Inpatient Rehabilitation Facility - Patient Assessment Instrument

July 2019

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

Prepared for
Center for Clinical Standards and Quality
and the Office of Minority Health
Centers for Medicare & Medicaid Services
Mail Stop C3-19-26
7500 Security Boulevard
Baltimore, MD 21244-1850
Prepared by
RTI International
3040 E. 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 Measure: Transfer of Health Information to the Provider–Post-Acute Care
Measure ..................................................................................................................................................... 2
Measure Description ............................................................................................................................. 2
Purpose/Rationale for the Quality Measure .......................................................................................... 3
Denominator ......................................................................................................................................... 5
Numerator ............................................................................................................................................. 5
Measure Time Window......................................................................................................................... 6
Items Included in the Quality Measure ................................................................................................. 6
Risk Adjustment.................................................................................................................................... 6
Quality Measure Calculation Steps ....................................................................................................... 6
Quality Measure Coding Steps ............................................................................................................. 7
Section 3. Cross-Setting Measure: Transfer of Health Information to the Patient–Post-Acute Care
Measure ..................................................................................................................................................... 8
Measure Description ............................................................................................................................. 8
Purpose/Rationale for the Quality Measure .......................................................................................... 8
Denominator ....................................................................................................................................... 10
Numerator ........................................................................................................................................... 10
Measure Time Window....................................................................................................................... 10
Items Included in the Quality Measure ............................................................................................... 10
Risk Adjustment.................................................................................................................................. 11
Quality Measure Calculation Steps ..................................................................................................... 11
Quality Measure Coding Steps ........................................................................................................... 11
Section 4. Update to the Discharge to Community–Post Acute Care (PAC) Inpatient Rehabilitation
Facility (IRF) Quality Reporting Program (QRP) Measure .................................................................... 12
Measure Update .................................................................................................................................. 12
Measure Description ........................................................................................................................... 12
Purpose/Rationale for the Measure ..................................................................................................... 12
Denominator ....................................................................................................................................... 16
Numerator ........................................................................................................................................... 16
Target Population and Measure Exclusions ........................................................................................ 18
Data Sources ....................................................................................................................................... 21
Measure Time Window....................................................................................................................... 21
Statistical Risk Model and Risk Adjustment Covariates .................................................................... 21

i

Measure Calculation Algorithm .......................................................................................................... 23
Chapter 2 Standardized Patient Assessment Data Elements ....................................................................... 25
Section 1: Introduction ............................................................................................................................ 25
Background ......................................................................................................................................... 25
National Beta Test............................................................................................................................... 26
Section 2: Cognitive Function................................................................................................................. 30
Brief Interview for Mental Status (BIMS) .......................................................................................... 31
Confusion Assessment Method (CAM©) ........................................................................................... 34
Mental Status (Depressed Mood) ........................................................................................................ 36
Patient Health Questionnaire-2 to 9 (PHQ-2 to 9) .............................................................................. 37
Section 3: Special Services, Treatments, and Interventions (Including Nutritional Approaches) .......... 43
Chemotherapy (IV, Oral, Other) ......................................................................................................... 43
Radiation ............................................................................................................................................. 46
Oxygen Therapy (Intermittent, Continuous, High-Concentration Oxygen Delivery System)............ 47
Suctioning (Scheduled, As Needed) ................................................................................................... 49
Tracheostomy Care ............................................................................................................................. 52
Non-invasive Mechanical Ventilation (Bilevel Positive Airway Pressure [BiPAP], Continuous
Positive Airway Pressure [CPAP]) ..................................................................................................... 54
Invasive Mechanical Ventilator .......................................................................................................... 56
IV Medications (Antibiotics, Anticoagulation, Vasoactive Medications, Other) ............................... 58
Transfusions ........................................................................................................................................ 60
Dialysis (Hemodialysis, Peritoneal dialysis) ...................................................................................... 62
IV Access (Peripheral IV, Midline, Central line) ............................................................................... 64
Parenteral/IV Feeding ......................................................................................................................... 66
Feeding Tube ...................................................................................................................................... 68
Mechanically Altered Diet .................................................................................................................. 70
Therapeutic Diet.................................................................................................................................. 72
High-Risk Drug Classes: Use and Indication...................................................................................... 75
Section 4: Medical Conditions and Co-Morbidities................................................................................ 79
Pain Interference ................................................................................................................................. 79
Section 5: Impairments ........................................................................................................................... 83
Hearing and Vision Impairments ........................................................................................................ 83
Hearing................................................................................................................................................ 83
Vision .................................................................................................................................................. 86
Section 6: New Category: Social Determinants of Health ...................................................................... 89
Standardized Data Elements to Assess for Social Determinants of Health ........................................ 89
Race and Ethnicity .............................................................................................................................. 89

ii

Preferred Language and Interpreter Services ...................................................................................... 91
Health Literacy.................................................................................................................................... 92
Transportation ..................................................................................................................................... 93
Social Isolation.................................................................................................................................... 94
APPENDIX A: Transfer of Health Information: Setting-Specific Language ......................................... 96
APPENDIX B: Discharge to Community–PAC IRF QRP Analyses ..................................................... 98
APPENDIX C: National Beta Test Supplementary Tables .................................................................. 111

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 the instrument as necessary
to enable such use. This requirement refers to the collection of such data by means of the IRF Patient
Assessment Instrument (IRF-PAI) for IRFs, the 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, and
the Outcome and Assessment Information Set (OASIS) for HHAs.
For more information on the statutory history of the IRF, LTCH, or SNF Quality Reporting
Program (QRP), please refer to the Fiscal Year (FY) 2016 final rules, and for the HH QRP, please refer to
the Calendar Year (CY) 2016 final rule. 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 two measures finalized for adoption for the IRF QRP through the FY 2020 IRF
Prospective Payment System (PPS) final 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.

1

Section 2. Cross-Setting Measure: Transfer of Health Information to the Provider–PostAcute Care Measure
Measure Description
This measure, the Transfer of Health Information to the Provider, assesses for and reports on the
timely transfer of health information, specifically transfer of a reconciled medication list. This measure
evaluates for the transfer of information when a patient/resident is discharged from their current setting to
a subsequent provider. For this 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 measure, developed under the IMPACT Act, has been developed conceptually for the IRF,
LTCH, SNF, and HHA settings. This measure is calculated by one standardized data element that asks,
“at the time of discharge, did the facility provide the patient’s/resident’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 IRFPAI 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 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 Appendix A.
The Reconciled Medication List
The Transfer of Health Information measures serve as a check to ensure that a reconciled
medication list is provided as the patient changes care settings at discharge. Defining the completeness of
that medication list is left to the discretion of the providers and patient who are coordinating this care.
An example of items that could be on a reconciled medication list can be but are not limited to a
list of the current prescribed and over-the-counter medications, nutritional supplements, vitamins, and/or
homeopathic and herbal products administered by any route at the time of discharge or transfer. A
reconciled medication could also include important information about: (1) the patient/resident, including
their name, date of birth, 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, and/or any special instructions.
However, this information serves as guidance and as stated prior, the completeness of the medication list
is left to the discretion of the providers and patient.
Documentation sources for reconciled medication list information include electronic and/or paper
records. Some examples of such records are discharge summary records, a Medication Administration
Record, an Intravenous Medication Administration Record, a home medication list, and physician orders.
The guidance on what to include in a reconciled medication list is aligned to the provisions in the
proposed Discharge Planning for Hospitals, Critical Access Hospital, and HHAs regulation, which
outlines discharge planning and the documentation of medications
(https://www.federalregister.gov/documents/2015/11/03/2015-27840/medicare-and-medicaid-programsrevisions-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
(https://www.federalregister.gov/documents/2016/10/04/2016-23503/medicare-and-medicaid-programsreform-of-requirements-for-long-term-care-facilities).

2

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 (HHA), 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 those, 20 percent
were discharged to a SNF, 18 percent were discharged to an HHA, 3 percent were discharged to an IRF,
and 1 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 takes several
forms, including verbal (e.g., clinician-to-clinician communication by telephone or in-person), paperbased (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
patient/resident safety risk, often life-threatening. 4 Poor communication and coordination across health
care settings contributes to patient complications, hospital readmissions, emergency department visits,
and medication errors. 5 Communication has been cited as the third-most-frequent root cause in sentinel
1

Tian, W. (2016, May). An all-payer view of hospital discharge to postacute care. Retrieved from https://www.hcupus.ahrq.gov/reports/statbriefs/sb205-Hospital-Discharge-Postacute-Care.jsp.
2 Ibid.
3 RTI International analysis of Medicare claims data for index stays in IRF 2016/2017. (RTI program reference: MM150).
4 Kwan, J. L., Lo, L., Sampson, M., & Shojania, K. G. (2013). Medication reconciliation during transitions of care as a patient
safety strategy: A systematic review. Annals of Internal Medicine, 158(5 Pt 2), 397–403. https://doi.org/10.7326/0003-4819158-5-201303051-00006
Boockvar, K. S., Blum, S., Kugler, A., Livote, E., Mergenhagen, K. A., Nebeker, J. R., . . . Yeh, J. (2011). Effect of admission
medication reconciliation on adverse drug events from admission medication changes. Archives of Internal Medicine, 171(9),
860–861. https://doi.org/10.1001/archinternmed.2011.163
Bell, C. M., Brener, S. S., Gunraj, N., Huo, C., Bierman, A. S., Scales, D. C., . . . Urbach, D. R. (2011). Association of ICU or
hospital admission with unintentional discontinuation of medications for chronic diseases. Journal of the American Medical
Association, 306(8), 840–847. https://doi.org/10.1001/jama.2011.1206
Basey, A. J., Krska, J., Kennedy, T. D., & Mackridge, A. J. (2014). Prescribing errors on admission to hospital and their potential
impact: A mixed-methods study. BMJ Quality & Safety, 23(1), 17–25. https://doi.org/10.1136/bmjqs-2013-001978
Desai, R., Williams, C. E., Greene, S. B., Pierson, S., & Hansen, R. A. (2011). Medication errors during patient transitions into
nursing homes: Characteristics and association with patient harm. The American Journal of Geriatric Pharmacotherapy, 9(6),
413–422. https://doi.org/10.1016/j.amjopharm.2011.10.005
Boling, P. A. (2009). Care transitions and home health care. Clinics in Geriatric Medicine, 25(1), 135–148.
https://doi.org/10.1016/j.cger.2008.11.005
5 Barnsteiner, J. H. (2005). Medication reconciliation: Transfer of medication information across settings-keeping it free from
error. The American Journal of Nursing, 105(3, Suppl), 31–36. https://doi.org/10.1097/00000446-200503001-00007
Arbaje, A. I., Kansagara, D. L., Salanitro, A. H., Englander, H. L., Kripalani, S., Jencks, S. F., & Lindquist, L. A. (2014).
Regardless of age: Incorporating principles from geriatric medicine to improve care transitions for patients with complex
needs. Journal of General Internal Medicine, 29(6), 932–939. https://doi.org/10.1007/s11606-013-2729-1
Jencks, S. F., Williams, M. V., & Coleman, E. A. (2009). Rehospitalizations among patients in the Medicare fee-for-service
program. The New England Journal of Medicine, 360(14), 1418–1428. https://doi.org/10.1056/NEJMsa0803563
Institute of Medicine. (2007). Preventing Medication Errors. Washington, DC: The National Academies Press.
https://doi.org/10.17226/11623.
Kitson, N. A., Price, M., Lau, F. Y., & Showler, G. (2013). Developing a medication communication framework across
continuums of care using the Circle of Care Modeling approach. BMC Health Services Research, 13(1), 418.
https://doi.org/10.1186/1472-6963-13-418
Mor, V., Intrator, O., Feng, Z., & Grabowski, D. C. (2010). The revolving door of rehospitalization from skilled nursing
facilities. Health Affairs, 29(1), 57–64. https://doi.org/10.1377/hlthaff.2009.0629

3

events, which The Joint Commission defines as a patient safety event that results in death, permanent
harm, or severe temporary harm. 6 Failed or ineffective patient handoffs are estimated to play a role in 20
percent of serious preventable adverse events. 7 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. 8 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 health care; care
coordination and person-centered care; and 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. 9 Individuals in PAC settings may be vulnerable to adverse health outcomes because of
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

Forster, A. J., Murff, H. J., Peterson, J. F., Gandhi, T. K., & Bates, D. W. (2003). The incidence and severity of adverse events
affecting patients after discharge from the hospital. Annals of Internal Medicine, 138(3), 161–167.
https://doi.org/10.7326/0003-4819-138-3-200302040-00007
King, B. J., Gilmore-Bykovskyi, A. L., Roiland, R. A., Polnaszek, B. E., Bowers, B. J., & Kind, A. J. (2013). The consequences
of poor communication during transitions from hospital to skilled nursing facility: A qualitative study. Journal of the
American Geriatrics Society, 61(7), 1095–1102. https://doi.org/10.1111/jgs.12328
6
The Joint Commission. (2017, June 29). Sentinel event policy and procedures. Retrieved from
https://www.jointcommission.org/sentinel_event_policy_and_procedures/
7 The Joint Commission. (2016, March 2). Sentinel event statistics updated, released through end of 2015Retrieved from
https://www.jointcommission.org/assets/1/23/jconline_Mar_2_2016.pdf
8 Mor, Intrator, Feng, & Grabowski, 2010.
Institute of Medicine, 2007.
Starmer, A. J., Sectish, T. C., Simon, D. W., Keohane, C., McSweeney, M. E., Chung, E. Y., . . . Landrigan, C. P. (2013). Rates
of medical errors and preventable adverse events among hospitalized children following implementation of a resident handoff
bundle. Journal of the American Medical Association, 310(21), 2262–2270. https://doi.org/10.1001/jama.2013.281961
Pronovost, P., Johns, M. M. E., Palmer, S., Bono, R. C., Fridsma, D. B., Gettinger, A., ... Wang, Y. C. (Eds.). (2018). Procuring
interoperability: Achieving high-quality, connected, and person-centered care. Washington, DC: National Academy of
Medicine. Retrieved from https://nam.edu/wp-content/uploads/2018/10/Procuring-Interoperability_web.pdf
Balaban, R. B., Weissman, J. S., Samuel, P. A., & Woolhandler, S. (2008). Redefining and redesigning hospital discharge to
enhance patient care: A randomized controlled study. Journal of General Internal Medicine, 23(8), 1228–1233.
https://doi.org/10.1007/s11606-008-0618-9
9 Starmer, A. J., Spector, N. D., Srivastava, R., West, D. C., Rosenbluth, G., Allen, A. D., . . . Landrigan, C. P., & the I-PASS
Study Group. (2014). Changes in medical errors after implementation of a handoff program. The New England Journal of
Medicine, 371(19), 1803–1812. https://doi.org/10.1056/NEJMsa1405556
Kruse, C. S., Marquez, G., Nelson, D., & Polomares, O. (2018). The use of health information exchange to augment patient
handoff in long-term care: A systematic review. Applied Clinical Informatics, 9(4), 752–771. https://doi.org/10.1055/s-00381670651
Brody, A. A., Gibson, B., Tresner-Kirsch, D., Kramer, H., Thraen, I., Coarr, M. E., & Rupper, R. (2016). 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, 64(11), e166–e170. https://doi.org/10.1111/jgs.14457

4

settings. 10 Preventable adverse drug events (ADEs) occur after hospital discharge in a variety of settings,
including PAC. 11
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. 12 The transfer of a medication list
between providers is necessary for medication reconciliation interventions, which have been shown to be
a cost-effective way to avoid ADEs by reducing errors, 13 especially when medications are reviewed by a
pharmacist and when it is done in conjunction with the use of electronic medical records. 14
Denominator
The denominator is the 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 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
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]).

10

Chhabra, P. T., Rattinger, G. B., Dutcher, S. K., Hare, M. E., Parsons, K. L., & Zuckerman, I. H. (2012). Medication
reconciliation during the transition to and from long-term care settings: A systematic review. Research in Social &
Administrative Pharmacy, 8(1), 60–75. https://doi.org/10.1016/j.sapharm.2010.12.002
Levinson, D. R. (2014). 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. Retrieved from
https://oig.hhs.gov/oei/reports/oei-06-11-00370.pdf
11 Battles J., Azam I., Grady M., & Reback K. (2017, August). Advances in patient safety and medical liability. AHRQ
Publication No. 17-0017-EF. Rockville, MD: Agency for Healthcare Research and Quality. Retrieved from
https://www.ahrq.gov/sites/default/files/publications/files/advances-complete_3.pdf
12 Patterson, M. E., Foust, J. B., Bollinger, S., Coleman, C., & Nguyen, D. (2019). Inter-facility communication barriers delay
resolving medication discrepancies during transitions of care. Research in Social and Administrative Pharmacy, 15(4), 366–
369. https://dx.doi.org/10.1016/j.sapharm.2018.05.124
13 Boockvar, et al., 2011.
Kwan, Lo, L., Sampson, & Shojania, 2013.
Chhabra et al., 2012.
14 Agrawal, A., & Wu, W. Y. (2009). Reducing medication errors and improving systems reliability using an electronic
medication reconciliation system. Joint Commission Journal on Quality and Patient Safety, 35(2), 106–114.
https://doi.org/10.1016/S1553-7250(09)35014-X

5

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.
Items Included in the Quality Measure
One data element will be included to calculate the measure. One data element will be collected to
inform internal measure consistency logic.
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 to Subsequent Provider

A2122. Route of Current Reconciled Medication List Transmission to Subsequent Provider
Indicate the route(s) of transmission of the current reconciled medication list to the subsequent
provider.
Check all that apply
Route of Transmission
↓
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:
Step 1.

Calculate the denominator count
Calculate the total number of patient stays with discharge to a subsequent
provider based on discharge location item 44D.

Step 2.

Calculate the numerator count
Calculate the total number of stays where a reconciled medication list was
transferred: A2121 = [1]

Step 3.

Calculate the facility observed score

6

Divide the facility’s numerator count by its denominator count; 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 whether the facility provided the reconciled medication list to the
subsequent provider.
A valid response for item 44D would trigger the coder to complete item A2121.
3. At discharge, code for the route of transmission.
A valid response for item A2121 [A2121 = 1] would send the coder to item A2122. This item
is used for internal measure consistency logic.

7

Section 3. Cross-Setting Measure: Transfer of Health Information to the Patient–PostAcute Care Measure
Measure Description
This measure, the Transfer of Health Information to the Patient, assesses for and reports on the
timely transfer of health information, specifically transfer of a reconciled medication list. This measure
evaluates for the transfer of information when a patient/resident is discharged from their current setting of
PAC 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 measure, developed under the IMPACT Act, has been developed conceptually for the IRF,
LTCH, SNF, and HHA settings. This measure is calculated by one standardized 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). The IRF-PAI, which tracks discharge
location status, will be used to track discharge to home. 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 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
Discussion related to what is a reconciled medication list is located in Chapter 1, Section 2. The
Transfer of Health Information measures serve as a check to ensure that a reconciled medication list is
provided as the patient changes care settings at discharge. Defining the completeness of that medication
list is left to the discretion of the providers and patient who are coordinating this care.
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 an HHA. 15 Of the Medicare FFS beneficiaries
with an IRF stay in FYs 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. 16
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 other community
settings. Incomplete or missing health information, such as medication information, increases the
likelihood of a patient safety risk, often life-threatening. 17 Individuals who use PAC settings are
particularly vulnerable to adverse health outcomes because of their higher likelihood of multiple
comorbid chronic conditions, polypharmacy, and complicated transitions between care settings. 18 Upon
discharge to home, individuals in PAC settings may be faced with numerous medication changes, new
15

Tian, 2016.
RTI International analysis of Medicare claims data for index stays in IRF 2016/2017. (RTI program reference: MM150).
17 Kwan et al., 2013.
Boockvar et al., 2011.
Bell et al., 2011.
Basey, Krska, Kennedy, & Mackridge, 2014.
Desai, Williams, Greene, Pierson& Hansen, 2011.
18 Brody et al., 2016.
Chhabra et al., 2012.
16

8

medication regimes, and follow-up details. 19 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. 20
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. 21 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 with patients, families, and caregivers supports the goals of highquality, personalized, and efficient health care; care coordination and person-centered care; and real-time,
data-driven clinical decision making.
Most PAC electronic health record systems generate a discharge medication list. Interventions to
promote patient participation in medication management have been shown to be acceptable and
potentially useful for improving patient outcomes and reducing costs. 22 Furthermore, provision of a
reconciled medication list to patients/residents and their caregivers can improve transitional care. 23
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) are as follows: 24
4. Provide the patient (or family as needed) with written information on the medications the patient
should be taking when leaving the organization’s care (for example, name, dose, route, frequency,
purpose).
5. Explain the importance of managing medication information to the patient.
The Agency for Healthcare Research and Quality (AHRQ) Project Re-Engineered Discharge
(RED) Toolkit includes several medication-related strategies (e.g., active medication reconciliation,

19

Brody et al., 2016.
Bell et al., 2011.
Sheehan, O. C., Kharrazi, H., Carl, K. J., Leff, B., Wolff, J. L., Roth, D. L., . . . Boyd, C. M. (2018). Helping older adults
improve their medication experience (HOME) by addressing medication regimen complexity in home healthcare. Home
Healthcare Now, 36(1), 10–19. https://doi.org/10.1097/NHH.0000000000000632.
20 Mor et al., 2010.
Starmer et al., 2013.
21 Director, Survey and Certification Group, CMS. (2013, May 17). Revision to state operations manual (SOM), Hospital
Appendix A - Interpretive Guidelines for 42 CFR 482.43, Discharge Planning. Retrieved from
https://www.cms.gov/Medicare/Provider-Enrollment-and-Certification/SurveyCertificationGenInfo/Downloads/Survey-andCert-Letter-13-32.pdf.
The State Operations Manual Guidance to Surveyors for Long-Term Care Facilities (Guidance §483.21(c)(1) Rev. 11-22-17) for
discharge planning. Retrieved from https://www.cms.gov/Regulations-andGuidance/Guidance/Manuals/downloads/som107ap_pp_guidelines_ltcf.pdf.
22 Greene, J., & Hibbard, J. H. (2012). Why does patient activation matter? An examination of the relationships between patient
activation and health-related outcomes. Journal of General Internal Medicine, 27(5), 520–526.
https://doi.org/10.1007/s11606-011-1931-2
Phatak, A., Prusi, R., Ward, B., Hansen, L. O., Williams, M. V., Vetter, E., . . . Postelnick, M. (2016). Impact of pharmacist
involvement in the transitional care of high-risk patients through medication reconciliation, medication education, and
postdischarge call-backs (IPITCH Study). Journal of Hospital Medicine, 11(1), 39–44. https://doi.org/10.1002/jhm.2493
23 Toles, M., Colón-Emeric, C., Naylor, M. D., Asafu-Adjei, J., & Hanson, L. C. (2017). Connect-home: Transitional care of
skilled nursing facility patients and their caregivers. Journal of the American Geriatric Society, 65(10), 2322–2328.
https://doi.org/10.1111/jgs.15015
24 The Joint Commission. (2018). 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

9

medication teaching for patients and caregivers, development of medication list for patients and their
health care providers). 25
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, board/care, assisted
living, group home, transitional living, or home under care of an organized home health service
organization or hospice. Discharge to one of these locations is 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
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 (A2123 = [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.
Items Included in the Quality Measure
One data element will be included to calculate the measure. One data element will be collected to
inform internal measure consistency logic.
Provision of Current Reconciled Medication List to Patient at Discharge
A2123. 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

25

Jack, B., Paasche-Orlow, M., Mitchell, S., Forsythe, S., Martin, J., & Brach, C. (n.d.). Re-Engineered Discharge (RED) toolkit.
Rockville, MD: Agency for Healthcare Research and Quality. Retrieved from
https://www.ahrq.gov/professionals/systems/hospital/red/toolkit/index.html, Last accessed November, 28, 2018.

10

Route of Current Medication List Transmission to Patient
A2124. Route of Current Reconciled Medication List Transmission to Patient
Indicate the route(s) of transmission of the current reconciled medication list to the
patient/family/caregiver.
Check all that apply
↓

Route of Transmission
A. Electronic Health Record (e.g., electronic access to patient
portal)
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:
Step 1.

Calculate the denominator count
Calculate the number of patient stays with discharge to home using discharge
location item 44D.

Step 2.

Calculate the numerator count
Calculate the number of stays where a reconciled medication list was transferred:
A2123 = [1]

Step 3.

Calculate the facility observed score
Divide the facility’s numerator count by its denominator count; 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 whether the facility provided the reconciled medication list to the
patient, family, and/or caregiver.
A valid response for item 44D would trigger the coder to complete item A2123.
3. At discharge, code for the route of transmission.
A valid response for item A2123 [A2123 = 1] would send the coder to item A2124. This item
is used for internal measure consistency logic.

11

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. 26 These draft specifications include a new
measure exclusion for baseline nursing facility (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 after an IRF stay, do not have an
unplanned readmission to an acute care hospital or LTCH in the 31 days following discharge to
community, and 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. 27,28,29
We adopted four discharge to community measures for IRF, LTCH, SNF, and home health (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 Discharge to Community–PAC IRF QRP measure is a meaningful
patient- and family-centered measure of successful community discharge.
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
26

The original measure specifications are available at https://www.cms.gov/Medicare/Quality-Initiatives-Patient-

Assessment-Instruments/IRF-Quality-Reporting/Downloads/Measure-Specifications-for-FY17-IRF-QRP-FinalRule.pdf.
27

American Hospital Association. (2017). National Uniform Billing Committee Official UB-04 Data Specifications Manual 2018
(Version 12). Chicago, IL: Author.
28
Patient discharge status codes 81 and 86 are intended for use on acute care claims only. However, because 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.
29
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.

12

episode, compared with those discharged to institutional settings. 30 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. 31 Also, providers have found
that successful discharge to community was a major driver of their ability to achieve savings, where
capitated payments for PAC were in place. 32 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. 33
Analyses conducted by the Medicare Payment Advisory Commission (MedPAC) using 2013
PAC data demonstrate the substantially higher costs of institutional PAC stays compared with HH
stays. 34 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, and SNF
stays) ranged from $13,948 to $17,506, depending on the position of the institutional PAC stay in a
sequence of PAC care. 35
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. 36
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. 37
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, region, urban or rural location), ownership (for example, for-profit or
nonprofit), and freestanding or hospital-based units; and across patient-level characteristics, such as race
30

Dobrez, D., Heinemann, A. W., Deutsch, A., Manheim, L., & Mallinson, T. (2010). Impact of Medicare’s prospective
payment system for inpatient rehabilitation facilities on stroke patient outcomes. American Journal of Physical Medicine &
Rehabilitation, 89(3), 198–204. https://doi.org/10.1097/PHM.0b013e3181c9fb40
Gage, B., Morley, M., Spain, P., Ingber, M. (2009). Examining post acute care relationships in an integrated hospital system.
Final Report. Research Triangle Park, NC: RTI International.
31
Gage, Morley, Spain, & Ingber, 2009.
32
Doran, J. P., & Zabinski, S. J. (2015). 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, 30(3), 353–355.
https://doi.org/10.1016/j.arth.2015.01.035
33
Newcomer, R. J., Ko, M., Kang, T., Harrington, C., Hulett, D., & Bindman, A. B. (2016). Health care expenditures after
initiating long-term services and supports in the community versus in a nursing facility. Medical Care, 54(3), 221–228.
https://doi.org/10.1097/MLR.0000000000000491
34
Medicare Payment Advisory Commission. (2018, June). Chapter 4: Paying for sequential stays in a unified prospective
payment system for post-acute care. In June 2018 Report to the Congress: Medicare and the Health Care Delivery System.
Retrieved from http://www.medpac.gov/docs/default-source/reports/jun18_ch4_medpacreport_sec.pdf?sfvrsn=0
35
Ibid.
36
Gage et al., 2009.
37
Ibid.

13

and gender. 38 Discharge to community rates in the IRF setting have been reported to range from about 60
to 80 percent. 39 Longer-term studies show that rates of discharge to community from IRFs have
decreased over time as IRF length of stay has decreased. 40 In the IRF Medicare FFS population, using
national unadjusted data from calendar years 2015 and 2016, 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. 41 A multi-center study of 23 LTCHs demonstrated that 28.8 percent of 1,061 patients who were

38

Reistetter, T. A., Karmarkar, A. M., Graham, J. E., Eschbach, K., Kuo, Y. F., Granger, C. V., . . . Ottenbacher, K. J. (2014).
Regional variation in stroke rehabilitation outcomes. Archives of Physical Medicine and Rehabilitation, 95(1), 29–38.
https://doi.org/10.1016/j.apmr.2013.07.018
El-Solh, A. A., Saltzman, S. K., Ramadan, F. H., & Naughton, B. J. (2000). Validity of an artificial neural network in predicting
discharge destination from a postacute geriatric rehabilitation unit. Archives of Physical Medicine and Rehabilitation, 81(10),
1388–1393. https://doi.org/10.1053/apmr.2000.16348
Medicare Payment Advisory Commission. (2018). March 2018 Report to the Congress: Medicare Payment Policy. Washington,
DC: Author Retrieved from http://www.medpac.gov/docs/default-source/reports/mar18_medpac_entirereport_sec.pdf
Bhandari, V. K., Kushel, M., Price, L., & Schillinger, D. (2005). Racial disparities in outcomes of inpatient stroke rehabilitation.
Archives of Physical Medicine and Rehabilitation, 86(11), 2081–2086. https://doi.org/10.1016/j.apmr.2005.05.008
Chang, P. F., Ostir, G. V., Kuo, Y. F., Granger, C. V., & Ottenbacher, K. J. (2008). Ethnic differences in discharge destination
among older patients with traumatic brain injury. Archives of Physical Medicine and Rehabilitation, 89(2), 231–236.
https://doi.org/10.1016/j.apmr.2007.08.143
Bergés, I. M., Kuo, Y. F., Ostir, G. V., Granger, C. V., Graham, J. E., & Ottenbacher, K. J. (2008). Gender and ethnic differences
in rehabilitation outcomes after hip-replacement surgery. American Journal of Physical Medicine & Rehabilitation, 87(7),
567–572. https://doi.org/10.1097/PHM.0b013e31817c143a
39
Galloway, R. V., Granger, C. V., Karmarkar, A. M., Graham, J. E., Deutsch, A., Niewczyk, P., . . . Ottenbacher, K. J. (2013).
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, 92(1), 14–27.
https://doi.org/10.1097/PHM.0b013e31827441bc
Morley, M. A., Coots, L. A., Forgues, A. L., & Gage, B. J. (2012). Inpatient rehabilitation utilization for Medicare beneficiaries
with multiple sclerosis. Archives of Physical Medicine and Rehabilitation, 93(8), 1377–1383.
https://doi.org/10.1016/j.apmr.2012.03.008
Reistetter, T. A., Graham, J. E., Deutsch, A., Granger, C. V., Markello, S., & Ottenbacher, K. J. (2010). Utility of functional
status for classifying community versus institutional discharges after inpatient rehabilitation for stroke. Archives of Physical
Medicine and Rehabilitation, 91(3), 345–350. https://doi.org/10.1016/j.apmr.2009.11.010
Gagnon, D., Nadeau, S., & Tam, V. (2005). Clinical and administrative outcomes during publicly-funded inpatient stroke
rehabilitation based on a case-mix group classification model. Journal of Rehabilitation Medicine, 37(1), 45–52.
https://doi.org/10.1080/16501970410015055
DaVanzo, J., El-Gamil, A., Li, J., Shimer, M., Manolov, N., & Dobson, A. (2014). Assessment of patient outcomes of
rehabilitative care provided in inpatient rehabilitation facilities (IRFs) and after discharge. Vienna, VA: Dobson DaVanzo &
Associates, LLC.
Kushner, D. S., Peters, K. M., & Johnson-Greene, D. (2015a). 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, 96(7), 1310–1318. https://doi.org/10.1016/j.apmr.2015.03.011
40
Galloway et al., 2013.
Mallinson, T., Deutsch, A., Bateman, J., Tseng, H. Y., Manheim, L., Almagor, O., & Heinemann, A. W. (2014). 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, 95(2), 209–217.
https://doi.org/10.1016/j.apmr.2013.05.031
41
El-Solh, Saltzman, Ramadan, & Naughton, 2000.
Hall, R. K., Toles, M., Massing, M., Jackson, E., Peacock-Hinton, S., O’Hare, A. M., & Colón-Emeric, C. (2015). Utilization of
acute care among patients with ESRD discharged home from skilled nursing facilities. Clinical Journal of the American
Society of Nephrology (CJASN), 10(3), 428–434. https://doi.org/10.2215/CJN.03510414
Stearns, S. C., Dalton, K., Holmes, G. M., & Seagrave, S. M. (2006). Using propensity stratification to compare patient outcomes
in hospital-based versus freestanding skilled-nursing facilities. Medical Care Research and Review: MCRR, 63(5), 599–622.
https://doi.org/10.1177/1077558706290944

14

ventilator-dependent on admission were discharged to home. 42 A single-center study found that 31
percent of LTCH hemodialysis patients were discharged to home. 43 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. 44 However, significant numbers of patients were
admitted to hospitals (29 percent) and lesser numbers to SNFs (7.6 percent), IRFs (1.5 percent), HHAs
(7.2 percent), or hospices (3.3 percent). 45
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. 46 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. 47 The effectiveness of these

Wodchis, W. P., Teare, G. F., Naglie, G., Bronskill, S. E., Gill, S. S., Hillmer, M. P., . . . Fries, B. E. (2005). Skilled nursing
facility rehabilitation and discharge to home after stroke. Archives of Physical Medicine and Rehabilitation, 86(3), 442–448.
https://doi.org/10.1016/j.apmr.2004.06.067
42
Scheinhorn, D. J., Hassenpflug, M. S., Votto, J. J., Chao, D. C., Epstein, S. K., Doig, G. S., . . . Petrak, R. A., & the
Ventilation Outcomes Study Group. (2007). Post-ICU mechanical ventilation at 23 long-term care hospitals: A multicenter
outcomes study. Chest, 131(1), 85–93. https://doi.org/10.1378/chest.06-1081
43
Thakar, C. V., Quate-Operacz, M., Leonard, A. C., & Eckman, M. H. (2010). Outcomes of hemodialysis patients in a longterm care hospital setting: A single-center study. American Journal of Kidney Diseases, 55(2), 300–306.
https://doi.org/10.1053/j.ajkd.2009.08.021
44
Wolff, J. L., Meadow, A., Weiss, C. O., Boyd, C. M., & Leff, B. (2008). Medicare home health patients’ transitions through
acute and post-acute care settings. Medical Care, 46(11), 1188–1193. https://doi.org/10.1097/MLR.0b013e31817d69d3
45
Ibid.
46
Kushner, Peters, & Johnson-Greene, 2015a.
Wodchis et al., 2005.
Berkowitz, R. E., Jones, R. N., Rieder, R., Bryan, M., Schreiber, R., Verney, S., & Paasche-Orlow, M. K. (2011). Improving
disposition outcomes for patients in a geriatric skilled nursing facility. Journal of the American Geriatrics Society, 59(6),
1130–1136. https://doi.org/10.1111/j.1532-5415.2011.03417.x
Kushner, D. S., Peters, K. M., & Johnson-Greene, D. (2015b). 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, 7(4):354-364. http://dx.doi.org/10.1016/j.pmrj.2014.10.010
O’Brien, S. R., & Zhang, N. (2018). Association between therapy intensity and discharge outcomes in aged Medicare skilled
nursing facilities admissions. Archives of Physical Medicine and Rehabilitation, 99(1), 107–115.
https://doi.org/10.1016/j.apmr.2017.07.012
47
Kushner, Peters, & Johnson-Greene2015a.
Wodchis et al., 2005.
Berkowitz et al., 2011.
Kushner, Peters, & Johnson-Greene, 2015b.
Jung, H. Y., Trivedi, A. N., Grabowski, D. C., & Mor, V. (2016). Does more therapy in skilled nursing facilities lead to better
outcomes in patients with hip fracture? Physical Therapy, 96(1), 81–89. https://doi.org/10.2522/ptj.20150090
Camicia, M., Wang, H., DiVita, M., Mix, J., & Niewczyk, P. (2016). Length of stay at inpatient rehabilitation facility and stroke
patient outcomes. Rehabilitation Nursing Journal, 41(2), 78–90. https://doi.org/10.1002/rnj.218
Buttke, D., Cooke, V., Abrahamson, K., Shippee, T., Davila, H., Kane, R., & Arling, G. (2018). A Statewide Model for assisting
nursing home residents to transition successfully to the community. Geriatrics, 3(2), 18.
https://doi.org/10.3390/geriatrics3020018
Logue, M. D., & Drago, J. (2013). Evaluation of a modified community based care transitions model to reduce costs and improve
outcomes. BMC Geriatrics, 13(1), 94. https://doi.org/10.1186/1471-2318-13-94
Carnahan, J. L., Slaven, J. E., Callahan, C. M., Tu, W., & Torke, A. M. (2017). Transitions from skilled nursing facility to home:
The relationship of early outpatient care to hospital readmission. Journal of the American Medical Directors Association,
18(10), 853–859. https://doi.org/10.1016/j.jamda.2017.05.007
Rodakowski, J., Rocco, P. B., Ortiz, M., Folb, B., Schulz, R., Morton, S. C., . . . James, A. E., III. (2017). Caregiver integration
during discharge planning for older adults to reduce resource use: A metaanalysis. Journal of the American Geriatrics
Society, 65(8), 1748–1755. https://doi.org/10.1111/jgs.14873

15

interventions suggests that improvement in discharge to community rates among PAC 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 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 remain alive during the postdischarge 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 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. 48
Table 1 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

48

01

Discharged to home/self-care (routine discharge)

06

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

81

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

86

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

American Hospital Association, 2017.

16

Patient discharge status codes 81 and 86 are intended for use on acute care claims only. However,
because 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
claims, we identify unplanned readmissions based on the CMS planned readmissions algorithm 49 used in
the following PAC readmission measures, endorsed by the National Quality Forum (NQF) and used in
several CMS programs: (1) NQF #2510: Skilled Nursing Facility 30-Day All-Cause Readmission
Measure (SNFRM); (2) NQF #2502: All-Cause Unplanned Readmission Measure for 30 Days Post
Discharge from Inpatient Rehabilitation Facilities; (3) NQF #2512: All-Cause Unplanned Readmission
Measure for 30 Days Post Discharge from Long-Term Care Hospitals; and (4) NQF #2380:
Rehospitalization During the First 30 Days of Home Health. 50 These readmission measures are based on
the Hospital-Wide All-Cause Readmission Measure (HWR) (CMS/Yale) (NQF #1789), 51 with some
additions made for the SNF, IRF, and LTCH setting measures. 52 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.
This measure was developed with International Classification of Diseases, Ninth Revision,
Clinical Modification (ICD-9-CM) procedure and diagnosis codes, and 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
49

Yale New Haven Health Services Corporation – Center for Outcomes Research & Evaluation (YNHHSC/CORE). (2018,
March). Appendix E. Planned Readmission Algorithm. In 2018 All-Cause Hospital Wide Measure Updates and
Specifications Report: Hospital-Level 30-Day Risk-Standardized Readmission Measure – Version 7.0. Prepared for the
Centers for Medicare & Medicaid Services. Retrieved from
https://www.qualitynet.org/dcs/ContentServer?c=Page&pagename=QnetPublic%2FPage%2FQnetTier4&cid=12190698558
41
50
NQF #2510: Skilled Nursing Facility 30-Day All-Cause Readmission Measure (SNFRM).
www.qualityforum.org/QPS/2510
NQF #2502: All-Cause Unplanned Readmission Measure for 30 Days Post Discharge from Inpatient Rehabilitation Facilities.
www.qualityforum.org/QPS/2502
NQF #2512: All-Cause Unplanned Readmission Measure for 30 Days Post Discharge from Long-Term Care Hospitals.
www.qualityforum.org/QPS/2512
NQF #2380: Rehospitalization During the First 30 Days of Home Health. www.qualityforum.org/QPS/2380
51
NQF #1789: Hospital-Wide All-Cause Readmission Measure (HWR) (CMS/Yale). www.qualityforum.org/QPS/1789
52 RTI International. (2016, July). Measure specifications for measures adopted in the FY 2017 IRF QRP Final Rule. Retrieved
from 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.

17

of discharge and the 31 days following day of discharge. Death in the post-discharge window is identified
using 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 long-term 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 before 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 than the rest of the Medicare population.
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 before 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 before 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
because of 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 be at higher
risk of post-discharge readmissions or death, depending on their medical condition or because of
potential non-adherence or non-compliance with care recommendations.
Data source: Patient discharge status code from Inpatient SAF IRF claim.

18

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 because of 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 than non-hospice
patients. For non-hospice patients, the primary goal of PAC is to return to baseline, independent
living in the community; death is an undesirable outcome in the non-hospice population. For
patients on hospice, the goal is to give 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 PAC setting, makes the final decision of discharge to hospice-home
or hospice-facility.
Data source: Discharge to hospice is based on the Inpatient SAF IRF claim. Post-discharge hospice
benefit is 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 before 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 before the
IRF admission date are excluded because risk adjustment for certain comorbidities requires
information on acute inpatient bills for one year before 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

19

53

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.
15) New 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)

53

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

20

Rationale: Baseline long-term NF residents did not live in the community before 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 Omnibus Budget Reconciliation Act (OBRA)-only
assessment (i.e., a non-SNF PPS assessment) with no intervening community discharge between the
OBRA assessment and acute care admission date flags the index IRF stay as a baseline long-term 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.

•

MDS: Documentation available at https://www.resdac.org/cms-data/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 calendar year (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
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

21

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 Y ij , denote the outcome (equal to 1 if patient i is discharged to community, 0 otherwise) for a
patient i at facility j; Z ij denotes a set of risk adjustment variables. We assume the outcome is related to
the risk adjusters via a logit function with dispersion:
logit(Prob(Y ij  = 1)) = α j + β*Z ij + ε ij
(1)
2
α j  = µ + ω j ; ω j ~ N(0, τ )
where Z ij = (Z 1 , Z 2 , ... Zk ) is a set of k patient-level risk adjustment variables; α j 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-to-expected 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 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). 54
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. 55
6. Indicator for ESRD status.

54

Documentation of the AHRQ Clinical Classifications Software groupings of ICD-9 codes is available at http://www.hcupus.ahrq.gov/toolssoftware/ccs/ccs.jsp.
Documentation of the AHRQ Clinical Classifications Software groupings of ICD-10 codes is available at https://www.hcupus.ahrq.gov/toolssoftware/ccs10/ccs10.jsp.
55
Ibid.

22

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. 56
10. Number of prior acute hospital discharges in the past year, not including the hospitalization in the
30 days before 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, pred j , for index
facility stays at facility j , we used the following equation:
pred j  = Σlogit-1(µ + ω i + β*Z ij )

(2)

where the sum is over all stays in facility j , and ωi is the random intercept.
To calculate the expected number, exp j , we used the following equation:
exp j  = Σlogit-1 (µ + β*Z ij )
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, SRR j , we used the following equation:

56

CMS-HCC Mappings of ICD-9 and ICD-10 Codes are included in the software at the following website:
http://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Risk-Adjustors.html.

23

SRR j  = pred j /exp j
Step 6:

(4)

Calculate the risk-standardized discharge to community rate for each facility.
To aid interpretation, the facility-wide SRR j , 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 (RSR j ).
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.

24

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
PAC settings. The four PAC settings specified in the IMPACT Act are HHAs, IRFs, LTCHs, and SNFs.
The goals of implementing cross-setting SPADEs are to facilitate care coordination and interoperability
and to 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
adopting SPADEs for five categories specified in the IMPACT Act:
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 comorbidities (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 finalized in the FY
2020 IRF PPS final rule. 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 these SPADEs comes from several sources, including the Post-Acute Care
Payment Reform Demonstration (PAC PRD), 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). IRR 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

25

extremely high or low, the resulting kappa statistic does not accurately convey the level of agreement. 57
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 these 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 at more than 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. 58 During a 6-year
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 and 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 IRR, 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/QualityInitiatives-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
adequate coverage of the clinical range of patients/residents receiving care nationally in each of the four
57 Cicchetti, D. V., & Feinstein, A. R. (1990). High agreement but low kappa: II. Resolving the paradoxes. Journal of Clinical

Epidemiology, 43(6), 551–558. https://doi.org/10.1016/0895-4356(90)90159-M
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. https://doi.org/10.1037/a0037489
Byrt, T., Bishop, J., & Carlin, J. B. (1993). Bias, prevalence and kappa. Journal of Clinical Epidemiology, 46(5), 423–429.
https://doi.org/10.1016/0895-4356(93)90018-V
McHugh, M. L. (2012). Interrater reliability: The kappa statistic. Biochemia Medica, 22(3), 276–282.
https://doi.org/10.11613/BM.2012.031
58 Saliba, D., & Buchanan, J. (2008a). Development and validation of a revised nursing home assessment tool: MDS 3.0. Santa
Monica, CA: RAND Corporation. Retrieved from
https://www.cms.hhs.gov/NursingHomeQualityInits/Downloads/MDS30FinalReport.pdf

26

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 case-mix 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 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).
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 in-depth
technical discussion of the design and methods of the National Beta Test can be found in the document
titled “Development and Evaluation Candidate Standardized Patient Assessment Data Elements: Findings
from the National Beta Test (Volume 2),” available at https://www.cms.gov/Medicare/Quality-InitiativesPatient-Assessment-Instruments/Post-Acute-Care-Quality-Initiatives/IMPACT-Act-of-2014/IMPACTAct-Downloads-and-Videos.html.
Design and sampling
The National Beta Test included PAC providers in 14 geographic/metropolitan areas, or
“markets,” across the country. This number was chosen to be similar to the design used for the PAC PRD.
A multistage stratified random sampling plan was used to obtain the sample of 14markets 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 2 hours of one another to
facilitate completion of assessments in a timely manner

Of 306 markets in the United States, 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. Because these markets are a random sample, they are
expected to be representative of the set of 64 eligible facilities and findings are therefore 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

27

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 enough 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 to enhance sample diversity in light of the
larger proportion of these setting types nationally. A total of 143 PAC facilities (35 HHAs, 22 IRFs, 26
LTCHs, 60 SNFs) were successfully recruited across 14 U.S. markets 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 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 PPSs 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 noncommunicative 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 non-communicative patients/residents for
testing of the non-communicative data elements, it precluded assessing these patients/residents with noninterview SPADEs at admission. The three data elements developed specifically for non-communicative
patients/residents are not included in this 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 states that facilities that deliver
PAC services under Medicare are required to provide qualified interpreters to their patients/residents with
limited English proficiency. 59 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 National 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 3–7) 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 rule (Brief Interview for Mental
Status [BIMS], Pain Interference, Patient Health Questionnaire [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.

59 For more information, see https://www.hhs.gov/civil-rights/for-individuals/section-1557/index.html

28

Data collection
Admission assessments were completed between admission days 3–7; discharge assessments
could be completed from 2 days before discharge through the discharge date. Trained research nurses and
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 IRR.
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 because of the larger
number of participating facilities/agencies. However, participating LTCHs and IRFs also 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 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 SPADEs.
Table 1.2 in Appendix C shows completion rates by National Beta Test 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 “Development and Evaluation Candidate Standardized Patient
Assessment Data Elements: Findings from the National Beta Test (Volume 2),” available at
https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/Post-Acute-CareQuality-Initiatives/IMPACT-Act-of-2014/IMPACT-Act-Downloads-and-Videos.html.

29

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 endocrine
imbalances, and delirium. 60 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. 61 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. 62, In addition, assessments help PAC providers 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.
1. The BIMS
2. The Confusion Assessment Method (CAM)
The data elements involve different aspects of cognition (e.g., short-term memory,
comprehension) and types of data (e.g., interview, performance-based). They are collected by various
modes (e.g., clinician assessed, patient reported).

60 National Institute on Aging. (2013). Assessing cognitive impairment in older patients. Retrieved from
https://www.nia.nih.gov/health/assessing-cognitive-impairment-older-patients
61 This estimate is based on responses to the BIMS in a study of patient/residents in the PAC PRD: Gage, B., Morley, M., Smith,
L., Ingber, M. J., Deutsch, A., Kline, T., ... & Kelleher, C. (2012). Post-acute care payment reform demonstration: Final
report (Vol 4). Research Triangle Park, NC: RTI International. Retrieved from https://www.cms.gov/Research-StatisticsData-and-Systems/Statistics-Trends-and-Reports/Reports/Research-ReportsItems/PAC_Payment_Reform_Demo_Final.html.
62 Casey, D. A., Antimisiaris, D., & O’Brien, J. (2010). Drugs for Alzheimer’s disease: Are they effective? P&T, 35(4), 208–
211.
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, 657508. https://doi.org/10.1155/2013/657508
Langa, K. M., & Levine, D. A. (2014). The diagnosis and management of mild cognitive impairment: A clinical review. Journal
of the American Medical Association, 312(23), 2551–2561. https://doi.org/10.1001/jama.2014.13806

30

Brief Interview for Mental Status (BIMS)
The 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. 63 Patients with brain injury and stroke are
commonly transferred to IRFs for intensive PAC: approximately 21 percent of IRF patients have a
primary diagnosis of stroke, and approximately 8 percent have a primary diagnosis of brain injury. 64 In
addition, cognitive impairments are associated with engagement in rehabilitation therapies, 65 and
individuals with severe cognitive impairment as measured by BIMS at IRF admission are more likely to
be readmitted after discharge. 66 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.
Data Elements for the Assessment of Cognitive Function: The BIMS
C0100. Should Brief Interview for Mental Status (C0200-C0500) be Conducted? (3-day assessment
period)
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

63 Gage, Morley, et al., 2012.
64 Medicare Payment Advisory Commission. (2016). Chapter 9: Inpatient rehabilitation facility services (pp. 235–269). In March

2016 Report to the Congress: Medicare Payment Policy. City, ST: Author. Retrieved from
http://www.medpac.gov/docs/default-source/reports/chapter-9-inpatient-rehabilitation-facility-services-march-2016-report.pdf?sfvrsn=0
65 Lenze, E. J., Munin, M. C., Dew, M. A., Rogers, J. C., Seligman, K., Mulsant, B. H., & Reynolds, C. F., III. (2004). Adverse
effects of depression and cognitive impairment on rehabilitation participation and recovery from hip fracture. International
Journal of Geriatric Psychiatry, 19(5), 472–478. https://doi.org/10.1002/gps.1116
66 Gage et al., 2012.

31

Brief Interview for Mental Status (BIMS)
C0200. Repetition of Three Words
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
3. Three
Enter Code
2. Two
1. One
0. None
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)
Ask patient: “Please tell me what year it is right now.”
A. Able to report correct year
3. Correct
Enter Code
2. Missed by 1 year
1. Missed by 2 - 5 years
0. Missed by > 5 years or no answer
Ask patient: “What month are we in right now?”
B. Able to report correct month
Enter Code
2. Accurate within 5 days
1. Missed by 6 days to 1 month
0. Missed by > 1 month or no answer
Ask patient: “What day of the week is today?”
Enter Code C. Able to report correct day of the week
1. Correct
0. Incorrect or no answer
C0400. Recall
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”
Enter Code
2. Yes, no cue required
1. Yes, after cueing ("something to wear")
0. No - could not recall
B. Able to recall “blue”
Enter Code
2. Yes, no cue required
1. Yes, after cueing ("a color")
0. No - could not recall
C. Able to recall “bed”
Enter Code
2. Yes, no cue required
1. Yes, after cueing ("a piece of furniture")
0. No - could not recall

32

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 the 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. 67 The
BIMS data elements were also included in the national MDS 3.0 test in nursing homes and showed almost
perfect reliability. 68 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. 69
Evidence supporting use of the BIMS from the National Beta Test
Assessing impairment: In the National Beta Test, the BIMS was administered at admission to 646
patients/residents in HHAs, 786 in IRFs, 496 in LTCHs, and 1,134 in SNFs (n = 3,062 overall). Overall, 5
percent of patients/residents met criteria for being severely impaired, 18 percent for being moderately
impaired, and 76 percent for being intact. In the IRF setting, 3 percent were severely impaired, 15 percent
were moderately impaired, and 82 percent were 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. Item-level missing
data ranged from 0.4 to 1.7 percent overall 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 HHAs, 537 in IRFs, 332 in LTCHs, and 494 in SNFs
(n = 1,808 overall). Overall mean time to complete the BIMS was 2.2 minutes (standard deviation
[SD] = 1.2 minutes). Time to complete in the IRF setting was 1.8 minutes (SD = 0.9 minutes).
Interrater reliability: The 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
IRFs). 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
67 Gage, Morley, et al., 2012.
68 Saliba, D., & Buchanan, J. (2008). Development and validation of a revised nursing home assessment tool: MDS 3.0:

Appendices. Santa Monica, CA: RAND Corporation. Retrieved from https://www.cms.gov/Medicare/Quality-InitiativesPatient-Assessment-Instruments/NursingHomeQualityInits/downloads/MDS30FinalReportAppendix.pdf.
69 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.
https://doi.org/10.1016/j.jamda.2012.06.004

33

within the BIMS ranged from 0.83 to 0.93 across all settings and ranged from 0.81 to 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 CAM is a widely used delirium screening tool. 70 Delirium, when undetected or untreated,
can increase the likelihood of complications, rehospitalization, and death relative to patients/residents
without delirium. 71
Although multiple versions of the CAM have been developed, CMS finalized that the short
version be adopted for SPADEs. The Short CAM contains only four items (i.e., items 1 to 4) from the
original CAM (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 an altered level of consciousness. 72 Delirium
may also interfere with functional recovery and a patient’s ability to actively participate in intensive
rehabilitation therapies, 73 which is required by IRFs. In addition, presence of delirium has implications
for administering and interpreting cognitive assessments, 74 which in turn 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.

70 De, J., & Wand, A. P. (2015). Delirium screening: A systematic review of delirium screening tools in hospitalized patients.

The Gerontologist, 55(6), 1079–1099. https://doi.org/10.1093/geront/gnv100
71 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. https://doi.org/10.1111/j.1532-5415.2005.53305.x
72 Unpublished data from the PAC PRD Public Comments sample, 2008–2010.
73 Marcantonio, E. R., Simon, S. E., Bergmann, M. A., Jones, R. N., Murphy, K. M., & Morris, J. N. (2003). Delirium symptoms
in post-acute care: Prevalent, persistent, and associated with poor functional recovery. Journal of the American Geriatrics
Society, 51(1), 4–9. https://doi.org/10.1034/j.1601-5215.2002.51002.x
Kiely, D. K., Jones, R. N., Bergmann, M. A., Murphy, K. M., Orav, E. J., & Marcantonio, E. R. (2006). Association between
delirium resolution and functional recovery among newly admitted postacute facility patients. The Journals of Gerontology.
Series A, Biological Sciences and Medical Sciences, 61(2), 204–208. https://doi.org/10.1093/gerona/61.2.204
74 Landi, F., Liperoti, R., & Bernabei, R. (2011). Postacute rehabilitation in cognitively impaired patients: Comprehensive
assessment and tailored interventions. Journal of the American Medical Directors Association, 12(6), 395–397.
McCusker, J., Cole, M., Dendukuri, N., Belzile, E., & Primeau, F. (2001). Delirium in older medical inpatients and subsequent
cognitive and functional status: A prospective study. Canadian Medical Association Journal (CMAJ), 165(5), 575–583.

34

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

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.

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

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. Although 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 an item added 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). 75 Based on a metaanalysis of diagnostic accuracy in nine studies, the CAM demonstrated moderate sensitivity (82 percent,
95 percent confidence interval: 69–91 percent) and high specificity (99 percent, 95 percent confidence
75 Saliba & Buchanan, 2008b.

35

interval: 87–100 percent), respectively, using a delirium diagnosis (Diagnostic and Statistical Manual of
Mental Disorders IV) as the standard. 76
Evidence supporting use of the CAM from the National Beta Test
Assessing impairment: In the National Beta Test, 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 difficulty 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 difficulty 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 HHAs, 472 in IRFs, 284 in LTCHs, and 405 in SNFs (n = 1,536 overall). Overall the
mean time to complete the CAM was 1.4 minutes (SD = 0.7 minutes). In the IRF setting, the mean time to
complete the CAM was 1.3 minutes (SD = 0.6 minutes).
Interrater reliability: The 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
IRFs). The kappa for the focusing attention item was good across settings (0.66) 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) and 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 underrecognized and
thus undertreated. Existing data show that depressed mood is relatively common in patients and residents
receiving PAC services. The PAC PRD found that about 9 percent of individuals in PAC were classified
as likely to have depression. 77 The prevalence varied from a low of 7 percent of beneficiaries in SNFs to a

76 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, 1359–1370.
https://doi.org/10.2147/NDT.S49520
77 This estimate is based on patient responses to a question about being sad in the two weeks before the assessment interview in a
study of patient/residents in the PAC PRD (Gage, Morley et al., 2012). If they responded “often” or “always,” they were
considered to have depression.

36

high of 11 percent of patients in IRFs. 78 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. 79
Older adults with depression may exhibit different symptoms than younger adults, including
fatigue, insomnia, irritable mood, confusion, and lack of focus. 80 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. 81 Psychosocial treatments of depression in older adults have been shown to
be more effective than no treatment, according to self-rated and clinician-rated measures of depression. 82
Assessments of the signs and symptoms of depression help PAC providers 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 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 the two cardinal criteria for
depression: depressed mood and anhedonia (inability to feel pleasure). 83 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
78 Gage, Morley, et al., 2012.
79 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.
https://doi.org/10.1002/gps.4723
80 National Institute on Aging. (2011). Depression and older adults. Retrieved from https://www.nia.nih.gov/health/depressionand-older-adults
81 Lebowitz, B. D., Pearson, J. L., Schneider, L. S., Reynolds, C. F., III, Alexopoulos, G. S., Bruce, M. L., . . . Parmelee, P.
(1997). Diagnosis and treatment of depression in late life. Consensus statement update. Journal of the American Medical
Association, 278(14), 1186–1190. https://doi.org/10.1001/jama.1997.03550140078045
82 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. https://doi.org/10.1037/0022-006X.62.1.69
Wei, W., Sambamoorthi, U., Olfson, M., Walkup, J. T., & Crystal, S. (2005). Use of psychotherapy for depression in older
adults. The American Journal of Psychiatry, 162(4), 711–717. https://doi.org/10.1176/appi.ajp.162.4.711
83 American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC:
American Psychiatric Association.

37

assessed by the PHQ-2, more than the other three PAC settings. 84 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, 85 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. 86
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.

84 Gage, Morley. et al., (2012).
85 Lenze et al., 2004.

Lequerica, A. H., & Kortte, K. (2010). Therapeutic engagement: A proposed model of engagement in medical rehabilitation.
American Journal of Physical Medicine & Rehabilitation, 89(5), 415–422. https://doi.org/10.1097/PHM.0b013e3181d8ceb2
86 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. Available at: https://www.cms.gov/Medicare/Quality-Initiatives-Patient-AssessmentInstruments/Post-Acute-Care-Quality-Initiatives/Downloads/The-Development-and-Testing-of-the-Continuity-AssessmentRecord-and-Evaluation-CARE-Item-Set-Final-Report-on-Reliability-Testing-Volume-2-of-3.pdf

38

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.
1. Symptom Presence
2. Symptom Frequency
1.
2.
0. No (enter 0 in column 2)
0. Never or 1 day
Symptom
Symptom
1. Yes (enter 0-3 in column 2)
1. 2-6 days (several days)
Presence
Frequency
9. No response (leave column 2
2. 7-11 days (half or more of
blank)
the days)
Enter Scores in Boxes
3. 12-14 days (nearly every
day)
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.
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 02 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.

39

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 an assessment of depression severity. 87 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 3) have been
higher and more stable, ranging between 74 percent and 97 percent (median value = 90 percent). 88 89 90 91
92 93 94 95 96 97 98 99 100 101 102It is thus a viable option for standardization, with the benefits of the shorter
assessment counterbalancing the limitation of the lower sensitivity.

87 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. https://doi.org/10.1016/j.jpsychores.2004.09.006
88 Arroll, B., Goodyear-Smith, F., Crengle, S., Gunn, J., Kerse, N., Fishman, T., ... & Hatcher, S. (2010). Validation of PHQ-2

and PHQ-9 to screen for major depression in the primary care population. Annals of Family Medicine 8(4): 348-353.
89 Bhana, A., Rathod, S. D., Selohilwe, O., Kathree, T., & Petersen, I. (2015). The validity of the Patient Health Questionnaire

for screening depression in chronic care patients in primary health care in South Africa. BMC Psychiatry 15(1): 118.
90 Boyle, L. L., Richardson, T. M., He, H., Xia, Y., Tu, X., Boustani, M., & Conwell, Y. (2011). How do the PHQ‐2, the PHQ‐9

perform in aging services clients with cognitive impairment? International Journal of Geriatric Psychiatry 26(9): 952-960.
DOI: 10.1002/gps.2632
91 Chagas, M. H., Crippa, J. A., Loureiro, S. R., Hallak, J. E., Meneses-Gaya, C. D., Machado-de-Sousa, J. P., ... & Tumas, V.
(2011). Validity of the PHQ-2 for the screening of major depression in Parkinson's disease: two questions and one important
answer. Aging & Mental Health 15(7): 838-843.
92 Chen, S., Chiu, H., Xu, B., Ma, Y., Jin, T., Wu, M., & Conwell, Y. (2010). Reliability and validity of the PHQ‐9 for screening
late‐life depression in Chinese primary care. International Journal of Geriatric Psychiatry 25(11): 1127-1133.
93 de Lima Osório, F., Vilela Mendes, A., Crippa, J. A., & Loureiro, S. R. (2009). Study of the discriminative validity of the
PHQ‐9 and PHQ‐2 in a sample of Brazilian women in the context of primary health care. Perspectives in Psychiatric Care
45(3): 216-227.
94 Hanwella, R., Ekanayake, S., & de Silva, V. A. (2014). The validity and reliability of the Sinhala translation of the Patient
Health Questionnaire (PHQ-9) and PHQ-2 Screener. Depression Research and Treatment, 2014.
95 Inagaki, M., Ohtsuki, T., Yonemoto, N., Kawashima, Y., Saitoh, A., Oikawa, Y., ... & Yamada, M. (2013). Validity of the
PHQ-9 and PHQ-2 in general internal medicine primary care at a Japanese rural hospital: a cross-sectional study. General
Hospital Psychiatry 35(6): 592-597.
96 Kroenke, K., Spitzer, R. L., & Williams, J. B. (2001). The PHQ‐9. Journal of General Internal Medicine 16(9): 606-613.
97 Anand, A., Li, Y., Wang, Y., Wu, J., Gao, S., Bukhari, L., ... & Lowe, M. J. (2005). Activity and connectivity of brain mood
regulating circuit in depression: a functional magnetic resonance study. Biological Psychiatry 57(10): 1079-1088.
98 Phelan, E., Williams, B., Meeker, K., Bonn, K., Frederick, J., LoGerfo, J., & Snowden, M. (2010). A study of the diagnostic
accuracy of the PHQ-9 in primary care elderly. BMC Family Practice 11(1): 63.
99 Suzuki, K., Kumei, S., Ohhira, M., Nozu, T., & Okumura, T. (2015). Screening for major depressive disorder with the Patient
Health Questionnaire (PHQ-9 and PHQ-2) in an outpatient clinic staffed by primary care physicians in Japan: a case control
study. PloS One, 10(3): e0119147.
100 Thombs, B. D., Ziegelstein, R. C., & Whooley, M. A. (2008). Optimizing detection of major depression among patients with
coronary artery disease using the patient health questionnaire: data from the heart and soul study. Journal of General
Internal Medicine 23(12): 2014-2017.
101 Xiong, N., Fritzsche, K., Wei, J., Hong, X., Leonhart, R., Zhao, X., ... & Fischer, F. (2015). Validation of patient health
questionnaire (PHQ) for major depression in Chinese outpatients with multiple somatic symptoms: a multicenter crosssectional study. Journal of Affective Disorders 174: 636-643.
102 Zuithoff, N. P., Vergouwe, Y., King, M., Nazareth, I., van Wezep, M. J., Moons, K. G., & Geerlings, M. I. (2010). The PHQ9 for detection of major depressive disorder in primary care: consequences of current thresholds in a cross-sectional study.
BMC Family Practice 11(1): 98.

40

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. 103 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). 104 In addition, the Staff Time and Resource Intensity
Verification (STRIVE) study, conducted in a national sample of nursing homes by CMS, concluded that
the PHQ-9 used in the MDS 3.0 was the “best measure” for identifying individuals with higher wageweighted staff time, defined as the time that nursing home staff spent caring for residents. 105
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 the
National Beta Test. 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 show signs of
depression in the PHQ-2 would not receive the seven additional elements contained in the PHQ-9. In the
National Beta Test, the PHQ-2 to 9 was administered to 646 patients/residents in HHAs, 786 in IRFs, 496
in LTCHs, and 1,134 in SNFs (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
IRF patients experienced feeling down, depressed, or hopeless nearly every day over the past 2 weeks.
Similarly, about 1 in 10 IRF patients experienced these symptoms on half or more of the days.
More than 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 was 2.4 across settings (SD = 1.7) and 2.3 (SD = 1.7) in the IRF setting. The average full PHQ9 score across settings was 11.9 (SD = 5.3), and the average score in the IRF setting was 11.8 (SD = 5.3).
The PHQ-9 has thresholds to indicate the severity of probable depression. 106 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 the mild (31 percent 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.

103 Saliba & Buchanan, 2008b.
104 Ibid.
105 Centers for Medicare & Medicaid Services. (2013). Analyses of data collected in CMS national nursing home time study used

to establish RUG-IV model. Retrieved from https://www.cms.gov/Medicare/Medicare-Fee-for-ServicePayment/SNFPPS/TimeStudy.html
106 Kroenke, K., Spitzer, R., & Williams, J. (2001). The PHQ-9 validity of a brief depression severity measure. Journal of
General Internal Medicine, 16, 606–613. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1495268/.

41

Time to complete: Time to complete was examined among 428 assessments in HHAs, 515 in
IRFs, 305 in LTCHs, and 479 in SNFs (n = 1,727 overall). Among patients/residents who only received
the PHQ-2, time to complete was an average of 1.7 minutes (SD = 1.1). The average time to complete the
PHQ-2 in the IRF setting was 1.5 minutes (SD = 0.9). Among patients receiving the full PHQ-9, the time
to complete was an average of 4.0 minutes (SD = 1.2). In the IRF setting, the 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: IRR was assessed for 196 patients/residents in HHAs, 254 in IRFs, 231 in
LTCHs, and 267 in SNFs (n = 948 overall). IRR for all symptom presence and frequency items was
excellent: kappas ranged from 0.95 to 1.00 for the four settings combined and from 0.87 to 1.00 in IRFs.
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 in IRFs. 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 IRFs).
Percent agreement was also nearly perfect, ranging from 97 percent to 100 percent overall and 93
percent to 100 percent in IRFs. For eligibility to complete the full PHQ-9, percent agreement was 99
percent across settings and in IRFs. For the sum of symptom frequencies, percent agreement was 95
percent across settings and 94 percent in IRFs. Please refer to Table 3.1.2 in Appendix C for kappa and
percent agreement statistics for all PHQ items.

42

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 to plan the provision of these important therapies, ensure the
continued appropriateness of care, and support care transitions. The assessment of special services,
treatments, and interventions may also help 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.
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. IV medications (antibiotics, anticoagulation, vasoactive medications, other)
9. Transfusions
10. Dialysis (hemodialysis, peritoneal dialysis)
11. IV access (peripheral IV, midline, central line)
12. Parenteral/IV feeding
13. Feeding tube
14. Mechanically altered diet
15. Therapeutic diet
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 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 resource intensive 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 commonly
43

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,
because of 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 currently not assessed in
the IRF-PAI. Patients in the rehabilitation setting 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 (e.g., lab work, nursing care) than some other populations of
patients. Individuals impaired by cancer or chemotherapy treatments have been shown to make functional
gains in the IRF setting. 107 Some cancer patients can benefit from 3 hours of therapy per day and benefit
from multimodal types of therapy to address heterogeneous needs that can include neurologic issues,
orthopedic problems, general conditioning, pain management, and lymphedema management. 108
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 35 percent. 109 Receipt of chemotherapy has implications for
care planning, assessing functional gains, and estimating patient length of stay and resource use in the 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.

107 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.
https://doi.org/10.1016/S0003-9993(96)90220-8
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.
https://doi.org/10.1097/00002060-200003000-00005
108 Fialka-Moser, V., Crevenna, R., Korpan, M., & Quittan, M. (2003). Cancer rehabilitation: Particularly with aspects on
physical impairments. Journal of Rehabilitation Medicine, 35(4), 153–162. https://doi.org/10.1080/16501970306129
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. https://doi.org/10.1002/jso.20777
109 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.
https://doi.org/10.1097/PHM.0b013e31817fb94e
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. https://doi.org/10.1016/j.pmrj.2014.01.009

44

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.
a.
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. 110 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. 111
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 intravenously, orally, or by another route. In the National Beta Test, the data elements were
administered to 629 patients/residents in HHAs, 762 in IRFs, 448 in LTCHs, and 1,087 in SNFs
(n = 2,926 overall). Across settings, the overwhelming majority of patients/residents (99 percent) did not
receive chemotherapy. In the IRF setting, specifically, only 3 percent of patients had chemotherapy
treatment noted. More-detailed rates of chemotherapy implementation across settings are shown in
Appendix C, Table 4.1.1.
Missing data: Overall, rates of missing responses for the Chemotherapy items were very low.
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.
Time to complete: Time to complete was examined among 422 assessments in HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the
Chemotherapy items was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).

110 Gage, Constantine, et al., 2012.
111 Saliba & Buchanan, 2008b.

45

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 and 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. 112 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, 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 than those who are not. 113
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 35 percent. 114 Receipt of radiation therapy has implications for
care planning, assessing functional gains, and estimating patient length of stay and resource use in the 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.

112 Yamada, Y. (2009). Principles of radiotherapy (pp. 73–80). In Stubblefield, Michael D. & O’Dell, Michael W. (Eds.),
Cancer rehabilitation: principles and practice. New York, NY: Demos Medical Publishing.
National Cancer Institute. (2010). Radiation therapy to treat cancer. Retrieved from https://www.cancer.gov/aboutcancer/treatment/types/radiation-therapy
113 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.
https://doi.org/10.1053/apmr.2001.21862
McKinley, Huang, & Tewksbury, 2000
114 Guo, Persyn, Palmer, & Bruera, 2008
Asher et al., 2014.

46

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.
a.
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. 115
Evidence supporting use of Radiation from the National Beta Test
Assessing Radiation: One item assessed whether radiation was performed during the assessment
period. In the National Beta Test, the data element was administered to 629 patients/residents in HHAs,
762 in IRFs, 448 in LTCHs, and 1,087 in SNFs (n = 2,926 overall). Across settings, only three
patients/residents (one in SNF, two in HHA; 0 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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554, overall). The average time to complete the Radiation
item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (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.
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
115 Saliba & Buchanan, 2008b.

47

(e.g., nasal cannula, various types of masks). Accessories are also required (regulators, filters, tubing,
etc.). 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 is 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. 116 In addition, patients with cardiac conditions (some of whom may require oxygen therapy)
represent approximately 5 percent of IRF cases. 117 Patients’ use of oxygen therapy has important
implications for their ability to participate in intensive rehabilitation therapies (3 hours per day, 5 days per
week) and their ability to make functional gains over the course of rehabilitation. These factors in turn
may affect their 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.
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.
a.
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

116 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. Retrieved from
http://rc.rcjournal.com/content/58/4/601.short
117 Medicare Payment Advisory Commission, 2016.

48

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. 118 In nursing homes, a checkbox for oxygen therapy
during the last 5 days was shown to have reliability ranging from 0.93 to 0.96 (kappas) in the national
MDS 3.0 test. 119
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 the National Beta Test, the data element Oxygen Therapy
(Intermittent, Continuous, High-Concentration Delivery System) was administered to 629
patients/residents in HHAs, 762 in IRFs, 448 in LTCHs, and 1,087 in SNFs (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 a
high-concentration delivery system. This pattern was similar in the IRF setting, where intermittent
therapy was the most common (11 percent). Continuous therapy (8 percent) and high-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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554, overall). The average time to complete the Oxygen
Therapy data element was 0.22 minutes overall (SD = 0.1). The average time to complete the data element
in the IRF setting was 0.25 minutes (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). The kappa for implementation of oxygen therapy was
substantial/good both overall (0.82) and in the IRF setting (0.80). The kappa for the intermittent therapy
sub-element was 0.81 overall and 0.76 in the IRF setting, and the kappa for the continuous therapy subelement was 0.55 overall and 0.68 in the IRF setting. Kappas are not reported for the high-concentration
therapy sub-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 for a variety of reasons, including excess production of secretions from a pulmonary

118 Gage, Constantine, et al., 2012.
119 Saliba & Buchanan, 2008b.

49

infectious process or neurological deficits that inhibit the ability to cough, swallow, and so on. Suction 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 treated 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. Pneumonia, a comorbidity that may occur
among lower extremity fracture patients in the IRF setting, is associated with longer length of stay, lower
discharge functional status ratings, and lower odds of home discharge. 120 Additionally, pneumonia 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). 121 The need for suctioning may affect patients’ ability to fully 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.

120 Ahmed, Graham, Karmarkar, Granger, & Ottenbacher, 2013
121 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. https://doi.org/10.1097/PHM.0000000000000629

50

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.
a.
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. 122 In nursing homes, a
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. 123
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 National Beta
Test, the data element Suctioning (Scheduled, As Needed) was administered to 629 patients/residents in
HHAs, 762 in IRFs, 448 in LTCHs, and 1,087s in SNFs (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.

122 Gage, Constantine, et al., 2012.
123 Saliba & Buchanan, 2008b.

51

Time to complete: Time to complete was examined among 422 assessments in HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the Suctioning
items was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (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 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, some indications for tracheostomy include 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); tumors of the upper airway; severe neck, mouth, or chest wall
injuries; degenerative neuromuscular diseases such as amyotrophic lateral sclerosis (ALS); spinal cord
injuries; and airway burns. Generally, suctioning is necessary to ensure that the tracheostomy is clear of
secretions, which can inhibit successful oxygenation. Often, individuals with tracheostomies also receive
supplemental oxygenation. The presence of a tracheostomy, permanent or temporary, warrants careful
monitoring and immediate intervention if the tracheostomy becomes occluded or, in the case of a
temporary tracheostomy, if the devices used become dislodged.
For patients with a tracheostomy, tracheostomy care, which primarily consists of cleaning,
dressing changes, and replacement of the tracheostomy cannula (tube), is a critical part of their care plans.
Regular cleaning is important to prevent infection, such as pneumonia, and to prevent any occlusions,
which create the risk of inadequate oxygenation. Although 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 need 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 Functional Independence Measure (FIM) scores, and more medical
complications. 124 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.
124 Roth, E. J., Lovell, L., Harvey, R. L., Bode, R. K., & Heinemann, A. W. (2002). Stroke rehabilitation: indwelling urinary

catheters, enteral feeding tubes, and tracheostomies are associated with resource use and functional outcomes. Stroke, 33(7),
1845–1850. https://doi.org/10.1161/01.STR.0000020122.30516.FF

52

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.
a.
On Admission
Check all that apply
Respiratory Therapies
E1. Tracheostomy Care
Current use
Tracheostomy care is currently assessed in the MDS. The data element first assesses whether the
resident received tracheostomy care while not a resident of the assessing facility and within the last 14
days, and then assesses 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. 125
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 the National Beta Test, the data element was administered to 629
patients/residents in HHA settings, 762 in IRFs, 448 in LTCHs, and 1,087 in SNFs (n = 2,926 overall).
Across 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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the
Tracheostomy Care item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting
(SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). The IRR was excellent for the Tracheostomy Care data
element, as measured by percent agreement of paired raters. The 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.

125 Saliba, & Buchanan, 2008b.

53

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 supports breathing by providing positive airway pressure that
prevents airways from collapsing 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 in 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), whereas 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). 126
For example, sleep-disordered breathing has been identified as common in stroke patients and is a risk
factor for stroke itself and stroke recurrence; treatment of stroke patients with obstructive sleep apnea
with CPAP has been associated with improved functional motor outcomes. 127 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.
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. 128 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. Furthermore, use may indicate 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.

126 Medicare Payment Advisory Commission, (2016).
127 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. https://doi.org/10.1016/j.apmr.2009.12.019
Brown, D. L. (2006). Sleep disorders and stroke. Seminars in Neurology, 26(1), 117–122. https://doi.org/10.1055/s-2006-933315
Davis, A. P., Billings, M. E., Longstreth, W. T., Jr., & 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.
https://doi.org/10.1212/CPJ.0b013e318296f274
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, 42(4), 1062–1067.
https://doi.org/10.1161/STROKEAHA.110.597468
128 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. https://doi.org/10.1016/j.rmed.2009.03.016
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.
https://doi.org/10.1159/000369862

54

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.
a.
On Admission
Check all that apply
Respiratory Therapies
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 whether 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
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. 129
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 BiPAP or CPAP. In the National Beta Test, the data
element was administered to 629 patients/residents in HHAs, 762 in IRFs, 448 in LTCHs, and 1,087 in
SNFs (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 (6 percent) was more common than
BiPAP (1 percent). Detailed findings regarding non-invasive mechanical ventilators 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.

129 Gage, Constantine, et al., 2012.

55

Time to complete: Time to complete was examined among 422 assessments in HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the Noninvasive Mechanical Ventilator items was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF
setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (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 Non-invasive
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 patients who are unable to support
their own respiration. Patients receiving closed-system ventilation include those receiving ventilation via
a tracheostomy and patients with an endotracheal tube (e.g., nasally or orally intubated). Depending on
the patient’s underlying diagnosis, clinical condition, and prognosis, the patient 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 patients can support their own respiration, whereas
chronic neurodegenerative diseases are likely to progress over time and therefore preclude patients from
weaning and eventually having the tube removed.
Ventilation in this manner is a resource-intensive therapy associated with life-threatening
conditions in which the patient would not survive without invasive ventilation. However, ventilator use
has inherent risks requiring close monitoring, and failure to adequately care for ventilator-dependent
patients can lead to death, pneumonia, sepsis, and other iatrogenic events. 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, 130 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. 131 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.

130 Make, B., Gilmartin, M., Brody, J. S., & Snider, G. L. (1984). Rehabilitation of ventilator-dependent subjects with lung

diseases. The concept and initial experience. Chest, 86(3), 358–365. https://doi.org/10.1378/chest.86.3.358
131 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 Medicine, 38(10), 1947–1953.
https://doi.org/10.1097/CCM.0b013e3181ef4460

56

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.
a.
On Admission
Check all that apply
Respiratory Therapies
F1. Invasive Mechanical Ventilator (ventilator or respirator)
Current use
Invasive mechanical ventilator use 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).
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. 132 A version of the item was tested in the MDS 3.0 National
Evaluation Study and had perfect agreement (100 percent). 133
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 the National Beta Test, the data element was
administered to 629 patients/residents in HHAs, 762 in IRFs, 448 in LTCHs, and 1,087 in SNFs
(n = 2,926 overall). Across settings overall, only 13 assessments (0 percent after rounding) noted use of an
invasive mechanical ventilator. One of these 13 patients was in the IRF setting (12 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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the Invasive
Mechanical Ventilator item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting
(SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). The IRR was excellent for the Invasive Mechanical Ventilator
data element, as measured by percent agreement of paired raters. The kappa was not estimated for the
Invasive Mechanical Ventilator data element because the proportion was out of range for a stable kappa
132 Gage, Constantine, et al., 2012.
133 Saliba, & Buchanan, 2008b.

57

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)
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 subtypes of IV medications (antibiotics, anticoagulants,
vasoactive, and other) are very different. IV antibiotics are used for severe infections when (1) the
bioavailability of the oral form of the medication would be inadequate to kill the pathogen, (2) an oral
form of the medication does not exist, or (3) the patient is unable to take the medication by mouth.
Because of growing concern about antimicrobial resistance, antibiotic stewardship initiatives are aimed at
increasing evidence-based antibiotic prescribing and decreasing antibiotic overuse. Although data on
which antibiotics are used 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 or subtypes thereof. Several classes
of patients with IRF qualifying conditions are at risk of infections that could require IV 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 IV anticoagulation. Patient groups at risk include those admitted after lower
extremity fracture, lower extremity joint replacement, major multiple trauma, or spinal cord injury;
traumatic brain injury patients; stroke patients; and other patients whose mobility has been limited by
other neurologic conditions. For example, incidence of deep vein thrombosis varies from 16.4 percent to
100 percent among stroke, spinal cord injury, and traumatic brain injury patients not receiving
prophylaxis, and incidence remains high when prophylactic measures (e.g., pneumatic compression,

58

compression stockings, mobilization, medication) are used. 134 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.
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.
a.
On Admission
Check all that apply
Other
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 at admission asking whether the patient is receiving any IV medications. 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.

134 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.
https://doi.org/10.1016/j.apmr.2003.10.023

59

In nursing homes, a checkbox for IV medications during the last 5 days was shown to have
reliability of 0.95 (kappa) in the national MDS 3.0 test. 135
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 the National Beta Test, the data element was administered to
629 patients/residents in HHAs, 762 in IRFs, 448 in LTCHs, and 1,087 in SNFs (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.
Missing data: Overall, there were very low rates of missing responses for the IV Medications
items. Across all settings, that is, when looking across respondents from all PAC providers, missingness
was less than 0.9 percent. In the IRF setting, missingness for the IV Medication items also 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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554, overall). The average time to complete the IV
Medications items was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). With the exception of the anticoagulation sub-element, 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 caused by 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 sub-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 affects 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 because of the need for added resources and to the extent that receipt of
transfusions indicates a more medically complex patient.

135 Saliba, & Buchanan, 2008b.

60

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. 136 As in other settings, blood transfusions are resource intensive, requiring
laboratory testing, coordination with the blood bank, and intensive bedside nursing care and monitoring.
Blood transfusions also 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.
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.
a.
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 5 days was shown to have reliability of
0.67 (kappa) in the national MDS 3.0 test. 137
Evidence supporting use of Transfusions from the National Beta Test
Assessing Transfusions: One item assessed whether transfusions were performed during the
assessment period. In the National Beta Test, the data element was administered to 629 patients/residents
in HHAs, 762 in IRFs, 448 in LTCHs, and 1,087 in SNFs (n = 2,926 overall). Across settings, only 14
patient/resident assessments (0 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.
136 Medicare Payment Advisory Commission, 2016.
137 Saliba, & Buchanan, 2008b.

61

Time to complete: Time to complete was examined among 422 assessments in HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554, overall). The average time to complete the Transfusion
item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (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 (99 percent).
Please refer to Table 4.9.2 in Appendix C for setting-specific percent agreement statistics for the
Transfusion item.
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 4 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 after. 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 on site. Close monitoring for fluid shifts, blood
pressure abnormalities, and other adverse effects is required before, during, and after 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. 138 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 139 and poorer function
performance outcomes. 140 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. 141
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). 142 Dialysis is a time-intensive service that requires
coordination with specialists and close monitoring of vital signs and laboratory studies. Dialysis also
carries risks of complications and infections. Accordingly, it may affect patients’ ability to participate in
138 Gage, Morley, et al., 2012
139 Forrest, G. P. (2004). Inpatient rehabilitation of patients requiring hemodialysis. Archives of Physical Medicine and

Rehabilitation, 85(1), 51–53. https://doi.org/10.1016/S0003-9993(03)00366-6
140 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.
https://doi.org/10.1016/S0003-9993(95)80661-X
141 Farragher, J., & Jassal, S. V., & the Blackwell Publishing Ltd. (2012). Rehabilitation of the geriatric dialysis patient.
Seminars in Dialysis, 25(6), 649–656. https://doi.org/10.1111/sdi.12014
142 Medicare Payment Advisory Commission, 2016.

62

an intensive rehabilitation program, resource use, and functional gains. 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.
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.
a.
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 whether 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 5 days was tested in the national
MDS 3.0 test and shown to have almost perfect reliability (kappas of 0.91 to 0.93). 143
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
the National Beta Test, the data element was administered to 629 patients/residents in HHAs setting, 762
in IRFs, 448 in LTCHs, and 1,087 in SNFs (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 0 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 than 1 percent. In the IRF setting specifically, missingness did
not exceed 0.9 percent. The low rate of missing data indicates feasibility of administration.

143 Saliba, & Buchanan, 2008b.

63

Time to complete: Time to complete was examined among 422 assessments in HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the Dialysis item
was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). Most kappas are not reported for the 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
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)
IV access refers to a catheter inserted into a vein for a variety of clinical reasons, including longterm medication treatment; hemodialysis; large volumes of blood or fluid; frequent access for blood
samples; IV fluid administration; total parenteral nutrition; or, in some instances, the measurement of
central venous pressure.
The sub-elements associated with IV access distinguish between peripheral access and central
access. In addition, 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 lifethreatening 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, and bleeding from an
open lumen.
Relevance to IRFs
The presence of IV access is not currently assessed in IRF-PAI, nor are specific subtypes of IV
access. The need for IV access in IRFs is common: in PAC PRD, 7.2 percent of IRF patients received
central line management. 144 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.

144 Gage, Morley, et al., 2012.

64

Data Element 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.
a.
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 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 the National Beta Test, the data elements were administered to 629
patients/residents in HHAs, 762 in IRFs, 448 in LTCHs, and 1,087 in SNFs (n = 2,926 overall). Across
settings, 24 percent of assessments noted use of IV access. The rate in the IRF setting was 22 percent. For
the specific type of IV access noted, a central line was most common across settings (13 percent),
followed closely by 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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554, overall). The average time to complete the IV Access
item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in IRFs (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). IRR was excellent across settings for the IV Access item
(kappa = 0.90) and the peripheral and central types of access (kappa = 0.81 and kappa = 0.85,
respectively). Similarly, IRR was substantial/good in the IRF specifically for the IV Access item (0.81)
and Peripheral sub-element (kappa = 0.81). Percent agreement for the data element was almost perfect.
Across settings, percent agreement was 96 percent for IV Access generally and the types of IV access (96
to 98 percent). In the IRF specifically, percent agreement was 94 percent for the general IV Access item,

65

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 after surgery, when feeding by mouth or digestive
system is not possible, when a patient's digestive system cannot absorb nutrients because of 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
nutritional needs enterally. Overall, parenteral/IV feeding is a form of nutritional support that can be used
to prevent or address malnutrition. 145 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. 146
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 more than 65 years of age. 147 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 used for
individuals with inflammatory bowel disease, a condition that is common in older adults. 148
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 IV 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. Parenteral feeding and tube feeding 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

145 National Collaborating Centre for Acute Care (UK). (2006). Nutrition support for adults: oral nutrition support, enteral tube

feeding and parenteral nutrition. Methods, Evidence & Guidance. London, UK: National Collaborating Centre for Acute
Care. Retrieved from https://www.nice.org.uk/guidance/cg32/evidence/full-guideline-194889853
146 Evans, C. (2005). Malnutrition in the elderly: A multifactorial failure to thrive. The Permanente Journal, 9(3), 38–41.
https://doi.org/10.7812/TPP/05-056
147 Corkins, M. R., Guenter, P., DiMaria-Ghalili, R. A., Jensen, G. L., Malone, A., Miller, S., . . . Resnick, H. E., & the
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. https://doi.org/10.1177/0148607113512154
148 Semrad, C. E. (2012). Use of parenteral nutrition in patients with inflammatory bowel disease. Gastroenterology &
Hepatology, 8(6), 393–395.
Mullady, D. K., & O’Keefe, S. J. (2006). Treatment of intestinal failure: Home parenteral nutrition. Nature Reviews.
Gastroenterology & Hepatology, 3(9), 492–504. https://doi.org/10.1038/ncpgasthep0580
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. https://doi.org/10.1093/ecco-jcc/jjv059

66

variety of complications. 149 Among IRF patients with stroke, an IRF qualifying condition, malnutrition
(which may or may not require parenteral/IV feeding), has been associated with poorer rehabilitation
outcomes, longer length of stay, and worse functional outcomes among stroke patients in some IRFs. 150
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.
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.
1.
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 MDS. The OASIS data element assesses whether the patient is receiving
parenteral nutrition at home. The IRF-PAI includes a checkbox data element to assess total parenteral
nutrition with a 3-day look-back period. The LCDS includes a checklist to assess whether the patient
receives total parenteral nutrition at admission. The MDS first assesses whether the patient received
parenteral/IV feeding while not a resident of the assessing facility and within the last 7 days, and then
whether the patient received parenteral/IV feeding while a resident and within the last 7 days.
Prior evidence supporting use of Parenteral/IV Feeding
A similar data element, Total Parenteral Nutrition, was tested in the PAC PRD and found to be
feasible across PAC settings. Parenteral/IV feeding in the last 5 days was shown to have almost perfect
reliability (kappa of 0.95) in the national MDS 3.0 test in nursing homes. 151
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 HHAs, 762 in IRFs,
448 in LTCHs, and 1,087 in SNFs (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.

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

nutritional intervention modify it? The American Journal of Clinical Nutrition, 47(2, Suppl), 352–356.
https://doi.org/10.1093/ajcn/47.2.352
150 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. https://doi.org/10.1016/S0003-9993(96)90081-7
151 Saliba, & Buchanan, 2008b.

67

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.
Time to complete: Time to complete was examined among 422 assessments in HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554, overall). The average time to complete the
Parenteral/IV Feeding item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting
(SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (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 and 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
whether 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. 152 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. 153
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 65 years of age. 154 Additionally, enteral nutrition can be used to provide nutrition for patients
with specific diseases. For example, tube feeding can be used for individuals with stroke 155 and those
with head and neck cancer, 156 conditions that are common in older adults. 157
Assessing use of a feeding tube can inform resource use, 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
152 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. https://doi.org/10.1177/0884533610378524
153 Evans, 2005.
154 Corkins et al., 2014.
155 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. https://doi.org/10.1177/0884533611405795
156 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. https://doi.org/10.1177/011542650702200168
157 Centers for Disease Control and Prevention (CDC). (2012). Prevalence of stroke—United States, 2006-2010. MMWR.
Morbidity and Mortality Weekly Report, 61(20), 379–382.
VanderWalde, N. A., Fleming, M., Weiss, J., & Chera, B. S. (2013). Treatment of older patients with head and neck cancer: A
review. The Oncologist, 18(5), 568–578. https://doi.org/10.1634/theoncologist.2012-0427

68

because of poor oral intake and inability to meet nutritional goals or because of aspiration risk. For IRF
patients, 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
patients 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 require tube feeding, has been associated with poorer rehabilitation outcomes
among geriatric stroke patients and length of stay and functional outcomes among stroke patients in some
IRFs. 158 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. 159 In addition, feeding tubes have
been associated with greater functional improvements over the course of IRF stays for severe stroke
patients. 160 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. 161
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.
1.
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 whether 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.

158 Finestone, Greene-Finestone, Wilson, & Teasell, 1996.

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. Retrieved from
https://www.ncbi.nlm.nih.gov/pubmed/8291969
159 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, Suppl 2), 82–92.
https://doi.org/10.1016/j.apmr.2005.07.314
160 James, R., Gines, D., Menlove, A., Horn, S. D., Gassaway, J., & Smout, R. J. (2005).
Roth, Lovell, Harvey, Bode, & Heinemann, 2002.
161 Dempsey, Mullen, & Buzby, 1988.

69

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 5 days, was shown to have almost perfect reliability (kappa of 0.89). 162
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 HHAs, 762 in IRFs, 448 in LTCHs, and
1,087 in SNFs (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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the Feeding
Tube item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (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 setting-specific 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, pureed 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. 163 In the absence of treatment, swallowing disorders can lead to malnutrition,
dehydration, aspiration pneumonia, poor overall health, chronic lung disease, choking, and death. 164
Other consequences can include lack of interest and enjoyment related to eating or drinking, and
embarrassment or isolation tied to social situations involving eating. 165
Dysphagia is highly prevalent 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 swallowing
disorder was 38 percent, and current prevalence of a swallowing disorder was 33 percent. 166 Additionally,
increasing age has been shown to be associated with a higher likelihood of swallowing problems in the
162 Saliba, & Buchanan, 2008b.
163 National Institute on Deafness and Other Communication Disorders. (2017). Dysphagia. Retrieved from

https://www.nidcd.nih.gov/health/dysphagia
164 American Speech-Language-Hearing Association. (n.d.). Adult dysphagia. Retrieved from

https://www.asha.org/PRPSpecificTopic.aspx?folderid=8589942550§ion=Overview
165 Ibid.
166 Roy, N., Stemple, J., Merrill, R. M., & Thomas, L. (2007). Dysphagia in the elderly: Preliminary evidence of prevalence, risk

factors, and socioemotional effects. The Annals of Otology, Rhinology, and Laryngology, 116(11), 858–865.
https://doi.org/10.1177/000348940711601112

70

previous year. 167 Beyond general aging effects on swallowing physiology, age-related disease is the main
risk factor for dysphagia in older adults. 168 Stroke and dementia are examples of common conditions
among the elderly that may contribute to issues with swallowing. 169
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 than patients without dysphagia. 170
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. 171 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 are IRF qualifying conditions). These conditions
accounted for 19.5 percent, 13.1 percent, and 8.7 percent, respectively, of patients in IRFs in 2014. 172
Because of 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 of clinical complexity,
complication risk, and resource use among key groups of IRF patients. Dysphagia commonly affects
stroke patients. Rates vary widely in the literature, from 37 percent to 78 percent of stroke patients,
depending upon the setting and screening instrument used. 173 Dysphagia also is a risk for malnutrition,
which has been found to be common among stroke patients and associated with worse functional
outcomes and more complications. 174 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. 175 Many other neurologic disorders, for which patients may be admitted to an IRF, may
feature dysphagia that may benefit from a mechanically altered diet. 176 Because a mechanically altered
diet can be a marker of clinical complexity and resource use, and because it can be related to the 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.

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

Surgery, 151(5), 765–769. https://doi.org/10.1177/0194599814549156
168 Sura, L., Madhavan, A., Carnaby, G., & Crary, M. A. (2012). Dysphagia in the elderly: Management and nutritional

considerations. Clinical Interventions in Aging, 7, 287–298.
169 Ibid.
170 Patel, D. A., Krishnaswami, S., Steger, E., Conover, E., Vaezi, M. F., Ciucci, M. R., & Francis, D. O. (2018). Economic and

survival burden of dysphagia among inpatients in the United States. Diseases of the Esophagus, 31(1), 1–7.
https://doi.org/10.1093/dote/dox131
171 Dempsey, Mullen, & Buzby, 1988.
172 Medicare Payment Advisory Commission, 2016.
173 Martino, R., Foley, N., Bhogal, S., Diamant, N., Speechley, M., & Teasell, R. (2005). Dysphagia after stroke: Incidence,
diagnosis, and pulmonary complications. Stroke, 36(12), 2756–2763. https://doi.org/10.1161/01.STR.0000190056.76543.eb
174 Finestone, H. M., & Greene-Finestone, L. S. (2003). Rehabilitation medicine: 2. Diagnosis of dysphagia and its nutritional

management for stroke patients. CMAJ, 169(10), 1041–1044. Retrieved from http://www.cmaj.ca/content/169/10/1041.full
175 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.
https://doi.org/10.1016/j.apmr.2007.11.063
176 Buchholz, D. W. (1994). Dysphagia associated with neurological disorders. Acta Oto-Rhino-Laryngologica Belgica, 48(2),
143–155. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/8209677

71

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.
1.
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). 177
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 HHAs, 762 in
IRFs, 448 in LTCHs, and 1,087 in SNFs (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 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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554, overall). The average time to complete the
Mechanically Altered Diet item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in IRFs (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). IRR for the 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.
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. This diet will
eliminate, decrease, or increase certain substances in the diet (e.g., sodium or potassium). Therapeutic
177 Saliba, & Buchanan, 2008b.

72

diets can include low cholesterol, renal, diabetic, and low salt diets, 178 the latter of which are most
commonly used. 179
Certain conditions, including diabetes, 180 chronic kidney disease, 181 hypertension, 182 and heart
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. 184 Additionally, 61.7 percent of adults 65 years of age or older have hypertension. 185 These
conditions may be treated with a therapeutic diet.
disease 183

The Therapeutic Diet data element is important to collect in the IRF setting 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. Communication among PAC settings on 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, 186 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 that 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
home. 187 Diabetes has also been shown to affect 20 to 22 percent of IRF knee replacement patients and 28
percent of stroke patients. 188 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,
178 Kamel, H. K., Malekgoudarzi, B., & Pahlavan, M. (2000). Inappropriate use of therapeutic diets in the nursing home. Journal

of the American Geriatrics Society, 48(7), 856–857. https://doi.org/10.1111/j.1532-5415.2000.tb04771.x
179 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. https://doi.org/10.1177/105477380201100308
180 Centers for Disease Control and Prevention. (2017a). 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
181 Centers for Disease Control and Prevention. (n.d.). Chronic kidney disease initiative [website]. Last reviewed March 12,
2019. Retrieved from http://www.cdc.gov/ckd.
182 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(7), 219–224. https://doi.org/10.15585/mmwr.mm6707a4
183 Centers for Disease Control and Prevention. (2017b). National Center for Health Statistics: Older persons’ health. Retrieved
from https://www.cdc.gov/nchs/fastats/older-american-health.htm
184 Centers for Disease Control and Prevention, 2017a.
185 Fang, Gillespie, Ayala, & Loustalot, 2018.
186 Medicare Payment Advisory Commission, 2016.
187Reistetter, 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. Retrieved from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3114361. https://doi.org/10.2337/dc10-2220
188 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. https://doi.org/10.1016/j.apmr.2008.11.021
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. https://doi.org/10.1161/01.STR.32.2.523

73

respectively. 189 As therapeutic diets may be a common requirement of many key IRF populations and
may be a marker of clinical complexity, standardized assessment of therapeutic diets is warranted in the
IRF setting.
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.
1.
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). 190
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 HHAs, 762 in IRFs, 448 in LTCHs,
and 1,087 in SNFs (n = 2,926 overall).
Across settings, more than 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.
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 HHAs, 457 in
IRFs, 244 in LTCHs, and 431 in SNFs (n = 1,554 overall). The average time to complete the Therapeutic
Diet item was 0.22 minutes overall (SD = 0.1) and 0.25 minutes in the IRF setting (SD = 0.1).
Interrater reliability: IRR was examined for 187 assessments in HHAs, 236 in IRFs, 203 in
LTCHs, and 256 in SNFs (n = 882 overall). The kappa for the Therapeutic Diet data element was
moderate across settings (0.60) and substantial/good in the IRF setting (0.70). 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.

189 Medicare Payment Advisory Commission, 2016.
190 Saliba, & Buchanan, 2008b.

74

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 (HHS) found that 31 percent of adverse events in 2008
among hospitalized Medicare beneficiaries were related to medication. 191 Adverse drug events (ADEs)
may be caused by medication errors such as drug omissions, errors in dosage, and errors in dosing
frequency. 192 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. 193 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 than
that among those younger than age 65. 194
Some classes of drugs are associated with more risk than others. 195 The six medication class
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 because of the adverse effects that may result from use. In particular,
anticoagulants and antiplatelets are associated with bleeding risk; 196 hypoglycemics are associated with
fluid retention, heart failure, and lactic acidosis; 197 opioids are associated with misuse; 198 antipsychotics
are associated with fractures and strokes; 199 and antimicrobials, the category of medications that includes
antibiotics, are associated with various adverse events, such as central nervous systems effects and
gastrointestinal intolerance. 200 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
191 Levinson, D. R. (2010). Adverse events in hospitals: National incidence among Medicare beneficiaries. OEI-06-09-00090.

Washington, DC: U. S. Department of Health and Human Services, Office of Inspector General.
192 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 & Safety in Health Care, 18(1), 32–36.
https://doi.org/10.1136/qshc.2007.025957
193 Barnsteiner, 2005.
Rozich, J., & Roger, R. (2001). Medication safety: One organization’s approach to the challenge. Journal of Clinical Outcomes
Management, 2001(8), 27–34.
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 Health-System Pharmacy,
61(16), 1689–1695. https://doi.org/10.1093/ajhp/61.16.1689
194 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(20), 2115–2125.
https://doi.org/10.1001/jama.2016.16201
195 Ibid.
196 Shoeb, M., & Fang, M. C. (2013). Assessing bleeding risk in patients taking anticoagulants. Journal of Thrombosis and
Thrombolysis, 35(3), 312–319. https://doi.org/10.1007/s11239-013-0899-7
Melkonian, M., Jarzebowski, W., Pautas, E., Siguret, V., Belmin, J., & Lafuente-Lafuente, C. (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 (JTH), 15(7), 1500–1510. https://doi.org/10.1111/jth.13697
197 Hamnvik, O. P., & McMahon, G. T. (2009). Balancing risk and benefit with oral hypoglycemic drugs. The Mount Sinai
Journal of Medicine, New York, 76(3), 234–243. https://doi.org/10.1002/msj.20116
198 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. https://doi.org/10.1016/j.cger.2016.06.006
199 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. https://doi.org/10.1111/jgs.12216
Wang, S., Linkletter, C., Dore, D., Mor, V., Buka, S., & Maclure, M. (2012). Age, antipsychotics, and the risk of ischemic stroke
in the Veterans Health Administration. Stroke, 43(1), 28–31. https://doi.org/10.1161/STROKEAHA.111.617191
200 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. https://doi.org/10.1086/428125

75

for use in older adults. 201 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 crucial. 202
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 5 percent experienced some type of medication-related adverse event over a 1month period, ranging in severity from a longer IRF stay to death. 203 In the same study, more than 8
percent of patients in IRFs experienced a medication-related “temporary harm event” during the 1-month
period, defined as requiring medical intervention but not causing lasting harm. 204 Of all adverse and
temporary harm events identified in IRFs, 46 percent were related to medication. 205 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. 206
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.

201 American Geriatrics Society 2019 Beers Criteria® Update Expert Panel. (2019). American Geriatrics Society 2019: Updated

Beers Criteria® for Potentially Inappropriate Medication Use in Older Adults. Journal of the American Geriatrics Society,
67(4), 674–694. https://doi.org/10.1111/jgs.15767
202 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, 2011, 768–776.
203 Levinson, D. R. (2016, July). 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. Available at:
https://oig.hhs.gov/oei/reports/oei-06-14-00110.pdf
204 Ibid.
205 Ibid.
206 Ibid.

76

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 by
pharmacological classification, not how it is used, in the
following 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)
Z. None of the above
Current use
The MDS currently assesses what classes of medication residents receive. The number of days
the resident received medications is assessed by category 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 are anticoagulants,
antiplatelets (excluding low-dose aspirin), hypoglycemics (including insulin), opioids, antipsychotics, and
antimicrobials (excluding topicals). In the National Beta Test, the data element was administered to 627
patients/residents in HHAs, 769 in IRFs, 459 in LTCHs, and 1,096 in SNFs (n = 2,951 overall).
In the four settings combined, the percentage 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
antiplatelets) to 92 percent (opioids) in the four settings combined, and in the IRF setting, the indication

77

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 HHAs, 446 in
IRFs, 271 in LTCHs, and 421 in SNFs (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: IRR was examined for 187 assessments in HHAs, 240 in IRFs, 212 in
LTCHs, and 261 in SNFs (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.

78

Section 4: Medical Conditions and Co-Morbidities
Pain Interference
Pain is a highly prevalent medical condition in the United States. A Centers for Disease Control
and Prevention (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. 207 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. 208
Conditions causing pain in older adults may be associated with depression, 209 sleep disturbance, 210 and
lower participation in rehabilitation activities. 211
A substantial percentage of older adults receiving services in a PAC setting experience pain.
According to 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 in HHAs
responding “yes” to this question. 212 According to the 2009 Medicare Current Beneficiary Survey, the
prevalence of moderate-to-severe pain 213 among residents of skilled and non-skilled nursing facilities was
22 percent, and the prevalence of persistent pain—defined as the same or worse pain over time—was 65
percent. 214
Pain in older adults can be treated with medications, complementary and alternative approaches,
or physical therapy. 215 Treatment of pain in older adults may be complicated by factors such as dementia;
207 Dahlhamer, J., Lucas, J., Zelaya, C., Nahin, R., Mackey, S., DeBar, L., . . . Helmick, C. (2018). Prevalence of chronic pain

and high-impact chronic pain among adults - United States, 2016. MMWR. Morbidity and Mortality Weekly Report, 67(36),
1001–1006. https://doi.org/10.15585/mmwr.mm6736a2
208 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–
1346. https://doi.org/10.1111/j.1532-5415.2009.02376.x
209 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. https://doi.org/10.1016/j.apmr.2014.02.001
210 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. https://doi.org/10.1002/gps.4349
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. https://doi.org/10.1002/gps.4839
211 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. https://doi.org/10.1007/s11999-013-2927-5
Brenner, I. & Marsella, A. (2008). Factors influencing exercise participation by clients in long-term care. Perspectives (Pre2012), 32(4), 5.
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, Suppl), S137–S144. https://doi.org/10.1016/j.apmr.2012.10.035
212 Gage, B. (2016). Data from the PAC PRD study, 2008-2010 [data file]. Available from Barbara Gage, August 16, 2016.
213 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.”
214 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, Biological
Sciences and Medical Sciences, 70(5), 598–603. https://doi.org/10.1093/gerona/glu226.
215 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

79

high rates of polypharmacy; end-of-life care; and patient expectations, attitudes, and fears related to pain
treatment. 216 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. 217
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 after discharge. Assessing pain in IRF patients during their stay can lead to
appropriate treatment and improved quality of life, reduce complications associated with immobility such as
skin breakdown and infection, and facilitate rehabilitation efforts and returning to community settings. Pain
assessment post-discharge can also be used to plan appropriate treatment and may reduce readmissions.
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
8. 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
8. Unable to answer

216 Molton, I. R., & Terrill, A. L. (2014). Overview of persistent pain in older adults. The American Psychologist, 69(2), 197–

207. https://doi.org/10.1037/a0035794
217 Institute of Medicine (IOM). (2011). Relieving pain in America: A blueprint for transforming prevention, care, education,

and research. Washington, DC: The National Academies Press.
American Geriatrics Society Panel on Pharmacological Management of Persistent Pain in Older Persons, 2009.

80

J0530. Pain Interference with Day-to-Day Activities
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
8. 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 has questions on whether pain has made it hard for the resident to sleep
at night and whether 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 IRR and showed strong IRR (weighted kappas of 0.836 and 0.789,
respectively). 218
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. 219
Evidence supporting use of Pain from the National Beta Test
Assessing Pain: In the National Beta Test, three pain interference data elements were assessed:
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 HHAs, 618 in IRFs, 375 in LTCHs,
and 872 in SNFs (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). Across settings, among these patients/residents, 73 percent
reported that pain rarely interfered with rehabilitation. Within the IRF setting, 76 percent of these patients
reported that pain rarely interfered with rehabilitation; about 1 in 14 (7 percent) had pain that interfered
with therapy “frequently” or “almost constantly.”
218 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 (Vol. 2). Research
Triangle Park, NC: RTI International. Retrieved from https://www.cms.gov/Medicare/Quality-Initiatives-Patient-AssessmentInstruments/Post-Acute-Care-Quality-Initiatives/Downloads/The-Development-and-Testing-of-the-Continuity-AssessmentRecord-and-Evaluation-CARE-Item-Set-Final-Report-on-Reliability-Testing-Volume-2-of-3.pdf
219 Saliba & Buchanan, 2008a.

81

Across settings, among those who reported experiencing any pain, 55 percent of patients/residents
reported pain 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 pain that
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 HHAs, 533 in IRFs, 321 in LTCHs, and
483 in SNFs (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 HHAs, 256 in IRFs, 232 in
LTCHs, and 268 in SNFs (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 both across settings and in the IRF setting specifically). More-detailed IRR statistics are shown in
Appendix C, Table 7.1.2.

82

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 patients’ and residents’ needs during PAC and
at discharge.
Assessments pertaining to sensory status aid PAC providers in understanding the needs of their
patients and residents by establishing a diagnosis of hearing or vision impairment, elucidating the
patients’ and residents’ ability and willingness to participate in treatments or use assistive devices during
their stays, 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.
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. 220 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. 221 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

220 Peelle, J. E., Troiani, V., Grossman, M., & Wingfield, A. (2011). Hearing loss in older adults affects neural systems

supporting speech comprehension. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience,
31(35), 12638–12643. https://doi.org/10.1523/JNEUROSCI.2559-11.2011
221 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

83

outcomes, 222 including falls, 223 dementia, 224 cognitive impairment, 225 anxiety, 226 emotional vitality, 227
and various medical conditions (e.g., arthritis, cancer, cardiovascular disease, diabetes, emphysema, high
blood pressure, and stroke). 228
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. 229 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. 230 Among older adults more
generally, reports on the prevalence of hearing loss vary. The National Institute on Deafness and Other
Communication Disorders 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. 231 Additionally, a study found that twothirds of individuals aged 70 years or older have bilateral hearing loss and approximately three-quarters
have hearing loss in at least one ear. 232
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 the Hearing item or any comparable hearing impairment
assessment items. In PAC PRD testing, 1.1 percent of IRF patients demonstrated severely impaired
hearing. 233 Hearing impairments can affect patient communication with providers, which has implications
for patient understanding of and adherence to treatment plans and rehabilitation goals. Hearing
222 Contrera, K. J., Wallhagen, M. I., Mamo, S. K., Oh, E. S., & Lin, F. R. (2016). Hearing loss health care for older adults.

Journal of the American Board of Family Medicine, 29(3), 394–403. https://doi.org/10.3122/jabfm.2016.03.150235
223 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. https://doi.org/10.1002/lary.25927
224 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. https://doi.org/10.1002/lio2.65
Deal, J. A., Betz, J., Yaffe, K., Harris, T., Purchase-Helzner, E., Satterfield, S., . . . Lin, F. R., & the Health ABC Study Group.
(2017). Hearing impairment and incident dementia and cognitive decline in older adults: The health ABC study. Journals of
Gerontology, Series A, Biological Sciences and Medical Sciences, 72(5), 703–709.
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. https://doi.org/10.1159/000485178
225 Wei et al., 2017.
226 Contrera, K. J., Betz, J., Deal, J., Choi, J. S., Ayonayon, H. N., Harris, T., . . . Lin, F. R., & the Health ABC Study. (2017).
Association of hearing impairment and anxiety in older adults. Journal of Aging and Health, 29(1), 172–184.
https://doi.org/10.1177/0898264316634571
227 Contrera, K. J., Betz, J., Deal, J. A., Choi, J. S., Ayonayon, H. N., Harris, T., . . . Lin, F. R., & the Health ABC Study. (2016).
Association of hearing impairment and emotional vitality in older adults. The Journals of Gerontology. Series B,
Psychological Sciences and Social Sciences, 71(3), 400–404. https://doi.org/10.1093/geronb/gbw005
228 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. https://doi.org/10.1016/j.dhjo.2017.05.007
229 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.
https://doi.org/10.1111/j.1532-5415.1992.tb01932.x
230 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, Morley, et al., 2012).
231 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
232 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. https://doi.org/10.2105/AJPH.2016.303299
233 Gage, Morley, et al., 2012.

84

impairments are also correlated with lower functional status and lower performance on measures of
cognitive functioning in older adults, 234 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.
Data Element for the Assessment of Impairments: Hearing
B0200. Hearing
Ability to hear (with hearing aid or hearing appliances if normally used)
0. Adequate – no difficulty in normal conversation, social interaction, listening to TV
Enter Code
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). 235
Evidence supporting use of Hearing from the National Beta Test
Assessing Hearing: In the National Beta Test, a Hearing assessment item (with hearing aids,
when applicable) was administered to 643 patients/residents in HHAs, 783 in IRFs, 498 in LTCHs, and
1,141 in SNFs (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 HHAs, 499 in
IRFs, 301 in LTCHs, and 456 in SNFs (n = 1,652 overall). Across all settings and in the IRF setting
specifically, the mean time to complete the Hearing item was 0.3 minutes (SD = 0.2 minutes).

234 Lin, F. R., Ferrucci, L., Metter, E. J., An, Y., Zonderman, A. B., & Resnick, S. M. (2011). Hearing loss and cognition in the

Baltimore Longitudinal Study of Aging. Neuropsychology, 25(6), 763–770. https://doi.org/10.1037/a0024238
Keller, B. K., Morton, J. L., Thomas, V. S., & Potter, J. F. (1999). The effect of visual and hearing impairments on functional
status. Journal of the American Geriatrics Society, 47(11), 1319–1325. https://doi.org/10.1111/j.1532-5415.1999.tb07432.x

235 Saliba, & Buchanan, 2008b.

85

Interrater reliability: IRR was assessed for the Hearing item for 197 patients/residents in HHAs,
258 in IRFs, 237 in LTCHs, and 268 in SNFs (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 also 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.
Vision
Visual impairment can be caused by not only age-related diseases (e.g., age-related macular
degeneration, cataracts, glaucoma, and diabetic retinopathy) but also nearsightedness, farsightedness, loss
of near vision with age, and/or untreated disease. 236 In addition to conditions affecting the eye itself,
visual deficits can be caused by other conditions, such as stroke and traumatic brain injury. Visual
impairment in older adults has been associated with depression and anxiety, 237 lower cognitive
function, 238 and poorer quality of life. 239
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, “No vision
or object identification questionable.” 240 Although most 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 United States who were visually
impaired, the largest proportions comprised those in older age categories: 80 years of age and older (50
percent), 70–79 years (24 percent), and 60–69 years (16 percent). 241 By 2050, the proportion of adults
with visual impairment will increase to 64 percent among individuals aged 80 years and older. 242
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, and this was associated with poorer outcomes with
respect to change in self-care and mobility. 243 Additionally, assessment of this information is useful for
ensuring safety in the IRF setting, as impaired vision increases the risk of falls. 244 Visual impairments are

236 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.
https://doi.org/10.1177/0269215511429162
237 Heesterbeek, T. J., van der Aa, H. P. A., van Rens, G. H. M. B., Twisk, J. W. R., & van Nispen, R. M. A. (2017). The
incidence and predictors of depressive and anxiety symptoms in older adults with vision impairment: A longitudinal
prospective cohort study. Ophthalmic & Physiological Optics, 37(4), 385–398. https://doi.org/10.1111/opo.12388
238 Chen, S. P., Bhattacharya, J., & Pershing, S. (2017). Association of vision loss with cognition in older adults. JAMA
Ophthalmology, 135(9), 963–970. https://doi.org/10.1001/jamaophthalmol.2017.2838
239 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, 27(8), 1957–1971. https://doi.org/10.1007/s11136-018-1799-2
240 Gage, Morley, et al., 2012.
241 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. https://doi.org/10.1001/jamaophthalmol.2016.1284
242 Ibid.
243 Ibid
244 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. https://doi.org/10.1093/aje/152.7.633

86

also associated with poorer rehabilitation outcomes among older IRF patients. 245 Visual impairments may
also affect patients’ participation in some rehabilitation therapies and their 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.
Data Element for the Assessment of Impairments: Vision
B1000. Vision
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
Enter Code
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. 246 The Vision data
element is also linked to performance with readily available materials (i.e., newspaper). In addition, the
Vision data element was tested in the PAC PRD assessment. The PAC PRD found substantial agreement
for IRR across settings for this data element (kappa of 0.74). 247
Evidence supporting use of Vision from the National Beta Test
Assessing Vision: In the National Beta Test, the Vision assessment item (with corrective lenses
when applicable) was administered to 643 patients/residents in HHAs, 783 in IRFs, 498 in LTCHs, and
1,141 in SNFs (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.
Missing data: There were very low rates of missing responses for the Vision item both overall
(0.6 percent) and in the IRF setting (0.6 percent), indicating feasibility of administration.

Freeman, E. E., Muñoz, 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. https://doi.org/10.1167/iovs.070326
245 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(5), 669–674.
https://doi.org/10.1682/JRRD.2003.11.0168
246 Saliba, & Buchanan, 2008b.
247 Gage, Smith, et al., 2012.

87

Time to complete: Time to complete was assessed among 396 patients/residents in HHAs, 499 in
IRFs, 301 in LTCHs, and 456 in SNFs (n = 1,652 overall). Across all settings and in the IRF setting
specifically, the mean time to complete the Vision item was 0.3 minutes (SD = 0.2 minutes).
Interrater reliability: IRR was assessed for the Vision item for 197 patients/residents in HHAs,
258 in IRFs, 237 in LTCHs, and 268 in SNFs (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.

88

Section 6: 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 data elements are as follows:
1. Race
2. Ethnicity
3. Preferred Language
4. Interpreter Services;
5. Health Literacy
6. Transportation
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 in PAC settings. 248 Although racial and ethnic disparities decrease when social factors are
controlled for, they often remain. The root causes of these disparities are 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 affect
Medicare beneficiaries.

248

Agency for Healthcare Research and Quality. (2018, September). 2017 National Healthcare Quality and Disparities Report.
AHRQ Pub. No. 18-0033-EF. Rockville, MD: Author.
Fiscella, K., & Sanders, M. R. (2016). Racial and ethnic disparities in the quality of health care. Annual Review of Public Health,
37(1), 375–394. https://doi.org/10.1146/annurev-publhealth-032315-021439
Centers for Medicare & Medicaid Services. (2018, February). 2018 National Impact Assessment of the Centers for Medicare &
Medicaid Services (CMS) Quality Measures Reports. Baltimore, MD: U.S. Department of Health and Human Services,
Centers for Medicare and Medicaid Services.
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.
Chase, J. D., Huang, L., Russell, D., Hanlon, A., O’Connor, M., Robinson, K. M., & Bowles, K. H. (2018). Racial/ethnic
disparities in disability outcomes among post-acute home care patients. Journal of Aging and Health, 30(9), 1406–1426.
https://doi.org/10.1177/0898264317717851

89

Data Elements for the Assessment of SDOH: Race and Ethnicity
Ethnicity
A1005. Ethnicity
Are you of 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
A Race and Ethnicity data element is currently collected in the MDS, LCDS, IRF-PAI, and
OASIS. The data element consists of a single question, which aligns with the 1997 Office of Management
and Budget (OMB) minimum data standards for federal data collection efforts. 249 The 1997 OMB
Standard lists five minimum categories of race: (1) American Indian or Alaska Native, (2) Asian, (3)
Black or African American, (4) Native Hawaiian or Other Pacific Islander, and (5) White. The 1997

249

Office of Management and Budget. (1997, October 30). Revisions to the Standards for the Classification of Federal Data on
Race and Ethnicity (Notice of Decision). Federal Register, 62(210), 58782–58790. Retrieved from
https://www.govinfo.gov/content/pkg/FR-1997-10-30/pdf/97-28653.pdf

90

OMB Standard also lists two minimum categories of ethnicity: (1) Hispanic or Latino, and (2) Not
Hispanic or Latino. 250 The current version uses a “Mark all that apply” response option.
Evidence supporting use of Race and Ethnicity
The modification will result in two separate data elements, one for race and one for ethnicity, that
will 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. Large federal surveys, such as the National Health Interview
Survey, the 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). 251 Individuals with
LEP have been shown to receive worse care and have poorer health outcomes, including higher
readmission rates. 252 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 affect the ability to identify and
address individual medical and non-medical care needs, to convey and understand clinical information,
and to convey and understand 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.
Data Elements for the Assessment of SDOH: 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

250

Ibid.
U.S. Census Bureau, 2013-2017 American Community Survey.
https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_17_5YR_S1601&prodType=table
252
Karliner, L. S., Kim, S. E., Meltzer, D. O., & Auerbach, A. D. (2010). Influence of language barriers on outcomes of hospital
care for general medicine inpatients. Journal of Hospital Medicine, 5(5), 276–282. https://doi.org/10.1002/jhm.658
Kim, E. J., Kim, T., Paasche-Orlow, M. K., Rose, A. J., & Hanchate, A. D. (2017). Disparities in hypertension associated with
limited English proficiency. Journal of General Internal Medicine, 32(6), 632–639. https://doi.org/10.1007/s11606-0173999-9
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.
251

91

Current use

The preferred language of residents and patients and the 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. 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. 253
Although 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, receipt of fewer preventive services, higher
medical costs, and higher rates of emergency department use. 254
Data Element for the Assessment of SDOH: 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
8. Patient unable to respond

253

Institute of Medicine. (2009). Race, ethnicity, and language data: Standardization for health care quality improvement.
Washington, DC: The National Academies Press.
254
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.

92

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. 255 NASEM’s 2016 report on
accounting for social risk factors in Medicare payment considers health literacy an individual risk factor
affected by other social risk factors. 256 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. 257 SILS is publicly available, and
shorter and easier to administer than the S-TOFHLA. Research found that a positive result on the SILS
demonstrates an increased likelihood that an individual has low health literacy. 258
Transportation
Relevance to IRFs
Transportation barriers can affect access to needed health care, causing missed appointments,
delayed care, and unfilled prescriptions, all of which can have a negative impact on health outcomes. 259
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 will facilitate the
connection to programs that can address identified needs.

255

Healthy People 2020. (2019, February). Social determinants of health. Retrieved from
https://www.healthypeople.gov/2020/topics-objectives/topic/social-determinants-of-health.
256
U.S. Department of Health & Human Services, Office of the Assistant Secretary for Planning and Evaluation. (2016,
December). Report to Congress: Social risk factors and performance under Medicare’s value-based purchasing programs.
Washington, DC: Author. Retrieved from https://aspe.hhs.gov/pdf-report/report-congress-social-risk-factors-andperformance-under-medicares-value-based-purchasing-programs.
257
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(1), 21. https://doi.org/10.1186/1471-2296-7-21
Note: At the beginning of this article, the researchers note they substitute the term health literacy with the phase “reading
ability” when discussing their results.
258
Brice, J. H., Foster, M. B., Principe, S., Moss, C., Shofer, F. S., Falk, R. J., . . . DeWalt, D. A. (2014). Single-item or two-item
literacy screener to predict the S-TOFHLA among adult hemodialysis patients. Patient Education and Counseling, 94(1), 71–
75. https://doi.org/10.1016/j.pec.2013.09.020
259
Syed, S. T., Gerber, B. S., & Sharp, L. K. (2013). Traveling towards disease: Transportation barriers to health care access.
Journal of Community Health, 38(5), 976–993. https://doi.org/10.1007/s10900-013-9681-1

93

Data Element for the Assessment of SDOH: 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
X. 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 data element uses the Transportation item from the Protocol for Responding to and Assessing
Patient Assets, Risks, and Experiences (PRAPARE) tool and is reflective of research on the importance of
addressing transportation as a critical SDOH. The national PRAPARE SDOH assessment protocol is
developed and owned by the National Association of Community Health Centers, 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 http://www.nachc.org/prapare. Items in the assessment tool are consistent with Healthy People
2020 priorities and ICD-10 coding. 260
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. 261 Social isolation tends to increase with age, is
a risk factor for physical and mental illness, and is a predictor of mortality. 262 PAC providers are wellsuited to design and implement programs to increase social engagement of patients while accounting for
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.

260

National Association of Community Health Centers. (2019). PRAPARE. Retrieved from http://www.nachc.org/research-anddata/prapare/.
261
Tomaka, J., Thompson, S., & Palacios, R. (2006). The relation of social isolation, loneliness, and social support to disease
outcomes among the elderly. Journal of Aging and Health, 18(3), 359–384. https://doi.org/10.1177/0898264305280993
Leading Age. (2019). Social Connectedness and Engagement Technology for Long-Term and Post-Acute Care: A Primer and
Provider Selection Guide. Washington, DC: Author. Available at https://www.leadingage.org/white-papers/socialconnectedness-and-engagement-technology-long-term-and-post-acute-care-primer-and
262
Landeiro, F., Barrows, P., Nuttall Musson, E., Gray, A. M., & Leal, J. (2017). Reducing social isolation and loneliness in
older people: A systematic review protocol. BMJ Open, 7(5), e013778. https://doi.org/10.1136/bmjopen-2016-013778
Ong, A. D., Uchino, B. N., & Wethington, E. (2016). Loneliness and health in older adults: A mini-review and synthesis.
Gerontology, 62(4), 443–449. https://doi.org/10.1159/000441651
Leigh-Hunt, N., Bagguley, D., Bash, K., Turner, V., Turnbull, S., Valtorta, N., & Caan, W. (2017). An overview of systematic
reviews on the public health consequences of social isolation and loneliness. Public Health, 152, 157–171.
https://doi.org/10.1016/j.puhe.2017.07.035

94

Data Element for the Assessment of SDOH: 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
8. 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 data element uses the social isolation item from the Accountable Health Communities (AHC)
Screening Tool, which was selected from the Patient-Reported Outcomes Measurement Information
System (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 SDOH, including social
isolation. More information about the AHC Screening Tool can be found at
https://innovation.cms.gov/Files/worksheets/ahcm-screeningtool.pdf.

95

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

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

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 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

A2122. Route of Current
Reconciled Medication List
Transmission to Subsequent
Provider
Indicate the route(s) of
transmission of the current
reconciled medication list to the
subsequent provider.
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)

A2122. Route of Current
Reconciled Medication List
Transmission to Subsequent
Provider
Indicate the route(s) of
transmission of the current
reconciled medication list to the
subsequent provider.
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)

A2122. Route of Current
Reconciled Medication List
Transmission to Subsequent
Provider
Indicate the route(s) of
transmission of the current
reconciled medication list to the
subsequent provider.
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)

96

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

LTCH

SNF

Discharge

Discharge

Discharge

A2123. 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

A2123. 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

A2123. 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?

A2124. Route of Current
Reconciled Medication List
Transmission to Patient
Indicate the route(s) of
transmission of the current
reconciled medication list to the
patient/family/caregiver.
A. Electronic Health Record (e.g.,
electronic access to patient portal)
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)

A2124. Route of Current
Reconciled Medication List
Transmission to Patient
Indicate the route(s) of
transmission of the current
reconciled medication list to the
patient/family/caregiver.
A. Electronic Health Record (e.g.,
electronic access to patient portal)
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)

A2124. Route of Current
Reconciled Medication List
Transmission to Resident
Indicate the route(s) of
transmission of the current
reconciled medication list to the
resident/family/caregiver.
A. Electronic Health Record (e.g.,
electronic access to patient portal)
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)

97

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

N

%

.

Odds
Ratio
(OR)

OR 95%
Lower
CL2

SE1

pvalue

3.060

0.038

<.0001

.

.

.

Estimate

OR 95%
Upper CL

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)

98

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

N

%

Estimate

SE1

pvalue

Odds Ratio OR 95%
(OR)
Lower CL2

OR 95%
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
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

Infectious & Parasitic Disease: Septicemia (2)

Dis Nerv Syst: Other Nervous System Disorders (95)

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

(continued)

99

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

Circ Syst: Coron Athero & Chest Pain (101, 102)
Circ Syst: Pulmonary Heart Disease (103)

N

%

Estimate

SE1

pvalue

Odds
Ratio
(OR)

OR 95%
Lower
CL2

OR 95%
Upper CL

10,491

1.8

−0.331

0.035

<.0001

0.718

0.670

0.769

2,767

0.5

−0.214

0.047

<.0001

0.807

0.736

0.886

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

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: Conduction Disorders & Cardiac Dysrhythmia (105, 106)
Circ Syst: CHF (108)
Circ Syst: CVD (109-111, 113)

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

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

(continued)

100

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)
Odds
OR 95%
Ratio
Lower
OR 95%
(OR)
CL2
Upper CL
0.749
0.694
0.810

%
Estimate
0.8
−0.288

SE1
0.040

pvalue
<.0001

0.3

−0.448

0.052

<.0001

0.639

0.577

0.708

4,150

0.7

−0.368

0.041

<.0001

0.692

0.640

0.750

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

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

27,998

4.7

−0.386

0.023

<.0001

0.680

0.649

0.711

Poison Psychotropic Agents, Poison Other Med, Poison Nonmed
(241-243)

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

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

N
4,479

Infective Arthritis and Osteomyelitis (201)

2,046

Rheumatoid Arthritis, Lupus, Other Connective Tissue Disease (202,
210)
Other Joint Disorders & Osteoporosis (204, 206)
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)

(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

N

%

Estimate

SE1

pvalue

Odds
Ratio
(OR)

Gangrene (248)
2,824
0.5
−0.674
0.049
<.0001
0.510
Mental Illness (650-670)
3,272
0.6
−0.289
0.049
<.0001
0.749
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
0.563
Stroke: (Motor <26.15 Age >84.5), (Motor > 22.35 Motor <26.15 Age
22,064
3.7
-1.481
0.028
<.0001
0.227
<84.5) (CMGs: 0108-0109)
Stroke: Motor Score <22.35 and Age <84.5 (CMG: 0110)
37,142
6.3
-1.982
0.027
<.0001
0.138
Traumatic Brain Injury: Motor Score >28.75 (CMGs: 0201-0205)
10,363
1.7
-0.543
0.039
<.0001
0.581
Traumatic Brain Injury: Motor Score <28.75 (CMGs: 0206-0207)
9,844
1.7
-1.405
0.038
<.0001
0.245
Non-traumatic Brain Injury: Motor Score >35.05 (CMGs: 0301-0302)
11,886
2.0
-0.473
0.036
<.0001
0.623
Non-traumatic Brain Injury: Motor Score <35.05 (CMGs: 0303-0304)
24,887
4.2
-1.190
0.031
<.0001
0.304
Traumatic Spinal Cord Injury: All (CMGs: 0401-0405)
4,262
0.7
-1.299
0.047
<.0001
0.273
Non-traumatic Spinal Cord Injury: Motor Score >31.25 (CMGs: 05019,265
1.6
-0.527
0.042
<.0001
0.591
0503)
Non-traumatic Spinal Cord Injury: Motor Score <31.25 (CMGs: 050415,463
2.6
-1.494
0.035
<.0001
0.224
0506)
Neurological: Motor Score >37.35 (CMGs: 0601-0602)
14,291
2.4
-0.495
0.035
<.0001
0.610
Neurological: Motor Score <37.35 (CMGs: 0603-0604)
54,240
9.1
-1.112
0.030
<.0001
0.329
Fracture of Lower Extremity: Motor Score >28.15 (CMGs: 070124,735
4.2
-0.386
0.036
<.0001
0.680
0703)
Fracture of Lower Extremity: Motor Score <28.15 (CMG: 0704)
47,974
8.1
-1.360
0.034
<.0001
0.257
Replacement of Lower Extremity Joint: Motor Score >28.65 (CMGs:
25,157
4.2
-0.382
0.039
<.0001
0.682
0801-0804)
Replacement of Lower Extremity Joint: Motor Score <28.65 (CMGs:
16,215
2.7
-1.191
0.038
<.0001
0.304
0805-0806)
Other Orthopedic: Motor Score >24.15 (CMGs: 0901-0903)
11,492
1.9
-0.453
0.039
<.0001
0.636
Other Orthopedic: Motor Score <24.15 (CMG: 0904)
29,110
4.9
-1.239
0.032
<.0001
0.290
Amputation, Lower Extremity: Motor Score >36.25 (CMGs:10013,408
0.6
-0.481
0.052
<.0001
0.618
1002)
Amputation, Lower Extremity: Motor Score <36.25 (CMG:1003) &
12,303
2.1
-1.146
0.039
<.0001
0.318
Amputation, Non-Lower Extremity (CMGs: 1101-1102)
Osteoarthritis: All (CMGs: 1201-1203)
866
0.2
-1.077
0.081
<.0001
0.341
Rheumatoid, Other Arthritis: All (CMGs: 1301-1303)
1,573
0.3
-1.003
0.063
<.0001
0.367
Cardiac: Motor Score >38.55 (CMGs: 1401-1402)
10,982
1.9
-0.473
0.038
<.0001
0.623

OR 95%
Lower
CL2

OR 95%
Upper CL

0.464
0.681

0.561
0.824

0.535
0.215

0.593
0.240

0.131
0.538
0.228
0.581
0.286
0.249
0.544

0.145
0.627
0.264
0.668
0.323
0.299
0.641

0.209

0.241

0.570
0.310
0.633

0.653
0.349
0.730

0.240
0.632

0.274
0.737

0.282

0.327

0.589
0.272
0.559

0.686
0.309
0.684

0.295

0.343

0.291
0.324
0.579

0.399
0.415
0.671
(continued)

102

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

N

%

Estimate

SE1

pvalue

Odds
Ratio
(OR)

OR 95%
Lower
CL2

OR 95%
Upper CL

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
10,641

4.9
0.7
1.5
0.4
1.8

-1.002
-0.544
-1.057
-1.025
-1.114

0.032
0.047
0.037
0.057
0.038

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

0.367
0.581
0.348
0.359
0.328

0.345
0.530
0.323
0.321
0.305

0.391
0.637
0.374
0.401
0.354

3,141

0.5

-1.036

0.050

<.0001

0.355

0.322

0.392

1,231
72,447

0.2
12.2

-1.180
-0.907

0.070
0.030

<.0001
<.0001

0.307
0.404

0.268
0.381

0.352
0.428

Cardiothoracic surgery

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)
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

Prior acute stay in psychiatric hospital

(continued)

103

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

N

%

Estimate

SE1

pvalue

Odds
Ratio
(OR)

OR 95%
Lower
CL2

OR 95%
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
2,497
0.4
−0.111
0.044
0.0109
0.895
0.822
Infections*

0.753
0.710
0.645
0.620
0.532

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

HCC6: Opportunistic Infections
HCC7: Other Infectious Diseases*

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

(continued)

104

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

N

%

Estimate

SE1

pvalue

Odds
Ratio
(OR)

OR 95%
Lower
CL2

OR 95%
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

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

HCC23: Other Significant Endocrine and Metabolic Disorders*
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)

105

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

N

%

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

HCC59: Reactive and Unspecified Psychosis
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)

106

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

N

%

Estimate

SE1

pvalue

Odds
Ratio
(OR)

OR 95%
Lower
CL2

OR 95%
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)

107

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

N

%

Estimate

SE1

pvalue

Odds
Ratio
(OR)

OR 95%
Lower
CL2

OR 95%
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)

108

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

N

%

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

1

SE = Standard Error; 2 CL = Confidence Limit.

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

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

Mean

SD

st

Min

1 pctl

5th
pctl

10th
pctl

25th
pctl

50th pctl
(Median) 75th pctl

90th
pctl

95th
pctl

99th
pctl

Max

Observed

65.10

7.83

0

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

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. SD = standard deviation, pctl = percentile. Source: RTI International analysis (program reference: MM130).

109

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

35%

Risk-Standardized Rate

30%

Observed Rate

400
350
300
Number of IRFs

Percent of IRFs

25%

250

20%

200

15%

150

10%

100

5%
0%

50
0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95 100

0

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).

110

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

IRF

LTCH

SNF

Overall

N = 35

N = 22

N = 26

N = 60

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 Percentage of Assessments Completed of Each Module
Module

Domains

Frequency

Percent

Communicative, N = 3121
A1-A2

Hearing and Vision

3065

98.2

B1

Brief Interview for Mental Status (BIMS)

3062

98.1

D

Pain Interview

3031

97.1

E1

PHQ-2 to 9

3010

96.4

B2

Confusion Assessment Method (CAM)

2973

95.3

I

Medication Reconciliation Protocol

2951

94.6

J

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.

111

Cognitive Status: Brief Interview for Mental Status (BIMS)
Table 2.1.1: Admission Response Distributions (in Percentages) for BIMS Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

646

786

496

1134

3062

Three

94

96

91

94

94

Two

4

3

4

4

4

One

1

1

2

1

1

None or no answer

0

1

3

1

1

Correct

89

94

88

87

89

Missed by 1 year

2

1

4

2

2

Missed by 2-5 years

1

1

1

2

1

Missed by >5 years or no answer

7

4

8

9

7

Accurate within 5 days

94

93

90

90

91

Missed by 6 days - 1 mo

3

3

2

4

3

Missed by >1 mo or no answer

4

4

8

6

5

Accurate

88

84

77

76

81

Incorrect or no answer

12

16

23

24

19

Yes, no cue required

80

84

78

76

79

Yes, after cue

9

5

9

9

8

No recall or answer

11

11

13

15

13

Yes, no cue required

84

85

78

79

81

Yes, after cue

11

11

12

13

12

No recall or answer

6

5

10

8

7

Yes, no cue required

73

75

64

66

70

Yes, after cue

12

10

12

14

12

No recall or answer

14

14

24

19

18

# 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)

(continued)

112

Table 2.1.1: Admission Response Distributions (in Percentages) for BIMS Items
(continued)
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

646

786

496

1134

3062

Intact

80

82

73

72

76

Moderately impaired

17

15

19

22

18

Severely impaired

4

3

7

7

5

BIMS Impairment Category

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

0.91

0.91

0.91

Recalls 'blue' (b1f)

0.84

0.82

0.87

0.78

0.83

Recalls 'bed' (b1g)

0.93

0.90

0.93

0.93

0.93

BIMS Impairment Category

0.94

0.85

0.91

0.91

0.91

# of words repeated after 1st attempt (b1a)

96

97

96

96

96

Recalls current year (b1b)

97

98

97

97

98

Recalls current month (b1c)

98

99

97

96

98

Recalls current day of week (b1d)

98

94

97

95

96

Recalls 'sock' (b1e)

94

97

95

96

95

Recalls 'blue' (b1f)

95

95

93

91

94

Recalls 'bed' (b1g)

96

95

95

96

96

BIMS Impairment Category

97

95

95

95

96

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

Percent agreement

113

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.

114

Cognitive Status: Confusion Assessment Method (CAM)
Table 2.2.1: Admission Response Distributions (in Percentages) 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)
Yes

5

6

5

4

5

89

85

89

90

88

Behavior continuously present

2

3

3

3

3

Behavior present, fluctuates

9

11

8

8

9

95

94

93

94

94

Behavior continuously present

1

2

2

1

1

Behavior present, fluctuates

4

5

4

6

5

98

95

94

96

96

Behavior continuously present

1

1

2

1

1

Behavior present, fluctuates

2

3

3

3

3

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)

97

93

98

97

96

Did patient have difficulty focusing attn (b2b)

91

89

93

93

91

Was patient thinking disorganized (b2c)

94

93

96

94

94

Did patient have altered consciousness (b2d)

98

97

95

96

96

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

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.

115

Mental Status: PHQ-2 to 9
Table 3.1.1: Admission Response Distribution (in Percentages) for PHQ-2 to 9 Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

639

776

479

1116

3010

65

61

56

65

62

0-1 day

4

4

5

3

4

2-6 days

15

16

13

13

14

7-11 days (half or more)

9

10

11

9

10

12-14 days (nearly all)

8

10

16

10

11

62

57

49

58

57

0-1 day

3

6

4

5

4

2-6 days

20

19

19

19

19

7-11 days (half or more)

7

9

13

8

9

12-14 days (nearly all)

8

8

16

11

10

2.2
(1.6)

2.3
(1.7)

2.7
(1.8)

24

27

38

27

28

153

209

182

306

850

30

34

34

33

33

0-1 day

2

3

1

2

2

2-6 days

15

15

13

16

15

7-11 days (half or more)

19

16

20

16

17

12-14 days (nearly all)

34

31

32

34

33

10

11

13

10

11

0-1 day

1

0

1

1

1

2-6 days

9

17

13

17

15

7-11 days (half or more)

27

26

23

28

26

12-14 days (nearly all)

52

46

50

44

48

50

43

34

46

44

1

2

2

1

1

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

(continued)

116

Table 3.1.1: Admission Response Distribution (in Percentages) for PHQ-2 to 9 Items
(continued)
Items
2-6 days

HHA

IRF

LTCH

SNF

Overall

9

11

10

9

10

7-11 days (half or more)

17

13

16

15

15

12-14 days (nearly all)

22

31

39

29

30

55

52

51

58

55

0-1 day

1

2

1

1

1

2-6 days

12

12

12

10

12

7-11 days (half or more)

15

16

10

12

13

12-14 days (nearly all)

17

17

26

18

19

54

47

44

48

48

0-1 day

1

1

1

1

1

2-6 days

15

16

9

16

14

7-11 days (half or more)

11

11

12

13

12

12-14 days (nearly all)

19

25

34

22

25

64

62

50

68

62

0-1 day

1

0

2

1

1

2-6 days

9

9

10

7

9

7-11 days (half or more)

8

13

13

10

11

18

16

25

14

18

82

78

77

80

79

0-1 day

2

4

3

2

3

2-6 days

9

7

7

9

8

7-11 days (half or more)

5

3

5

5

4

12-14 days (nearly all)

3

7

7

4

5

11.4
(5.0)

11.8
(5.3)

13.0
(5.8)

11.5
(5.1)

11.9
(5.3)

None (0 – 4)

10

4

6

7

6

Mild (5 – 9)

27

36

27

33

31

Moderate (10 – 14)

37

32

25

34

32

Moderately severe (15 – 19)

20

19

28

18

21

6

9

14

8

9

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)

117

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

HHA

IRF

LTCH

SNF

Overall

196

254

231

267

948

Symptom present: little interest or pleasure (e1a1)

0.95

0.99

0.99

0.98

0.98

Symptom frequency: little interest or pleasure (e1a2)

0.98

1.00

0.98

0.98

0.99

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

0.99

0.98

1.00

0.99

0.99

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

0.93

0.98

0.98

0.99

0.98

Eligible for PHQ-9 per PHQ-2

0.96

0.98

0.98

0.98

0.98

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

0.90

1.00

1.00

1.00

0.98

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

1.00

0.98

0.90

0.96

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

1.00

1.00

1.00

1.00

Symptom present: feel bad about self (e1f1)

1.00

1.00

1.00

1.00

1.00

Symptom frequency: feel bad about self (e1f2)

1.00

1.00

0.95

1.00

0.98

Symptom present: trouble concentrating (e1g1)

1.00

1.00

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

1.00

0.97

Symptom present: suicidal thoughts (e1i1)

1.00

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

0.95

0.97

0.96

Symptom present: little interest or pleasure (e1a1)

97

100

100

99

99

Symptom frequency: little interest or pleasure (e1a2)

99

100

98

98

99

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

99

99

100

100

100

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

95

98

98

99

98

Eligible for PHQ-9 per PHQ-2

98

99

99

99

99

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

96

100

100

100

99

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

100

98

94

96

97

Symptom present: tired / no energy (e1d1)

100

98

99

99

99

Symptom frequency: tired / no energy (e1d2)

100

96

99

100

99

98

97

97

100

98

Symptom frequency: poor appetite or overeating (e1e2)

100

100

100

100

100

Symptom present: feel bad about self (e1f1)

100

100

100

100

100

Kappa/weighted kappa

Percent Agreement

Symptom present: poor appetite or overeating (e1e1)

(continued)

118

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

HHA

IRF

LTCH

SNF

Overall

Symptom frequency: feel bad about self (e1f2)

100

100

95

100

98

Symptom present: trouble concentrating (e1g1)

100

100

100

99

100

96

97

97

100

98

Symptom present: moving or speaking slowly (e1h1)

100

97

95

100

98

Symptom frequency: moving or speaking slowly (e1h2)

100

93

100

100

98

Symptom present: suicidal thoughts (e1i1)

100

100

98

100

99

Symptom frequency: suicidal thoughts (e1i2)

93

100

95

100

97

Sum of all symptom frequencies (PHQ-9)*

96

94

94

96

95

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, Treatments, and Interventions (SSTI)
Table 4.1.1: Admission Response Distributions (in Percentages) for SSTI–Chemotherapy
Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Chemotherapy (j2a)

1

3

0

1

1

Chemo treatment performed: IV (j2a2a)

0

1

0

0

0

Chemo treatment performed: oral (j2a3a)

0

2

0

1

1

Chemo treatment performed: other (j2a10a)

0

0

0

0

0

Table 4.1.2: IRR Kappa and Percent Agreement for SSTI–Chemotherapy Items
Items
# of patients

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)

-

-

-

-

-

99

100

100

99

100

Noted chemo treatment performed: IV (j2a2a)

100

100

100

99

100

Noted chemo treatment performed: oral (j2a3a)

100

100

100

100

100

Kappa/Weighted kappa

Percent Agreement
Noted treatment performed: Chemotherapy (j2a)

119

Items
Noted chemo treatment performed: other (j2a10a)

HHA

IRF

100

LTCH

100

SNF

100

Overall

100

100

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 Percentages) for SSTI–Radiation
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

0

0

0

0

0

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

-

-

-

-

-

99

100

100

100

100

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

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 4.3.1: Admission Response Distributions (in Percentages) for SSTI–Oxygen Therapy
Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

13

17

44

16

20

Type of O2 therapy performed: intermittent (j2c2a)

7

11

37

11

14

Type of O2 therapy performed: continuous (j2c3a)

6

8

5

5

6

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

0

1

6

0

1

Treatment performed: Oxygen Therapy (j2c)

120

Table 4.3.2: IRR Kappa and Percent Agreement for SSTI–Oxygen Therapy Items
Items
# of patients

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)

96

94

93

91

93

Type of O2 therapy performed: intermittent (j2c2a)

98

95

92

95

95

Type of O2 therapy performed: continuous (j2c3a)

97

95

92

93

94

100

100

97

100

99

Kappa
Treatment performed: Oxygen Therapy (j2c)

Percent Agreement

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

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 4.4.1: Admission Response Distributions (in Percentages) for SSTI–Suctioning Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Suctioning (j2d)

0

1

5

1

1

Type of suctioning performed: scheduled (j2d2a)

0

0

1

0

0

Type of suctioning performed: as needed (j2d3a)

0

1

5

1

1

121

Table 4.4.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI–Suctioning
Items
Items
# of patients

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)

-

-

-

-

-

99

99

98

96

98

Type of suctioning performed: scheduled (j2d2a)

100

99

99

99

99

Type of suctioning performed: as needed (j2d3a)

99

100

98

96

98

Kappa

Percent Agreement
Treatment performed: Suctioning (j2d)

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 4.5.1: Admission Response Distributions (in Percentages) for SSTI–Tracheostomy
Care Item
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

0

1

5

0

1

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

100

99

100

100

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

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.

122

Table 4.6.1: Admission Response Distributions (in Percentages) for SSTI–Noninvasive
Mechanical Ventilator (NIMV)
Items

HHA

# of assessments

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Treatment performed: Non-invasive Mechanical Ventilator
(j2g)

4

6

9

4

5

Type of NIMV performed: BiPAP (j2g2a)

1

1

7

1

2

Type of NIMV performed: CPAP (j2g3a)

2

6

2

3

3

Table 4.6.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI–Noninvasive
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)

96

98

96

98

97

Type of NIMV performed: BiPAP (j2g2a)

96

100

97

100

98

Type of NIMV performed: CPAP (j2g3a)

98

98

98

98

98

# of patients
Kappa

Percent Agreement

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 4.7.1: Admission Response Distributions (in Percentages) 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

0

0

3

0

0

123

Table 4.7.2: IRR Kappa and Percent Agreement for SSTI–Invasive Mechanical Ventilator
Item
Items
# of patients

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

-

-

-

-

-

100

100

100

100

100

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

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 4.8.1: Admission Response Distributions (in Percentages) for SSTI–IV Meds Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Other treatment performed: IV Meds (j2h)

15

17

77

16

25

Type of IV meds given: antibiotics (j2h3a)

4

8

64

9

16

Type of IV meds given: anticoagulation (j2h4a)

8

6

17

6

8

Type of IV meds given: other (j2h10a)

6

5

20

4

7

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

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

-

0.13

Type of IV meds given: other (j2h10a)

-

-

0.46

-

0.46

Other treatment performed: IV Meds (j2h)

83

91

89

87

88

Type of IV meds given: antibiotics (j2h3a)

98

97

93

96

96

Type of IV meds given: anticoagulation (j2h4a)

90

94

82

92

90

Type of IV meds given: other (j2h10a)

93

98

79

94

91

Kappa

Percent Agreement

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.

124

Table 4.9.1: Admission Response Distributions (in Percentages) for SSTI–Transfusions
Item
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

0

1

2

0

0

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

99

99

100

100

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

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 4.10.1: Admission Response Distributions (in Percentages) for SSTI–Dialysis Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Other treatment performed: Dialysis (j2j)

3

5

15

3

5

Type of dialysis performed: hemodialysis (j2j2a)

3

4

15

3

5

Type of dialysis performed: peritoneal (j2j3a)

0

0

0

0

0

125

Table 4.10.2: IRR Kappa/Weighted Kappa and Percent Agreement for SSTI–Dialysis
Items
Items
# of patients

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)

98

98

98

99

98

Type of dialysis performed: hemodialysis (j2j2a)

98

98

97

99

98

100

100

100

100

100

Kappa

Percent Agreement

Type of dialysis performed: peritoneal (j2j3a)

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 4.11.1: Admission Response Distributions (in Percentages) for SSTI–IV Access Items
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

Other treatment performed: IV Access (j2k)

4

22

91

10

24

Type of IV access: peripheral IV (j2k2a)

0

14

40

2

11

Type of IV access: midline (j2k3a)

0

1

13

0

2

Type of IV access: central line (j2k4a)

3

6

54

7

13

Type of IV access: other (j2k10a)

0

2

3

1

1

126

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

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)

-

-

-

-

-

97

94

99

95

96

Type of IV access: peripheral IV (j2k2a)

100

96

89

97

96

Type of IV access: midline (j2k3a)

100

99

94

100

98

Type of IV access: central line (j2k4a)

98

98

89

97

96

Type of IV access: other (j2k10a)

97

98

95

99

97

Kappa

Percent Agreement
Other treatment performed: IV Access (j2k)

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.

Nutritional Approaches
Table 5.1.1: Admission Response Distributions (in Percentages) for Nutritional
Approaches–Parenteral/IV Feeding
Items
# of assessments
Nutritional approach performed: parenteral/IV (j1a)

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

0

1

4

0

1

127

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

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

-

-

-

-

-

100

100

99

100

100

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

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 5.2.1: Admission Response Distributions (in Percentages) for Nutritional
Approaches–Feeding Tube
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

0

3

8

2

3

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

100

98

100

100

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

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 5.3.1: Admission Response Distributions (in Percentages) for Nutritional
Approaches–Mechanically Altered Diet
Items
# of assessments
Nutritional approach performed: mechanically altered diet (j1c)

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

2

15

14

11

10

128

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

89

92

94

93

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

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 5.4.1: Admission Response Distributions (in Percentages) for Nutritional
Approaches–Therapeutic Diet
Items
# of assessments
Nutritional approach performed: therapeutic diet (j1d)

HHA

IRF

LTCH

SNF

Overall

629

762

448

1087

2926

54

49

59

49

52

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

HHA

IRF

LTCH

SNF

Overall

187

236

203

256

882

0.43

0.70

0.62

0.61

0.60

71

85

82

80

80

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

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.

129

High-Risk Drug Classes: Use and Indication Items
Table 6.1.1: Admission Response Distributions (in Percentages) for Medication Class
Taking and Indication Items

Medication Class

HHA
(627)

IRF
(769)

LTCH
(459)

SNF
(1096)

Overall
(2951)

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

Anticoagulants

29

47

61

29

66

20

42

77

48

45

Antiplatelets

15

52

19

31

16

10

12

77

15

45

Hypoglycemics

29

47

30

49

48

52

26

72

31

56

Opioids

39

87

51

91

64

90

52

96

51

92

Antipsychotics

9

73

9

33

14

30

16

89

12

66

Antimicrobials

13

57

23

60

73

22

27

84

30

53

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

130

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

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)

91

93

94

93

93

Is patient taking: antiplatelets (i1a2)

92

91

95

91

92

Is patient taking: hypoglycemics (i1a3)

92

92

99

96

95

Is patient taking: opioids (i1a4)

92

93

96

92

93

Is patient taking: antipsychotics (i1a5)

96

95

94

93

94

Is patient taking: antimicrobials (i1a6)

94

91

97

93

94

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

79

89

91

96

90

Indication noted for anticoagulants (i1b1)

77

85

94

95

89

Indication noted for antiplatelets (i1b2)

84

93

100

95

94

Indication noted for hypoglycemics (i1b3)

69

82

85

90

82

Indication noted for opioids (i1b4)

87

96

89

100

94

Indication noted for antipsychotics (i1b5)

63

100

95

89

90

Indication noted for antimicrobials (i1b6)

88

81

91

98

91

Indication noted for opioids (i1b4)

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.

131

Pain: Pain Interference
Table 7.1.1: Admission Response Distributions (in Percentages) for Pain Interference Items
Among Patients/Residents Reporting Any Pain in the Last 3 Days or 5 Days
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

489

618

375

872

2354

Rarely or not at all

40

32

29

37

35

Occasionally

29

30

24

28

28

Frequently

19

26

29

23

24

Almost constantly

12

13

17

13

13

78

98

81

93

89

379

606

302

803

2090

Rarely or not at all

74

76

62

73

73

Occasionally

14

17

17

16

16

Frequently

7

5

14

8

8

Almost constantly

5

2

7

3

4

Rarely or not at all

40

55

42

41

45

Occasionally

26

18

19

26

23

Frequently

17

16

20

21

19

Almost constantly

16

11

19

12

14

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)

132

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

0.98

0.99

0.98

How often limited rehab due to pain (d4b)

0.95

0.96

0.98

0.97

0.97

How often limited daily activities due to pain (d4c)

0.97

0.98

0.99

0.98

0.98

How often pain made it hard to sleep (d3)

95

98

98

100

98

How often limited rehab due to pain (d4b)

97

98

98

99

98

How often limited daily activities due to pain (d4c)

97

98

99

99

98

Kappa

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. *Pearson correlation for rating
of worst pain, which is on a 0-10 scale

Impairments: Hearing
Table 8.1.1: Admission Response Distributions (in Percentages) for Hearing Item
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

643

783

498

1141

3065

Adequate

65

75

81

76

74

Minimal difficulty

24

18

13

15

17

Moderate difficulty

11

6

4

8

8

0

1

1

1

1

Ability to hear (a1)

Highly impaired

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

83

87

84

83

84

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.

133

Impairments: Vision
Table 9.2.1: Admission Response Distributions (in Percentages) for Vision Item
Items
# of assessments

HHA

IRF

LTCH

SNF

Overall

643

783

498

1141

3065

Adequate

73

85

76

78

78

Impaired

21

12

16

16

16

Moderately impaired

4

2

6

4

4

Highly impaired

1

1

1

1

1

Severely impaired

1

0

1

1

1

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

83

90

75

83

83

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.

134