was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25
minutes (SD = 0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). IRR for Mechanically Altered Diet data element
was substantial/good across settings (0.65) and moderate in the IRF specifically (0.53). Percent agreement
for the data element was 93 percent across settings and 89 percent in the IRF setting. Please refer to Table
5.3.2 in Appendix C for setting-specific kappa and percent agreement statistics for the Mechanically
Altered Diet item.

255 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf

Therapeutic Diet
A therapeutic diet is a diet intervention ordered by a health care practitioner as part of the
treatment for a disease or clinical condition manifesting an altered nutritional status, to eliminate,
decrease, or increase certain substances in the diet (e.g., sodium or potassium). Therapeutic diets can
include low cholesterol, renal, diabetic, and low salt diets, 256 the latter of which are most commonly
used. 257
Certain conditions, including diabetes, 258 chronic kidney disease, 259 hypertension, 260 and heart
disease 261 are highly prevalent among older adults who may receive services in a PAC setting. For
example, the percentage of adults with diabetes is 25.2 percent among individuals 65 years of age or
older. 262 Additionally, 61.7 percent of adults aged 65 or older have hypertension. 263 These conditions
may be treated utilizing a therapeutic diet.
The Therapeutic Diet data element is important to collect in the IRF setting in order to distinguish
therapeutic diet from various other nutritional approaches. It is less resource intensive from the bedside
nursing perspective but does signify one or more underlying clinical conditions that preclude the patient
from eating a regular diet. The communication among PAC settings of whether a patient is receiving a
particular therapeutic diet is critical to ensure safe transitions of care.
Relevance to IRFs
Therapeutic diets are not currently assessed in IRF-PAI and data are lacking regarding the
prevalence of therapeutic diets in the IRF setting. However, therapeutic diets are part of the treatment and
lifestyle changes required for patients with chronic conditions, which are common in IRF populations. In
2013 and 2014, more than 5 percent of IRF cases were for cardiac conditions, 264 many of which require
therapeutic diets (e.g., fluid restriction, low-fat, low sodium) for successful management of that condition
while the patient undergoes rehabilitation services. Similarly, diabetes, a condition which requires a
carbohydrate controlled therapeutic diet, has been found to affect 23 percent of patients in IRFs after hip
fracture and result in longer lengths of stay, lower functional status ratings, and reduced odds of discharge

256 Kamel, H. K., Louis, S., Malekgoudarzi, B., & Pahlavan, M. (2000). Inappropriate use of therapeutic diets in the nursing

home. Journal of the American Geriatrics Society, 48(7), 856-857.

257 Crogan, N. L., Corbett, C. F., & Short, R. A. (2002). The minimum data set: predicting malnutrition in newly admitted

nursing home residents. Clinical Nursing Research, 11(3), 341-353.

258 Centers for Disease Control and Prevention. (2017). National Diabetes Statistics Report, 2017: Estimates of Diabetes and Its

Burden in the United States. Retrieved from https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statisticsreport.pdf
259 Centers for Disease Control and Prevention. Chronic Kidney Disease Issue Brief. Retrieved from
https://www.cdc.gov/diabetes/pdfs/programs/CKDBrief.pdf
260 Fang, J., Gillespie, C., Ayala, C., Loustalot, F. (2018). Prevalence of Self-Reported Hypertension and Antihypertensive
Medication Use Among Adults Aged ≥18 Years — United States, 2011–2015. MMWR Morbidity and Mortality Weekly
Report, 67, 219–224.
261 Centers for Disease Control and Prevention. (2017). National Center for Health Statistics: Older Persons’ Health. Retrieved
from https://www.cdc.gov/nchs/fastats/older-american-health.htm
262 Centers for Disease Control and Prevention. (2017). National Diabetes Statistics Report, 2017: Estimates of Diabetes and Its
Burden in the United States. Retrieved from https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statisticsreport.pdf
263 Fang J, Gillespie C, Ayala C, Loustalot F. (2018). Prevalence of Self-Reported Hypertension and Antihypertensive
Medication Use Among Adults Aged ≥18 Years — United States, 2011–2015. MMWR Morbidity and Mortality Weekly
Report, 67, 219–224.
264 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9 Inpatient
rehabilitation facility services. Retrieved from http://www.medpac.gov/-documents-/reports

home. 265 Diabetes has also been shown to affect 20 to 22 percent of IRF knee replacement patients and 28
percent of stroke patients. 266 267 Fractures of the lower extremity, major joint replacements of the lower
extremity, and stroke accounted for 12.2 percent, 7.8 percent, and 19.5 percent of IRF cases in 2014,
respectively. 268 As therapeutic diets may be a common requirement of many key IRF populations and
may be a marker of clinical complexity, the standardized assessment of Therapeutic Diets is warranted in
the IRF setting.
Proposed Data Element for the Assessment of Special Services, Treatments, and Interventions:
Therapeutic Diet
K0520. Nutritional Approaches
Check all of the following nutritional approaches that apply on admission.
Check all that apply
D. Therapeutic diet (e.g., low salt, diabetic, low cholesterol)
Current use
Therapeutic Diet is currently assessed in the MDS. It first assesses whether the resident received a
therapeutic diet while not a resident and within the last 7 days, and then whether the resident received a
therapeutic diet while a resident and within the last 7 days.
Prior evidence supporting use of Therapeutic Diet
In the national MDS 3.0 test in nursing homes, the Therapeutic Diet data element was shown to
have substantial to almost perfect reliability (kappas from 0.89 to 0.93). 269
Evidence supporting use of Therapeutic Diet from the National Beta Test
Assessing Therapeutic Diet: The Therapeutic Diet data element was included in the National Beta
Test. This data element was administered to 629 patients/residents in the HHA setting, 762 in IRF, 448 in
LTCH, and 1,087 in SNF settings (n = 2,926 overall).
Across settings, over half of assessments (52 percent) indicated Therapeutic Diet. In the IRF
setting, 49 percent of assessments noted Therapeutic Diet. Detailed Therapeutic Diet implementation is
shown in Appendix C, Table 5.4.1 for all four settings.

265Reistetter, T. A., Graham, J. E., Deutsch, A., Markello, S. J., Granger, C. V., & Ottenbacher, K. J. (2011). Diabetes

comorbidity and age influence rehabilitation outcomes after hip fracture. Diabetes Care 34(6): 1375-1377. Available at
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3114361
266 DeJong, G., Hsieh, C. H., Gassaway, J., Horn, S. D., Smout, R. J., Putman, K., ... & Foley, M. P. (2009). Characterizing
rehabilitation services for patients with knee and hip replacement in skilled nursing facilities and inpatient rehabilitation
facilities. Archives of Physical Medicine and Rehabilitation 90(8): 1269-1283. Available at
http://www.sciencedirect.com/science/article/pii/S0003999309003025
267 Roth, E. J., Lovell, L., Harvey, R. L., Heinemann, A. W., Semik, P., & Diaz, S. (2001). Incidence of and risk factors for
medical complications during stroke rehabilitation. Stroke 32(2): 523-529. Available at
http://stroke.ahajournals.org/content/32/2/523.short
268 Medicare Payment Advisory Commission. (2016). Report to the Congress: Medicare Payment Policy. Chapter 9 Inpatient
rehabilitation facility services. Available at http://www.medpac.gov/-documents-/reports
269 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.
Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf

Missing data: There were low levels of missing data for the Therapeutic Diet data element both in
the four settings combined (0.6 percent) and in the IRF setting specifically (0.8 percent).
Time to complete: Time to complete was examined among 422 assessments in HHA, 457 in IRF,
244 in LTCH, and 431 in SNF (n = 1,554 overall). Average time to complete the Therapeutic Diet item
was 0.22 minutes overall (SD = 0.1). Average time to complete in the IRF setting was 0.25 minutes (SD =
0.1).
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 236 in
IRF, 203 in LTCH, and 256 in SNF (n = 882 overall). The kappa for the Therapeutic Diet data element
was substantial/good at 0.60 across settings and substantial/good (0.70) in the IRF setting. Percent
agreement for the data element was 80 percent across settings and 85 percent in the IRF setting. Please
refer to Table 5.4.2 in Appendix C for setting-specific kappa and percent agreement statistics for the
Therapeutic Diet item.
High-Risk Drug Classes: Use and Indication
Most patients receiving PAC services depend on short- and long-term medications to manage
their medical conditions. However, medications are a leading cause of adverse events. A study by the
U.S. Department of Health and Human Services found that 31 percent of adverse events that occurred in
2008 among hospitalized Medicare beneficiaries were related to medication. 270 Adverse drug events may
be caused by medication errors such as drug omissions, errors in dosage, and errors in dosing
frequency. 271 In addition, approximately half of all hospital-related medication errors and 20 percent of
ADEs occur during transitions within, admission to, transfer to, or discharge from a hospital. 272 273 274
ADEs are more common among older adults, who make up most patients receiving PAC services. The
rate of emergency department visits for ADEs is three times higher among adults 65 years of age and
older compared to that among those younger than age 65. 275
Some classes of drugs are associated with more risk than others.276 The six medication classes
proposed as response options in the High-Risk Drug Classes: Use and Indication data element are
anticoagulants; antiplatelets; hypoglycemics (including insulin); opioids; antipsychotics; and antibiotics.
These drug classes are considered high-risk due to the adverse effects that may result from use. In
particular, anticoagulants and antiplatelets are associated with bleeding risk; 277 278 hypoglycemics are

270 U.S. Department of Health and Human Services. Office of Inspector General. Daniel R. Levinson. Adverse Events in

Hospitals: National Incidence Among Medicare Beneficiaries. OEI-06-09-00090. November 2010.

271 Boockvar, K. S., Liu, S., Goldstein, N., Nebeker, J., Siu, A., & Fried, T. (2009). Prescribing discrepancies likely to cause

adverse drug events after patient transfer. Quality and Safety in Health Care, 18(1):32–6.

272 Barnsteiner, J. H. (2005). Medication reconciliation: transfer of medication information across settings-keeping it free from

error. Journal of Infusion Nursing, 28(2 Suppl):31-36.

273 Rozich, J., Roger, R. (2001). Medication safety: one organization’s approach to the challenge. Journal of Clinical Outcomes

Management, 2001(8):27-34.

274 Gleason, K. M., Groszek, J. M., Sullivan, C., Rooney, D., Barnard, C., & Noskin, G. A. (2004). Reconciliation of

discrepancies in medication histories and admission orders of newly hospitalized patients. American Journal of HealthSystem Pharmacy, 61(16): 1689-1695
275 Shehab, N., Lovegrove, M. C., Geller, A. I., Rose, K. O., Weidle, N. J., & Budnitz, D. S. (2016). US emergency department
visits for outpatient adverse drug events, 2013–2014. Journal of the American Medical Association,316:2115–25.
276 Ibid.
277 Shoeb, M., & Fang, M. C. (2013). Assessing bleeding risk in patients taking anticoagulants. Journal of Thrombosis and
Thrombolysis, 35(3):312–319.
278 Melkonian, M., Jarzebowski, W., & Pautas, E. (2017). Bleeding risk of antiplatelet drugs compared with oral anticoagulants
in older patients with atrial fibrillation: a systematic review and meta‐analysis. Journal of Thrombosis and Haemostasis,
15:1500–1510.

associated with fluid retention, heart failure, and lactic acidosis; 279 opioids are associated with misuse; 280
antipsychotics are associated with fractures and strokes; 281 282 and antimicrobials, the category of
medications that includes antibiotics, are associated with various adverse events such as central nervous
systems effects and gastrointestinal intolerance. 283 Moreover, some medications in the six drug classes in
this group of data elements are included in the 2019 Updated Beers Criteria® list as potentially
inappropriate medications for use in older adults. 284 Although a complete medication list should record
several important attributes of each medication (e.g., dosage, route, stop date), recording an indication for
the drug is of crucial importance. 285
Relevance to IRFs
Many patients treated in the IRF setting have one or more conditions that require treatment with a
medication in a high-risk drug class. In a nationally representative sample of Medicare beneficiaries in
IRFs in 2012, almost five percent of Medicare patients in IRFs experienced some type of medicationrelated adverse event over a one-month period, ranging in severity from a longer IRF stay to contributing
to death. 286 In the same study, over eight percent of patients in IRFs experienced a medication-related
“temporary harm event” during the one-month period, defined as requiring medical intervention, but not
causing lasting harm. 287 Of all adverse and temporary harm events identified in IRFs, 46 percent were
related to medication. 288 The top three categories of adverse or temporary harm events related to
medications in IRFs were delirium and other changes in mental status due to medication, hypoglycemic
events related to medication, and hypotension secondary to medication. 289
Assessing use of high-risk medications by IRF patients and indications for each medication
would provide important information related to patient safety in IRFs and care transitions between IRFs
and other settings. The IRF-PAI does not currently contain data elements that document the use of any
medication or the indication or reason for the patient taking the medication. The standardized assessment
of high-risk medication use and ensuring that indications are noted in the medical record are important
steps toward overall medication safety within and between PAC provider settings.

279 Hamnvik, O. P., & McMahon, G. T. (2009). Balancing Risk and Benefit with Oral Hypoglycemic Drugs. The Mount Sinai

Journal of Medicine, New York. 76:234–243.

280 Naples, J. G., Gellad, W. F., Hanlon, J. T. (2016). The Role of Opioid Analgesics in Geriatric Pain Management. Clinics in

Geriatric Medicine, 32(4):725-735.

281 Rigler, S. K., Shireman, T. I., Cook-Wiens, G. J., Ellerbeck, E. F., Whittle, J. C., Mehr, D. R., & Mahnken, J. D. (2013).

Fracture risk in nursing home residents initiating antipsychotic medications. Journal of the American Geriatrics Society,
61(5):715–722.
282 Wang, S., Linkletter, C., Dore, D. et al. (2012). Age, antipsychotics, and the risk of ischemic stroke in the Veterans Health
Administration. Stroke, 43:28–31.
283 Faulkner, C. M., Cox, H. L., & Williamson, J. C. (2005). Unique aspects of antimicrobial use in older adults. Clinical
Infectious Diseases, 40(7):997–1004.
284 American Geriatrics Society 2019 Beers Criteria Update Expert Panel. American Geriatrics Society 2019: Updated Beers
Criteria for Potentially Inappropriate Medication Use in Older Adults. Journal of the American Geriatrics Society, 00:1-21.
285 Li, Y., Salmasian, H., Harpaz, R., Chase, H., & Friedman, C. (2011). Determining the reasons for medication prescriptions in
the EHR using knowledge and natural language processing. AMIA Annual Symposium Proceedings, 768-76.
286 Levinson, D. R., “Adverse events in rehabilitation hospitals: national incidence among Medicare beneficiaries.” Washington,
DC: U.S. Department of Health and Human Services, Office of the Inspector General, July 2016. Available at:
https://oig.hhs.gov/oei/reports/oei-06-14-00110.pdf.
287 Ibid.
288 Ibid.
289 Ibid.

Proposed Data Element for the Assessment of High-risk Drug Classes: Use and Indication
N0415. High-Risk Drug Classes: Use and Indication
1. Is taking
Check if the patient is taking any medications in the
following drug classes
2. Indication noted
If Column 1 is checked, check if there is an indication
noted for all medications in the drug class

1. Is taking

2. Indication noted

Check all that apply Check all that apply
↓
↓

A. Antipsychotic
E. Anticoagulant
F. Antibiotic
H. Opioid
I. Antiplatelet
J. Hypoglycemic (including insulin)
Current use
Medication classes received is currently assessed in the MDS. The number of days the resident
received the following medications by pharmacological classification during the last seven days is
assessed for: antipsychotic, antianxiety, antidepressant, hypnotic, anticoagulant, antibiotic, diuretic, and
opioid medications.
Prior evidence supporting use of High-Risk Drug Classes: Use and Indication
The High-Risk Drug Classes: Use and Indication data element was not tested in prior
demonstration efforts. However, the use of similar data elements in the MDS 3.0 speak to the feasibility
of collecting data on patient medications in a standardized assessment.
Evidence supporting use of High-Risk Drug Classes: Use and Indication from the National Beta
Test
Assessing High-Risk Drug Classes: Use and Indication: As part of the assessment of the
Medication Reconciliation process, the National Beta Test included a data element that assesses whether
the patient/resident was taking any medications in each of the six high-risk drug classes, and for each
medication, whether there was a corresponding indication noted. The six classes include: anticoagulants,
antiplatelets (excluding low-dose aspirin), hypoglycemics (including insulin), opioids, antipsychotics, and
antimicrobials (excluding topicals). In Beta testing, the data element was administered to 627
patients/residents in the HHA setting, 769 in IRF, 459 in LTCH, and 1,096 in SNF settings (n = 2,951
overall).
In the four settings combined, the percent of patients/residents taking medications in each of the
six classes ranged from 12 percent (Antipsychotics) to 51 percent (Opioids). In the IRF setting, these
percentages ranged from 9 percent (Antipsychotics) to 61 percent (Anticoagulants). The presence of
indications for noted medications in the various classes ranged from 45 percent (Anticoagulants and
80

Antiplatelets) to 92 percent (Opioids) in the four settings combined, and in the IRF setting the indication
percentages ranged from 29 percent (Anticoagulants) to 91 percent (Opioids). The overall and settingspecific findings for each high-risk drug class are detailed in Table 6.1.1 in Appendix C.
Missing data: There were very low rates of missing responses for the medication use items. In the
four settings combined, missingness rates did not exceed 4.2 percent for any of the six drug class items.
Similarly, in the IRF setting missingness rates did not exceed 3.9 percent for the six drug class items.
Missing data was also very low for indication items. Missingness rates did not exceed 1.2 percent in the
four settings combined and did not exceed 2.1 percent in the IRF setting. In general, the low rate of
missing data indicates feasibility of administration.
Time to complete: Time to complete was examined among 406 assessments in the HHA setting,
446 in the IRF setting, 271 in the LTCH setting, and 421 in the SNF setting (n = 1,544 overall). Average
time to complete the high-risk drug classes: use and indication items was approximately 1.0 minute (SD =
0.6 minutes) in the four settings combined and 1.1 minutes (SD = 0.6 minutes) in the IRF setting.
Interrater reliability: Interrater reliability was examined for 187 assessments in HHA, 240 in
IRF, 212 in LTCH, and 261 in SNF (n = 900 overall). Kappas were not estimated within or across settings
for items assessing antipsychotic use and indication of opioids because the proportions were out of range
for stable kappa estimates.
In the four settings combined, IRRs across settings ranged from substantial/good to
excellent/almost perfect (kappas = 0.72 to 0.89) for medication use items. In the IRF setting, kappas for
medication use were also substantial/good to excellent/almost perfect (kappas = 0.71 to 0.86). For
indication items, kappas ranged from substantial/good to excellent/almost perfect, both in the four settings
combined (kappa = 0.65 to 0.87) and in the IRF setting (0.62 to 1.00).
Percent agreement was very high for the medication use items, both in the four settings combined
(92 to 95 percent) and in the IRF setting (91 to 95 percent). Similarly, percent agreement was generally
high for indication items, both in the four settings combined (82 to 94 percent) and in the IRF setting (81
to 100 percent). More detailed IRR statistics are shown in Appendix C, Table 6.1.2.

Section 4: Medical Conditions and Co-Morbidities
Pain Interference
Pain is a highly prevalent medical condition in the United States. A CDC analysis of 2016
National Health Interview Study data found that 8 percent of Americans report high-impact chronic pain,
that is, pain that limits life or work activities on most days or every day in the past 6 months. 290 Pain in
older adults occurs in conjunction with many acute and chronic conditions, such as osteoarthritis, leg pain
during the night, cancer and associated treatment, neuralgia from diabetes mellitus, infections such as
herpes zoster/shingles, and peripheral vascular disease. 291 Conditions causing pain in older adults may be
associated with depression, 292 sleep disturbance, 293 294 and lower participation in rehabilitation activities. 295
296 297

A substantial percentage of older adults receiving services in a PAC setting experience pain.
Based on assessment testing performed in the PAC PRD, more than half of patients in the PAC settings
reported having experienced “pain or hurting at any time during the last two days”, with 55 percent in
LTCHs, 65 percent in SNFs, 68 percent in IRFs, and 70 percent of patients receiving HHA services
responding “yes” to this question. 298 Based on the 2009 Medicare Current Beneficiary Survey, the
prevalence of moderate-to-severe pain 299 among residents of skilled and non-skilled nursing facilities was
22 percent, while the prevalence of persistent pain – defined as the same or worse pain over time – was 65
percent. 300
Pain in older adults can be treated with medications, complementary and alternative approaches,
or physical therapy. 301 Treatment of pain in older adults may be complicated by issues such as dementia,
high rates of polypharmacy, end-of-life care, and patient expectations, attitudes, and fears related to pain
290 Dahlhamer J, Lucas J, Zelaya, C, et al. Prevalence of Chronic Pain and High-Impact Chronic Pain Among Adults — United

States, 2016. MMWR Morb Mortal Wkly Rep 2018;67:1001–1006. DOI: http://dx.doi.org/10.15585/mmwr.mm6736a2

291 American Geriatrics Society Panel on Pharmacological Management of Persistent Pain in Older Persons. (2009.)

Pharmacological management of persistent pain in older persons. Journal of the American Geriatrics Society, 57(8), 1331.

292 Sullivan-Singh, S. J., Sawyer, K., Ehde, D. M., Bell, K. R., Temkin, N., Dikmen, S., ... & Hoffman, J. M. (2014).

Comorbidity of pain and depression among persons with traumatic brain injury. Archives of Physical Medicine and
Rehabilitation, 95(6), 1100-1105.
293 Eslami, V., Zimmerman, M. E., Grewal, T., Katz, M., & Lipton, R. B. (2016). Pain grade and sleep disturbance in older
adults: evaluation the role of pain, and stress for depressed and non‐depressed individuals. International Journal of Geriatric
Psychiatry, 31(5), 450-457.
294 Blytt, K. M., Bjorvatn, B., Husebo, B., & Flo, E. (2018). Effects of pain treatment on sleep in nursing home patients with
dementia and depression: A multicenter placebo‐controlled randomized clinical trial. International Journal of Geriatric
Psychiatry, 33(4), 663-670.
295 Chin, R. P. H., Ho, C. H., & Cheung, L. P. C. (2013). Scheduled analgesic regimen improves rehabilitation after hip fracture
surgery. Clinical Orthopaedics and Related Research®, 471(7), 2349-2360.
296 Brenner, I. & Marsella, A. (2008). Factors influencing exercise participation by clients in long-term care. Perspectives (Pre2012), 32(4), 5.
297 Zanca, J. M., Dijkers, M. P., Hammond, F. M., & Horn, S. D. (2013). Pain and its impact on inpatient rehabilitation for acute
traumatic spinal cord injury: analysis of observational data collected in the SCIRehab study. Archives of Physical Medicine
and Rehabilitation, 94(4), S137-S144.
298 Gage, B. (2016). Data from the PAC PRD study, 2008-2010 [data file]. Available from Barbara Gage, August 16, 2016.
299 In this study, pain was measured based on two MDS items that assess pain frequency and intensity, with “moderate
pain…defined as having daily mild to moderate pain” and “severe pain … as having daily pain at times horrible or
excruciating”.
300 Shen, X., Zuckerman, I. H., Palmer, J. B., & Stuart, B. (2015). Trends in prevalence for moderate-to-severe pain and
persistent pain among Medicare beneficiaries in nursing homes, 2006–2009. Journals of Gerontology Series A: Biomedical
Sciences and Medical Sciences, 70(5), 598-603.
301 National Institute on Aging. (2018, February 28). Pain: You Can Get Help. Retrieved from
https://www.nia.nih.gov/health/pain-you-can-get-help

treatment. 302 Untreated pain is an often-debilitating condition that is associated with a host of adverse
physical consequences, including loss of function, poor quality of life, disruption of sleep and appetite,
inactivity, and weakness, as well as psychological effects such as depression, anxiety, fear, and anger. 303
304

Relevance to IRFs
Many patients in the IRF setting report having pain and experiencing it often. From the 2018
National Beta Test, 79 percent of patients in the IRF setting reported having “pain or hurting.” Of those
who reported pain, 64 percent experienced pain “frequently” or “almost constantly.”
Pain among IRF patients can interfere with rehabilitation and has potential secondary
complications. The potential effects of pain on patient health are myriad, and it is critical to assess pain
during hospitalization and post discharge. Assessing pain in IRF patients during their stay can lead to
appropriate treatment and improved quality of life, can reduce complications associated with immobility
such as skin breakdown and infection, and can facilitate rehabilitation efforts and the ability to return to
community settings. Pain assessment post-discharge can also be used to plan appropriate treatment and may
potentially reduce readmissions.
Proposed Data Elements for Assessment of Pain Interference
J0510. Pain Effect on Sleep
Enter Code

Ask patient: “Over the past 5 days, how much of the time has pain made it hard for you
to sleep at night?”
0. Does not apply – I have not had any pain or hurting in the past 5 days  Skip to
XXXX
1. Rarely or not at all
2. Occasionally
3. Frequently
4. Almost Constantly
9. Unable to answer

J0520. Pain Interference with Therapy Activities
Enter Code

Ask patient: “Over the past 5 days, how often have you limited your participation in
rehabilitation therapy sessions due to pain?”
0. Does not apply – I have not received rehabilitation therapy in the past 5 days
1. Rarely or not at all
2. Occasionally
3. Frequently
4. Almost Constantly
9. Unable to answer

J0530. Pain Interference with Day-to-Day Activities
302 Molton, I. R., & Terrill, A. L. (2014). Overview of persistent pain in older adults. American Psychologist, 69(2), 197.

303 Institute of Medicine (IOM). (2011). Relieving Pain in America: A Blueprint for Transforming Prevention, Care, Education,

and Research. Washington, DC, The National Academies Press.

304 American Geriatrics Society Panel on Pharmacological Management of Persistent Pain in Older Persons. (2009).

Pharmacological management of persistent pain in older persons. Journal of the American Geriatrics Society, 57(8), 1331.

Enter Code

Ask patient: “Over the past 5 days, how often have you limited your day-to-day
activities (excluding rehabilitation therapy sessions) because of pain?”
1. Rarely or not at all
2. Occasionally
3. Frequently
4. Almost Constantly
9. Unable to answer

Current use
Data elements on the topic of pain are currently assessed in OASIS and MDS. The OASIS
assesses the frequency of pain interfering with patient’s activity or movement. A pain assessment
interview is included in MDS and includes questions on whether pain has made it hard for the resident to
sleep at night and if pain has limited day-to-day activities.
Prior evidence supporting use of Pain Interference data elements
Two interview-based data elements, pain effect on sleep and pain effect on activities, were
included in the PAC PRD testing of interrater reliability, and. showed strong interrater reliability
(weighted kappas of 0.836 and 0.789, respectively). 305
In a national test to develop and validate the MDS 3.0, two items (pain made it hard to sleep, pain
limited day-to-day activities) were validated for measuring the effect of pain on function. 306
Evidence supporting use of Pain from the National Beta Test
Assessing Pain: In Beta testing, three pain interference data elements were assessed, including
Effect of Pain on Sleep, Pain Interference with Rehabilitation Therapies (If Applicable), and Pain
Interference with Daily Activities. A total of 489 patients/residents in HHA, 618 in IRF, 375 in LTCH,
and 872 in SNF settings (n = 2,354 overall) reported experiencing any pain and were administered the
three pain interference items. Setting-specific frequencies are shown in Appendix C, Table 7.1.1.
Across settings, among the 78 percent of patients/residents who reported experiencing any pain,
pain interfered with sleep more often than rarely for two of three patients/residents (65 percent); 37
percent of patients/residents with pain had pain that made it difficult to sleep “frequently” or “almost
constantly.” In the IRF setting, among the 79 percent of patients who reported experiencing any pain, pain
interfered with sleep more than rarely for two of three patients (68 percent); 39 percent of patients with
pain in the IRF experienced pain that interfered with sleep “frequently” or “almost constantly.”
Among the patients/residents who reported experiencing any pain, most had been offered
rehabilitation therapies (e.g., physical therapy, occupational therapy, speech therapy), both across settings
(89 percent) and in the IRF (98 percent). Among these patients/residents overall, 73 percent reported that
pain rarely interfered with rehabilitation. Within the IRF setting, 76 percent of these patient of IRF
patients who had pain and were offered therapy reported that pain rarely interfered with rehabilitation;

305 Gage, B., Smith, L., Ross, J., Coots, L., Kline, T., Shamsuddin, K., , ... & Gage-Croll, Z. (2012). The Development and

Testing of the Continuity Assessment Record and Evaluation (CARE) Item Set: Final Report on Reliability Testing. Volume
2 of 3. Research Triangle Park, NC: RTI International. Retrieved from https://www.cms.gov/Medicare/Quality-InitiativesPatient-Assessment-Instruments/Post-Acute-Care-Quality-Initiatives/Downloads/The-Development-and-Testing-of-theContinuity-Assessment-Record-and-Evaluation-CARE-Item-Set-Final-Report-on-Reliability-Testing-Volume-2-of-3.pdf
306 Saliba, D., & Buchanan, J. (2008). Development and validation of a revised nursing home assessment tool: MDS 3.0. Santa
Monica, CA: RAND Corporation. 2008. Available at
http://www.cms.hhs.gov/NursingHomeQualityInits/Downloads/MDS30FinalReport.pdf

about one in fourteen (7 percent) had pain that interfered with therapy “frequently” or “almost
constantly.”
Among those who reported experiencing any pain, fifty-five percent of patients/residents across
settings reported limiting their daily activities (not including rehabilitation) more often than “rarely or not
at all.” About one in three of these patients/residents (33 percent) had pain that limited activities
“frequently” or “almost constantly.” In the IRF setting, 45 percent of patients with pain had pain that
interfered more often than rarely. About one of four IRF patients with pain (27 percent) had limited
activities “frequently” or “almost constantly” due to pain.
Fifty-five percent of patients/residents across settings reported limiting their daily activities (not
including rehabilitation) more often than “rarely or not at all.” About one in three patients/residents (33
percent) had pain that limited activities “frequently” or “almost constantly.” In the IRF setting, 45 percent
of patients had pain that interfered more often than rarely. About one of four IRF patients (27 percent)
had limited activities “frequently” or “almost constantly.”
Missing data: Overall, there were low rates of missing data for pain data elements. Across all
settings, missing data did not exceed 2.4 percent for any data element. Similarly, in the IRF setting,
missing data did not exceed 2.6 percent for any data element. In general, the low rate of missing data
indicates feasibility of administration.
Time to complete: The length of time to administer the pain data elements was examined as
another indicator of feasibility among 440 patients/residents in HHA, 533 in IRF, 321 in LTCH, and 483
in SNF (n = 1,777 overall). Across settings, the average time to complete the three interference items was
1.3 minutes (SD = 0.6). In the IRF setting, time to complete was similar at 1.2 minutes (SD = 0.5).
Interrater reliability: IRR was assessed for 197 patients/residents in HHA, 256 in IRF, 232 in
LTCH, and 268 in SNF (n = 953 overall). IRR statistics were generally excellent/perfect, indicating high
levels of agreement in responses to the data elements across assessment staff. For the pain interference
data elements across settings, kappas were excellent/almost perfect, with values of either 0.97 or 0.98.
The same was true in the IRF setting, where excellent/almost perfect kappas ranged from 0.96 to 0.98.
Percent agreement was similarly high, with nearly perfect or perfect agreement, 98 percent for all items in
the four settings combined and in IRF specifically. More detailed IRR statistics are shown in Appendix C,
Table 7.1.2.

Section 5: Impairments
Hearing and Vision Impairments
Hearing and vision impairments are common conditions that, if unaddressed, affect patients’ and
residents’ activities of daily living, communication, physical functioning, rehabilitation outcomes, and
overall quality of life. Sensory limitations can lead to confusion in new settings, increase isolation,
contribute to mood disorders, and impede accurate assessment of other medical conditions such as
cognition. Hearing impairments may cause difficulty in communication of important information
concerning the patient’s or resident’s condition, preferences, and care transitions; vision impairments
have been associated with increased risk of falls. Both types of impairment can also interfere with
comprehension of and adherence to discharge plans. Onset of hearing and vision impairments can be
gradual, so accurate screening tools and follow-up evaluations are essential to determining which patients
and residents need hearing- or vision-specific medical attention or assistive devices, and to ensuring that
person-directed care plans are developed to accommodate a patient or resident’s needs during post-acute
care and at discharge.
Assessments pertaining to sensory status aids PAC providers in better understanding the needs of
their patients and residents by establishing a diagnosis of hearing or vision impairment, elucidating the
patient or resident’s ability and willingness to participate in treatments or use assistive devices during

their stay, and identifying appropriate ongoing therapy and support needs at the time of discharge. The
standardized assessment of vision impairment among PAC patients and residents supports clinical
decision-making, early clinical intervention, person-centered care, and improved care continuity and
coordination. The use of valid and reliable standardized assessments can aid in the communication of
information within and across providers, further enabling the transfer of accurate health information.
Standardized Data Elements to Assess Hearing and Vision Impairments
CMS has identified two data elements for cross-setting standardized assessment of hearing and vision
impairment. The proposed data elements are:
1. Hearing
2. Vision
Hearing
Hearing impairment is one of the most common complaints in adults over the age of 60 and is a
major contributor to difficulties in speech comprehension. 307 Causes of hearing loss can include noise,
earwax or fluid buildup, a punctured ear drum, viruses and bacteria, certain health conditions (e.g., stroke,
cardiac conditions, and brain injury), medications, heredity, and aging. 308 Age-related hearing loss is
caused by presbycusis and occurs gradually over time as an individual ages. It is typically hereditary and
usually affects both ears. Hearing impairment in older adults has been associated with a myriad of
outcomes, 309 including falls, 310 dementia, 311 312 313 cognitive impairment, 314 anxiety, 315 emotional
vitality, 316 and various medical conditions (e.g., arthritis, cancer, cardiovascular disease, diabetes,
emphysema, high blood pressure, and stroke). 317

307

Peelle, J. E., Troiani, V., Grossman, M., & Wingfield, A. (2011). “Hearing loss in older adults affects neural
systems supporting speech comprehension.” Journal of Neuroscience 31(35): 12638-12643.
308 National Institute on Aging. (2018). Hearing Loss: A Common Problem for Older Adults. Retrieved from
https://www.nia.nih.gov/health/hearing-loss-common-problem-older-adults
309 Contrera, K. J., Wallhagen, M. I., Mamo, S. K., Oh, E. S., & Lin, F. R. (2016). Hearing loss health care for older
adults. The Journal of the American Board of Family Medicine, 29(3), 394-403.
310 Jiam, N. T. L., Li, C., & Agrawal, Y. (2016). Hearing loss and falls: A systematic review and meta‐analysis. The
Laryngoscope, 126(11), 2587-2596.
311 Thomson, R. S., Auduong, P., Miller, A. T., & Gurgel, R. K. (2017). Hearing loss as a risk factor for dementia: a
systematic review. Laryngoscope Investigative Otolaryngology, 2(2), 69-79.
312 Deal, J. A., Betz, J., Yaffe, K., Harris, T., Purchase-Helzner, E., Satterfield, S., ... & Health ABC Study Group.
(2016). Hearing impairment and incident dementia and cognitive decline in older adults: the health ABC study.
Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences, 72(5), 703-709.
313 Wei, J., Hu, Y., Zhang, L., Hao, Q., Yang, R., Lu, H., ... & Chandrasekar, E. K. (2017). Hearing Impairment,
Mild Cognitive Impairment, and Dementia: A Meta-Analysis of Cohort Studies. Dementia and Geriatric
Cognitive Disorders Extra, 7(3), 440-452.
314 Wei, J., Hu, Y., Zhang, L., Hao, Q., Yang, R., Lu, H., ... & Chandrasekar, E. K. (2017). Hearing Impairment,
Mild Cognitive Impairment, and Dementia: A Meta-Analysis of Cohort Studies. Dementia and Geriatric
Cognitive Disorders Extra, 7(3), 440-452.
315 Contrera, K. J., Betz, J., Deal, J., Choi, J. S., Ayonayon, H. N., Harris, T., ... & Rubin, S. M. (2017). Association
of hearing impairment and anxiety in older adults. Journal of Aging and Health, 29(1), 172-184.
316 Contrera, K. J., Betz, J., Deal, J. A., Choi, J. S., Ayonayon, H. N., Harris, T., ... & Rubin, S. M. (2016). Association of

hearing impairment and emotional vitality in older adults. Journals of Gerontology Series B: Psychological Sciences and
Social Sciences, 71(3), 400-404.
317 McKee, M. M., Stransky, M. L., & Reichard, A. (2018). Hearing loss and associated medical conditions among individuals
65 years and older. Disability and Health Journal, 11(1), 122-125.

A high proportion of older adults receiving services in a PAC setting experience hearing
impairment. About 51 percent of nursing facility patients and residents are estimated to have moderate to
severe hearing impairment. 318 Data from the PAC PRD suggest that severe hearing impairment affects 1
to 2 percent of Medicare FFS beneficiaries in the four types of PAC. 319 320 Among older adults more
generally, reports on the prevalence of hearing loss vary. The National Institute on Deafness and Other
Communication Disorders (NIDCD) has stated that one third of people between ages 65 and 74 have
hearing loss and roughly half of those older than 75 are hearing-impaired. 321 Additionally, a study found
that two thirds of individuals aged 70 years or older have bilateral hearing loss and approximately three
quarters have hearing loss in at least one ear. 322
Assessing hearing impairment is critical to improving patient outcomes, safety, and quality of
life. In addition, assessment can inform future care planning and care transitions.
Relevance to IRFs
The IRF-PAI does not currently include Hearing or any comparable hearing impairment
assessment items. In PAC PRD testing, 1.1 percent of IRF patients demonstrated severely impaired
hearing. 323 Hearing impairments can impact the effectiveness of patient communication with providers,
which has implications for patient understanding of and adherence to treatment plans and rehabilitation
goals. Hearing impairments are also correlated with lower functional status and lower performance on
measures of cognitive functioning in older adults, 324 325 which has implications for monitoring patient
progress toward goals for some IRF patients and may also affect participation in some intensive
rehabilitation therapies (e.g., speech and language therapies, cognitive rehabilitation). Assessing Hearing
would provide important information for communication, ensuring safety, care planning, care transitions,
and resource use in IRFs.

318 Garahan, M. B., Waller, J. A., Houghton, M., Tisdale, W. A., & Runge, C. F. (1992). “Hearing loss prevalence and

management in nursing home residents.” Journal of the American Geriatrics Society 40(2): 130-134.

319 Hearing impairments were classified into categories from mildly impaired to severely impaired. The percentages reported

here refer to severe impairment of hearing, defined as “Absence of useful hearing” (Gage et al., 2012).

320 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform

demonstration: Final report. Research Triangle Park, NC: RTI International. 2012. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.
321 National Institute on Deafness and Other Communication Disorders. (2018). Hearing Loss and Older Adults. Retrieved from
https://www.nidcd.nih.gov/health/hearing-loss-older-adults
322 Goman, A. M., & Lin, F. R. (2016). Prevalence of hearing loss by severity in the United States. American Journal of Public
Health, 106(10), 1820-1822.
323 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. (2012). Post-acute care payment
reform demonstration: Final report. Research Triangle Park, NC: RTI International. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.
324 Lin FR, Ferrucci L, Metter EJ, An Y, Zonderman AB, Resnick SM. (2011). Hearing loss and cognition in the Baltimore
Longitudinal Study of Aging. Neuropsychology 25(6):763.
325 Keller BK, Morton JL, Thomas VS, Potter JF. (1999). The effect of visual and hearing impairments on functional status.
Journal of the American Geriatrics Society 1;47(11):1319-25.

Proposed Data Element for the Assessment of Impairments: Hearing
B0200. Hearing
Enter Code Ability to hear (with hearing aid or hearing appliances if normally used)
0. Adequate – no difficulty in normal conversation, social interaction, listening to TV
1. Minimal difficulty – difficulty in some environments (e.g., when person speaks softly or
setting is noisy)
2. Moderate difficulty – speaker has to increase volume and speak distinctly
3. Highly impaired – absence of useful hearing
Current use
The Hearing data element is currently collected in the MDS, and is assessed with the use of a
hearing aid, if applicable.
Prior evidence supporting use of Hearing
The Hearing data element tested in the PAC PRD includes one question regarding hearing ability,
which showed high reliability across PAC settings (unweighted kappa = 0.78). The MDS 3.0 version of
the Hearing data element also had almost perfect agreement in the MDS 3.0 national test in nursing
homes (weighted kappas = 0.94 and 0.89). 326
Evidence supporting use of Hearing from the National Beta Test
Assessing Hearing: In the Beta testing, a Hearing assessment item (with hearing aids, when
applicable) was administered to 643 patients/residents in HHA, 783 in IRF, 498 in LTCH, and 1141 in
SNF (n = 3,065 overall). Overall, 74 percent of patients/residents had adequate hearing, 17 percent had
minimal difficulty hearing, 8 percent had moderate difficulty hearing, and 1 percent were highly
impaired. In the IRF setting, 75 percent of patients had adequate hearing, 18 percent had minimal
difficulty hearing, 6 percent had moderate difficulty hearing, and 1 percent were highly impaired. See
Appendix C, Table 8.1.1 for setting-specific response frequencies for the Hearing data element.
Missing data: There were very low rates of missing responses for the Hearing data element both
overall (0.3 percent) and in the IRF setting (0.4 percent), indicating feasibility of administration.
Time to complete: Time to complete was assessed among 396 patients/residents in HHA, 499 in
IRF, 301 in LTCH, and 456 in SNF (n = 1,652 overall). Across all settings, the mean time to complete the
Hearing item was 0.3 minutes (SD=0.2 minutes). Likewise, in the IRF setting, mean time to complete the
Hearing item was 0.3 minutes (SD=0.2 minutes).
Interrater reliability: Interrater reliability (IRR) was assessed for the Hearing item for 197
patients/residents in HHA, 258 in IRF, 237 in LTCH, and 268 in SNF (n = 960 overall). Across all
settings, kappa for the Hearing item was substantial/good (0.65). In the IRF setting, kappa for the Hearing
item was substantial/good (0.67). Percent agreement was high for the Hearing item both across settings
(84 percent) and in the IRF setting (87 percent). More detailed IRR statistics are shown in Appendix C,
Table 8.1.2.

326 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf

Vision
Visual impairment can be caused not only by age-related diseases (e.g., age-related macular
degeneration [AMD], cataract, glaucoma, and diabetic retinopathy) but also due to nearsightedness,
farsightedness, loss of near vision with age, and/or untreated disease. 327 In addition to conditions
affecting the eye itself, visual deficits can also be caused by other conditions such as stroke and traumatic
brain injury. Visual impairment in older adults has been associated with depression and anxiety, 328 lower
cognitive function, 329 and poorer quality of life. 330
The PAC PRD study found that between 1 and 3 percent of Medicare FFS beneficiaries among
the four types of PAC providers had the most extreme category of visual impairment assessed, having
“No vision or object identification questionable.” 331 While the majority of patients and residents in the
PAC settings do not exhibit severely impaired vision, visual impairment affects a substantial proportion
of older adults and is predicted to increase substantially over time. A study examining visual impairment
among adults in the United States found that in 2015, among the 3.22 million persons in the US who were
visually impaired, the largest proportions comprised those in older age categories: aged 80 years and older
(50 percent), 70–79 years (24 percent), and 60–69 years (16 percent). 332 By 2050, the proportion of adults
with visual impairment will increase to 64 percent among individuals aged 80 years and older. 333
Assessing visual impairment is critical to improving patient outcomes, safety, and quality of life.
Additionally, assessment can inform future care planning and care transitions.
Relevance to IRFs
The IRF-PAI does not currently assess vision impairment. In PAC PRD testing, 1.7 percent of
IRF patients demonstrated severely impaired vision, 334 and this was associated with poorer outcomes
with respect to change in self-care and mobility. 335 Additionally, assessment of this information is useful
for ensuring safety in the IRF setting, as impaired vision increases the risk of falls. 336 337 Visual

327 Cimarolli, V. R., Boerner, K., Brennan-Ing, M., Reinhardt, J. P., & Horowitz, A. (2012). “Challenges faced by older adults

with vision loss: a qualitative study with implications for rehabilitation.” Clinical Rehabilitation 26(8): 748-757.

328 Heesterbeek, T. J., van der Aa, H. P., van Rens, G. H., Twisk, J. W., & van Nispen, R. M. (2017). The incidence and

predictors of depressive and anxiety symptoms in older adults with vision impairment: a longitudinal prospective cohort
study. Ophthalmic and Physiological Optics, 37(4), 385-398.
329 Chen, S. P., Bhattacharya, J., & Pershing, S. (2017). Association of vision loss with cognition in older adults. JAMA
Ophthalmology, 135(9), 963-970.
330 Tseng, Y. C., Liu, S. H. Y., Lou, M. F., & Huang, G. S. (2018). Quality of life in older adults with sensory impairments: a
systematic review. Quality of Life Research, 1-15.
331 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform
demonstration: Final report. Research Triangle Park, NC: RTI International. 2012. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.
332 Varma, R., Vajaranant, T. S., Burkemper, B., Wu, S., Torres, M., Hsu, C., ... & McKean-Cowdin, R. (2016). Visual
impairment and blindness in adults in the United States: demographic and geographic variations from 2015 to 2050. JAMA
Ophthalmology, 134(7), 802-809.
333 Varma, R., Vajaranant, T. S., Burkemper, B., Wu, S., Torres, M., Hsu, C., ... & McKean-Cowdin, R. (2016). Visual
impairment and blindness in adults in the United States: demographic and geographic variations from 2015 to 2050. JAMA
Ophthalmology, 134(7), 802-809.
334 Ibid
335 Ibid
336 Ivers, R. Q., Norton, R., Cumming, R. G., Butler, M., & Campbell, A. J. (2000). Visual impairment and risk of hip fracture.
American Journal of Epidemiology 152(7): 633-639.
337 Freeman, E. E., Munoz, B., Rubin, G., West, S. K. (2007). Visual field loss increases the risk of falls in older adults: the
Salisbury eye evaluation. Investigative Ophthalmology & Visual Science 48(10): 4445-4450.

impairments are also associated with poorer rehabilitation outcomes among older IRF patients. 338 Visual
impairments may also affect patients’ participation in some rehabilitation therapies and/or ability to
complete cognitive assessment tools (e.g., performance on visual-motor tasks). Assessing Vision would
provide important information for patient safety, communication, care planning, care transitions, and
resource use in IRFs.
Proposed Data Element for the Assessment of Impairments: Vision
B1000. Vision
Enter Code Ability to see in adequate light (with glasses or other visual appliances)
0. Adequate – sees fine detail, such as regular print in newspapers/books
1. Impaired – sees large print, but not regular print in newspapers/books
2. Moderately impaired – limited vision; not able to see newspaper headlines but can
identify objects
3. Highly impaired – object identification in question, but eyes appear to follow objects
4. Severely impaired – no vision or sees only light, colors or shapes; eyes do not appear
to follow objects
Current use
Vision is currently assessed in the OASIS and MDS, with corrective lenses when applicable.
Vision is assessed in OASIS with three response options ranging from 0 (normal vision) to 2 (severely
impaired). The Vision data element (Ability to See in Adequate Light) in the MDS contains five response
options ranging from 0 (adequate) to 4 (severely impaired).
Prior evidence supporting use of Vision
The MDS 3.0 Vision data element has been shown to perform reliably in screening for vision
impairment (weighted kappa = 0.917) in the national MDS 3.0 test in nursing homes. 339 The Vision data
element is also linked to performance with readily available materials (i.e., newspaper). Finally, the
Vision data element was tested in the PAC PRD assessment. The PAC PRD found substantial agreement
for interrater reliability across settings for this data element (kappa of 0.74). 340
Evidence supporting use of Vision from the National Beta Test
Assessing Vision: In the Beta testing, the Vision assessment item (with corrective lenses when
applicable) was administered to 643 patients/residents in HHA, 783 in IRF, 498 in LTCH, and 1141 in
SNF (n = 3,065 overall).
Overall, 78 percent of patients/residents had adequate vision, 16 percent had impaired vision, and
6 percent had moderately to severely impaired vision. In the IRF setting, 85 percent of patients/residents
had adequate vision, 12 percent had impaired vision, and 3 percent had moderately to severely impaired
vision. Setting-specific frequencies are shown in Appendix C, Table 9.2.1.

338 Lieberman, D., Friger, M., Lieberman, D. (2004). Visual and hearing impairment in elderly patients hospitalized for

rehabilitation following hip fracture. Journal of Rehabilitation Research and Development 41:669–674.

339 Saliba, D., & Buchanan, J. Development and validation of a revised nursing home assessment tool: MDS 3.0. Appendices.

Santa Monica, CA: RAND Corporation. 2008. Available at https://www.cms.gov/Medicare/Quality-Initiatives-PatientAssessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf
340 Gage, B., Morley, M., Smith, L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. Post-acute care payment reform
demonstration: Final report. Research Triangle Park, NC: RTI International. 2012. Available at
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.

Missing data: There were very low rates of missing responses for the Vision item both overall
(0.6 percent) and the IRF setting (0.6 percent), indicating feasibility of administration.
Time to complete: Time to complete was assessed among 396 patients/residents in HHA, 499 in
IRF, 301 in LTCH, and 456 in SNF (n = 1,652 overall). Across all settings, the mean time to complete the
Vision item was 0.3 minutes (SD = 0.2 minutes). Likewise, in the IRF setting, mean time to complete the
Vision item was 0.3 minutes (SD = 0.2 minutes).
Interrater reliability: Interrater reliability (IRR) was assessed for the Vision item for 197
patients/residents in HHA, 258 in IRF, 237 in LTCH, and 268 in SNF (n = 960). Across all settings,
kappa for the Vision item was moderate (0.56). In the IRF setting, kappa for the Vision item was also
moderate (0.50). Percent agreement was high for the Vision item across settings (83 percent). Agreement
for the Vision items in the IRF setting was slightly higher (90 percent). More detailed IRR statistics are
shown in Appendix C, Table 9.2.2.

Section 6: Proposed New Category: Social Determinants of Health
Standardized Data Elements to Assess for Social Determinants of Health
CMS has identified data elements for cross-setting standardization of assessment for seven social
determinants of health (SDOH). The proposed data elements are:
1. Race;
2. Ethnicity;
3. Preferred Language;
4. Interpreter Services;
5. Health Literacy;
6. Transportation; and
7. Social Isolation.
Race and Ethnicity
Relevance to IRFs
The persistence of racial and ethnic disparities in health and health care is widely documented,
including but not limited to PAC settings. 341,342,343,344,345 Although racial and ethnic disparities decrease
when social factors are controlled for, they often remain. The root cause of these disparities is not always
clear because data on many SDOH are not collected. Measuring SDOH in IRF settings is an important
step to addressing these avoidable differences in health outcomes. Collecting data on race and ethnicity
supports patient-centered care, and informs understanding of patient complexity and risk factors that may
affect payment, quality measurement, and care outcomes for IRFs. Improving how race and ethnicity data
are collected is an important component of improving quality by identifying and addressing health
disparities that impact Medicare beneficiaries.

341 2017 National Healthcare Quality and Disparities Report. Rockville, MD: Agency for Healthcare Research and Quality;

September 2018. AHRQ Pub. No. 18-0033-EF.

342 Fiscella, K. and Sanders, M.R. Racial and Ethnic Disparities in the Quality of Health Care. (2016). Annual Review of Public

Health. 37:375-394.

343 2018 National Impact Assessment of the Centers for Medicare & Medicaid Services (CMS) Quality Measures Reports.

Baltimore, MD: US Department of Health and Human Services, Centers for Medicare and Medicaid Services; February 28,
2018.
344 Smedley, B.D., Stith, A.Y., & Nelson, A.R. (2003). Unequal treatment: confronting racial and ethnic disparities in health
care. Washington, D.C., National Academy Press.
345 Chase, J., Huang, L. and Russell, D. (2017). Racial/ethnic disparities in disability outcomes among post-acute home care
patients. J of Aging and Health. 30(9):1406-1426.

Proposed Data Elements for the Assessment of Social Determinants of Health: Race and Ethnicity
Ethnicity
A1005. Ethnicity
Are you Hispanic, Latino/a, or Spanish origin?
Check all that apply
A. No, not of Hispanic, Latino/a, or Spanish origin
B. Yes, Mexican, Mexican American, Chicano/a
C. Yes, Puerto Rican
D. Yes, Cuban
E. Yes, Another Hispanic, Latino, or Spanish origin
X. Patient unable to respond

Race
A1010. Race
What is your race?
Check all that apply
A. White
B. Black or African American
C. American Indian or Alaska Native
D. Asian Indian
E. Chinese
F. Filipino
G. Japanese
H. Korean
I. Vietnamese
J. Other Asian
K. Native Hawaiian
L. Guamanian or Chamorro
M. Samoan
N. Other Pacific Islander
X. Patient unable to respond

Current use
There is currently a Race and Ethnicity data element, collected in the MDS, LCDS, IRF-PAI, and
OASIS that consists of a single question, which aligns with the 1997 Office of Management and Budget
(OMB) minimum data standards for federal data collection efforts. 346 The 1997 OMB Standard lists five
minimum categories of race: (1) American Indian or Alaska Native; (2) Asian; (3) Black or African
346 “Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity (Notice of Decision)”. Federal

Register 62:210 (October 30, 1997) pp. 58782-58790. Available from: https://www.govinfo.gov/content/pkg/FR-1997-1030/pdf/97-28653.pdf.

American; (4) Native Hawaiian or Other Pacific Islander; and, (5) White. The 1997 OMB Standard also
lists two minimum categories of ethnicity: (1) Hispanic or Latino, and (2) Not Hispanic or Latino. 347 The
current version uses a “Mark all that apply” response option.
Evidence supporting use of Race and Ethnicity
The proposed modification would result in two separate data elements, one for Race and one for
Ethnicity that would conform with the 2011 HHS Data Standards for person-level data collection and the
1997 OMB Standards. The 2011 HHS Data Standards permit the collection of more detailed information
on population groups provided additional categories can be aggregated into the OMB minimum standard
set of categories. The 2011 HHS Data Standards require a two-question format when self-identification is
used to collect data on race and ethnicity, as would be required by these two data elements. Large federal
surveys, such as the National Health Interview Survey, Behavioral Risk Factor Surveillance System, and
the National Survey on Drug Use and Health have implemented the 2011 HHS Data Standards. CMS has
similarly updated the Medicare Current Beneficiary Survey, the Medicare Health Outcomes Survey, and
the Health Insurance Marketplace Application for Health Coverage with the 2011 HHS data standards.
Preferred Language and Interpreter Services
Relevance to IRFs
More than 64 million people in the United States speak a language other than English at home,
and nearly 40 million of those individuals have limited English proficiency (LEP). 348 Individuals with
LEP have been shown to receive worse care and have poorer health outcomes, including higher
readmission rates. 349,350,351 Communication with individuals with LEP is an important component of
quality health care, which starts by understanding the population in need of language services.
Unaddressed language barriers between a patient and provider care team negatively affects the ability to
identify and address individual medical and non-medical care needs, to convey and understand clinical
information, as well as discharge and follow up instructions, all of which are necessary for providing high
quality care. Understanding the communication assistance needs of residents and patients with LEP,
including individuals who are Deaf or hard of hearing, is critical for ensuring good outcomes.

347 Office of Management and Budget. Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity.

Washington, DC. Federal Register Notice, October 30, 1997.

348 U.S. Census Bureau, 2009-2013 American Community Survey.
349 Karliner LS, Kim SE, Meltzer DO, Auerbach AD. Influence of language barriers on outcomes of hospital care for general

medicine inpatients. J Hosp Med. 2010 May-Jun;5(5):276-82. doi: 10.1002/jhm.658.

350 Kim EJ, Kim T, Paasche-Orlow MK, et al. Disparities in Hypertension Associated with Limited English Proficiency. J Gen

Intern Med. 2017 Jun;32(6):632-639. doi: 10.1007/s11606-017-3999-9.

351 National Academies of Sciences, Engineering, and Medicine. 2016. Accounting for social risk factors in Medicare payment:

Identifying social risk factors. Washington, DC: The National Academies Press.

Proposed Data Elements for the Assessment of Social Determinants of Health: Preferred Language
and Interpreter Services
A1110. Language
A. What is your preferred language?

Enter Code

B. Do you need or want an interpreter to communicate with a doctor or health care
staff?
0. No
1. Yes
9. Unable to determine

Current use
The preferred language of residents and patients and need for interpreter services are assessed in
two PAC assessment tools. The LCDS and the MDS use the same two data elements to assess preferred
language and whether a patient or resident needs or wants an interpreter to communicate with health care
staff. The current preferred language data element in LCDS and MDS is open-ended, allowing the patient
or resident to identify their preferred language, including American Sign Language (ASL). The MDS
initially implemented preferred language and interpreter services data elements to assess the needs of SNF
residents and patients and inform care planning. For alignment purposes, the LCDS later adopted the
same data elements for LTCHs.
Evidence supporting use of Preferred Language and Interpreter Services
The 2009 National Academies of Sciences, Engineering, and Medicine (NASEM) report on
standardizing data for health care quality improvement emphasizes that language and communication
needs should be assessed as a standard part of health care delivery and quality improvement strategies. 352
While the 2011 HHS Primary Language Data Standard recommends a two-part question to assess spoken
language, the need to improve the assessment of language preferences and communication needs across
PAC settings should be balanced with the provider and patient assessment burden. In addition, preferred
spoken language would not allow information to be collected on American Sign Language, as is
accounted for by the preferred language and interpreter services data elements currently in the MDS and
LCDS.
Health Literacy
Relevance to IRFs
Similar to language barriers, low health literacy can interfere with communication between the
provider and resident or patient and the ability for residents and patients or their caregivers to understand
and follow treatment plans, including medication management. Poor health literacy is linked to lower
levels of knowledge about health, worse health outcomes, and the receipt of fewer preventive services but
higher medical costs and rates of emergency department use. 353

352 IOM (Institute of Medicine). 2009. Race, Ethnicity, and Language Data: Standardization for Health Care Quality

Improvement. Washington, DC: The National Academies Press.

353 National Academies of Sciences, Engineering, and Medicine. 2016. Accounting for social risk factors in Medicare payment:

Identifying social risk factors. Washington, DC: The National Academies Press.

Proposed Data Element for the Assessment of Social Determinants of Health: Health Literacy
B1300. Health Literacy
How often do you need to have someone help you when you read instructions, pamphlets, or other
written material from your doctor or pharmacy?
Enter Code

0. Never
1. Rarely
2. Sometimes
3. Often
4. Always
9. Patient unable to respond

Current use
A health literacy data element is not currently used in any of the PAC assessment tools.
Evidence supporting use of Health Literacy
Health literacy is prioritized by Healthy People 2020 as an SDOH. 354 Health literacy is also
considered an individual risk factor impacted by other social risk factors in NASEM’s 2016 report on
accounting for social risk factors in Medicare payment. 355 The Single Item Literacy Screener (SILS)
question, which assesses reading ability (a primary component of health literacy), tested reasonably well
against the 36 item Short Test of Functional Health Literacy in Adults (S-TOFHLA), a thoroughly vetted
and widely adopted health literacy test, in assessing the likelihood of low health literacy in an adult
sample from primary care practices participating in the Vermont Diabetes Information System. 356,357 SILS
is publicly available, and shorter and easier to administer than the S-TOFHLA , and research found that a
positive result on the SILS demonstrates an increased likelihood that an individual has low health literacy.
Transportation
Relevance to IRFs
Transportation barriers can affect access to needed health care, causing missed appointments,
delayed care, unfilled prescriptions, all of which can have a negative impact on health outcomes. 358
Access to transportation for ongoing health care and medication access needs, particularly for those with
chronic diseases, is essential to successful chronic disease management. Adopting a data element to
collect and analyze information regarding transportation needs across PAC settings would facilitate the
connection to programs that can address identified needs.

354 Social Determinants of Health. Healthy People 2020. https://www.healthypeople.gov/2020/topics-objectives/topic/social-

determinants-of-health. (February 2019).

355 U.S. Department of Health & Human Services, Office of the Assistant Secretary for Planning and Evaluation. Report to

Congress: Social Risk Factors and Performance Under Medicare’s Value-Based Purchasing Programs. Available at:
https://aspe.hhs.gov/pdf-report/report-congress-social-risk-factors-and-performance-under-medicares-value-basedpurchasing-programs. Washington, DC: 2016.
356 Morris, N. S., MacLean, C. D., Chew, L. D., & Littenberg, B. (2006). The Single Item Literacy Screener: evaluation of a
brief instrument to identify limited reading ability. BMC family practice, 7, 21. doi:10.1186/1471-2296-7-21.
357 Brice, J.H., Foster, M.B., Principe, S., Moss, C., Shofer, F.S., Falk, R.J., Ferris, M.E., DeWalt, D.A. (2013). Single-item or
two-item literacy screener to predict the S-TOFHLA among adult hemodialysis patients. Patient Educ Couns. 94(1):71-5.
358 Syed, S.T., Gerber, B.S., and Sharp, L.K. (2013). Traveling Towards Disease: Transportation Barriers to Health Care Access.
J Community Health. 38(5): 976–993.

Proposed Data Element for the Assessment of Social Determinants of Health: Transportation
A1250. Transportation
Has lack of transportation kept you from medical appointments, meetings, work, or from getting
things needed for daily living?
Check all that apply
A. Yes, it has kept me from medical appointments or from getting my medications
B. Yes, it has kept me from non-medical meetings, appointments, work, or from getting things
that I need
C. No
D. Patient unable to respond
Current use
A transportation data element is not currently used in any of the PAC assessment tools.
Evidence supporting use of Transportation
The proposed data element uses the Transportation item from the Protocol for Responding to and
Assessing Patient Assets, Risks, and Experiences (PRAPARE) tool and is responsive to research on the
importance of addressing transportation as a critical SDOH. The national PRAPARE social determinants
of health assessment protocol is developed and owned by the National Association of Community Health
Centers (NACHC), in partnership with the Association of Asian Pacific Community Health Organization,
the Oregon Primary Care Association, and the Institute for Alternative Futures. More information about
development of the PRAPARE tool can be found at: https://protect2.fireeye.com/url?k=7cb6eb4420e2f238-7cb6da7b-0cc47adc5fa2-1751cb986c8c2f8c&u=http://www.nachc.org/prapare. Items in the
assessment tool are consistent with Healthy People 2020 priorities and ICD-10 coding. 359
Social Isolation
Relevance to IRFs
Distinct from loneliness, social isolation refers to an actual or perceived lack of contact with other
people, such as living alone or residing in a remote area. 360,361 Social isolation tends to increase with age,
is a risk factor for physical and mental illness, and a predictor of mortality. 362,363,364 Post-acute care
providers are well-suited to design and implement programs to increase social engagement of patients,
while also taking into account individual needs and preferences. Adopting a data element to collect and
analyze information about social isolation in IRFs and across PAC settings would facilitate the
identification of patients who are socially isolated and who may benefit from engagement efforts.
359 PRAPARE. National Association of Community Health Centers. http://www.nachc.org/research-and-data/prapare/.

(December 2018).

360 Tomaka, J., Thompson, S., and Palacios, R. (2006). The Relation of Social Isolation, Loneliness, and Social Support to

Disease Outcomes Among the Elderly. J of Aging and Health. 18(3): 359-384.

361 Social Connectedness and Engagement Technology for Long-Term and Post-Acute Care: A Primer and Provider Selection

Guide. (2019). Leading Age. Available at https://www.leadingage.org/white-papers/social-connectedness-and-engagementtechnology-long-term-and-post-acute-care-primer-and#1.1
362 Landeiro, F., Barrows, P., Nuttall Musson, E., Gray, A.M., and Leal, J. (2017). Reducing Social Loneliness in Older People:
A Systematic Review Protocol. BMJ Open. 7(5): e013778.
363 Ong, A.D., Uchino, B.N., and Wethington, E. (2016). Loneliness and Health in Older Adults: A Mini-Review and Synthesis.
Gerontology. 62:443-449.
364 Leigh-Hunt, N., Bagguley, D., Bash, K., Turner, V., Turnbull, S., Valtorta, N., and Caan, W. (2017). An overview of
systematic reviews on the public health consequences of social isolation and loneliness. Public Health. 152:157-171.

Proposed Data Element for the Assessment of Social Determinants of Health: Social Isolation
D0700. Social Isolation
How often do you feel lonely or isolated from those around you?
Enter Code

0. Never
1. Rarely
2. Sometimes
3. Often
4. Always
9. Patient unable to respond

Current use
A social isolation data element is not currently used in any of the PAC assessment tools.
Evidence supporting use of Social Isolation
The proposed data element uses the social isolation item from the Accountable Health
Communities (AHC) Screening Tool, which was selected from the PROMIS Item Bank on Emotional
Distress. The AHC Screening Tool was developed by a panel of interdisciplinary experts that looked at
evidence-based ways to measure social determinants of health, including social isolation. More
information about the AHC Screening Tool can be found at:
https://innovation.cms.gov/Files/worksheets/ahcm-screeningtool.pdf.

APPENDIX A:
Transfer of Health Information – Setting-Specific Language
Tables A-1 and A-2 below summarize the setting specific language used to describe the resident
or patient within each PAC setting. There are no other differences in the content or language within each
Transfer of Health Information to the Provider-Post-Acute Care quality measure data element and within
each Transfer of Health Information to the Patient–Post-Acute Care quality measure data element.
Table A-1
Transfer of Health Information to the Provider–Post-Acute Care - Setting-Specific Language
IRF

LTCH

SNF

Discharge

A2121. Provision of Current
Reconciled Medication List to
Subsequent Provider at
Discharge
At the time of discharge to another
provider, did your facility provide
the patient’s current reconciled
medication list to the subsequent
provider?
Enter Code: 
0. No - Current reconciled
medication list not provided to the
subsequent provider
1. Yes - Current reconciled
medication list provided to the
subsequent provider

A2121. Provision of Current
Reconciled Medication List to
Subsequent Provider at
Discharge
At the time of discharge to another
provider, did your facility provide
the resident’s current reconciled
medication list to the subsequent
provider?
Enter Code: 
0. No - Current reconciled
medication list not provided to the
subsequent provider
1. Yes - Current reconciled
medication list provided to the
subsequent provider

A2123. Route of Current
Reconciled Medication List
Transmission (column 1)
Indicate the route(s) of
transmission of the current
reconciled medication list to the
subsequent provider and/or
patient/family/caregiver.
A. Electronic Health Record
B. Health Information Exchange
Organization
C. Verbal (e.g., in-person,
telephone, video conferencing)
D. Paper-based (e.g., fax, copies,
printouts)
E. Other Methods (e.g., texting,
email, CDs)

Table 2
Transfer of Health Information to the Patient–Post-Acute Care - Setting-Specific Language
IRF

LTCH

SNF

Discharge

A2122. Provision of Current
Reconciled Medication List to
Patient at Discharge
At the time of discharge, did your
facility provide the patient’s
current reconciled medication list
to the patient, family and/or
caregiver?
Enter Code: 
0. No - Current reconciled
medication list not provided to the
patient, family and/or caregiver
1. Yes - Current reconciled
medication list provided to the
patient, family and/or caregiver

A2122. Provision of Current
Reconciled Medication List to
Resident at Discharge
At the time of discharge, did your
facility provide the resident’s
current reconciled medication list
to the resident, family and/or
caregiver?
Enter Code: 
0. No - Current reconciled
medication list not provided to the
resident, family and/or caregiver
1. Yes - Current reconciled
medication list provided to the
resident, family and/or caregiver?

A2123. Route of Current
Reconciled Medication List
Transmission (column 2)
Indicate the route(s) of
transmission of the current
reconciled medication list to the
subsequent provider and/or
patient/family/caregiver.
A. Electronic Health Record
B. Health Information Exchange
Organization
C. Verbal (e.g., in-person,
telephone, video conferencing)
D. Paper-based (e.g., fax, copies,
printouts)
E. Other Methods (e.g., texting,
email, CDs)

100

APPENDIX B:
Discharge to Community–PAC IRF QRP Analyses
Table B-1.
Logistic Regression Model Results for Discharge to Community–Post Acute Care (PAC) Inpatient Rehabilitation Facility
(IRF) Quality Reporting Program (QRP), Calendar Year 2015–2016
Number of beneficiaries included in the model = 594,733
Observed number (percent) of beneficiaries in the sample who were discharged to community = 383,703 (64.52%).
Model c-statistic = 0.708
Based on Medicare fee-for-service claims data from CY 2015–2016. These model estimates only apply to CY 2015–2016 IRF data. We will re-estimate the
regression model for each measurement period to allow the estimated effects of patient characteristics to vary over time.
Risk Adjuster

Intercept

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

3.060

0.038

<.0001

Age and Sex Groupings (Reference: Female, age 18–64 years)
Male, age 18–64 years

36,396

6.1

−0.034

0.017

0.0486

0.967

0.935

1.000

Male, age 65–74 years

91,141

15.3

−0.050

0.015

0.0009

0.951

0.923

0.980

Male, age 75–79 years

49,544

8.3

−0.163

0.017

<.0001

0.849

0.822

0.877

Male, age 80–84 years

43,991

7.4

−0.280

0.017

<.0001

0.756

0.731

0.781

Male, age 85–89 years

31,807

5.4

−0.468

0.018

<.0001

0.626

0.604

0.649

Male, age 90–94 years

12,492

2.1

−0.623

0.024

<.0001

0.536

0.512

0.562

2,360

0.4

−0.660

0.045

<.0001

0.517

0.473

0.565

Female, age 65–74 years

98,577

16.6

−0.010

0.015

0.4982

0.990

0.961

1.019

Female, age 75–79 years

58,311

9.8

−0.149

0.016

<.0001

0.862

0.835

0.890

Female, age 80–84 years

57,036

9.6

−0.299

0.016

<.0001

0.742

0.718

0.766

Female, age 85–89 years

49,097

8.3

−0.504

0.017

<.0001

0.604

0.584

0.624

Female, age 90–94 years

23,430

3.9

−0.616

0.020

<.0001

0.540

0.520

0.562

5,493

0.9

−0.714

0.032

<.0001

0.490

0.460

0.521

76,090

12.8

−0.108

0.009

<.0001

0.898

0.882

0.914

Male, age ≥ 95 years

Female, age ≥ 95 years
Original Reason for Entitlement
Age ≥ 65 at IRF admission and original reason for entitlement was
disability or ESRD

(continued)

101

Table B-1.
Logistic Regression Model Results for Discharge to Community–Post Acute Care (PAC) Inpatient Rehabilitation Facility
(IRF) Quality Reporting Program (QRP), Calendar Year 2015–2016 (continued)
Risk Adjuster

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

Principal Diagnosis Clinical Classifications Software (CCS) Groupings Based on Prior Acute Stay (Reference: includes all CCS numbers not listed as
risk adjusters)
Other Infectious and Parasitic Diseases (1, 3-10)
1,367
0.2
−0.323
0.063
<.0001
0.724
0.640
0.818
Infectious & Parasitic Disease: Septicemia (2)

31,937

5.4

−0.263

0.026

<.0001

0.769

0.731

0.809

Neoplasms - Liver, Pancreas, Bronchus, Lung, Ovary, Brain &
Nervous System (16, 17, 19, 27, 35)

2,958

0.5

−0.593

0.048

<.0001

0.553

0.503

0.607

Secondary Malignant Neoplasm (42)

2,181

0.4

−0.638

0.053

<.0001

0.528

0.476

0.586

Neoplasms-Benign (44-47); Neoplasms-Low (22-26, 28-31, 36);
Neoplasms-Medium (11-15, 18, 20-21, 32-34, 37-41, 43)

7,911

1.3

−0.418

0.035

<.0001

0.658

0.615

0.705

Endocrine Disorders (48, 51, 53, 54)

1,745

0.3

−0.166

0.058

0.0041

0.847

0.756

0.949

Diabetes with and without Complications (49, 50)

8,230

1.4

−0.401

0.034

<.0001

0.670

0.626

0.716

Nutritional Defic and Other Nutritional Disorders (52, 58)

1,021

0.2

−0.503

0.071

<.0001

0.605

0.527

0.695

Fluid/Electrolyte Disorders (55)

4,236

0.7

−0.411

0.040

<.0001

0.663

0.613

0.717

Diseases of Blood and Blood-Forming Organs (56-57, 59-64)

2,298

0.4

−0.457

0.050

<.0001

0.633

0.575

0.698

Dis Nerv Syst: Meningitis, Encephalitis, Other CNS infection (76-78)

1,577

0.3

−0.531

0.059

<.0001

0.588

0.524

0.660

Dis Nerv Syst: Parkinson's, MS, Other Hered CNS Disease, Paralysis
(79-82)

5,732

1.0

−0.498

0.037

<.0001

0.608

0.565

0.654

Dis Nerv Syst: Epilepsy; Convulsions (83)

3,790

0.6

−0.221

0.043

<.0001

0.802

0.737

0.872

10,013

1.7

−0.364

0.032

<.0001

0.695

0.653

0.740

Circ Syst: Heart Valve Disorders (96)

7,951

1.3

−0.300

0.039

<.0001

0.741

0.687

0.799

Circ Syst: Carditis & Other Heart Disease (97, 104)

1,088

0.2

−0.420

0.070

<.0001

0.657

0.573

0.753

Circ Syst: HTN & HTN Complication (98, 99)

5,209

0.9

−0.322

0.038

<.0001

0.725

0.673

0.780

11,904

2.0

−0.395

0.032

<.0001

0.674

0.633

0.717

Dis Nerv Syst: Other Nervous System Disorders (95)

Circ Syst: Acute MI & Cardiac Arrest (100, 107)

(continued)

102

Circ Syst: Coron Athero & Chest Pain (101, 102)

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

10,491

1.8

−0.331

0.035

<.0001

0.718

0.670

0.769

Circ Syst: Pulmonary Heart Disease (103)

2,767

0.5

−0.214

0.047

<.0001

0.807

0.736

0.886

Circ Syst: Conduction Disorders & Cardiac Dysrhythmia (105, 106)

9,441

1.6

−0.300

0.032

<.0001

0.741

0.696

0.789

13,531

2.3

−0.388

0.030

<.0001

0.679

0.640

0.720

105,186

17.7

−0.582

0.027

<.0001

0.559

0.530

0.589

Circ Syst: TIA (112)

2,964

0.5

−0.082

0.049

0.0954

0.921

0.837

1.014

Circ Syst: Peripheral and Visceral Atherosclerosis (114)

3,891

0.7

−0.532

0.042

<.0001

0.588

0.541

0.638

Circ Syst: Aneurysm (115)

2,407

0.4

−0.366

0.051

<.0001

0.693

0.627

0.766

Circ Syst: Arterial Embolism & Other Circul Disease (116, 117)

3,418

0.6

−0.424

0.043

<.0001

0.654

0.601

0.712

Circ Syst: Phlebitis, Varicose Veins, Hemorrhoids, Other Vein Disease
(118-121)

2,182

0.4

−0.343

0.051

<.0001

0.709

0.642

0.784

Resp Syst: Pneumonia, Influenza, Acute Bronchitis, Other Upper Resp
(122,123, 125-126)

11,289

1.9

−0.212

0.031

<.0001

0.809

0.762

0.859

Resp Syst: Tonsillitis, Pleurisy, Lung Disease, Other Lower or Upper
Resp (124, 130, 132-134)

2,509

0.4

−0.343

0.049

<.0001

0.709

0.645

0.780

Resp Syst: COPD & Asthma (127, 128)

6,501

1.1

−0.410

0.037

<.0001

0.664

0.617

0.713

Resp Syst: Aspiration Pneumonia (129)

2,988

0.5

−0.374

0.045

<.0001

0.688

0.630

0.752

Resp Syst: Adult Respiratory Failure (131)

7,560

1.3

−0.299

0.034

<.0001

0.741

0.693

0.793

11,510

1.9

−0.332

0.031

<.0001

0.717

0.676

0.762

Digestive System - Intestinal Obstruction without Hernia (145)

2,948

0.5

−0.188

0.046

<.0001

0.829

0.757

0.907

Biliary Disease, Liver Disease, Pancreatic disorders (149-152)

4,161

0.7

−0.335

0.041

<.0001

0.716

0.661

0.775

GI Hemorrhage (153)

3,846

0.6

−0.301

0.041

<.0001

0.740

0.683

0.803

Genitourinary: Other (156, 160-166, 168-175)

1,721

0.3

−0.274

0.059

<.0001

0.760

0.678

0.853

Genitourinary: Acute or Chronic Renal Failure (157,158)

9,630

1.6

−0.417

0.032

<.0001

0.659

0.620

0.702

Genitourinary: UTI (159)

8,102

1.4

−0.356

0.033

<.0001

0.700

0.656

0.747

Circ Syst: CHF (108)
Circ Syst: CVD (109-111, 113)

Diseases of Digestive System (135-144, 146-148, 154,155)

(continued)

103

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

Diseases of the Skin and Subcutaneous Tissue (167, 197-200)

4,479

0.8

−0.288

0.040

<.0001

0.749

0.694

0.810

Infective Arthritis and Osteomyelitis (201)

2,046

0.3

−0.448

0.052

<.0001

0.639

0.577

0.708

Rheumatoid Arthritis, Lupus, Other Connective Tissue Disease (202,
210)

4,150

0.7

−0.368

0.041

<.0001

0.692

0.640

0.750

Other Joint Disorders & Osteoporosis (204, 206)

1,104

0.2

−0.253

0.073

0.0006

0.777

0.673

0.897

30,099

5.1

−0.171

0.026

<.0001

0.843

0.802

0.886

Pathological Fracture (207)

5,536

0.9

−0.426

0.037

<.0001

0.653

0.608

0.702

Other Bone Disease (212)

1,957

0.3

−0.193

0.058

0.0008

0.824

0.737

0.923

Congenital Anomalies (213-217)

914

0.2

−0.226

0.082

0.0056

0.798

0.680

0.936

Joint Injury (225)

616

0.1

−0.427

0.092

<.0001

0.652

0.544

0.782

57,263

9.6

−0.466

0.028

<.0001

0.627

0.594

0.662

1,527

0.3

−0.760

0.063

<.0001

0.468

0.414

0.529

21,298

3.6

−0.324

0.028

<.0001

0.723

0.684

0.764

Fracture of Upper Limb (229)

5,236

0.9

−0.468

0.038

<.0001

0.626

0.581

0.674

Fracture of Lower Limb (230)

12,229

2.1

−0.622

0.030

<.0001

0.537

0.506

0.569

1,963

0.3

−0.174

0.056

0.0018

0.841

0.754

0.938

16,147

2.7

−0.538

0.034

<.0001

0.584

0.546

0.625

Crush Injury (234)

1,746

0.3

−0.069

0.060

0.244

0.933

0.830

1.048

Open Wound Head & Extremities, Burns, Other Injuries due to
External Causes (235, 236, 240, 244)

2,256

0.4

−0.311

0.052

<.0001

0.733

0.661

0.811

27,998

4.7

−0.386

0.023

<.0001

0.680

0.649

0.711

Poison Psychotropic Agents, Poison Other Med, Poison Nonmed (241243)

1,093

0.2

−0.175

0.071

0.014

0.840

0.730

0.965

Symptoms, Signs & Ill-Defined Conditions & Factors influencing
health status (245-247, 249-259)

5,237

0.9

−0.277

0.038

<.0001

0.758

0.704

0.816

Back Problem (205)

Fracture of Hip (226)
Spinal Cord Injury (227)
Skull and Face Fractures & Other Fractures (228, 231)

Sprains and Strains & Superficial Injury/Contusion (232, 239)
Intracranial Injury (233)

Complications of Device, Procedures, or Medical Care (237-238)

(continued)

104

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

Gangrene (248)
2,824
0.5
−0.674
0.049
<.0001
Mental Illness (650-670)
3,272
0.6
−0.289
0.049
<.0001
IRF Case-Mix Groups (CMGs) (Reference: Stroke: Motor Score > 44.45 (CMGs: 0101-0103); Expired (CMGs: 5101-5104))
Stroke: Motor Score 26.15-44.45 (CMGs: 0104-0107)
51,933
8.7
-0.574
0.027
<.0001
Stroke: (Motor <26.15 Age >84.5), (Motor > 22.35 Motor <26.15 Age
<84.5) (CMGs: 0108-0109)
22,064
3.7
-1.481
0.028
<.0001
Stroke: Motor Score <22.35 and Age <84.5 (CMG: 0110)
37,142
6.3
-1.982
0.027
<.0001
Traumatic Brain Injury: Motor Score >28.75 (CMGs: 0201-0205)
10,363
1.7
-0.543
0.039
<.0001
Traumatic Brain Injury: Motor Score <28.75 (CMGs: 0206-0207)
9,844
1.7
-1.405
0.038
<.0001
Non-traumatic Brain Injury: Motor Score >35.05 (CMGs: 0301-0302)
11,886
2.0
-0.473
0.036
<.0001
Non-traumatic Brain Injury: Motor Score <35.05 (CMGs: 0303-0304)
24,887
4.2
-1.190
0.031
<.0001
Traumatic Spinal Cord Injury: All (CMGs: 0401-0405)
4,262
0.7
-1.299
0.047
<.0001
Non-traumatic Spinal Cord Injury: Motor Score >31.25 (CMGs: 05010503)
9,265
1.6
-0.527
0.042
<.0001
Non-traumatic Spinal Cord Injury: Motor Score <31.25 (CMGs: 05040506)
15,463
2.6
-1.494
0.035
<.0001
Neurological: Motor Score >37.35 (CMGs: 0601-0602)
14,291
2.4
-0.495
0.035
<.0001
Neurological: Motor Score <37.35 (CMGs: 0603-0604)
54,240
9.1
-1.112
0.030
<.0001
Fracture of Lower Extremity: Motor Score >28.15 (CMGs: 0701-0703)
24,735
4.2
-0.386
0.036
<.0001
Fracture of Lower Extremity: Motor Score <28.15 (CMG: 0704)
47,974
8.1
-1.360
0.034
<.0001
Replacement of Lower Extremity Joint: Motor Score >28.65 (CMGs:
0801-0804)
25,157
4.2
-0.382
0.039
<.0001
Replacement of Lower Extremity Joint: Motor Score <28.65 (CMGs:
0805-0806)
16,215
2.7
-1.191
0.038
<.0001
Other Orthopedic: Motor Score >24.15 (CMGs: 0901-0903)
11,492
1.9
-0.453
0.039
<.0001
Other Orthopedic: Motor Score <24.15 (CMG: 0904)
29,110
4.9
-1.239
0.032
<.0001
Amputation, Lower Extremity: Motor Score >36.25 (CMGs:10011002)
3,408
0.6
-0.481
0.052
<.0001
Amputation, Lower Extremity: Motor Score <36.25 (CMG:1003) &
Amputation, Non-Lower Extremity (CMGs: 1101-1102)
12,303
2.1
-1.146
0.039
<.0001
Osteoarthritis: All (CMGs: 1201-1203)
866
0.2
-1.077
0.081
<.0001
Rheumatoid, Other Arthritis: All (CMGs: 1301-1303)
1,573
0.3
-1.003
0.063
<.0001
Cardiac: Motor Score >38.55 (CMGs: 1401-1402)
10,982
1.9
-0.473
0.038
<.0001

0.510
0.749

0.464
0.681

0.561
0.824

0.563

0.535

0.593

0.227
0.138
0.581
0.245
0.623
0.304
0.273

0.215
0.131
0.538
0.228
0.581
0.286
0.249

0.240
0.145
0.627
0.264
0.668
0.323
0.299

0.591

0.544

0.641

0.224
0.610
0.329
0.680
0.257

0.209
0.570
0.310
0.633
0.240

0.241
0.653
0.349
0.730
0.274

0.682

0.632

0.737

0.304
0.636
0.290

0.282
0.589
0.272

0.327
0.686
0.309

0.618

0.559

0.684

0.318
0.341
0.367
0.623

0.295
0.291
0.324
0.579

0.343
0.399
0.415
0.671
(continued)

105

Cardiac: Motor Score <38.55 (CMGs: 1403-1404)
Pulmonary: Motor Score >39.05 (CMGs: 1501-1502)
Pulmonary: Motor Score <39.05 (CMGs: 1503-1504)
Pain Syndrome: All (CMGs: 1601-1603)
Major Multiple Trauma Without Brain or Spinal Cord Injury (CMGs:
1701-1704)
Major Multiple Trauma With Brain or Spinal Cord Injury (CMGs:
1801-1803)
Guillain Barre (CMGs: 1901-1903)
Miscellaneous (CMGs: 2001-2004), Burns (CMG 2101), Short-stay
cases (CMG: 5001)
Surgical Categories Based on Prior Acute Stay

29,247
4,083
8,948
2,207

Cardiothoracic surgery

4.9
0.7
1.5
0.4

Estimate

-1.002
-0.544
-1.057
-1.025

SE1

0.032
0.047
0.037
0.057

pvalue

<.0001
<.0001
<.0001
<.0001

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

0.367
0.581
0.348
0.359

0.345
0.530
0.323
0.321

0.391
0.637
0.374
0.401

10,641

1.8

-1.114

0.038

<.0001

0.328

0.305

0.354

3,141
1,231

0.5
0.2

-1.036
-1.180

0.050
0.070

<.0001
<.0001

0.355
0.307

0.322
0.268

0.392
0.352

72,447

12.2

-0.907

0.030

<.0001

0.404

0.381

0.428

31,341

5.3

0.211

0.020

<.0001

1.235

1.189

1.283

1,652

0.3

0.043

0.055

0.4294

1.044

0.938

1.162

Neurosurgery

27,063

4.6

0.016

0.017

0.3397

1.016

0.984

1.049

General surgery

44,588

7.5

0.105

0.015

<.0001

1.110

1.078

1.143

930

0.2

−0.085

0.072

0.24

0.919

0.797

1.058

170,827

28.7

0.066

0.013

<.0001

1.068

1.041

1.097

12,880

2.2

−0.083

0.021

<.0001

0.920

0.884

0.958

Urologic surgery

4,018

0.7

0.066

0.036

0.069

1.068

0.995

1.147

Vascular surgery

14,347

2.4

0.100

0.020

<.0001

1.106

1.063

1.150

1.0

−0.086

0.029

0.0033

0.918

0.867

0.972

Otolaryngology

Obstetrics/Gynecology
Orthopedic surgery
Plastic surgery

Dialysis in Prior Acute Stay where End-Stage Renal Disease not Indicated
Dialysis Where HCC133 (End-Stage Renal Disease) Not Indicated

6,099

Prior Acute Length of Stay in Non-Psychiatric Hospital or Prior Stay in Psychiatric Hospital (Reference: 1-3 days in Non-Psychiatric Hospital)
Prior acute stay in psychiatric hospital

683

0.1

−0.538

0.091

<.0001

0.584

0.489

0.698

4-5 days in non-psychiatric hospital

150,262

25.3

−0.089

0.008

<.0001

0.915

0.900

0.930

6-8 days in non-psychiatric hospital

123,559

20.8

−0.224

0.009

<.0001

0.799

0.785

0.814
(continued)

106

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

9-10 days in non-psychiatric hospital
41,459
7.0
−0.310
0.013
<.0001
0.734
0.715
11-14 days in non-psychiatric hospital
45,891
7.7
−0.369
0.013
<.0001
0.692
0.674
15-20 days in non-psychiatric hospital
28,005
4.7
−0.470
0.016
<.0001
0.625
0.606
21-30 days in non-psychiatric hospital
14,369
2.4
−0.520
0.021
<.0001
0.595
0.570
30+ days in non-psychiatric hospital
6,038
1.0
−0.691
0.031
<.0001
0.501
0.471
Comorbidities - Hierarchical Condition Categories (HCCs) (* indicates that the HCC is based on the most recent acute care claim only. HCCs not
preceded by * are based on acute care claims from the past 365 days (including the most recent acute care claim)).
HCC3: Bacterial, Fungal, and Parasitic Central Nervous System
Infections*
2,497
0.4
−0.111
0.044
0.0109
0.895
0.822

0.753
0.710
0.645
0.620
0.532

HCC6: Opportunistic Infections

0.975

2,395

0.4

−0.093

0.044

0.032

0.911

0.836

0.992

64,584

10.9

−0.026

0.010

0.0117

0.975

0.956

0.994

HCC8: Metastatic Cancer and Acute Leukemia*

8,774

1.5

−0.288

0.024

<.0001

0.750

0.716

0.785

HCC9: Lung and Other Severe Cancers*

7,099

1.2

−0.182

0.026

<.0001

0.834

0.793

0.877

HCC10: Lymphoma and Other Cancers*

6,990

1.2

−0.101

0.026

0.0001

0.904

0.859

0.952

HCC11: Colorectal, Bladder, and Other Cancers*

3,107

0.5

−0.022

0.039

0.5774

0.979

0.907

1.056

HCC12: Breast, Prostate, and Other Cancers and Tumors*

8,071

1.4

−0.042

0.025

0.0895

0.959

0.913

1.007

HCC13; HCC14: Other Respiratory and Heart Neoplasms; Other
Digestive and Urinary Neoplasms*

3,190

0.5

−0.021

0.039

0.5843

0.979

0.908

1.056

HCC17: Diabetes with Acute Complications

1,887

0.3

−0.143

0.050

0.0045

0.867

0.785

0.957

214,610

36.1

−0.047

0.006

<.0001

0.954

0.942

0.966

4,232

0.7

−0.063

0.034

0.067

0.939

0.878

1.004

HCC7: Other Infectious Diseases*

HCC18; HCC19: Diabetes with Chronic Complications; Diabetes
without Complication
HCC20: Type I Diabetes Mellitus*

(continued)

107

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

HCC21: Protein-Calorie Malnutrition

40,992

6.9

−0.170

0.012

<.0001

0.844

0.825

0.863

HCC22: Morbid Obesity

45,288

7.6

−0.067

0.012

<.0001

0.935

0.914

0.956

HCC23: Other Significant Endocrine and Metabolic Disorders*

24,160

4.1

−0.044

0.015

0.0027

0.957

0.930

0.985

208,632

35.1

−0.071

0.007

<.0001

0.931

0.919

0.943

HCC27: End-Stage Liver Disease

3,796

0.6

−0.368

0.035

<.0001

0.692

0.646

0.742

HCC28: Cirrhosis of Liver

5,487

0.9

−0.268

0.029

<.0001

0.765

0.722

0.810

HCC29: Chronic Hepatitis

1,930

0.3

−0.030

0.051

0.5499

0.970

0.879

1.071

HCC36: Peptic Ulcer, Hemorrhage, Other Specified Gastrointestinal
Disorders*

31,245

5.3

−0.046

0.013

0.0005

0.956

0.931

0.980

HCC39: Bone/Joint/Muscle Infections/Necrosis*

12,511

2.1

−0.083

0.022

0.0002

0.921

0.882

0.961

HCC40: Rheumatoid Arthritis and Inflammatory Connective Tissue
Disease

31,201

5.3

−0.013

0.013

0.3136

0.987

0.962

1.013

3,468

0.6

−0.148

0.037

<.0001

0.862

0.803

0.927

50,705

8.5

−0.039

0.011

0.0004

0.962

0.942

0.983

207,568

34.9

−0.030

0.007

<.0001

0.970

0.957

0.983

HCC50: Delirium and Encephalopathy*

65,056

10.9

−0.047

0.010

<.0001

0.954

0.936

0.972

HCC51; HCC52: Dementia With Complications; Dementia Without
Complication

62,160

10.5

−0.159

0.009

<.0001

0.853

0.837

0.869

9,747

1.6

−0.089

0.023

<.0001

0.915

0.875

0.957

HCC54; HCC55: Drug/Alcohol Psychosis; Drug/Alcohol
Dependence*

16,762

2.8

−0.040

0.018

0.0263

0.961

0.928

0.995

HCC56: Drug/Alcohol Abuse, Without Dependence*

57,695

9.7

−0.046

0.010

<.0001

0.956

0.936

0.975

3,293

0.6

−0.223

0.039

<.0001

0.800

0.742

0.863

13,759

2.3

−0.109

0.019

<.0001

0.896

0.863

0.931

HCC24: Disorders of Fluid/Electrolyte/Acid-Base Balance*

HCC46: Severe Hematological Disorders*
HCC48: Coagulation Defects and Other Specified Hematological
Disorders*
HCC49: Iron Deficiency and Other/Unspecified Anemias and Blood
Disease*

HCC53: Nonpsychotic Organic Brain Syndromes/Conditions

HCC57: Schizophrenia
HCC58: Major Depressive, Bipolar, and Paranoid Disorders

(continued)

108

HCC59: Reactive and Unspecified Psychosis

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

2,825

0.5

−0.124

0.041

0.0024

0.883

0.815

0.957

309

0.1

−0.249

0.124

0.0435

0.779

0.612

0.993

79,813

13.4

−0.057

0.009

<.0001

0.944

0.929

0.961

HCC64 - HCC67: Profound Mental Retardation/Developmental
Disability; Severe Mental Retardation/Developmental Disability;
Moderate Mental Retardation/Developmental Disability; Mild Mental
Retardation, Autism, Down Syndrome

2,088

0.4

−0.141

0.049

0.004

0.869

0.790

0.956

HCC68; HCC69: Other Developmental Disability; Attention Deficit
Disorder

1,497

0.3

−0.144

0.058

0.0136

0.866

0.772

0.971

HCC70: Quadriplegia

3,073

0.5

−0.294

0.040

<.0001

0.745

0.690

0.806

HCC71: Paraplegia

4,102

0.7

−0.213

0.035

<.0001

0.808

0.755

0.866

HCC72: Spinal Cord Disorders/Injuries

7,962

1.3

−0.112

0.027

<.0001

0.894

0.849

0.942

316

0.1

−0.130

0.122

0.2874

0.878

0.691

1.116

1,148

0.2

−0.224

0.066

0.0006

0.799

0.703

0.909

HCC78: Parkinson's and Huntington's Diseases

19,341

3.3

−0.058

0.016

0.0004

0.943

0.913

0.974

HCC80: Coma, Brain Compression/Anoxic Damage

19,430

3.3

−0.144

0.018

<.0001

0.866

0.836

0.897

HCC82: Respirator Dependence/Tracheostomy Status

3,322

0.6

−0.052

0.038

0.1666

0.949

0.882

1.022

147,151

24.7

−0.085

0.008

<.0001

0.918

0.905

0.932

12,259

2.1

−0.100

0.020

<.0001

0.905

0.871

0.941

HCC87: Unstable Angina and Other Acute Ischemic Heart Disease*

9,721

1.6

−0.046

0.023

0.0413

0.955

0.914

0.998

HCC90: Heart Infection/Inflammation, Except Rheumatic*

5,209

0.9

−0.088

0.031

0.0039

0.916

0.862

0.972

53,766

9.0

−0.026

0.010

0.0103

0.974

0.955

0.994

177,247

29.8

−0.088

0.007

<.0001

0.916

0.904

0.928

9,314

1.6

−0.151

0.024

<.0001

0.860

0.821

0.901

HCC60: Personality Disorders
HCC61: Depression

HCC73: Amyotrophic Lateral Sclerosis and Other Motor Neuron
Disease*
HCC74: Cerebral Palsy

HCC85: Congestive Heart Failure
HCC86: Acute Myocardial Infarction*

HCC91: Valvular and Rheumatic Heart Disease*
HCC96: Specified Heart Arrhythmias
HCC99: Cerebral Hemorrhage*

(continued)

109

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

HCC100: Ischemic or Unspecified Stroke*

16,066

2.7

−0.193

0.020

<.0001

0.824

0.793

0.857

HCC103: Hemiplegia/Hemiparesis

82,270

13.8

−0.146

0.011

<.0001

0.864

0.846

0.883

2,549

0.4

−0.030

0.045

0.5036

0.971

0.889

1.059

25,984

4.4

−0.073

0.015

<.0001

0.929

0.902

0.957

8,107

1.4

−0.233

0.029

<.0001

0.792

0.748

0.838

HCC107: Vascular Disease with Complications*

10,636

1.8

−0.065

0.022

0.0025

0.937

0.898

0.977

HCC108: Vascular Disease*

70,461

11.9

−0.042

0.009

<.0001

0.959

0.941

0.976

HCC109: Other Circulatory Disease*

52,711

8.9

−0.004

0.010

0.6912

0.996

0.976

1.016

119,087

20.0

−0.086

0.008

<.0001

0.918

0.904

0.932

HCC114: Aspiration and Specified Bacterial Pneumonias*

20,928

3.5

−0.062

0.016

0.0001

0.940

0.911

0.970

HCC116: Viral and Unspecified Pneumonia, Pleurisy*

35,314

5.9

−0.007

0.013

0.5576

0.993

0.968

1.017

HCC117: Pleural Effusion/Pneumothorax*

23,621

4.0

−0.041

0.015

0.0073

0.960

0.932

0.989

151

0.03

−0.149

0.174

0.3918

0.862

0.613

1.211

HCC132: Kidney Transplant Status

4,575

0.8

−0.300

0.034

<.0001

0.741

0.694

0.791

HCC133: End-Stage Renal Disease

27,548

4.6

−0.364

0.015

<.0001

0.695

0.675

0.716

445

0.1

−0.202

0.102

0.0472

0.817

0.670

0.998

101,623

17.1

−0.134

0.009

<.0001

0.875

0.860

0.890

638

0.1

−0.353

0.084

<.0001

0.703

0.596

0.828

6,523

1.1

−0.202

0.027

<.0001

0.817

0.775

0.862

HCC138: Chronic Kidney Disease, Moderate (Stage 3)

31,128

5.2

−0.026

0.013

0.0525

0.975

0.950

1.000

HCC139: Chronic Kidney Disease, Mild or Unspecified (Stages 1-2 or
Unspecified)

24,845

4.2

−0.067

0.015

<.0001

0.935

0.909

0.962

HCC142: Urinary Obstruction and Retention*

53,230

9.0

−0.051

0.010

<.0001

0.951

0.932

0.970

HCC104: Monoplegia, Other Paralytic Syndromes
HCC105: Late Effects of Cerebrovascular Disease, Except Paralysis
HCC106: Atherosclerosis of the Extremities with Ulceration or
Gangrene*

HCC110 - HCC112: Cystic Fibrosis; Chronic Obstructive Pulmonary
Disease; Fibrosis of Lung and Other Chronic Lung Disorders

HCC120: Major Eye Infections/Inflammations*

HCC134: Dialysis Status
HCC135; HCC140: Acute Renal Failure; Unspecified Renal Failure
HCC136: Chronic Kidney Disease (Stage 5)
HCC137: Chronic Kidney Disease, Severe (Stage 4)

(continued)

110

Estimate

SE1

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

HCC143: Urinary Incontinence*

22,227

3.7

−0.020

0.015

0.1931

0.980

0.951

1.010

HCC144: Urinary Tract Infection*

82,781

13.9

−0.058

0.009

<.0001

0.944

0.927

0.961

HCC157; HCC158: Pressure Ulcer of Skin with Necrosis Through to
Muscle, Tendon, or Bone; Pressure Ulcer of Skin with Full Thickness
Skin Loss*

3,998

0.7

−0.269

0.034

<.0001

0.764

0.715

0.817

HCC159: Pressure Ulcer of Skin with Partial Thickness Skin Loss*

5,120

0.9

−0.283

0.030

<.0001

0.753

0.711

0.798

HCC160: Pressure Pre-Ulcer Skin Changes or Unspecified Stage*

4,739

0.8

−0.225

0.031

<.0001

0.799

0.752

0.849

11,066

1.9

−0.124

0.022

<.0001

0.883

0.845

0.922

196

0.03

−0.071

0.153

0.6453

0.932

0.690

1.258

21,169

3.6

−0.070

0.016

<.0001

0.933

0.904

0.963

8,872

1.5

−0.104

0.026

<.0001

0.901

0.856

0.948

HCC169: Vertebral Fractures without Spinal Cord Injury*

13,949

2.3

−0.056

0.020

0.005

0.945

0.909

0.983

HCC170: Hip Fracture/Dislocation*

11,610

2.0

−0.169

0.022

<.0001

0.844

0.809

0.882

8,797

1.5

−0.201

0.025

<.0001

0.818

0.779

0.859

165,242

27.8

−0.093

0.010

<.0001

0.912

0.895

0.929

11,794

2.0

−0.087

0.021

<.0001

0.917

0.880

0.954

HCC178: Major Symptoms, Abnormalities*

264,309

44.4

−0.052

0.007

<.0001

0.950

0.937

0.962

HCC179: Minor Symptoms, Signs, Findings*

101,062

17.0

−0.048

0.009

<.0001

0.953

0.938

0.970

7,624

1.3

−0.148

0.025

<.0001

0.862

0.821

0.905

HCC189; HCC190: Amputation Status, Lower Limb/Amputation
Complications; Amputation Status, Upper Limb

10,966

1.8

−0.028

0.022

0.2082

0.972

0.931

1.016

HCC197: Supplemental Oxygen

19,741

3.3

−0.089

0.017

<.0001

0.915

0.886

0.945

HCC199: Patient Lifts, Power Operated Vehicles, Beds*

1,344

0.2

−0.061

0.058

0.2928

0.941

0.840

1.054

HCC200: Wheelchairs, Commodes*

3,362

0.6

−0.087

0.038

0.02

0.917

0.852

0.986

HCC161: Chronic Ulcer of Skin, Except Pressure
HCC162: Severe Skin Burn or Condition
HCC164: Cellulitis, Local Skin Infection*
HCC166; HCC167: Severe or Major Head Injury*

HCC171: Major Fracture, Except of Skull, Vertebrae, or Hip*
HCC174: Other Injuries*
HCC176: Complications of Specified Implanted Device or Graft*

HCC188: Artificial Openings for Feeding or Elimination*

(continued)

111

Estimate

SE1

Acute History: Number of Hospital Stays in Past Year, Excluding Most Recent Stay (Reference: No stays)
1 Stay - Acute history
138,336
23.3
−0.423
0.007

pvalue

Odds Ratio OR 95%
OR 95%
(OR)
Lower CL2 Upper CL

<.0001

0.655

0.646

0.664

2 Stays - Acute history

42,380

7.1

−0.426

0.011

<.0001

0.653

0.639

0.668

3 Stays - Acute history

34,795

5.9

−0.849

0.012

<.0001

0.428

0.418

0.438

4 Stays - Acute history

11,464

1.9

−0.694

0.020

<.0001

0.499

0.480

0.520

5 Stays - Acute history

11,014

1.9

−1.042

0.021

<.0001

0.353

0.339

0.367

6 Stays - Acute history

6,150

1.0

−1.096

0.027

<.0001

0.334

0.317

0.352

7 Stays - Acute history

4,818

0.8

−1.223

0.031

<.0001

0.294

0.277

0.313

8 Stays - Acute history

2,268

0.4

−1.027

0.044

<.0001

0.358

0.328

0.390

9 Stays - Acute history

2,874

0.5

−1.370

0.041

<.0001

0.254

0.235

0.275

10+ Stays - Acute history

7,413

1.2

−1.476

0.026

<.0001

0.229

0.217

0.241

SE = Standard Error; 2 CL = Confidence Limit.

Source: RTI International analysis of Medicare claims data (program reference: MM130 Model 3)

112

Table B-2.
Inpatient Rehabilitation Facility: Facility-Level Observed and Risk-Standardized Discharge to Community Rates, 2015-2016
Discharge to
Community Rate

Mean

Min

1st pctl

5th
pctl

10th
pctl

25th
pctl

50th pctl
(Median) 75th pctl

90th
pctl

95th
pctl

99th
pctl

Max

Observed

65.10

7.83

45.51

52.44

55.97

60.52

65.35

69.95

73.82

77.20

82.23

100.00

Risk-Standardized

64.46

5.35

42.44

50.19

55.39

57.79

60.97

64.74

67.94

70.99

72.79

76.50

84.26

113

Figure B-1.
Inpatient Rehabilitation Facility: Facility-Level Observed and Risk-Standardized Discharge to Community Rates, 2015-2016

35%

Risk-Standardized Rate

30%

400
350

Observed Rate

300
Number of IRFs

Percent of IRFs

25%

250

20%

200
15%

150
10%

100

5%
0%

100

IRF Facility-Level Discharge to Community Rate

NOTE: Based on CY 2015-2016 Medicare fee-for-service claims data from 1,158 IRFs. Facility-level number of IRFs stays ranged from 1 to 6,469 with a
mean of 514.6 and median of 341.0. Solid bars represent the observed rate distribution; striped bars represent the risk-standardized rate distribution; the
overlap between solid and striped bars represents the overlap between observed and risk-standardized rate distributions. Source: RTI International analysis
(program reference: MM130).

114

APPENDIX C:
National Beta Test Supplementary Tables
The reference tables in this appendix refer to the SPADEs tested in the National Field Test.
Alphanumeric item numbers (Example: b1a, b1b, b1c) refer to the items as labeled in the assessment
protocols, which are available for download here: https://www.cms.gov/Medicare/Quality-InitiativesPatient-Assessment-Instruments/Post-Acute-Care-Quality-Initiatives/IMPACT-Act-of-2014/-IMPACTAct-Standardized-Assessment-National-Testing-.html

Table 1.1: Assessment counts for National Beta Test results
HHA
N=35

IRF
N=22

LTCH
N=26

SNF
N=60

Overall
N=143

Admission

653

794

507

1167

3121

Time to Complete (Facility/Agency Staff only)

469

549

386

565

1969

IRR

198

261

242

274

976

Table 1.2: Frequency and percent of assessments completed of each module
Module

Domains

Frequency

Percent

Communicative, N=3121
A1-A2

Hearing and Vision

3065

98.2

Brief Interview for Mental Status (BIMS)

3062

98.1

Pain Interview

3031

97.1

PHQ-2 to 9

3010

96.4

Confusion Assessment Method (CAM)

2973

95.3

Medication Reconciliation Protocol

2951

94.6

Special Services, Treatments, and Interventions (SSTI)

2926

93.8

All modules

At least one response in each module

2795

89.2

NOTE: Percentage of assessments are based on assessments used in the frequency tables where “completed” means
responded to at least one data element.

115

COGNITIVE STATUS: Brief Interview for Mental Status (BIMS)
Table 2.1.1: Admission response distributions (in percent) for BIMS items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

646

786

496

1134

3062

Three

Two

One

None or no answer

Missed by 1 year

Missed by 2-5 years

Missed by >5 years or no answer

Accurate within 5 days

Missed by 6 days - 1 mo

Missed by >1 mo or no answer

Accurate

Incorrect or no answer

Yes, no cue required

Yes after cue

Yes, no cue required

Yes, after cue

No recall or answer

# of words repeated after 1st attempt (b1a)

Recalls current year (b1b)
Correct

Recalls current month (b1c)

Recalls current day of week (b1d)

Recalls 'sock' (b1e)
Yes, no cue required
Yes, after cue
No recall or answer
Recalls 'blue' (b1f)

No recall or answer
Recalls 'bed' (b1g)

116

Items

HHA

IRF

LTCH

SNF

Overall

BIMS Impairment Category
Intact

Moderately impaired

Severely impaired

Table 2.1.2: IRR Kappa/Weighted Kappa and Percent Agreement for BIMS Items
Items

HHA

IRF

LTCH

SNF

Overall

# of patients

199

259

238

270

966

Kappa/weighted kappa
-

0.88

0.90

0.93

0.90

0.89

0.86

Recalls current day of week (b1d)

0.92

0.81

0.91

0.86

0.88

Recalls 'sock' (b1e)

0.87

0.91

Recalls 'blue' (b1f)

0.84

0.82

0.87

0.78

0.83

Recalls 'bed' (b1g)

0.93

0.90

0.93

BIMS Impairment Category

0.94

0.85

0.91

# of words repeated after 1st attempt (b1a)

Recalls current year (b1b)

Recalls current month (b1c)

Recalls current day of week (b1d)

Recalls 'sock' (b1e)

Recalls 'blue' (b1f)

Recalls 'bed' (b1g)

BIMS Impairment Category

# of words repeated after 1st attempt (b1a)
Recalls current year (b1b)
Recalls current month (b1c)

Percent agreement

NOTE: Interrater reliability not shown for items with proportions out of range for stable kappa estimate (per study
power calculations). Interpretation of kappa or weighted kappa is as follows: 0.00-0.20: slight/poor; 0.21-0.40: fair;
0.41-0.60: moderate; 0.61-0.80: substantial/good; 0.81-1.00: excellent/almost perfect.

117

COGNITIVE STATUS: Confusion Assessment Method (CAM)
Table 2.2.1: Admission response distributions (in percent) for CAM items
Items

HHA

IRF

LTCH

SNF

Overall

# of assessments

630

771

471

1101

2973

Evidence of change in mental status from baseline (b2a)
5

Behavior continuously present

Behavior present, fluctuates

Behavior continuously present

Behavior present, fluctuates

Behavior continuously present

Behavior present, fluctuates

Yes
Did patient have difficulty focusing attn (b2b)
Behavior not present

Was patient thinking disorganized (b2c)
Behavior not present

Did patient have altered consciousness (b2d)
Behavior not present

Table 2.2.2: IRR Kappa/Weighted Kappa and Percent Agreement for CAM items
Items

HHA

IRF

LTCH

SNF

Overall

# of patients

189

245

223

257

914

Kappa/weighted kappa
-

0.60

0.66

0.55

0.75

0.70

0.66

Was patient thinking disorganized (b2c)

0.68

Did patient have altered consciousness (b2d)

Evidence of change in mental status from baseline (b2a)

Did patient have difficulty focusing attn (b2b)

Was patient thinking disorganized (b2c)

Did patient have altered consciousness (b2d)

Evidence of change in mental status from baseline (b2a)
Did patient have difficulty focusing attn (b2b)

Percent agreement

118

Mental Status: PHQ-2 to 9
Table 3.1.1: Admission response distribution (in percent) for PHQ-2 to 9 items
Items

HHA

IRF

LTCH

SNF

Overall

639

776

479

1116

3010

0-1 day

2-6 days

7-11 days (half or more)

12-14 days (nearly all)

0-1 day

2-6 days

7-11 days (half or more)

12-14 days (nearly all)

2.2
(1.6)

2.3
(1.7)

2.7
(1.8)

153

209

182

306

850

0-1 day

2-6 days

7-11 days (half or more)

12-14 days (nearly all)

0-1 day

2-6 days

7-11 days (half or more)

12-14 days (nearly all)

# of assessments
Symptom presence & frequency: little interest or pleasure (e1a)
No

Symptom presence & frequency: feeling down, depressed,
hopeless (e1b)
No

PHQ-2
Mean (SD)

2.4
2.4 (1.7)
(1.7)

Eligible for PHQ-9 per PHQ-2
Yes
# of assessments eligible for PHQ-9 per PHQ-2
Symptom presence & frequency: too little/too much sleep (e1c)
No

Symptom presence & frequency: tired / no energy (e1d)
No

Symptom presence & frequency: poor appetite or overeating
(e1e)
No
0-1 day

119

Items

HHA

IRF

LTCH

SNF

Overall

7-11 days (half or more)

12-14 days (nearly all)

0-1 day

2-6 days

7-11 days (half or more)

12-14 days (nearly all)

0-1 day

2-6 days

7-11 days (half or more)

12-14 days (nearly all)

0-1 day

2-6 days

7-11 days (half or more)

0-1 day

2-6 days

7-11 days (half or more)

12-14 days (nearly all)

11.4
(5.0)

11.8
(5.3)

13.0
(5.8)

11.5
(5.1)

11.9
(5.3)

None (0 – 4)

Mild (5 – 9)

Moderate (10 – 14)

Moderately severe (15 – 19)

2-6 days

Symptom presence & frequency: feel bad about self (e1f)
No

Symptom presence & frequency: trouble concentrating (e1g)
No

Symptom presence & frequency: moving or speaking slowly
(e1h)
No

12-14 days (nearly all)
Symptom presence & frequency: suicidal thoughts (e1i)
No

PHQ-9
Mean (SD)
Depression categorization (PHQ-9)

Severe (20 – 27)

120

Table 3.1.2: IRR Kappa/Weighted Kappa and Percent Agreement for PHQ-2 to 9 items
Items

HHA

IRF

LTCH

SNF

Overall

196

254

231

267

948

Symptom present: little interest or pleasure (e1a1)

0.95

0.99

0.98

Symptom frequency: little interest or pleasure (e1a2)

0.98

1.00

0.98

0.99

Symptom present: feeling down, depressed, hopeless (e1b1)

0.99

0.98

1.00

0.99

Symptom frequency: feeling down, depressed, hopeless
(e1b2)

0.93

0.98

0.99

0.98

Eligible for PHQ-9 per PHQ-2

0.96

0.98

Symptom present: too little/too much sleep (e1c1)

0.90

1.00

0.98

Symptom frequency: too little/too much sleep (e1c2)

1.00

0.98

0.90

0.96

Symptom present: tired / no energy (e1d1)

1.00

0.91

0.95

0.94

0.95

Symptom frequency: tired / no energy (e1d2)

1.00

0.93

0.98

1.00

0.98

Symptom present: poor appetite or overeating (e1e1)

0.96

0.93

0.95

1.00

0.96

Symptom frequency: poor appetite or overeating (e1e2)

1.00

Symptom present: feel bad about self (e1f1)

1.00

Symptom frequency: feel bad about self (e1f2)

1.00

0.95

1.00

0.98

Symptom present: trouble concentrating (e1g1)

1.00

0.97

0.99

Symptom frequency: trouble concentrating (e1g2)

0.96

0.97

0.94

1.00

0.97

Symptom present: moving or speaking slowly (e1h1)

1.00

0.94

0.90

1.00

0.95

Symptom frequency: moving or speaking slowly (e1h2)

1.00

0.87

1.00

0.97

Symptom present: suicidal thoughts (e1i1)

1.00

0.94

1.00

0.98

Symptom frequency: suicidal thoughts (e1i2)

0.93

1.00

0.95

1.00

0.97

Sum of all symptom frequencies (PHQ-9) *

0.97

0.95

0.97

0.96

Symptom present: little interest or pleasure (e1a1)

100

Symptom frequency: little interest or pleasure (e1a2)

100

Symptom present: feeling down, depressed, hopeless (e1b1)

100

Symptom frequency: feeling down, depressed, hopeless
(e1b2)

Eligible for PHQ-9 per PHQ-2

Symptom present: too little/too much sleep (e1c1)

100

Symptom frequency: too little/too much sleep (e1c2)

100

Symptom present: tired / no energy (e1d1)

100

Symptom frequency: tired / no energy (e1d2)

100

Symptom frequency: poor appetite or overeating (e1e2)

100

Symptom present: feel bad about self (e1f1)

100

Symptom frequency: feel bad about self (e1f2)

100

# of patients
Kappa/weighted kappa

Percent Agreement

Symptom present: poor appetite or overeating (e1e1)

121

Items

HHA

IRF

LTCH

SNF

Overall

100

Symptom present: moving or speaking slowly (e1h1)

100

Symptom frequency: moving or speaking slowly (e1h2)

100

Symptom present: suicidal thoughts (e1i1)

100

Symptom frequency: suicidal thoughts (e1i2)

100

Sum of all symptom frequencies (PHQ-9)*

Symptom present: trouble concentrating (e1g1)
Symptom frequency: trouble concentrating (e1g2)

NOTE: As classified into the five categories shown in Table 3.1.1. Interpretation of kappa or weighted kappa is as
follows: 0.00-0.20: slight/poor; 0.21-0.40: fair; 0.41-0.60: moderate; 0.61-0.80: substantial/good; 0.81-1.00:
excellent/almost perfect.

Special Services, Treatment and Interventions (SSTI)
Table 4.1.1: Admission response distributions (in percent) for SSTI - Chemotherapy items
Items

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Chemotherapy (j2a)

Chemo treatment performed: IV (j2a2a)

Chemo treatment performed: oral (j2a3a)

Chemo treatment performed: other (j2a10a)

# of assessments

122

Table 4.1.2: IRR Kappa and Percent Agreement for SSTI – Chemotherapy items
Items

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

Noted treatment performed: Chemotherapy (j2a)

Noted chemo treatment performed: IV (j2a2a)

Noted chemo treatment performed: oral (j2a3a)

Noted chemo treatment performed: other (j2a10a)

100

Noted chemo treatment performed: IV (j2a2a)

100

Noted chemo treatment performed: oral (j2a3a)

100

Noted chemo treatment performed: other (j2a10a)

100

# of patients
Kappa/Weighted kappa

Percent Agreement
Noted treatment performed: Chemotherapy (j2a)

NOTE: Based on dichotomized never noted vs. noted any day. Interrater reliability not shown for items with
proportions out of range for stable kappa estimate (per study power calculations). Interpretation of kappa or
weighted kappa is as follows: 0.00-0.20: slight/poor; 0.21-0.40: fair; 0.41-0.60: moderate; 0.61-0.80:
substantial/good; 0.81-1.00: excellent/almost perfect.

Table 4.2.1: Admission response distributions (in percent) for SSTI - Radiation
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Radiation (j2b)

Table 4.2.2: IRR Kappa and Percent Agreement for SSTI – Radiation item
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

100

Kappa/Weighted kappa
Noted treatment performed: Radiation (j2b)
Percent Agreement
Noted treatment performed: Radiation (j2b)

123

Table 4.3.1: Admission response distributions (in percent) for SSTI – Oxygen Therapy
items
Items

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Type of O2 therapy performed: intermittent (j2c2a)

Type of O2 therapy performed: continuous (j2c3a)

Type of O2 therapy performed: high-concentration (j2c4a)

# of assessments
Treatment performed: Oxygen Therapy (j2c)

Table 4.3.2: IRR Kappa and Percent Agreement for SSTI – Oxygen Therapy items
Items

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

0.82

0.80

0.86

0.71

0.82

Type of O2 therapy performed: intermittent (j2c2a)

0.76

0.82

0.75

0.81

Type of O2 therapy performed: continuous (j2c3a)

0.68

0.35

0.55

Type of O2 therapy performed: high-concentration (j2c4a)

Treatment performed: Oxygen Therapy (j2c)

Type of O2 therapy performed: intermittent (j2c2a)

Type of O2 therapy performed: continuous (j2c3a)

100

# of patients
Kappa
Treatment performed: Oxygen Therapy (j2c)

Percent Agreement

Type of O2 therapy performed: high-concentration (j2c4a)

Table 4.4.1: Admission response distributions (in percent) for SSTI – Suctioning items
Items

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Suctioning (j2d)

Type of suctioning performed: scheduled (j2d2a)

Type of suctioning performed: as needed (j2d3a)

# of assessments

124

Table 4.4.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI – Suctioning
items
Items

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

Treatment performed: Suctioning (j2d)

Type of suctioning performed: scheduled (j2d2a)

Type of suctioning performed: as needed (j2d3a)

Type of suctioning performed: scheduled (j2d2a)

100

Type of suctioning performed: as needed (j2d3a)

100

# of patients
Kappa

Percent Agreement
Treatment performed: Suctioning (j2d)

Table 4.5.1: Admission response distributions (in percent) for SSTI – Tracheostomy Care
item
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Tracheostomy Care (j2e)

Table 4.5.2: IRR Kappa and Percent Agreement for SSTI – Tracheostomy Care item
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

100

Kappa/Weighted kappa
Treatment performed: Tracheostomy Care (j2e)
Percent Agreement
Treatment performed: Tracheostomy Care (j2e)

125

Table 4.6.1: Admission response distributions (in percent) for SSTI - Non-invasive
Mechanical Ventilator (NIMV)
Items

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Non-invasive Mechanical Ventilator
(j2g)

Type of NIMV performed: BiPAP (j2g2a)

Type of NIMV performed: CPAP (j2g3a)

# of assessments

Table 4.6.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI – Non-invasive
Mechanical Ventilator (NIMV) items
Items

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

Treatment performed: Non-invasive Mechanical Ventilator
(j2g)

0.77

Type of NIMV performed: BiPAP (j2g2a)

Type of NIMV performed: CPAP (j2g3a)

Treatment performed: Non-invasive Mechanical Ventilator
(j2g)

Type of NIMV performed: BiPAP (j2g2a)

100

Type of NIMV performed: CPAP (j2g3a)

# of patients
Kappa

Percent Agreement

Table 4.7.1: Admission response distributions (in percent) for SSTI - Invasive Mechanical
Ventilator item
Items
# of assessments
Treatment performed: Invasive Mechanical Ventilator (j2f)

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Table 4.7.2: IRR Kappa and Percent Agreement for SSTI – Invasive Mechanical Ventilator
126

item
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

100

Kappa
Treatment performed: Invasive Mechanical Ventilator (j2f)
Percent Agreement
Treatment performed: Invasive Mechanical Ventilator (j2f)

Table 4.8.1: Admission response distributions (in percent) for SSTI – IV Meds items
Items

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Other treatment performed: IV Meds (j2h)

Type of IV meds given: antibiotics (j2h3a)

Type of IV meds given: anticoagulation (j2h4a)

Type of IV meds given: other (j2h10a)

# of assessments

127

Table 4.8.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI – IV Meds
items
Items

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

Other treatment performed: IV Meds (j2h)

0.15

0.61

0.68

0.52

0.70

Type of IV meds given: antibiotics (j2h3a)

0.84

0.78

0.88

Type of IV meds given: anticoagulation (j2h4a)

0.13

Type of IV meds given: other (j2h10a)

0.46

Other treatment performed: IV Meds (j2h)

Type of IV meds given: antibiotics (j2h3a)

Type of IV meds given: anticoagulation (j2h4a)

Type of IV meds given: other (j2h10a)

# of patients
Kappa

Percent Agreement

Table 4.9.1: Admission response distributions (in percent) for SSTI – Transfusions item
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Other treatment performed: Transfusions (j2i)

Table 4.9.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI – Transfusions
item
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

100

Kappa
Other treatment performed: Transfusions (j2i)
Percent Agreement
Other treatment performed: Transfusions (j2i)

128

Table 4.10.1: Admission response distributions (in percent) for SSTI – Dialysis items
Items

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Other treatment performed: Dialysis (j2j)

Type of dialysis performed: hemodialysis (j2j2a)

Type of dialysis performed: peritoneal (j2j3a)

# of assessments

Table 4.10.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI – Dialysis
items
Items

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

Other treatment performed: Dialysis (j2j)

0.92

Type of dialysis performed: hemodialysis (j2j2a)

0.90

Type of dialysis performed: peritoneal (j2j3a)

Other treatment performed: Dialysis (j2j)

Type of dialysis performed: hemodialysis (j2j2a)

100

# of patients
Kappa

Percent Agreement

Type of dialysis performed: peritoneal (j2j3a)

Table 4.11.1: Admission response distributions (in percent) for SSTI – IV Access
Items

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Other treatment performed: IV Access (j2k)

Type of IV access: peripheral IV (j2k2a)

Type of IV access: midline (j2k3a)

Type of IV access: central line (j2k4a)

Type of IV access: other (j2k10a)

# of assessments

129

Table 4.11.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI – IV Access
items
Items

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

Other treatment performed: IV Access (j2k)

0.81

0.74

0.90

Type of IV access: peripheral IV (j2k2a)

0.81

0.77

0.81

Type of IV access: midline (j2k3a)

0.75

Type of IV access: central line (j2k4a)

0.78

0.85

Type of IV access: other (j2k10a)

Type of IV access: peripheral IV (j2k2a)

100

Type of IV access: midline (j2k3a)

100

Type of IV access: central line (j2k4a)

Type of IV access: other (j2k10a)

# of patients
Kappa

Percent Agreement
Other treatment performed: IV Access (j2k)

Nutritional Approaches
Table 5.1.1: Admission response distributions (in percent) for Nutritional Approaches –
Parenteral / IV Feeding Tube
Items
# of assessments
Nutritional approach performed: parenteral/IV (j1a)

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

130

Table 5.1.2: IRR Kappa/Weighted Kappa and Percent Agreement for Nutritional
Approaches – Parenteral/IV Feeding Tube
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

100

Kappa
Nutritional approach performed: parenteral/IV (j1a)
Percent Agreement
Nutritional approach performed: parenteral/IV (j1a)

Table 5.2.1: Admission response distributions (in percent) for Nutritional Approaches –
Feeding Tube
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Nutritional approach performed: feeding tube (j1b)

Table 5.2.2: IRR Kappa/Weighted Kappa and Percent Agreement for Nutritional
Approaches – Feeding Tube
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

100

Kappa
Nutritional approach performed: feeding tube (j1b)
Percent Agreement
Nutritional approach performed: feeding tube (j1b)

131

Table 5.3.1: Admission response distributions (in percent) for Nutritional Approaches –
Mechanically Altered Diet
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Nutritional approach performed: mechanically altered diet (j1c)

Table 5.3.2: IRR Kappa/Weighted Kappa and Percent Agreement for Nutritional
Approaches – Mechanically Altered Diet
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

0.53

0.69

0.70

0.65

100

Kappa
Nutritional approach performed: mechanically altered diet
(j1c)
Percent Agreement
Nutritional approach performed: mechanically altered diet
(j1c)

132

Table 5.4.1: Admission response distributions (in percent) for Nutritional Approaches –
Therapeutic Diet
Items

HHA

# of assessments
Nutritional approach performed: therapeutic diet (j1d)

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Table 5.4.2: IRR Kappa/Weighted Kappa and Percent Agreement for Nutritional
Approaches – Therapeutic Diet
Items

HHA

# of patients

IRF

LTCH

SNF

Overall

187

236

203

256

882

0.43

0.70

0.62

0.61

0.60

Kappa
Nutritional approach performed: therapeutic diet (j1d)
Percent Agreement
Nutritional approach performed: therapeutic diet (j1d)

High-Risk Drug Classes: Use and Indication items
Table 6.1.1: Admission response distributions (in percent) for Medication Class Taking and
Indication Items
Medication Class

HHA
(627)

IRF
(769)

LTCH
(459)

Overall
(2951)

SNF
(1096)

Taking Indication Taking Indication Taking Indication Taking Indication Taking Indication
(Percent) (Percent) (Percent) (Percent) (Percent) (Percent) (Percent) (Percent) (Percent) (Percent)

Anticoagulants

Antiplatelets

Hypoglycemics

Opioids

Antipsychotics

Antimicrobials

NOTE: Indication (percent) reflects percent with indication among those taking medications in that class

133

Table 6.1.2: IRR Kappa and Percent Agreement for Medication Class Taking and
Indication items
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

240

212

261

900

Kappa
Is patient taking: anticoagulants (i1a1)

0.78

0.84

0.87

0.85

Is patient taking: antiplatelets (i1a2)

0.69

0.71

0.83

0.72

Is patient taking: hypoglycemics (i1a3)

0.83

0.80

0.97

0.90

0.89

Is patient taking: opioids (i1a4)

0.84

0.86

0.90

0.85

0.86

Is patient taking: antipsychotics (i1a5)

Is patient taking: antimicrobials (i1a6)

0.76

0.93

0.82

0.86

Indication noted for anticoagulants (i1b1)

0.54

0.64

0.80

0.87

0.78

Indication noted for antiplatelets (i1b2)

0.69

0.85

0.89

0.87

Indication noted for hypoglycemics (i1b3)

0.39

0.62

0.70

0.75

0.65

Indication noted for antipsychotics (i1b5)

0.33

1.00

0.88

0.73

0.81

Indication noted for antimicrobials (i1b6)

0.74

0.63

0.72

0.81

Is patient taking: anticoagulants (i1a1)

Is patient taking: antiplatelets (i1a2)

Is patient taking: hypoglycemics (i1a3)

Is patient taking: opioids (i1a4)

Is patient taking: antipsychotics (i1a5)

Is patient taking: antimicrobials (i1a6)

Indication noted for all meds in class (i1b1-6)

Indication noted for anticoagulants (i1b1)

Indication noted for antiplatelets (i1b2)

100

Indication noted for hypoglycemics (i1b3)

Indication noted for opioids (i1b4)

100

Indication noted for antipsychotics (i1b5)

100

Indication noted for antimicrobials (i1b6)

Indication noted for opioids (i1b4)

Percent Agreement

134

PAIN: Pain Interference
Table 7.1.1: Admission response distributions (in percent) for Pain Interference items
among patients/residents reporting any pain in the last 3 days or 5 days
Items

HHA

IRF

LTCH

SNF

Overall

489

618

375

872

2354

Rarely or not at all

Occasionally

Frequently

Almost constantly

379

606

302

803

2090

Rarely or not at all

Occasionall

Frequently

Almost constantly

Rarely or not at all

Occasionally

Frequently

Almost constantly

# of assessments
How often pain made it hard to sleep
(d3)

Offered rehab therapies (d4a)
Yes
Yes N
How often limited rehab due to pain
(d4b)

How often limited daily activities due to
pain (d4c)

135

Table 7.1.2: IRR Kappa/Weighted Kappa and Percent Agreement for Pain Interference
items
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

197

256

232

268

953

How often pain made it hard to sleep (d3)

0.96

0.98

0.99

0.98

How often limited rehab due to pain (d4b)

0.95

0.96

0.98

0.97

How often limited daily activities due to pain (d4c)

0.97

0.98

0.99

0.98

How often pain made it hard to sleep (d3)

100

How often limited rehab due to pain (d4b)

How often limited daily activities due to pain (d4c)

Kappa

Percent Agreement

IMPAIRMENTS: Hearing
Table 8.1.1: Admission response distributions (in percent) for Hearing item
Items

HHA

IRF

LTCH

SNF

Overall

643

783

498

1141

3065

Adequate

Minimal difficulty

Moderate difficulty

# of assessments
Ability to hear (a1)

Highly impaired

136

Table 8.1.2: IRR Weighted Kappa and Percent Agreement for Hearing item
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

197

258

237

268

960

0.71

0.67

0.58

0.62

0.65

Weighted kappa
Ability to hear (a1)
Percent agreement
Ability to hear (a1)

NOTE: Interpretation of kappa or weighted kappa is as follows: 0.00-0.20: slight/poor; 0.21-0.40: fair; 0.41-0.60:
moderate; 0.61-0.80: substantial/good; 0.81-1.00: excellent/almost perfect.

IMPAIRMENTS: Vision
Table 9.2.1: Admission response distributions (in percent) for Vision item
Items

HHA

IRF

LTCH

SNF

Overall

643

783

498

1141

3065

Adequate

Impaired

Moderately impaired

Highly impaired

Severely impaired

# of assessments
Ability to see (a2)

Table 9.2.2: IRR Weighted Kappa and Percent Agreement for Vision item
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

197

258

237

268

960

0.67

0.50

0.47

0.57

0.56

Weighted kappa
Ability to see (a2)
Percent agreement
Ability to see (a2)

NOTE: Interpretation of kappa or weighted kappa is as follows: 0.00-0.20: slight/poor; 0.21-0.40: fair; 0.41-0.60:
moderate; 0.61-0.80: substantial/good; 0.81-1.00: excellent/almost perfect.

137

File Type	application/pdf
File Modified	0000-00-00
File Created	0000-00-00