2016-18113. Medicare Program; Prospective Payment System and Consolidated Billing for Skilled Nursing Facilities for FY 2017, SNF Value-Based Purchasing Program, SNF Quality Reporting Program, and SNF Payment Models Research

AGENCY:

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

ACTION:

Final rule.

SUMMARY:

This final rule updates the payment rates used under the prospective payment system (PPS) for skilled nursing facilities (SNFs) for fiscal year (FY) 2017. In addition, it specifies a potentially preventable readmission measure for the Skilled Nursing Facility Value-Based Purchasing Program (SNF VBP), and implements requirements for that program, including performance standards, a scoring methodology, and a review and correction process for performance information to be made public, aimed at implementing value-based purchasing for SNFs. Additionally, this final rule includes additional polices and measures in the Skilled Nursing Facility Quality Reporting Program (SNF QRP). This final rule also responds to comments on the SNF Payment Models Research (PMR) project.

DATES:

These regulations are effective on October 1, 2016.

FOR FURTHER INFORMATION CONTACT:

 

Penny Gershman, [phone], for information related to SNF PPS clinical issues.

John Kane, [phone], for information related to the development of the payment rates and case-mix indexes.

Kia Sidbury, [phone], for information related to the wage index.

Bill Ullman, [phone], for information related to level of care determinations, consolidated billing, and general information.

Stephanie Frilling, [phone], for information related to skilled nursing facility value-based purchasing.

Charlayne Van, [phone], for information related to skilled nursing facility quality reporting.

SUPPLEMENTARY INFORMATION:

Availability of Certain Tables Exclusively Through the Internet on the CMS Web Site

As discussed in the FY 2017 SNF PPS proposed rule (81 FR 24230), tables setting forth the Wage Index for Urban Areas Based on CBSA Labor Market Areas and the Wage Index Based on CBSA Labor Market Areas for Rural Areas are no longer published in the Federal Register. Instead, these tables are available exclusively through the Internet on the CMS Web site. The wage index tables for this final rule can be accessed on the SNF PPS Wage Index home page, at http://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​WageIndex.html.

Readers who experience any problems accessing any of these online SNF PPS wage index tables should contact Kia Sidbury at [phone].

To assist readers in referencing sections contained in this document, we are providing the following Table of Contents.

Table of Contents

I. Executive Summary

A. Purpose

B. Summary of Major Provisions

C. Summary of Cost and Benefits

II. Background on SNF PPS

A. Statutory Basis and Scope

B. Initial Transition for the SNF PPS

C. Required Annual Rate Updates

III. Analysis of and Responses to Public Comments on the FY 2017 SNF PPS Proposed Rule

A. General Comments on the FY 2017 SNF PPS Proposed Rule

B. SNF PPS Rate Setting Methodology and FY 2017 Update

1. Federal Base Rates

2. SNF Market Basket Update

3. Case-Mix Adjustment

4. Wage Index Adjustment

5. Adjusted Rate Computation Example

C. Additional Aspects of the SNF PPS

1. SNF Level of Care—Administrative Presumption

2. Consolidated Billing

3. Payment for SNF-Level Swing-Bed Services

D. Other Issues

1. Skilled Nursing Facility Value-Based Purchasing Program (SNF VBP)

2. Skilled Nursing Facility (SNF) Quality Reporting Program (QRP)

3. SNF Payment Models Research

IV. Collection of Information Requirements

V. Economic Analyses

Acronyms

In addition, because of the many terms to which we refer by acronym in this final rule, we are listing these abbreviations and their corresponding terms in alphabetical order below:

AIDS Acquired Immune Deficiency Syndrome

ARD Assessment reference date

BBA Balanced Budget Act of 1997, Pub. L. 105-33

BBRA Medicare, Medicaid, and SCHIP Balanced Budget Refinement Act of 1999, Pub. L. 106-113

BIPA Medicare, Medicaid, and SCHIP Benefits Improvement and Protection Act of 2000, Pub. L. 106-554

CAH Critical access hospital

CASPER Certification and Survey Provider Enhanced Reporting

CBSA Core-based statistical area

CCN CMS Certification Number

CFR Code of Federal Regulations

CMI Case-mix index

CMS Centers for Medicare & Medicaid Services

FFS Fee-for-service

FR Federal Register

FY Fiscal year

HCPCS Healthcare Common Procedure Coding System

HIQR Hospital Inpatient Quality Reporting

HOQR Hospital Outpatient Quality Reporting

HRRP Hospital Readmissions Reduction Program

HVBP Hospital Value-Based Purchasing

IGI IHS (Information Handling Services) Global Insight, Inc.

IMPACT Improving Medicare Post-Acute Care Transformation Act of 2014, Pub. L. 113-185

IPPS Inpatient prospective payment system

IRF Inpatient Rehabilitation Facility

LTC Long-term care

LTCH Long-term care hospital

MAP Measures Application Partnership

MDS Minimum data set

MFP Multifactor productivity

MMA Medicare Prescription Drug, Improvement, and Modernization Act of 2003, Pub. L. 108-173

MSA Metropolitan statistical area

NF Nursing facility

NQF National Quality Forum

OMB Office of Management and Budget

PAC Post-acute care

PAMA Protecting Access to Medicare Act of 2014, Pub. L. 113-93

PBJ Payroll-Based Journal

PMR Payment Models Research

PPS Prospective Payment System

PQRS Physician Quality Reporting System

QIES Quality Improvement Evaluation System

QIES ASAP Quality Improvement and Evaluation System Assessment Submission and Processing

QRP Quality Reporting Program

RAI Resident assessment instrument

RAVEN Resident assessment validation entry

RFA Regulatory Flexibility Act, Pub. L. 96-354

RIA Regulatory impact analysis

RUG-III Resource Utilization Groups, Version 3

RUG-IV Resource Utilization Groups, Version 4

RUG-53 Refined 53-Group RUG-III Case-Mix Classification System

SCHIP State Children's Health Insurance Program

sDTI Suspected deep tissue injuries

SNF Skilled nursing facility

SNF QRP Skill nursing facility quality reporting program

SNFRM Skilled Nursing Facility 30-Day All-Cause Readmission Measure

STM Staff time measurement

STRIVE Staff time and resource intensity verification

TEP Technical expert panel

UMRA Unfunded Mandates Reform Act, Pub. L. 104-4

VBP Value-based purchasing

I. Executive Summary

A. Purpose

This final rule updates the SNF prospective payment rates for FY 2017 as required under section 1888(e)(4)(E) of the Social Security Act (the Act). It also responds to section 1888(e)(4)(H) of the Act, which requires the Secretary to provide for publication in the Federal Register before the August 1 that precedes the start of each fiscal year (FY) certain specified information relating to the payment update (see section II.C.). This final rule also includes an update on the SNF PMR project. In addition, it specifies a potentially preventable readmission measure for the Skilled Nursing Facility (SNF) Value-Based Purchasing (VBP) Program and finalizes other requirements related to that Program's implementation, including performance standards, a scoring methodology, and a review and correction process for performance information to be made public under the Program. We are also including four new quality and resource use measures for the SNF QRP and new SNF review and correction procedures for performance data that are to be publicly reported.

B. Summary of Major Provisions

In accordance with sections 1888(e)(4)(E)(ii)(IV) and 1888(e)(5) of the Act, the federal rates in this final rule reflect an update to the rates that we published in the SNF PPS final rule for FY 2016 (80 FR 46390), which reflects the SNF market basket index, as adjusted by the multifactor productivity (MFP) adjustment, for FY 2017. We are also finalizing various requirements for the SNF VBP Program, including a potentially preventable readmission measure, performance standards, and a scoring methodology, among other policies. In addition, we are adopting and implementing four new quality and resource use measures for the SNF QRP and new SNF review and correction procedures for performance data that are to be publicly reported as we continue to implement this program and meet the requirements of the IMPACT Act.

C. Summary of Cost and Benefits

Provision description	Total transfers
FY 2017 SNF PPS payment rate update	The overall economic impact of this final rule would be an estimated increase of $920 million in aggregate payments to SNFs during FY 2017.

II. Background on SNF PPS

A. Statutory Basis and Scope

As amended by section 4432 of the Balanced Budget Act of 1997 (BBA, Pub. L. 105-33, enacted on August 5, 1997), section 1888(e) of the Act provides for the implementation of a PPS for SNFs. This methodology uses prospective, case-mix adjusted per diem payment rates applicable to all covered SNF services defined in section 1888(e)(2)(A) of the Act. The SNF PPS is effective for cost reporting periods beginning on or after July 1, 1998, and covers all costs of furnishing covered SNF services (routine, ancillary, and capital-related costs) other than costs associated with approved educational activities and bad debts. Under section 1888(e)(2)(A)(i) of the Act, covered SNF services include post-hospital extended care services for which benefits are provided under Part A, as well as those items and services (other than a small number of excluded services, such as physician services) for which payment may otherwise be made under Part B and which are furnished to Medicare beneficiaries who are residents in a SNF during a covered Part A stay. A comprehensive discussion of these provisions appears in the May 12, 1998 interim final rule (63 FR 26252). In addition, a detailed discussion of the legislative history of the SNF PPS is available online at http://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​Downloads/​Legislative_​History_​07302013.pdf.

Section 215(a) of PAMA added section 1888(g) to the Act requiring the Secretary to specify an all-cause all-condition hospital readmission measure and a resource use measure, an all-condition risk-adjusted potentially preventable hospital readmission measure, for the SNF setting. Additionally, section 215(b) of PAMA added section 1888(h) to the Act requiring the Secretary to implement a VBP program for SNFs. Finally, section 2(a) of the IMPACT Act added section 1899B to the Act that, among other things, requires SNFs to report standardized data for measures in specified quality and resource use domains. In addition, the IMPACT Act added section 1888(e)(6) to the Act, which requires the Secretary to implement a quality reporting program for SNFs, which includes a requirement that SNFs report certain data to receive their full payment under the SNF PPS.

B. Initial Transition for the SNF PPS

Under sections 1888(e)(1)(A) and 1888(e)(11) of the Act, the SNF PPS included an initial, three-phase transition that blended a facility-specific rate (reflecting the individual facility's historical cost experience) with the federal case-mix adjusted rate. The transition extended through the facility's first 3 cost reporting periods under the PPS, up to and including the one that began in FY 2001. Thus, the SNF PPS is no longer operating under the transition, as all facilities have been paid at the full federal rate effective with cost reporting periods beginning in FY 2002. As we now base payments for SNFs entirely on the adjusted federal per diem rates, we no longer include adjustment factors under the transition related to facility-specific rates for the upcoming FY.

C. Required Annual Rate Updates

Section 1888(e)(4)(E) of the Act requires the SNF PPS payment rates to be updated annually. The most recent annual update occurred in a final rule that set forth updates to the SNF PPS payment rates for FY 2016 (80 FR 46390, August 4, 2015).

Section 1888(e)(4)(H) of the Act specifies that we provide for publication annually in the Federal Register of the following:

• The unadjusted federal per diem rates to be applied to days of covered SNF services furnished during the upcoming FY.
• The case-mix classification system to be applied for these services during the upcoming FY.
• The factors to be applied in making the area wage adjustment for these services.

Along with other revisions discussed later in this preamble, this final rule would provide the required annual updates to the per diem payment rates for SNFs for FY 2017.

III. Analysis of and Responses to Public Comments on the FY 2017 SNF PPS Proposed Rule

In response to the publication of the FY 2017 SNF PPS proposed rule, we received 95 public comments from individuals, providers, corporations, government agencies, private citizens, trade associations, and major organizations. The following are brief summaries of each proposed provision, a summary of the public comments that we received related to that proposal, and our responses to the comments.

A. General Comments on the FY 2017 SNF PPS Proposed Rule

In addition to the comments we received on specific proposals contained within the proposed rule (which we address later in this final rule), commenters also submitted the following, more general, observations on the SNF PPS and SNF care generally. A discussion of these comments, along with our responses, appears below.

Comment: One commenter stated that there is a significant amount of fraud and abuse in the SNF PPS. The commenter further stated that, often times, non-licensed professionals will dictate the type of care beneficiaries receive, specifically referring to the number of therapy minutes a beneficiary receives. This commenter also stated that if a health care professional tries to speak about these issues, his or her job may be in jeopardy.

Response: We appreciate this commenter raising these concerns. While outside the scope of this rule, we will pass these concerns along to our colleagues in the Center for Program Integrity, who are responsible for identifying and addressing instances of fraud, waste and abuse in the Medicare program. Additionally, information on areas of potential waste, fraud or abuse may be reported to the Office of the Inspector General Hotline by calling 1-800-HHS-TIPS (1-800-447-8477).

Comment: A number of commenters raised concerns regarding the cost of care for the beneficiary. One commenter discussed how the individual beneficiary cost for living in a nursing home seemed to greatly exceed the cost of living in the community. A few commenters referenced the pace and breadth of potential changes to conditions of participation for long-term care facilities, notably those contained in rulemaking such as the 2015 proposed rule entitled “Medicare and Medicaid Programs: Reform of Requirements for Long-Term Care Facilities” (80 FR 42168), as well as noted that the cost of implementing these provisions is not covered by Medicaid or Medicare.

Response: While we appreciate the commenters raising these concerns, these comments and the provisions of the proposed rule referenced by commenters are outside the scope of this final rule. That being said, we will share these comments with the appropriate team within CMS responsible for these provisions.

Comment: A few commenters raised concerns regarding decisions made by Medicare Administrative Contractors. One commenter requested that we instruct these contractors to refrain from denying coverage and payment for SNF Part B claims in which physician visits occur more frequently than the minimum standards set by the conditions of participation at § 483.40. Another commenter requested that we examine potential instances in which contractors might unnecessarily target speech-language pathology services by making revisions to Medicare manuals which might affect coverage of these services.

Response: With regard to our instructing the contractors to refrain from denying coverage or payment for SNF Part B claims in which physician visits occur more frequently than the minimum standard set by the conditions of participation, this comment is outside the scope of this final rule. However, we will forward these comments to the appropriate division within CMS for consideration. With regard to contractors targeting speech-language pathology services, we are not aware of such targeting. We will continue to educate the contractors to ensure compliance with all federal guidance and regulations.

B. SNF PPS Rate Setting Methodology and FY 2017 Update

1. Federal Base Rates

Under section 1888(e)(4) of the Act, the SNF PPS uses per diem federal payment rates based on mean SNF costs in a base year (FY 1995) updated for inflation to the first effective period of the PPS. We developed the federal payment rates using allowable costs from hospital-based and freestanding SNF cost reports for reporting periods beginning in FY 1995. The data used in developing the federal rates also incorporated a Part B add-on, which is an estimate of the amounts that, prior to the SNF PPS, would have been payable under Part B for covered SNF services furnished to individuals during the course of a covered Part A stay in a SNF.

In developing the rates for the initial period, we updated costs to the first effective year of the PPS (the 15-month period beginning July 1, 1998) using a SNF market basket index, and then standardized for geographic variations in wages and for the costs of facility differences in case mix. In compiling the database used to compute the federal payment rates, we excluded those providers that received new provider exemptions from the routine cost limits, as well as costs related to payments for exceptions to the routine cost limits. Using the formula that the BBA prescribed, we set the federal rates at a level equal to the weighted mean of freestanding costs plus 50 percent of the difference between the freestanding mean and weighted mean of all SNF costs (hospital-based and freestanding) combined. We computed and applied separately the payment rates for facilities located in urban and rural areas, and adjusted the portion of the federal rate attributable to wage-related costs by a wage index to reflect geographic variations in wages.

2. SNF Market Basket Update

a. SNF Market Basket Index

Section 1888(e)(5)(A) of the Act requires us to establish a SNF market basket index that reflects changes over time in the prices of an appropriate mix of goods and services included in covered SNF services. Accordingly, we have developed a SNF market basket index that encompasses the most commonly used cost categories for SNF routine services, ancillary services, and capital-related expenses. We use the SNF market basket index, adjusted in the manner described below, to update the federal rates on an annual basis. In the SNF PPS final rule for FY 2014 (78 FR 47939 through 47946), we revised and rebased the market basket, which included updating the base year from FY 2004 to FY 2010.

For the FY 2017 proposed rule, the FY 2010-based SNF market basket growth rate was estimated to be 2.6 percent, which was based on the IHS Global Insight Inc. (IGI) first quarter 2016 forecast, with historical data through fourth quarter 2015. However, as discussed in the FY 2017 SNF PPS proposed rule (81 FR 24234), we proposed that if more recent data become available (for example, a more recent estimate of the FY 2010 based SNF market basket and/or MFP adjustment), we would use such data, if appropriate, to determine the FY 2017 SNF market basket percentage change, labor-related share relative importance, forecast error adjustment, and MFP adjustment in this final rule. Since that time, we have received an updated FY 2017 market basket percentage increase, which is based on the second quarter 2016 IGI forecast of the FY 2010-based SNF market basket. The revised market basket growth rate is 2.7 percent. In section III.B.2.e. of this final rule, we discuss the specific application of this adjustment to the forthcoming annual update of the SNF PPS payment rates.

b. Use of the SNF Market Basket Percentage

Section 1888(e)(5)(B) of the Act defines the SNF market basket percentage as the percentage change in the SNF market basket index from the midpoint of the previous FY to the midpoint of the current FY. For the federal rates set forth in this final rule, we use the percentage change in the SNF market basket index to compute the update factor for FY 2017. This is based on the IGI second quarter 2016 forecast (with historical data through the first quarter 2016) of the FY 2017 percentage increase in the FY 2010-based SNF market basket index for routine, ancillary, and capital-related expenses, which is used to compute the update factor in this final rule. As discussed in sections III.B.2.c. and III.B.2.d. of this final rule, this market basket percentage change is reduced by the applicable forecast error correction (as described in § 413.337(d)(2)) and by the MFP adjustment as required by section 1888(e)(5)(B)(ii) of the Act. Finally, as discussed in section II.B. of this final rule, we no longer compute update factors to adjust a facility-specific portion of the SNF PPS rates, because the initial three-phase transition period from facility-specific to full federal rates that started with cost reporting periods beginning in July 1998 has expired.

c. Forecast Error Adjustment

As discussed in the June 10, 2003 supplemental proposed rule (68 FR 34768) and finalized in the August 4, 2003, final rule (68 FR 46057 through 46059), § 413.337(d)(2) provides for an adjustment to account for market basket forecast error. The initial adjustment for market basket forecast error applied to the update of the FY 2003 rate for FY 2004, and took into account the cumulative forecast error for the period from FY 2000 through FY 2002, resulting in an increase of 3.26 percent to the FY 2004 update. Subsequent adjustments in succeeding FYs take into account the forecast error from the most recently available FY for which there are final data, and apply the difference between the forecasted and actual change in the market basket when the difference exceeds a specified threshold. We originally used a 0.25 percentage point threshold for this purpose; however, for the reasons specified in the FY 2008 SNF PPS final rule (72 FR 43425, August 3, 2007), we adopted a 0.5 percentage point threshold effective for FY 2008 and subsequent FYs. As we stated in the final rule for FY 2004 that first issued the market basket forecast error adjustment (68 FR 46058, August 4, 2003), the adjustment will reflect both upward and downward adjustments, as appropriate.

For FY 2015 (the most recently available FY for which there is final data), the estimated increase in the market basket index was 2.5 percentage points, while the actual increase for FY 2015 was 2.3 percentage points, resulting in the actual increase being 0.2 percentage point lower than the estimated increase. Accordingly, as the difference between the estimated and actual amount of change in the market basket index does not exceed the 0.5 percentage point threshold, the FY 2017 market basket percentage change of 2.7 percent will be not adjusted to account for the forecast error. Table 1 shows the forecasted and actual market basket amounts for FY 2015.

Table 1—Difference Between the Forecasted and Actual Market Basket Increases for FY 2015

Index	Forecasted FY 2015 increase *	Actual FY 2015 increase **	FY 2015 difference
SNF	2.5	2.3	0.2
* Published in Federal Register; based on second quarter 2014 IGI forecast (2010-based index).
** Based on second quarter 2016 IGI forecast, with historical data through the first quarter 2016 (2010-based index).

d. Multifactor Productivity Adjustment

Section 3401(b) of the Affordable Care Act requires that, in FY 2012 (and in subsequent FYs), the market basket percentage under the SNF payment system as described in section 1888(e)(5)(B)(i) of the Act is to be reduced annually by the productivity adjustment described in section 1886(b)(3)(B)(xi)(II) of the Act. Section 1886(b)(3)(B)(xi)(II) of the Act, added by section 3401(a) of the Affordable Care Act, sets forth the definition of this productivity adjustment. The statute defines the productivity adjustment to be equal to the 10-year moving average of changes in annual economy-wide private nonfarm business multi-factor productivity (as projected by the Secretary for the 10-year period ending with the applicable FY, year, cost-reporting period, or other annual period) (the MFP adjustment). The Bureau of Labor Statistics (BLS) is the agency that publishes the official measure of private nonfarm business MFP. We refer readers to the BLS Web site at http://www.bls.gov/​mfp for the BLS historical published MFP data.

MFP is derived by subtracting the contribution of labor and capital inputs growth from output growth. The projections of the components of MFP are currently produced by IGI, a nationally recognized economic forecasting firm with which CMS contracts to forecast the components of the market baskets and MFP. To generate a forecast of MFP, IGI replicates the MFP measure calculated by the BLS, using a series of proxy variables derived from IGI's U.S. macroeconomic models. For a discussion of the MFP projection methodology, we refer readers to the FY 2012 SNF PPS final rule (76 FR 48527 through 48529) and the FY 2016 SNF PPS final rule (80 FR 46395). A complete description of the MFP projection methodology is available on our Web site at http://www.cms.gov/​Research-Statistics-Data-and-Systems/​Statistics-Trends-and-Reports/​MedicareProgramRatesStats/​MarketBasketResearch.html.

(i) Incorporating the MFP Adjustment Into the Market Basket Update

Per section 1888(e)(5)(A) of the Act, the Secretary shall establish a SNF market basket index that reflects changes over time in the prices of an appropriate mix of goods and services included in covered SNF services. Section 1888(e)(5)(B)(ii) of the Act, added by section 3401(b) of the Affordable Care Act, requires that for FY 2012 and each subsequent FY, after determining the market basket percentage described in section 1888(e)(5)(B)(i) of the Act, the Secretary shall reduce such percentage by the productivity adjustment described in section 1886(b)(3)(B)(xi)(II) (which we refer to as the MFP adjustment). Section 1888(e)(5)(B)(ii) of the Act further states that the reduction of the market basket percentage by the MFP adjustment may result in the market basket percentage being less than zero for a FY, and may result in payment rates under section 1888(e) of the Act for a FY being less than such payment rates for the preceding FY. Thus, if the application of the MFP adjustment to the market basket percentage calculated under section 1888(e)(5)(B)(i) of the Act results in an MFP-adjusted market basket percentage that is less than zero, then the annual update to the unadjusted federal per diem rates under section 1888(e)(4)(E)(ii) of the Act would be negative, and such rates would decrease relative to the prior FY.

For the FY 2017 update, the MFP adjustment is calculated as the 10-year moving average of changes in MFP for the period ending September 30, 2017. In the FY 2017 SNF PPS proposed rule, this adjustment was calculated to be 0.5 percent. However, as discussed in the FY 2017 SNF PPS proposed rule (81 FR 24234), we proposed that if more recent data become available (for example, a more recent estimate of the FY 2010-based SNF market basket and/or MFP adjustment), we would use such data, if appropriate, to determine, among other things, the FY 2017 SNF market basket percentage change and the MFP adjustment in this final rule. Therefore, based on IGI's most recent second quarter 2016 forecast (with historical data through first quarter 2016), the MFP adjustment for FY 2017 is 0.3 percent. Consistent with section 1888(e)(5)(B)(i) of the Act and § 413.337(d)(2) of the regulations, the market basket percentage for FY 2017 for the SNF PPS is based on IGI's second quarter 2016 forecast of the SNF market basket update, which is estimated to be 2.7 percent, as adjusted by any applicable forecast error adjustment (as discussed above, in this final rule, we are not applying a forecast error adjustment to the SNF market basket update). In accordance with section 1888(e)(5)(B)(ii) of the Act (as added by section 3401(b) of the Affordable Care Act) and § 413.337(d)(3), this market basket percentage is then reduced by the MFP adjustment (the 10-year moving average of changes in MFP for the period ending September 30, 2017) of 0.3 percent, which is calculated as described above and based on IGI's second quarter 2016 forecast. The resulting MFP-adjusted SNF market basket update is equal to 2.4 percent, or 2.7 percent less 0.3 percentage point.

e. Market Basket Update Factor for FY 2017

Sections 1888(e)(4)(E)(ii)(IV) and 1888(e)(5)(i) of the Act require that the update factor used to establish the FY 2017 unadjusted federal rates be at a level equal to the market basket index percentage change. Accordingly, we determined the total growth from the average market basket level for the period of October 1, 2015 through September 30, 2016 to the average market basket level for the period of October 1, 2016 through September 30, 2017. This process yields a percentage change in the market basket of 2.7 percent.

As further explained in section III.B.2.c. of this final rule, as applicable, we adjust the market basket percentage change by the forecast error from the most recently available FY for which there is final data and apply this adjustment whenever the difference between the forecasted and actual percentage change in the market basket exceeds a 0.5 percentage point threshold. Since the difference between the forecasted FY 2015 SNF market basket percentage change and the actual FY 2015 SNF market basket percentage change (FY 2015 is the most recently available FY for which there is historical data) did not exceed the 0.5 percentage point threshold, the FY 2017 market basket percentage change of 2.7 percent will not be adjusted by the forecast error correction.

For FY 2017, section 1888(e)(5)(B)(ii) of the Act requires us to reduce the market basket percentage change by the MFP adjustment (the 10-year moving average of changes in MFP for the period ending September 30, 2017) of 0.3 percent, as described in section III.B.2.d. of this final rule. The resulting net SNF market basket update would equal 2.4 percent, or 2.7 percent less the 0.3 percentage point MFP adjustment. A discussion of the general comments that we received on the market basket update factor for FY 2017, and our responses to those comments, appears below.

Comment: We received a number of comments in relation to applying the FY 2017 market basket update factor in the determination of the FY 2017 unadjusted federal per diem rates, with some commenters supporting its application in determining the FY 2017 unadjusted per diem rates, while others opposed its application. In their March 2016 report (available at http://medpac.gov/​documents/​reports/​chapter-7-skilled-nursing-facility-services-(march-2016-report).pdf?​sfvrsn=​0) and in their comment on the FY 2017 SNF PPS proposed rule, MedPAC recommended that we eliminate the market basket update for SNFs altogether and implement revisions to the SNF PPS.

Response: We appreciate all of the comments received on the proposed market basket update for FY 2017. In response to those comments opposing the application of the FY 2017 market basket update factor in determining the FY 2017 unadjusted federal per diem rates, specifically MedPAC's proposal to eliminate the market basket update for SNFs, under section 1888(e)(4)(E)(ii)(IV) and (e)(5)(B) of the Act, we are required to update the unadjusted Federal per diem rates each fiscal year by the SNF market basket percentage change, as reduced by the MFP adjustment.

Comment: Several commenters recommended that the SNF market basket be reweighted more frequently. They stated that due to the rapidly changing long term care environment, SNFs have and will continue to make significant modifications to their operations, including the need to respond to alternative payment models, managed care, and emerging quality requirements. One specific recommendation was to update the SNF market basket cost weights in accordance with the hospital market basket update schedule in order to increase the accuracy of the SNF market basket—particularly if the SNF wage index continues to be directly linked to the hospital wage index.

Response: We appreciate the commenter's suggestion for a more frequent rebasing of the SNF market basket. In the past, we have rebased the SNF market basket roughly every 5 to 7 years. In accordance with section 404 of Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (MMA, Pub. L. 108-173), we determined that the frequency for rebasing the hospital market basket would be every 4 years. The SNF market basket was last rebased and revised 3 years ago in the FY 2014 SNF PPS final rule (reflecting 2010 base year expenditures), and was effective beginning in FY 2014. We will continue to review the most recent SNF Medicare cost report data and resulting market basket cost weights for any notable changes, and determine if we need to rebase the SNF market basket more frequently than roughly every 5 to 7 years. Should we determine that the SNF market basket would be improved by updating the base year, such an update would be proposed in rulemaking and be subject to public comment.

Comment: One commenter requested that we engage in an ongoing dialogue with the commenter's association on their market basket research. The goal of such discussions would be to inform us and support any analogous CMS reform efforts.

Response: We appreciate the commenter's review of the market basket and continued dialogue regarding their research. Additionally, the commenter is encouraged to submit any research to [e-mail].

Comment: One commenter identified a potential error in our calculation of the proposed FY 2017 unadjusted federal per diem rates. Specifically, the commenter stated that the FY 2017 unadjusted federal per diem rates published in the FY 2017 SNF PPS proposed rule (81 FR 24234) did not appear to reflect the full, proposed FY 2017 market basket update factor of 2.1 percent.

Response: We appreciate this comment and, after review of the calculations used to determine the FY 2017 unadjusted federal per diem rates, we have determined that there was an error in our calculation of the proposed FY 2017 unadjusted federal per diem rates. Specifically, when performing the calculation of the FY 2017 unadjusted federal per diem rates, we begin with the FY 2016 unadjusted federal per diem rates which are updated by the FY 2017 MFP-adjusted market basket update factor in accordance with section 1888(e)(4)(E)(ii)(IV) and (e)(5)(B) of the Act. However, in performing the calculation, we inadvertently made an error in transcribing the FY 2016 unadjusted federal per diem rates (though we applied the correct FY 2017 proposed market basket update factor of 2.1 percent). Specifically, for the FY 2017 SNF PPS proposed rule, we inadvertently used the following rates as the FY 2016 unadjusted urban federal per diem rates in the calculation of the proposed FY 2017 urban unadjusted federal per diem rates: $171.12 (nursing case-mix), $128.90 (therapy case-mix), $16.97 (therapy non-case-mix), and $87.33 (non-case-mix). We inadvertently used the following rates as the FY 2016 unadjusted rural federal per diem rates in the calculation of the proposed FY 2017 unadjusted rural federal per diem rates: $163.48 (nursing case-mix), $148.62 (therapy case-mix), $18.14 (therapy non-case-mix), and $88.95 (non-case-mix). The correct FY 2016 urban and rural unadjusted federal per diem rates which should have been used in this calculation, and which have been used in the calculation of the final FY 2017 urban and rural unadjusted federal per diem rates provided in Tables 2 and 3 below, are those in Tables 2 and 3 of the FY 2016 SNF PPS final rule (80 FR 46397).

Additionally, as further discussed in section III.B.4., we also discovered an error in the calculation of the proposed FY 2017 wage index budget neutrality factor, which also impacted the calculation of the proposed FY 2017 unadjusted federal per diem rates set forth in the proposed rule (81 FR 24234) (as well as the impact analysis provided in Table 19 of the FY 2017 SNF PPS proposed rule (81 FR 24278), as further discussed in section VI.A.4. of this final rule).

We appreciate the commenter bringing this error to our attention. The corrected final FY 2017 SNF PPS unadjusted federal per diem rates are set forth below in Tables 2 and 3. We further note that, as described previously in this section, the FY 2017 market basket update factor and MFP adjustment were both updated in advance of the final rule. As such, the FY 2017 unadjusted federal per diem rates provided in Tables 2 and 3 reflect the updated FY 2017 market basket increase factor and MFP adjustment, as well as the corrected FY 2016 unadjusted federal per diem rates and corrected wage index budget neutrality factor which serve as the foundation for calculating the FY 2017 unadjusted federal per diem rates.

Accordingly, for the reasons specified in this final rule and in the FY 2017 SNF PPS proposed rule (81 FR 24230), we are applying the FY 2017 market basket factor, as adjusted by the MFP adjustment as described above, in our determination of the FY 2017 unadjusted federal per diem rates. We used the SNF market basket, adjusted as described previously, to adjust each per diem component of the federal rates forward to reflect the change in the average prices for FY 2017 from average prices for FY 2016. We further adjusted the rates by a wage index budget neutrality factor, described later in this section. Tables 2 and 3 reflect the updated components of the unadjusted federal rates for FY 2017, prior to adjustment for case-mix. As discussed previously in this section, the unadjusted federal per diem rates provided below reflect the updated FY 2017 market basket update factor, as adjusted by the updated MFP adjustment, and the corrections to the FY 2016 unadjusted federal per diem rates and the FY 2017 wage index budget neutrality factor described previously.

Table 2—FY 2017 Unadjusted Federal Rate per Diem Urban

Rate component	Nursing—case-mix	Therapy—case-mix	Therapy— non-case-mix	Non-case-mix
Per Diem Amount	$175.28	$132.03	$17.39	$89.46

Table 3—FY 2017 Unadjusted Federal Rate per Diem Rural

Rate component	Nursing—case-mix	Therapy—case-mix	Therapy— non-case-mix	Non-case-mix
Per Diem Amount	$167.45	$152.24	$18.58	$91.11

3. Case-Mix Adjustment

Under section 1888(e)(4)(G)(i) of the Act, the federal rate also incorporates an adjustment to account for facility case-mix, using a classification system that accounts for the relative resource utilization of different patient types. The statute specifies that the adjustment is to reflect both a resident classification system that the Secretary establishes to account for the relative resource use of different patient types, as well as resident assessment data and other data that the Secretary considers appropriate. In the interim final rule with comment period that initially implemented the SNF PPS (63 FR 26252, May 12, 1998), we developed the RUG-III case-mix classification system, which tied the amount of payment to resident resource use in combination with resident characteristic information. Staff time measurement (STM) studies conducted in 1990, 1995, and 1997 provided information on resource use (time spent by staff members on residents) and resident characteristics that enabled us not only to establish RUG-III, but also to create case-mix indexes (CMIs). The original RUG-III grouper logic was based on clinical data collected in 1990, 1995, and 1997. As discussed in the SNF PPS proposed rule for FY 2010 (74 FR 22208), we subsequently conducted a multi-year data collection and analysis under the Staff Time and Resource Intensity Verification (STRIVE) project to update the case-mix classification system for FY 2011. The resulting Resource Utilization Groups, Version 4 (RUG-IV) case-mix classification system reflected the data collected in 2006-2007 during the STRIVE project, and was finalized in the FY 2010 SNF PPS final rule (74 FR 40288) to take effect in FY 2011 concurrently with an updated new resident assessment instrument, version 3.0 of the Minimum Data Set (MDS 3.0), which collects the clinical data used for case-mix classification under RUG-IV.

We note that case-mix classification is based, in part, on the beneficiary's need for skilled nursing care and therapy services. The case-mix classification system uses clinical data from the MDS to assign a case-mix group to each patient that is then used to calculate a per diem payment under the SNF PPS. As discussed in section IV.A. of the FY 2017 SNF PPS proposed rule (81 FR 24241 through 24242), the clinical orientation of the case-mix classification system supports the SNF PPS's use of an administrative presumption that considers a beneficiary's initial case-mix classification to assist in making certain SNF level of care determinations. Further, because the MDS is used as a basis for payment, as well as a clinical assessment, we have provided extensive training on proper coding and the time frames for MDS completion in our Resident Assessment Instrument (RAI) Manual. For an MDS to be considered valid for use in determining payment, the MDS assessment must be completed in compliance with the instructions in the RAI Manual in effect at the time the assessment is completed. For payment and quality monitoring purposes, the RAI Manual consists of both the Manual instructions and the interpretive guidance and policy clarifications posted on the appropriate MDS Web site at http://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​MDS30RAIManual.html.

In addition, we note that section 511 of the MMA, amended section 1888(e)(12) of the Act, to provide for a temporary increase of 128 percent in the PPS per diem payment for any SNF residents with Acquired Immune Deficiency Syndrome (AIDS), effective with services furnished on or after October 1, 2004. This special add-on for SNF residents with AIDS was to remain in effect until the Secretary certifies that there is an appropriate adjustment in the case mix to compensate for the increased costs associated with such residents. The add-on for SNF residents with AIDS is also discussed in Program Transmittal #160 (Change Request #3291), issued on April 30, 2004, which is available online at www.cms.gov/​transmittals/​downloads/​r160cp.pdf. In the SNF PPS final rule for FY 2010 (74 FR 40288), we did not address this certification in that final rule's implementation of the case-mix refinements for RUG-IV, thus allowing the add-on payment required by section 511 of the MMA to remain in effect. For the limited number of SNF residents that qualify for this add-on, there is a significant increase in payments. For example, using FY 2014 data (which still used ICD-9-CM coding), we identified fewer than 4,800 SNF residents with a diagnosis code of 042 (Human Immunodeficiency Virus (HIV) Infection). As explained in the FY 2016 SNF PPS final rule (80 FR 46397 through 46398), on October 1, 2015 (consistent with section 212 of PAMA), we converted to using ICD-10-CM code B20 to identify those residents for whom it is appropriate to apply the AIDS add-on established by section 511 of the MMA. For FY 2017, an urban facility with a resident with AIDS in RUG-IV group “HC2” would have a case-mix adjusted per diem payment of $438.13 (see Table 4) before the application of the MMA adjustment. After an increase of 128 percent, this urban facility would receive a case-mix adjusted per diem payment of approximately $998.94.

Under section 1888(e)(4)(H) of the Act, each update of the payment rates must include the case-mix classification methodology applicable for the upcoming FY. The payment rates set forth in this final rule reflect the use of the RUG-IV case-mix classification system from October 1, 2016, through September 30, 2017. We list the case-mix adjusted RUG-IV payment rates, provided separately for urban and rural SNFs, in Tables 4 and 5 with corresponding case-mix values. We use the revised OMB delineations adopted in the FY 2015 SNF PPS final rule (79 FR 45632, 45634) to identify a facility's urban or rural status for the purpose of determining which set of rate tables would apply to the facility. Tables 4 and 5 do not reflect the add-on for SNF residents with AIDS enacted by section 511 of the MMA, which we apply only after making all other adjustments (such as wage index and case-mix). We would note that the case mix adjusted rates provided below are based on the FY 2017 unadjusted federal per diem rates provided in Tables 2 and 3 of this section, which reflect the updated FY 2017 SNF market basket update factor and updated MFP adjustment, as well as corrections to the errors associated with the unadjusted federal per diem rates published in the FY 2017 SNF PPS proposed rule (81 FR 24234) described previously in this section.

Table 4—RUG-IV Case-Mix Adjusted Federal Rates and Associated Indexes—Urban

RUG-IV Category	Nursing index	Therapy index	Nursing component	Therapy component	Non-case-mix therapy comp	Non-case-mix component	Total rate
RUX	2.67	1.87	$468.00	$246.90		$89.46	$804.36
RUL	2.57	1.87	450.47	246.90		89.46	786.83
RVX	2.61	1.28	457.48	169.00		89.46	715.94
RVL	2.19	1.28	383.86	169.00		89.46	642.32
RHX	2.55	0.85	446.96	112.23		89.46	648.65
RHL	2.15	0.85	376.85	112.23		89.46	578.54
RMX	2.47	0.55	432.94	72.62		89.46	595.02
RML	2.19	0.55	383.86	72.62		89.46	545.94
RLX	2.26	0.28	396.13	36.97		89.46	522.56
RUC	1.56	1.87	273.44	246.90		89.46	609.80
RUB	1.56	1.87	273.44	246.90		89.46	609.80
RUA	0.99	1.87	173.53	246.90		89.46	509.89
RVC	1.51	1.28	264.67	169.00		89.46	523.13
RVB	1.11	1.28	194.56	169.00		89.46	453.02
RVA	1.10	1.28	192.81	169.00		89.46	451.27
RHC	1.45	0.85	254.16	112.23		89.46	455.85
RHB	1.19	0.85	208.58	112.23		89.46	410.27
RHA	0.91	0.85	159.50	112.23		89.46	361.19
RMC	1.36	0.55	238.38	72.62		89.46	400.46
RMB	1.22	0.55	213.84	72.62		89.46	375.92
RMA	0.84	0.55	147.24	72.62		89.46	309.32
RLB	1.50	0.28	262.92	36.97		89.46	389.35
RLA	0.71	0.28	124.45	36.97		89.46	250.88
ES3	3.58		627.50		$17.39	89.46	734.35
ES2	2.67		468.00		17.39	89.46	574.85
ES1	2.32		406.65		17.39	89.46	513.50
HE2	2.22		389.12		17.39	89.46	495.97
HE1	1.74		304.99		17.39	89.46	411.84
HD2	2.04		357.57		17.39	89.46	464.42
HD1	1.60		280.45		17.39	89.46	387.30
HC2	1.89		331.28		17.39	89.46	438.13
HC1	1.48		259.41		17.39	89.46	366.26
HB2	1.86		326.02		17.39	89.46	432.87
HB1	1.46		255.91		17.39	89.46	362.76
LE2	1.96		343.55		17.39	89.46	450.40
LE1	1.54		269.93		17.39	89.46	376.78
LD2	1.86		326.02		17.39	89.46	432.87
LD1	1.46		255.91		17.39	89.46	362.76
LC2	1.56		273.44		17.39	89.46	380.29
LC1	1.22		213.84		17.39	89.46	320.69
LB2	1.45		254.16		17.39	89.46	361.01
LB1	1.14		199.82		17.39	89.46	306.67
CE2	1.68		294.47		17.39	89.46	401.32
CE1	1.50		262.92		17.39	89.46	369.77
CD2	1.56		273.44		17.39	89.46	380.29
CD1	1.38		241.89		17.39	89.46	348.74
CC2	1.29		226.11		17.39	89.46	332.96
CC1	1.15		201.57		17.39	89.46	308.42
CB2	1.15		201.57		17.39	89.46	308.42
CB1	1.02		178.79		17.39	89.46	285.64
CA2	0.88		154.25		17.39	89.46	261.10
CA1	0.78		136.72		17.39	89.46	243.57
BB2	0.97		170.02		17.39	89.46	276.87
BB1	0.90		157.75		17.39	89.46	264.60
BA2	0.70		122.70		17.39	89.46	229.55
BA1	0.64		112.18		17.39	89.46	219.03
PE2	1.50		262.92		17.39	89.46	369.77
PE1	1.40		245.39		17.39	89.46	352.24
PD2	1.38		241.89		17.39	89.46	348.74
PD1	1.28		224.36		17.39	89.46	331.21
PC2	1.10		192.81		17.39	89.46	299.66
PC1	1.02		178.79		17.39	89.46	285.64
PB2	0.84		147.24		17.39	89.46	254.09
PB1	0.78		136.72		17.39	89.46	243.57
PA2	0.59		103.42		17.39	89.46	210.27
PA1	0.54		94.65		17.39	89.46	201.50

Table 5—RUG-IV Case-Mix Adjusted Federal Rates and Associated Indexes—Rural

RUG-IV Category	Nursing index	Therapy index	Nursing component	Therapy component	Non-case-mix therapy comp	Non-case-mix component	Total rate
RUX	2.67	1.87	$447.09	$284.69		$91.11	$822.89
RUL	2.57	1.87	430.35	284.69		91.11	806.15
RVX	2.61	1.28	437.04	194.87		91.11	723.02
RVL	2.19	1.28	366.72	194.87		91.11	652.70
RHX	2.55	0.85	427.00	129.40		91.11	647.51
RHL	2.15	0.85	360.02	129.40		91.11	580.53
RMX	2.47	0.55	413.60	83.73		91.11	588.44
RML	2.19	0.55	366.72	83.73		91.11	541.56
RLX	2.26	0.28	378.44	42.63		91.11	512.18
RUC	1.56	1.87	261.22	284.69		91.11	637.02
RUB	1.56	1.87	261.22	284.69		91.11	637.02
RUA	0.99	1.87	165.78	284.69		91.11	541.58
RVC	1.51	1.28	252.85	194.87		91.11	538.83
RVB	1.11	1.28	185.87	194.87		91.11	471.85
RVA	1.10	1.28	184.20	194.87		91.11	470.18
RHC	1.45	0.85	242.80	129.40		91.11	463.31
RHB	1.19	0.85	199.27	129.40		91.11	419.78
RHA	0.91	0.85	152.38	129.40		91.11	372.89
RMC	1.36	0.55	227.73	83.73		91.11	402.57
RMB	1.22	0.55	204.29	83.73		91.11	379.13
RMA	0.84	0.55	140.66	83.73		91.11	315.50
RLB	1.50	0.28	251.18	42.63		91.11	384.92
RLA	0.71	0.28	118.89	42.63		91.11	252.63
ES3	3.58		599.47		$18.58	91.11	709.16
ES2	2.67		447.09		18.58	91.11	556.78
ES1	2.32		388.48		18.58	91.11	498.17
HE2	2.22		371.74		18.58	91.11	481.43
HE1	1.74		291.36		18.58	91.11	401.05
HD2	2.04		341.60		18.58	91.11	451.29
HD1	1.60		267.92		18.58	91.11	377.61
HC2	1.89		316.48		18.58	91.11	426.17
HC1	1.48		247.83		18.58	91.11	357.52
HB2	1.86		311.46		18.58	91.11	421.15
HB1	1.46		244.48		18.58	91.11	354.17
LE2	1.96		328.20		18.58	91.11	437.89
LE1	1.54		257.87		18.58	91.11	367.56
LD2	1.86		311.46		18.58	91.11	421.15
LD1	1.46		244.48		18.58	91.11	354.17
LC2	1.56		261.22		18.58	91.11	370.91
LC1	1.22		204.29		18.58	91.11	313.98
LB2	1.45		242.80		18.58	91.11	352.49
LB1	1.14		190.89		18.58	91.11	300.58
CE2	1.68		281.32		18.58	91.11	391.01
CE1	1.50		251.18		18.58	91.11	360.87
CD2	1.56		261.22		18.58	91.11	370.91
CD1	1.38		231.08		18.58	91.11	340.77
CC2	1.29		216.01		18.58	91.11	325.70
CC1	1.15		192.57		18.58	91.11	302.26
CB2	1.15		192.57		18.58	91.11	302.26
CB1	1.02		170.80		18.58	91.11	280.49
CA2	0.88		147.36		18.58	91.11	257.05
CA1	0.78		130.61		18.58	91.11	240.30
BB2	0.97		162.43		18.58	91.11	272.12
BB1	0.90		150.71		18.58	91.11	260.40
BA2	0.70		117.22		18.58	91.11	226.91
BA1	0.64		107.17		18.58	91.11	216.86
PE2	1.50		251.18		18.58	91.11	360.87
PE1	1.40		234.43		18.58	91.11	344.12
PD2	1.38		231.08		18.58	91.11	340.77
PD1	1.28		214.34		18.58	91.11	324.03
PC2	1.10		184.20		18.58	91.11	293.89
PC1	1.02		170.80		18.58	91.11	280.49
PB2	0.84		140.66		18.58	91.11	250.35
PB1	0.78		130.61		18.58	91.11	240.30
PA2	0.59		98.80		18.58	91.11	208.49
PA1	0.54		90.42		18.58	91.11	200.11

4. Wage Index Adjustment

Section 1888(e)(4)(G)(ii) of the Act requires that we adjust the federal rates to account for differences in area wage levels, using a wage index that the Secretary determines appropriate. Since the inception of the SNF PPS, we have used hospital inpatient wage data in developing a wage index to be applied to SNFs. We proposed to continue this practice for FY 2017, as we continue to believe that in the absence of SNF-specific wage data, using the hospital inpatient wage index data is appropriate and reasonable for the SNF PPS. As explained in the update notice for FY 2005 (69 FR 45786), the SNF PPS does not use the hospital area wage index's occupational mix adjustment, as this adjustment serves specifically to define the occupational categories more clearly in a hospital setting; moreover, the collection of the occupational wage data also excludes any wage data related to SNFs. Therefore, we believe that using the updated wage data exclusive of the occupational mix adjustment continues to be appropriate for SNF payments. For FY 2017, the updated wage data are for hospital cost reporting periods beginning on or after October 1, 2012 and before October 1, 2013 (FY 2013 cost report data).

We note that section 315 of the Medicare, Medicaid, and SCHIP Benefits Improvement and Protection Act of 2000 (BIPA, Pub. L. 106-554, enacted on December 21, 2000) authorized us to establish a geographic reclassification procedure that is specific to SNFs, but only after collecting the data necessary to establish a SNF wage index that is based on wage data from nursing homes. However, to date, this has proven to be unfeasible due to the volatility of existing SNF wage data and the significant amount of resources that would be required to improve the quality of that data.

In addition, we proposed to continue to use the same methodology discussed in the SNF PPS final rule for FY 2008 (72 FR 43423) to address those geographic areas in which there are no hospitals, and thus, no hospital wage index data on which to base the calculation of the FY 2017 SNF PPS wage index. For rural geographic areas that do not have hospitals, and therefore, lack hospital wage data on which to base an area wage adjustment, we would use the average wage index from all contiguous Core-Based Statistical Areas (CBSAs) as a reasonable proxy. For FY 2017, there are no rural geographic areas that do not have hospitals, and thus, this methodology would not be applied. For rural Puerto Rico, we would not apply this methodology due to the distinct economic circumstances that exist there (for example, due to the close proximity to one another of almost all of Puerto Rico's various urban and non-urban areas, this methodology would produce a wage index for rural Puerto Rico that is higher than that in half of its urban areas); instead, we would continue to use the most recent wage index previously available for that area. For urban areas without specific hospital wage index data, we would use the average wage indexes of all of the urban areas within the state to serve as a reasonable proxy for the wage index of that urban CBSA. For FY 2017, the only urban area without wage index data available is CBSA 25980, Hinesville-Fort Stewart, GA. The wage index applicable to FY 2017 is set forth in Tables A and B available on the CMS Web site at http://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​WageIndex.html.

Once calculated, we would apply the wage index adjustment to the labor-related portion of the federal rate. Each year, we calculate a revised labor-related share, based on the relative importance of labor-related cost categories (that is, those cost categories that are labor-intensive and vary with the local labor market) in the input price index. In the SNF PPS final rule for FY 2014 (78 FR 47944 through 47946), we finalized a proposal to revise the labor-related share to reflect the relative importance of the FY 2010-based SNF market basket cost weights for the following cost categories: Wages and salaries; employee benefits; the labor-related portion of nonmedical professional fees; administrative and facilities support services; all other: Labor-related services; and a proportion of capital-related expenses.

We calculate the labor-related relative importance from the SNF market basket, and it approximates the labor-related portion of the total costs, after taking into account historical and projected price changes between the base year and FY 2017. The price proxies that move the different cost categories in the market basket do not necessarily change at the same rate, and the relative importance captures these changes. Accordingly, the relative importance figure more closely reflects the cost share weights for FY 2017 than the base year weights from the SNF market basket.

We calculate the labor-related relative importance for FY 2017 in four steps. First, we compute the FY 2017 price index level for the total market basket and each cost category of the market basket. Second, we calculate a ratio for each cost category by dividing the FY 2017 price index level for that cost category by the total market basket price index level. Third, we determine the FY 2017 relative importance for each cost category by multiplying this ratio by the base year (FY 2010) weight. Finally, we add the FY 2017 relative importance for each of the labor-related cost categories (wages and salaries, employee benefits, the labor-related portion of non-medical professional fees, administrative and facilities support services, all other: Labor-related services, and a portion of capital-related expenses) to produce the FY 2017 labor-related relative importance. Table 6 summarizes the updated labor-related share for FY 2017, compared to the labor-related share that was used for the FY 2016 SNF PPS final rule. In the FY 2017 SNF PPS proposed rule, the labor-related share for FY 2017 was proposed to be 68.9 percent. However, as discussed in the FY 2017 SNF PPS proposed rule (81 FR 24234), we proposed that if more recent data become available, we would use such data, if appropriate, to determine, among other things, the FY 2017 SNF labor related share. Therefore, based on IGI's most recent second quarter 2016 forecast (with historical data through first quarter 2016), the labor-related share for FY 2017 is 68.8 percent.

We invited public comments on these proposals. A discussion of the comments we received on these proposals, as well as a discussion of the general comments we received on the wage index adjustment, and our responses to those comments, appears below.

Comment: One commenter is concerned with the significant drop in the wage index for Great Falls, Montana (CBSA 24500). The commenter mentioned that Montana is a frontier state as defined in the Affordable Care Act and that the Affordable Care Act, specifically section 10324 of the Affordable Care Act, establishes a wage index floor of 1.0 for frontier state hospitals. The commenter recommends that CMS use its authority to apply the ACA-mandated frontier floor for hospitals to SNFs.

Response: We appreciate the commenter's concern regarding the application of a floor on area wage indexes for SNFs in frontier states. Section 10324 of the Affordable Care Act requires that hospitals in frontier states cannot be assigned a wage index of less than 1.0000. We do not believe it would be prudent at this time to adopt such a policy under the SNF PPS. As we stated in the FY 2016 SNF PPS final rule (80 FR 46401), MedPAC has recommended eliminating the rural floor policy (which actually sets a floor for urban hospitals) from the calculation of the IPPS wage index (see, for example, Chapter 3 of MedPAC's March 2013 Report to Congress on Medicare Payment Policy, available at http://medpac.gov/​documents/​reports/​mar13_​entirereport.pdf, which notes on page 65 that in 2007, MedPAC had “. . . recommended eliminating these special wage index adjustments and adopting a new wage index system to avoid geographic inequities that can occur due to current wage index policies (Medicare Payment Advisory Commission 2007b.”) We stated in the FY 2016 SNF PPS final rule that if we adopted the rural floor at that time under the SNF PPS, we believed that the SNF PPS wage index could become vulnerable to problems similar to those that MedPAC identified in its March 2013 Report to Congress. Similarly, we have concerns regarding adopting a frontier state floor at this time under the SNF PPS as we are concerned that the frontier state floor could produce vulnerabilities for the SNF PPS wage index similar to those discussed by MedPAC in its report. As stated above, under section 1888(e)(4)(G)(ii) of the Act and § 413.337(a)(1)(ii) of the regulations, we adjust the SNF PPS rates to account for differences in area wage levels. We believe that applying a floor to those facilities located in frontier states would make the wage index for those areas less reflective of the area wage levels.

Comment: Several commenters recommend that we continue exploring potential approaches for collecting SNF-specific wage data to establish a SNF-specific wage index. These commenters stated that the hospital wage index does not provide a reasonable proxy for SNF wages and occupational mix and should be replaced by use of SNF-specific data as soon as is practicable. One commenter recommended that we consider collecting base-hourly wage data as part of the Payroll-Based Journal (PBJ) initiative, which may be used in developing a SNF-specific wage index.

Response: We appreciate the commenters raising these concerns regarding the use of the hospital wage index data under the SNF PPS, and the commenter's recommendation to continue exploring potential approaches for collecting SNF-specific wage data to establish a SNF-specific wage index. However, we note that, consistent with our previous responses to these recurring comments (most recently published in the FY 2016 SNF PPS final rule (80 FR 46401)), developing such a wage index would require a resource-intensive audit process similar to that used for IPPS hospital data, to improve the quality of the SNF cost report data, in order for it to be used as part of this analysis. We would further note that, as this audit process is quite extensive in the case of approximately 3,300 hospitals, it would be significantly more so in the case of approximately 15,000 SNFs. Therefore, while we continue to review all available data and contemplate the potential methodological approaches for a SNF-specific wage index in the future, we continue to believe that in the absence of the appropriate SNF-specific wage data, using the pre-reclassified hospital inpatient wage data (without the occupational mix adjustment) is appropriate and reasonable for the SNF PPS. With regard to the PBJ recommendation, we will pass this comment to our colleagues managing that initiative for further consideration.

Comment: A few commenters suggested that we modify the use of hospital wage data used to construct the SNF PPS wage index, specifically calling for us to remove certain labor categories and data that are specific to hospitals only. These commenters also suggested that this modified methodology could further be tailored to SNFs by weighting it by occupational mix data for SNFs published by the Bureau of Labor Statistics (BLS).

Response: We appreciate these commenters' suggestion that we modify the current hospital wage data used to construct the SNF PPS wage index to reflect the SNF environment more accurately. While we consider whether or not such an approach may constitute an interim step in the process of developing a SNF-specific wage index, we would note that other provider types also use the hospital wage index as the basis for their associated wage index. As such, we believe that such a recommendation should be part of a broader discussion of wage index reform across Medicare payment systems.

Comment: A few commenters raised concerns around evolving minimum wage standards across the country and recommended that we consider ways to incorporate increasing minimum wage standards into the SNF PPS wage index. One commenter recommended that we should modify the wage index adjustment in the future to identify “living wages” across the country and that wage index policies should ensure that facilities pay their staff such a living wage. This commenter also recommended that we reward facilities that invest in their workforce.

Response: With regard to rising minimum wage standards, we would note that such increases would likely be reflected in future data used to create the hospital wage index, to the extent these changes to state minimum wage standards are reflected in increased wages to hospital staff. Therefore, such standards would already be incorporated into the calculation of the SNF PPS wage index to the extent that these standards impact on facility wages. With regard to the comment that we should modify the wage index adjustment to identify and support facilities that pay a living wage to their staff, the purpose of the wage index adjustment is to reflect the actual wages being paid to staff, not to influence the wages being paid to staff. Therefore, we do not believe that we should make modifications to the wage index to reflect an ideal standard of wages that does not currently exist.

Accordingly, after considering the comments received and for the reasons discussed previously in this section and in the FY 2017 SNF PPS proposed rule (81 FR 24237 through 24241), we are finalizing the FY 2017 wage index adjustment and related policies as proposed in the FY 2017 SNF PPS proposed rule. For FY 2017, the updated wage data are for hospital cost reporting periods beginning on or after October 1, 2012 and before October 1, 2013 (FY 2013 cost report data). Table 6 summarizes the updated labor-related share for FY 2017, compared to the labor-related share that was used in the FY 2016 SNF PPS final rule.

Table 6—Labor-Related Relative Importance, FY 2016 and FY 2017

	Relative importance, labor-related, FY 2016 15:2 forecast 1	Relative importance, labor-related, FY 2017 16:2 forecast 2
Wages and salaries	48.8	48.8
Employee benefits	11.3	11.1
Nonmedical Professional fees: Labor-related	3.5	3.4
Administrative and facilities support services	0.5	0.5
All Other: Labor-related services	2.3	2.3
Capital-related (.391)	2.7	2.7
Total	69.1	68.8
1 Published in the Federal Register; based on second quarter 2015 IGI forecast.
2 Based on second quarter 2016 IGI forecast, with historical data through first quarter 2016.

Tables 7 and 8 show the RUG-IV case-mix adjusted federal rates by labor-related and non-labor-related components.

Table 7—RUG-IV Case-Mix Adjusted Federal Rates for Urban SNFs By Labor and Non-Labor Component

RUG-IV category	Total rate	Labor portion	Non-labor portion
RUX	804.36	$553.40	$250.96
RUL	786.83	541.34	245.49
RVX	715.94	492.57	223.37
RVL	642.32	441.92	200.40
RHX	648.65	446.27	202.38
RHL	578.54	398.04	180.50
RMX	595.02	409.37	185.65
RML	545.94	375.61	170.33
RLX	522.56	359.52	163.04
RUC	609.80	419.54	190.26
RUB	609.80	419.54	190.26
RUA	509.89	350.80	159.09
RVC	523.13	359.91	163.22
RVB	453.02	311.68	141.34
RVA	451.27	310.47	140.80
RHC	455.85	313.62	142.23
RHB	410.27	282.27	128.00
RHA	361.19	248.50	112.69
RMC	400.46	275.52	124.94
RMB	375.92	258.63	117.29
RMA	309.32	212.81	96.51
RLB	389.35	267.87	121.48
RLA	250.88	172.61	78.27
ES3	734.35	505.23	229.12
ES2	574.85	395.50	179.35
ES1	513.50	353.29	160.21
HE2	495.97	341.23	154.74
HE1	411.84	283.35	128.49
HD2	464.42	319.52	144.90
HD1	387.30	266.46	120.84
HC2	438.13	301.43	136.70
HC1	366.26	251.99	114.27
HB2	432.87	297.81	135.06
HB1	362.76	249.58	113.18
LE2	450.40	309.88	140.52
LE1	376.78	259.22	117.56
LD2	432.87	297.81	135.06
LD1	362.76	249.58	113.18
LC2	380.29	261.64	118.65
LC1	320.69	220.63	100.06
LB2	361.01	248.37	112.64
LB1	306.67	210.99	95.68
CE2	401.32	276.11	125.21
CE1	369.77	254.40	115.37
CD2	380.29	261.64	118.65
CD1	348.74	239.93	108.81
CC2	332.96	229.08	103.88
CC1	308.42	212.19	96.23
CB2	308.42	212.19	96.23
CB1	285.64	196.52	89.12
CA2	261.10	179.64	81.46
CA1	243.57	167.58	75.99
BB2	276.87	190.49	86.38
BB1	264.60	182.04	82.56
BA2	229.55	157.93	71.62
BA1	219.03	150.69	68.34
PE2	369.77	254.40	115.37
PE1	352.24	242.34	109.90
PD2	348.74	239.93	108.81
PD1	331.21	227.87	103.34
PC2	299.66	206.17	93.49
PC1	285.64	196.52	89.12
PB2	254.09	174.81	79.28
PB1	243.57	167.58	75.99
PA2	210.27	144.67	65.60
PA1	201.50	138.63	62.87

Table 8—RUG-IV Case-Mix Adjusted Federal Rates for Rural SNFs by Labor and Non-Labor Component

RUG-IV Category	Total rate	Labor portion	Non-labor portion
RUX	822.89	$566.15	$256.74
RUL	806.15	554.63	251.52
RVX	723.02	497.44	225.58
RVL	652.70	449.06	203.64
RHX	647.51	445.49	202.02
RHL	580.53	399.40	181.13
RMX	588.44	404.85	183.59
RML	541.56	372.59	168.97
RLX	512.18	352.38	159.80
RUC	637.02	438.27	198.75
RUB	637.02	438.27	198.75
RUA	541.58	372.61	168.97
RVC	538.83	370.72	168.11
RVB	471.85	324.63	147.22
RVA	470.18	323.48	146.70
RHC	463.31	318.76	144.55
RHB	419.78	288.81	130.97
RHA	372.89	256.55	116.34
RMC	402.57	276.97	125.60
RMB	379.13	260.84	118.29
RMA	315.50	217.06	98.44
RLB	384.92	264.82	120.10
RLA	252.63	173.81	78.82
ES3	709.16	487.90	221.26
ES2	556.78	383.06	173.72
ES1	498.17	342.74	155.43
HE2	481.43	331.22	150.21
HE1	401.05	275.92	125.13
HD2	451.29	310.49	140.80
HD1	377.61	259.80	117.81
HC2	426.17	293.20	132.97
HC1	357.52	245.97	111.55
HB2	421.15	289.75	131.40
HB1	354.17	243.67	110.50
LE2	437.89	301.27	136.62
LE1	367.56	252.88	114.68
LD2	421.15	289.75	131.40
LD1	354.17	243.67	110.50
LC2	370.91	255.19	115.72
LC1	313.98	216.02	97.96
LB2	352.49	242.51	109.98
LB1	300.58	206.80	93.78
CE2	391.01	269.01	122.00
CE1	360.87	248.28	112.59
CD2	370.91	255.19	115.72
CD1	340.77	234.45	106.32
CC2	325.70	224.08	101.62
CC1	302.26	207.95	94.31
CB2	302.26	207.95	94.31
CB1	280.49	192.98	87.51
CA2	257.05	176.85	80.20
CA1	240.30	165.33	74.97
BB2	272.12	187.22	84.90
BB1	260.40	179.16	81.24
BA2	226.91	156.11	70.80
BA1	216.86	149.20	67.66
PE2	360.87	248.28	112.59
PE1	344.12	236.75	107.37
PD2	340.77	234.45	106.32
PD1	324.03	222.93	101.10
PC2	293.89	202.20	91.69
PC1	280.49	192.98	87.51
PB2	250.35	172.24	78.11
PB1	240.30	165.33	74.97
PA2	208.49	143.44	65.05
PA1	200.11	137.68	62.43

Section 1888(e)(4)(G)(ii) of the Act also requires that we apply this wage index in a manner that does not result in aggregate payments under the SNF PPS that are greater or less than would otherwise be made if the wage adjustment had not been made. For FY 2017 (federal rates effective October 1, 2016), we will apply an adjustment to fulfill the budget neutrality requirement. We meet this requirement by multiplying each of the components of the unadjusted federal rates by a budget neutrality factor equal to the ratio of the weighted average wage adjustment factor for FY 2016 to the weighted average wage adjustment factor for FY 2017. For this calculation, we use the same FY 2015 claims utilization data for both the numerator and denominator of this ratio. We define the wage adjustment factor used in this calculation as the labor share of the rate component multiplied by the wage index plus the non-labor share of the rate component. The budget neutrality factor stated in the FY 2017 SNF PPS proposed rule was 1.0000. However, we discovered that in calculating the FY 2017 proposed wage index budget neutrality factor, we inadvertently failed to update the wage index data used in the calculation with the most recently available FY 2017 data. This resulted in a budget neutrality factor of 1.000, whereas, using the most recently available wage index data at the time of the proposed rule, the proposed factor should have been 0.9997. Moreover, because the wage index data used were incorrect and because the wage index is the primary source of variation in the impacts calculated in the regulatory impact analysis, the error which caused the incorrect calculation of the wage index budget neutrality factor in the proposed rule also affected the wage index impacts in Table 19 of the FY 2017 SNF PPS proposed rule (Projected Impact to the SNF PPS for FY 2017) (81 FR 24278). These impacts are discussed further in section V.A.4. of this final rule. We have recalculated the wage index budget neutrality factor for FY 2017 utilizing updated wage index data, and the final budget neutrality factor for FY 2017 is 1.0000.

In the SNF PPS final rule for FY 2006 (70 FR 45026, August 4, 2005), we adopted the changes discussed in the OMB Bulletin No. 03-04 (June 6, 2003), available online at www.whitehouse.gov/​omb/​bulletins/​b03-04.html,, which announced revised definitions for MSAs and the creation of micropolitan statistical areas and combined statistical areas.

In adopting the CBSA geographic designations, we provided for a 1-year transition in FY 2006 with a blended wage index for all providers. For FY 2006, the wage index for each provider consisted of a blend of 50 percent of the FY 2006 MSA-based wage index and 50 percent of the FY 2006 CBSA-based wage index (both using FY 2002 hospital data). We referred to the blended wage index as the FY 2006 SNF PPS transition wage index. As discussed in the SNF PPS final rule for FY 2006 (70 FR 45041), since the expiration of this 1-year transition on September 30, 2006, we have used the full CBSA-based wage index values.

Generally, OMB issues major revisions to statistical areas every 10 years, based on the results of the decennial census. In the FY 2015 SNF PPS final rule (79 FR 45644 through 45646), we finalized changes to the SNF PPS wage index based on the newest OMB delineations, as described in OMB Bulletin No. 13-01, beginning in FY 2015, including a 1-year transition with a blended wage index for FY 2015. OMB Bulletin No. 13-01 established revised delineations for Metropolitan Statistical Areas, Micropolitan Statistical Areas, and Combined Statistical Areas in the United States and Puerto Rico based on the 2010 Census, and provided guidance on the use of the delineations of these statistical areas using standards published in the June 28, 2010 Federal Register (75 FR 37246 through 37252). In addition, OMB occasionally issues minor updates and revisions to statistical areas in the years between the decennial censuses. On July 15, 2015, OMB issued OMB Bulletin No. 15-01, which provides minor updates to and supersedes OMB Bulletin No. 13-01 that was issued on February 28, 2013. The attachment to OMB Bulletin No. 15-01 provides detailed information on the update to statistical areas since February 28, 2013. The updates provided in OMB Bulletin No. 15-01 are based on the application of the 2010 Standards for Delineating Metropolitan and Micropolitan Statistical Areas to Census Bureau population estimates for July 1, 2012 and July 1, 2013. A copy of this bulletin may be obtained on the Web site at https://www.whitehouse.gov/​sites/​default/​files/​omb/​bulletins/​2015/​15-01.pdf. As we previously stated in the FY 2008 SNF PPS proposed and final rules (72 FR 25538 through 25539, and 72 FR 43423), we again wish to clarify that this and all subsequent SNF PPS rules and notices are considered to incorporate any such updates and revisions set forth in the most recent OMB bulletin that applies to the hospital wage data used to determine the current SNF PPS wage index. As noted previously in this section, the wage index applicable to FY 2017 is set forth in Tables A and B available on the CMS Web site at http://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​WageIndex.html.

5. Adjusted Rate Computation Example

Using the hypothetical SNF XYZ described below, Table 9 shows the adjustments made to the federal per diem rates to compute the provider's actual per diem PPS payment. We derive the Labor and Non-labor columns from Table 7. The wage index used in this example is based on the final wage index, which may be found in Table A as referenced previously in this section. As illustrated in Table 9, SNF XYZ's total PPS payment would equal $46,861.86.

Chart 9—Adjusted Rate Computation Example

SNF XYZ: Located in Frederick, MD (Urban CBSA 43524)

Wage Index: 0.9797

[See Wage Index in Table A] ¹

RUG-IV group	Labor	Wage index	Adjusted labor	Non-labor	Adjusted rate	Percent adjustment	Medicare days	Payment
RVX	$492.57	0.9797	$482.57	$223.37	$705.94	$705.94	14	$9,883.16
ES2	395.50	0.9797	387.47	179.35	566.82	566.82	30	17,004.60
RHA	248.50	0.9797	243.46	112.69	356.15	356.15	16	5,698.40
CC2 *	229.08	0.9797	224.43	103.88	328.31	748.55	10	7,485.50
BA2	157.93	0.9797	154.72	71.62	226.34	226.34	30	6,790.20
							100	46,861.86
* Reflects a 128 percent adjustment from section 511 of the MMA.
1 Available on the CMS Web site at http://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​WageIndex.html.

C. Additional Aspects of the SNF PPS

1. SNF Level of Care—Administrative Presumption

The establishment of the SNF PPS did not change Medicare's fundamental requirements for SNF coverage. However, because the case-mix classification is based, in part, on the beneficiary's need for skilled nursing care and therapy, we have attempted, where possible, to coordinate claims review procedures with the existing resident assessment process and case-mix classification system discussed in section III.B.3. of this final rule. This approach includes an administrative presumption that utilizes a beneficiary's initial classification in one of the upper 52 RUGs of the 66-group RUG-IV case-mix classification system to assist in making certain SNF level of care determinations.

In accordance with section 1888(e)(4)(H)(ii) of the Act and the regulations at § 413.345, we include in each update of the federal payment rates in the Federal Register the designation of those specific RUGs under the classification system that represent the required SNF level of care, as provided in § 409.30. As set forth in the FY 2011 SNF PPS update notice (75 FR 42910), this designation reflects an administrative presumption under the 66-group RUG-IV system that beneficiaries who are correctly assigned to one of the upper 52 RUG-IV groups on the initial 5-day, Medicare-required assessment are automatically classified as meeting the SNF level of care definition up to and including the assessment reference date (ARD) on the 5-day Medicare-required assessment.

A beneficiary assigned to any of the lower 14 RUG-IV groups is not automatically classified as either meeting or not meeting the definition, but instead receives an individual level of care determination using the existing administrative criteria. This presumption recognizes the strong likelihood that beneficiaries assigned to one of the upper 52 RUG-IV groups during the immediate post-hospital period require a covered level of care, which would be less likely for those beneficiaries assigned to one of the lower 14 RUG-IV groups.

In the July 30, 1999 final rule (64 FR 41670), we indicated that we would announce any changes to the guidelines for Medicare level of care determinations related to modifications in the case-mix classification structure. In this final rule, we continue to designate the upper 52 RUG-IV groups for purposes of this administrative presumption, consisting of all groups encompassed by the following RUG-IV categories:

• Rehabilitation plus Extensive Services.
• Ultra High Rehabilitation.
• Very High Rehabilitation.
• High Rehabilitation.
• Medium Rehabilitation.
• Low Rehabilitation.
• Extensive Services.
• Special Care High.
• Special Care Low.
• Clinically Complex.

However, we note that this administrative presumption policy does not supersede the SNF's responsibility to ensure that its decisions relating to level of care are appropriate and timely, including a review to confirm that the services prompting the beneficiary's assignment to one of the upper 52 RUG-IV groups (which, in turn, serves to trigger the administrative presumption) are themselves medically necessary. As we explained in the FY 2000 SNF PPS final rule (64 FR 41667), the administrative presumption:

. . . is itself rebuttable in those individual cases in which the services actually received by the resident do not meet the basic statutory criterion of being reasonable and necessary to diagnose or treat a beneficiary's condition (according to section 1862(a)(1) of the Act). Accordingly, the presumption would not apply, for example, in those situations in which a resident's assignment to one of the upper . . . groups is itself based on the receipt of services that are subsequently determined to be not reasonable and necessary.

Moreover, we want to stress the importance of careful monitoring for changes in each patient's condition to determine the continuing need for Part A SNF benefits after the ARD of the 5-day assessment.

2. Consolidated Billing

Sections 1842(b)(6)(E) and 1862(a)(18) of the Act (as added by section 4432(b) of the BBA) require a SNF to submit consolidated Medicare bills to its Medicare Administrative Contractor for almost all of the services that its residents receive during the course of a covered Part A stay. In addition, section 1862(a)(18) of the Act places the responsibility with the SNF for billing Medicare for physical therapy, occupational therapy, and speech-language pathology services that the resident receives during a noncovered stay. Section 1888(e)(2)(A) of the Act excludes a small list of services from the consolidated billing provision (primarily those services furnished by physicians and certain other types of practitioners), which remain separately billable under Part B when furnished to a SNF's Part A resident. These excluded service categories are discussed in greater detail in section V.B.2. of the May 12, 1998 interim final rule (63 FR 26295 through 26297).

A detailed discussion of the legislative history of the consolidated billing provision is available on the SNF PPS Web site at http://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​Downloads/​Legislative_​History_​07302013.pdf. In particular, section 103 of the Medicare, Medicaid, and SCHIP Balanced Budget Refinement Act of 1999 (BBRA) (Pub. L. 106-113, enacted on November 29, 1999) amended section 1888(e)(2)(A) of the Act by further excluding a number of individual high-cost, low probability services, identified by Healthcare Common Procedure Coding System (HCPCS) codes, within several broader categories (chemotherapy items, chemotherapy administration services, radioisotope services, and customized prosthetic devices) that otherwise remained subject to the provision. We discuss this BBRA amendment in greater detail in the SNF PPS proposed and final rules for FY 2001 (65 FR 19231 through 19232, April 10, 2000, and 65 FR 46790 through 46795, July 31, 2000), as well as in Program Memorandum AB-00-18 (Change Request #1070), issued March 2000, which is available online at www.cms.gov/​transmittals/​downloads/​ab001860.pdf.

As explained in the FY 2001 proposed rule (65 FR 19232), the amendments enacted in section 103 of the BBRA not only identified for exclusion from this provision a number of particular service codes within four specified categories (that is, chemotherapy items, chemotherapy administration services, radioisotope services, and customized prosthetic devices), but also gave the Secretary the authority to designate additional, individual services for exclusion within each of the specified service categories. In the proposed rule for FY 2001, we also noted that the BBRA Conference report (H.R. Rep. No. 106-479 at 854 (1999) (Conf. Rep.)) characterizes the individual services that this legislation targets for exclusion as high-cost, low probability events that could have devastating financial impacts because their costs far exceed the payment SNFs receive under the PPS. According to the conferees, section 103(a) of the BBRA is an attempt to exclude from the PPS certain services and costly items that are provided infrequently in SNFs. By contrast, we noted that the Congress declined to designate for exclusion any of the remaining services within those four categories (thus, leaving all of those services subject to SNF consolidated billing), because they are relatively inexpensive and are furnished routinely in SNFs.

As we further explained in the final rule for FY 2001 (65 FR 46790), and according to our longstanding policy, any additional service codes that we might designate for exclusion under our discretionary authority must meet the same statutory criteria used in identifying the original codes excluded from consolidated billing under section 103(a) of the BBRA: They must fall within one of the four service categories specified in the BBRA; and they also must meet the same standards of high cost and low probability in the SNF setting, as discussed in the BBRA Conference report. Accordingly, we characterized this statutory authority to identify additional service codes for exclusion as essentially affording the flexibility to revise the list of excluded codes in response to changes of major significance that may occur over time (for example, the development of new medical technologies or other advances in the state of medical practice) (65 FR 46791). In the FY 2017 SNF PPS proposed rule (81 FR 24242), we specifically invited public comments identifying HCPCS codes in any of these four service categories (chemotherapy items, chemotherapy administration services, radioisotope services, and customized prosthetic devices) representing recent medical advances that might meet our criteria for exclusion from SNF consolidated billing. We stated that we may consider excluding a particular service if it meets our criteria for exclusion as specified above. We also asked that commenters identify in their comments the specific HCPCS code that is associated with the service in question, as well as their rationale for requesting that the identified HCPCS code(s) be excluded.

Commenters submitted the following comments related to the proposed rule's discussion of the consolidated billing aspects of the SNF PPS. A discussion of these comments, along with our responses, appears below.

Comment: One commenter suggested excluding all high-cost oral chemotherapy drugs from consolidated billing, and proposed a threshold of $50 or more per tablet to define “high-cost” for this purpose. Another commenter specifically recommended for exclusion the oral chemotherapy drug Revlimid® (lenalidomide). Still another suggested that we conduct an analysis with a view toward excluding a broader range of expensive drugs beyond the category of chemotherapy alone, citing anecdotal evidence that leaving such drugs within the SNF PPS bundle may create a disincentive for admitting those patients who require them.

Response: When the Congress carved out certain exceptionally intensive chemotherapy drugs from the SNF PPS bundle in section 103 of the BBRA, it characterized those drugs as “high-cost” and “low probability.” This legislation did not categorically exclude all high-cost oral chemotherapy drugs from SNF consolidated billing. The accompanying Conference Report explained that this provision

. . . is an attempt to exclude from the PPS certain services and costly items that are provided infrequently in SNFs. For example, in the case of chemotherapy drugs, [this provision has] excluded specific chemotherapy drugs from the PPS because these drugs are not typically administered in a SNF, or are exceptionally expensive, or are given as infusions, thus requiring special staff expertise to administer. Some chemotherapy drugs, which are relatively inexpensive and are administered routinely in SNFs, were excluded from this provision” (H. Conf. Rep. No. 106-479 at 854) (emphasis added).

Accordingly, we decline to exclude all high-cost oral chemotherapy drugs as a class from consolidated billing, because any such drugs that are capable of being “administered routinely in SNFs” are not reasonably characterized as “requiring special staff expertise to administer.” We note that in the SNF PPS final rules for FYs 2009 (73 FR 46436, August 8, 2008) and 2010 (74 FR 40353, August 11, 2009), we declined to exclude certain oral medications suggested by commenters for the same reason. In addition, the BBRA Conference Report language (H. Conf. Rep. No. 106-479 at 854) further indicates that the term “high-cost” in this context would not serve to encompass a routinely-used chemotherapy drug merely because its cost somewhat exceeds the typical range of drug costs encountered in this setting; rather, this provision is directed specifically at those uncommon chemotherapy drugs that are so exceptionally expensive as to “. . . have devastating financial impacts because their costs far exceed the payment [SNFs] receive under the prospective payment system” (emphasis added). With specific reference to Revlimid®, we note that we already received a similar exclusion recommendation during the public comment period on the FY 2015 SNF PPS proposed rule, and we discussed our decision not to exclude this particular drug in that year's final rule (79 FR 45641 through 45642, August 5, 2014). Finally, in response to the suggestion that we exclude a broader range of expensive drugs beyond the category of chemotherapy alone, as we have noted repeatedly in previous rulemaking—most recently, in the FY 2016 SNF PPS final rule (80 FR 46406, August 4, 2015)—the statutory authority to designate additional services for exclusion applies solely to the four service categories (chemotherapy items, chemotherapy administration services, radioisotope services, and customized prosthetic devices) that are specified in the law. Accordingly, expanding the existing exclusion authority to encompass additional categories (such as non-chemotherapy drugs) is not provided for in current law.

Comment: Several commenters noted the importance of continuing to exclude prosthetic devices from consolidated billing. They suggested that the following four HCPCS codes should be added to the list of codes excluded from consolidated billing: L5010—Partial foot, molded socket, ankle height, with toe filler; L5020—Partial foot, molded socket, tibial tubercle height, with toe filler; L5969—Addition, endoskeletal ankle-foot or ankle system, power assist, includes any type motor(s); and L5987—All lower extremity prosthesis, shank foot system with vertical loading pylon. Some also advocated excluding custom orthotics from consolidated billing as well. They stated that the custom orthotic and prosthetic professions are closely aligned, with a sizable percentage of patients who require prosthetic care also requiring custom orthotics to address orthopedic impairments of the arms, legs, spine, and neck. They further suggested that the same factors that justify exempting prosthetic devices also apply to custom orthotics, as custom orthotics are typically a high-cost, low frequency service for patients in SNFs.

Response: The recommendation to exclude certain particular prosthetics essentially reiterates a comment made during last year's SNF PPS rulemaking cycle, which recommended for exclusion certain prosthetic device codes that were already in existence—but not excluded—upon the original 1999 enactment of the customized prosthetic device exclusion in the BBRA. In response, we reiterated in the FY 2016 SNF PPS final rule our longstanding position that if a particular prosthetic code was already in existence as of the BBRA enactment date but was not designated in the BBRA for exclusion, this meant that it was intended to remain within the SNF PPS bundle, subject to a GAO review that was conducted the following year (80 FR 46407, August 4, 2015). This would apply to three of the prosthetic codes (L5010, L5020, and L5987) cited in the current comments. Regarding the fourth prosthetic code (L5969), we also noted in last year's final rule (80 FR 46407) that code L5969 actually appears already on the exclusion list under Major Category III.D. (“Customized Prosthetic Devices”), where this particular L code has, in fact, been listed ever since its initial assignment in January 2014.

With reference to orthotics, in the FY 2016 SNF PPS final rule (80 FR 46407, August 4, 2015), we explained that while the law does specify customized prosthetic devices as one of the exclusion categories, this is a separate and distinct category from orthotics and does not encompass orthotics. Moreover, as already noted in this and previous final rules, the statutory authority to designate additional services for exclusion applies solely to the four service categories (chemotherapy items, chemotherapy administration services, radioisotope services, and customized prosthetic devices) that are specified in the law. Accordingly, expanding the existing exclusion authority to encompass additional categories (such as orthotics) is not provided for in current law.

3. Payment for SNF-Level Swing-Bed Services

Section 1883 of the Act permits certain small, rural hospitals to enter into a Medicare swing-bed agreement, under which the hospital can use its beds to provide either acute- or SNF-level care, as needed. For critical access hospitals (CAHs), Part A pays on a reasonable cost basis for SNF-level services furnished under a swing-bed agreement. However, in accordance with section 1888(e)(7) of the Act, these services furnished by non-CAH rural hospitals are paid under the SNF PPS, effective with cost reporting periods beginning on or after July 1, 2002. As explained in the FY 2002 final rule (66 FR 39562), this effective date is consistent with the statutory provision to integrate swing-bed rural hospitals into the SNF PPS by the end of the transition period, June 30, 2002.

Accordingly, all non-CAH swing-bed rural hospitals have now come under the SNF PPS. Therefore, all rates and wage indexes outlined in earlier sections of this proposed rule for the SNF PPS also apply to all non-CAH swing-bed rural hospitals. A complete discussion of assessment schedules, the MDS, and the transmission software (RAVEN-SB for Swing Beds) appears in the FY 2002 final rule (66 FR 39562) and in the FY 2010 final rule (74 FR 40288). As finalized in the FY 2010 SNF PPS final rule (74 FR 40356 through 40357), effective October 1, 2010, non-CAH swing-bed rural hospitals are required to complete an MDS 3.0 swing-bed assessment which is limited to the required demographic, payment, and quality items. The latest changes in the MDS for swing-bed rural hospitals appear on the SNF PPS Web site at http://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​index.html.

D. Other Issues

1. Skilled Nursing Facility Value-Based Purchasing Program (SNF VBP)

a. Background

Section 215 of the Protecting Access to Medicare Act of 2014 (PAMA) authorizes the SNF VBP Program by adding sections 1888(g) and (h) to the Act. These sections provide structure for the development of the SNF VBP Program, including, among other things, the requirement of only two measures—an all-cause, all-condition hospital readmission measure, which is to be replaced as soon as practicable by an all-condition risk-adjusted potentially preventable hospital readmission measure—and confidential and public reporting requirements for the SNF VBP Program. We began development of the SNF VBP Program in the FY 2016 SNF PPS final rule with, among other things, the adoption of an all-cause, all-condition hospital readmission measure, as required under section 1888(g)(1) of the Act. We will continue the process in this final rule with our adoption of an all-condition risk-adjusted potentially preventable hospital readmission measure for SNFs, which the Secretary is required to specify no later than October 1, 2016 under section 1888(g)(2) of the Act. The Act requires that the SNF VBP apply to payments for services furnished on or after October 1, 2018. The SNF VBP Program applies to freestanding SNFs, SNFs affiliated with acute care facilities, and all non-CAH swing-bed rural hospitals. We believe the implementation of the SNF VBP Program is an important step toward transforming how care is paid for, moving increasingly toward rewarding better value, outcomes, and innovations instead of merely volume.

For additional background information on the SNF VBP Program, including an overview of the SNF VBP Report to Congress and a summary of the Program's statutory requirements, we refer readers to the FY 2016 SNF PPS final rule (80 FR 46409 through 46410).

We received a number of general comments on the Program.

Comment: Some commenters urged us to broaden the SNF VBP Program to include other post-acute care outcome measures, such as measures of care transitions, resource use over care episodes, and beneficiary functional change. Commenters noted that these measures are required of all PAC providers, though implementation dates vary.

Response: We thank commenters for this feedback. However, as we stated in the FY 2016 SNF PPS final rule (80 FR 46410), we do not believe we have the authority to adopt measures covering additional clinical topics beyond those specified in sections 1888(g)(1) and (2) of the Act at this time.

Comment: Commenters urged us to monitor the Program's impact on facilities' delivery of care quality and on beneficiaries' quality of life in nursing homes.

Response: We thank the commenters for this suggestion. We intend to monitor the Program's effects on beneficiaries, care quality, and other factors carefully.

Comment: One commenter offered several general suggestions for the Program based on New York's experience with the Nursing Home VBP Demonstration (https://innovation.cms.gov/​initiatives/​Nursing-Home-Value-Based-Purchasing/​) including incomparability of specialty and general facilities, narrowly-structured measures for participating facilities, regional adjustments, measure and calculation information provided to facilities to assist with quality improvement, a focus on preventable hospitalizations, and incentive payments large enough and close enough to the performance period to maximize behavioral changes.

Response: We thank the commenter for these suggestions. We proposed to adopt a performance period that is as close as we feasibly can set it to the payment year in order to establish a clear link between quality measurement and value-based payment. We note also that the methodology for determining the size of the pool available to fund the value-based incentive payments that we will disburse under the Program is specified in the statute. We intend to provide SNFs with information to assist with quality improvement efforts, and will work with stakeholders to ensure that all SNFs are able to improve the quality of care that they provide to Medicare beneficiaries. However, we do not agree with the commenter that we should perform regional adjustments to the measures adopted under the Program. Our experience with achievement thresholds and benchmarks based on national data in the Hospital Value-Based Purchasing Program has given us confidence that regional adjustments are not necessary to ensure that achievement thresholds and benchmarks for this program are balanced, appropriate standards of high quality. Some groups of facilities may perform better or worse than other facilities on certain measures, but we do not believe it would be appropriate to raise or lower the performance standards or measured performance for a facility based on regional differences in quality measurement, because such adjustments would seem to indicate that some areas of the country should be held to higher or lower standards of care quality. We intend to monitor SNFs' performance on the measures adopted under the Program carefully and may consider further adjustments to the measures or to the scoring methodology in the future.

Comment: Commenter also suggested that we factor managed care expansions into our measure calculations, noting that many states are rapidly expanding into managed care for Medicare and Medicaid beneficiaries and that managed care delivery could affect quality measurements. Commenter also recommended that we consider major care innovations that are being developed and tested across state lines to ensure that the interventions with the greatest potential for quality improvement may proliferate among SNFs.

Response: We thank the commenter for the suggestion. However, the SNF VBP Program is limited by statute to payments made under Medicare's SNF PPS, not payments to managed-care organizations, and we therefore believe the Program is appropriately focused on Medicare quality data at this time. We may consider incorporating quality information related to care provided by managed-care organizations in the Program in the future. However, we do not have the authority to make value-based incentive payments to SNFs based on their performance with patients enrolled in managed care plans. We will monitor clinical research on the effects of managed care in comparison to care delivered under fee-for-service systems, however.

We will consider major care innovations as they arise in clinical literature and in care delivery and will work with SNFs and stakeholders in order to encourage their proliferation.

We thank the commenters for this feedback.

b. Measures

i. SNF 30-Day All-Cause Readmission Measure (SNFRM) (NQF #2510)

Per the requirement at section 1888(g)(1) of the Act, in the FY 2016 SNF PPS final rule (80 FR 46419), we finalized our proposal to specify the SNF 30-Day All-Cause Readmission Measure (SNFRM) (NQF #2510) as the SNF all-cause, all-condition hospital readmission measure for the SNF VBP Program. The SNFRM assesses the risk-standardized rate of all-cause, all-condition, unplanned inpatient hospital readmissions of Medicare fee-for-service (FFS) SNF patients within 30 days of discharge from an admission to an inpatient prospective payment system (IPPS) hospital, CAH, or psychiatric hospital. The measure is claims-based, requiring no additional data collection or submission burden for SNFs. For additional details on the SNFRM, including our responses to public comments, we refer readers to the FY 2016 SNF PPS final rule (80 FR 46411 through 46419).

We received one comment on the SNFRM.

Comment: One commenter urged us to provide more timely feedback to SNFs on their performance on the SNFRM in order to better enable performance improvement.

Response: We intend to provide as much feedback on the SNFRM as is operationally possible to SNFs, and to do so as quickly as possible. As required by section 1888(g)(5) of the Act and as discussed further below, we will provide quarterly confidential feedback reports to SNFs beginning October 1, 2016, and will continue providing as much information to SNFs on their performance on the SNFRM as possible using those reports.

ii. Skilled Nursing Facility 30-Day Potentially Preventable Readmission Measure (SNFPPR)

We proposed to specify the SNF 30-Day Potentially Preventable Readmission Measure (SNFPPR) as the SNF all-condition risk-adjusted potentially preventable hospital readmission measure to meet the requirements of section 1888(g)(2) of the Act. This proposed measure assesses the facility-level risk-standardized rate of unplanned, potentially preventable hospital readmissions for SNF patients within 30 days of discharge from a prior admission to an IPPS hospital, CAH, or psychiatric hospital. Hospital readmissions include readmissions to a short-stay acute-care hospital or CAH, with a diagnosis considered to be unplanned and potentially preventable. This proposed measure is claims-based, requiring no additional data collection or submission burden for SNFs.

Hospital readmissions among the Medicare population, including beneficiaries that utilize post-acute care, are common, costly, and often preventable.^[1 2] The Medicare Payment Advisory Commission (MedPAC) and a study by Jencks et al. estimated that 17 to 20 percent of Medicare beneficiaries discharged from the hospital were readmitted within 30 days. MedPAC found that more than 75 percent of 30-day and 15-day readmissions and 84 percent of 7-day readmissions were considered potentially preventable.^[3] In addition, MedPAC calculated that annual Medicare spending on potentially preventable readmissions would be $12B for 30-day, $8B for 15-day, and $5B for 7-day readmissions.^[4] For hospital readmissions from SNFs, MedPAC deemed 76 percent of readmissions as potentially avoidable—associated with $12B in Medicare expenditures.^[5] Mor et al. analyzed 2006 Medicare claims and SNF assessment data (Minimum Data Set), and reported a 23.5 percent readmission rate from SNFs, associated with $4.3B in expenditures.^[6]

We have addressed the high rates of hospital readmissions in the acute care setting, as well as in PAC by developing the SNF 30-Day All-Cause Readmission Measure (NQF #2510), as well as similar measures for other PAC providers (NQF #2502 for IRFs and NQF #2512 for LTCHs).^[7] These measures are endorsed by the National Quality Forum (NQF), and the NQF-endorsed measure (NQF #2510) was adopted for the SNF VBP program in the FY 2016 SNF PPS final rule (80 FR 46411 through 46419). These NQF-endorsed measures assess all-cause unplanned readmissions.

Several general methods and algorithms have been developed to assess potentially avoidable or preventable hospitalizations and readmissions for the Medicare population. These include the Agency for Healthcare Research and Quality's (AHRQ) Prevention Quality Indicators, approaches developed by MedPAC, and proprietary approaches, such as the 3M^[TM] algorithm for Potentially Preventable Readmissions (PPR).^[8 9 10] Recent work led by Kramer et al. for MedPAC identified 13 conditions for which readmissions were deemed as potentially preventable among SNF and IRF populations ^[11 12] ; however, these conditions did not differ by PAC setting or readmission window (that is, readmissions during the PAC stay or post-PAC discharge). Although much of the existing literature addresses hospital readmissions more broadly and potentially avoidable hospitalizations for specific settings like skilled nursing facilities, these findings are relevant to the development of potentially preventable readmission measures for PAC.^{[13 14 15]}

Based on the evidence discussed above and to meet PAMA requirements, we proposed to specify this measure, entitled, SNF 30-Day Potentially Preventable Readmission Measure (SNFPPR), for the SNF VBP Program. The SNFPPR measure was developed by CMS to harmonize with the NQF-endorsed SNF 30-Day All-Cause Readmission Measure (NQF #2510) ^[16] adopted in the FY 2016 SNF final rule (80 FR 46411 through 46419) and the Hospital-Wide Risk-Adjusted All-Cause Unplanned Readmission Measure (NQF #1789) (Hospital-Wide Readmission or HWR measure ^[17] ), finalized for the Hospital IQR Program in the FY 2013 IPPS/LTCH PPS final rule (77 FR 53521 through 53528). Although these existing measures focus on all-cause unplanned readmissions and the SNFPPR measure assesses potentially preventable hospital readmissions, the SNFPPR will use the same statistical approach, the same time window as NQF measure #2510 (that is, 30 days post-hospital discharge), and a similar set of patient characteristics for risk adjustment. As appropriate, the potentially preventable hospital readmission measure for SNFs is being harmonized with similar measures being finalized for LTCHs, IRFs, and HHAs to meet the requirements of the Improving Medicare Post-Acute Care Transformation Act of 2014 (IMPACT Act) (Pub. L. 113-185).

The SNFPPR measure estimates the risk-standardized rate of unplanned, potentially preventable hospital readmissions for Medicare FFS beneficiaries that occur within 30 days of discharge from the prior proximal hospitalization. This is a departure from readmission measures in other PAC settings, such as the two measures being adopted in the Inpatient Rehabilitation Facility (IRF) Quality Reporting Program, one of which assesses readmissions that take place during the IRF stay and the other that assesses readmissions within 30 days following discharge from the IRF. The SNFPPR measure is distinct because section 1888(h)(2) of the Act requires that only a single quality measure be implemented in the SNF VBP program at one time. A purely within-stay measure (that is, a measure that assesses readmission rates only when those readmissions occurred during a SNF stay) would perversely incentivize the premature discharge of residents from SNFs to avoid penalty. Conversely, limiting the measure to readmissions that occur within 30-days post-discharge from the SNF would not capture readmissions that occur during the SNF stay. In order to qualify for this measure, the SNF admission must take place within 1 day of discharge from a prior proximal hospital stay. The prior proximal hospital stay is defined as an inpatient admission to an acute care hospital (including IPPS, CAH, or a psychiatric hospital). Because the measure denominator is based on SNF admissions, a single Medicare beneficiary could be included in the measure multiple times within a given year. Readmissions counted in this measure are identified by examining Medicare FFS claims data for readmissions to either acute care hospitals (IPPS or CAH) that occur within 30 days of discharge from the prior proximal hospitalization, regardless of whether the readmission occurs during the SNF stay or takes place after the patient is discharged from the SNF. Because patients differ in complexity and morbidity, the measure is risk-adjusted for case-mix. Our approach for defining potentially preventable readmissions is described below.

Potentially Preventable Readmission Measure Definition: We conducted a comprehensive environmental scan, analyzed claims data, and obtained input from a technical expert panel (TEP) to develop a working conceptual definition and list of conditions for which hospital readmissions may be considered potentially preventable. The Ambulatory Care Sensitive Conditions (ACSC)/Prevention Quality Indicators (PQI), developed by AHRQ, served as the starting point in this work. For the purposes of the SNFPPR measure, the definition of potentially preventable readmissions differs based on whether the resident is admitted to the SNF (referred to as “within-stay”) or in the post-SNF discharge period; however, there is considerable overlap of the definitions. For patients readmitted to a hospital during within the SNF stay, potentially preventable readmissions (PPR) should be avoidable with sufficient medical monitoring and appropriate treatment. The within-stay list of PPR conditions includes the following, which are categorized by 4 clinical rationale groupings: (1) Inadequate management of chronic conditions; (2) Inadequate management of infections; (3) Inadequate management of other unplanned events; and (4) Inadequate injury prevention. For individuals in the post-SNF discharge period, a potentially preventable readmission refers to a readmission in which the probability of occurrence could be minimized with adequately planned, explained, and implemented post discharge instructions, including the establishment of appropriate follow-up ambulatory care. Our list of PPR conditions in the post-SNF discharge period includes the following, categorized by 3 clinical rationale groupings: (1) Inadequate management of chronic conditions; (2) Inadequate management of infections; and (3) Inadequate management of other unplanned events. Additional details regarding the definitions of potentially preventable readmissions are available in our Measure Specification (available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNF-VBP.html).

This SNFPPR measure focuses on readmissions that are potentially preventable and also unplanned. Similar to the SNF 30-Day All-Cause Readmission Measure (SNFRM) (NQF #2510), this measure uses the CMS Planned Readmission Algorithm to define planned readmissions. In addition to the CMS Planned Readmission Algorithm, this measure incorporates procedures that are considered planned in post-acute care settings, as identified in consultation with TEPs. Full details on the planned readmissions criteria used, including the additional procedures considered planned for post-acute care, can be found in the Measure Specifications (available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNF-VBP.html).

This measure assesses potentially preventable readmission rates while accounting for patient or resident demographics, principal diagnosis in the prior hospital stay, comorbidities, and other patient factors. The model also estimates a facility-specific effect, common to patients or residents treated in each facility. This measure is calculated for each SNF based on the ratio of the predicted number of risk-adjusted, unplanned, potentially preventable hospital readmissions that occurred within 30 days of discharge from the prior proximal hospitalization, including the estimated facility effect, to the estimated predicted number of risk-adjusted, unplanned hospital readmissions for the same individuals receiving care at the average SNF. A ratio above 1.0 indicates a higher than expected readmission rate (worse), while a ratio below 1.0 indicates a lower than expected readmission rate (better). This ratio is referred to as the standardized risk ratio or SRR. The SRR is then multiplied by the overall national raw rate of potentially preventable readmissions for all SNF stays. The resulting rate is the risk-standardized readmission rate (RSRR) of potentially preventable readmissions. The full methodology is detailed in the Measure Specifications (available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNF-VBP.html).

Eligible SNF stays in the measure are assessed until: (1) The 30-day period ends; or (2) the patient is readmitted to an acute care hospital (IPPS or CAH). If the readmission is classified as unplanned and potentially preventable, it is counted as a readmission in the measure calculation. If the readmission is planned or not preventable, the readmission is not counted in the measure rate.

Readmission rates are risk-adjusted for case-mix characteristics. The risk adjustment modeling estimates the effects of patient/resident characteristics, comorbidities, and select health care variables on the probability of readmission. More specifically, the risk-adjustment model for SNFs accounts for sociodemographic characteristics (age, sex, original reason for entitlement), principal diagnosis during the prior proximal hospital stay, body system specific surgical indicators, comorbidities, length of stay during the resident's prior proximal hospital stay, intensive care utilization, end-stage renal disease status, and number of prior acute care hospitalizations in the preceding 365 days. This measure is calculated using one full calendar year of data. The full measure specifications and results of the reliability testing can be found in the Measure Specifications (available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNF-VBP.html).

Our measure development contractor convened a TEP, which provided input on the technical specifications of this measure, including the development of an approach to define potentially preventable hospital readmissions for a number of PAC settings, including SNFs. Details from the TEP meetings, including TEP members' ratings of conditions proposed as being potentially preventable, are available in the TEP Summary Report available on the CMS Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html. We also solicited stakeholder feedback on the development of this measure through a public comment period held from November 2 through December 1, 2015. A summary of the public comments we received is also available on the CMS Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

In addition to our TEP and public comment feedback, we also considered input from the Measures Application Partnership (MAP) on the SNFPPR. The MAP is composed of multi-stakeholder groups convened by the NQF. The MAP provides input on the measures we are considering for implementation in certain quality reporting and pay-for-performance programs. In general, the MAP has noted the need for care transition measures in PAC/LTC performance measurement programs and stated that setting-specific admission and readmission measures would address this need.^[18] The SNFPPR measure that we proposed, and that we are adopting for the SNF VBP Program in this final rule, was included in the List of Measures under Consideration (MUC List) for December 1, 2015.^[19]

The MAP encouraged continued development of the measure in the SNF VBP Program to meet the mandate of PAMA. Specifically, the MAP stressed the need to promote shared accountability and ensure effective care transitions. More information about the MAP's recommendations for this measure is available at http://www.qualityforum.org/​Publications/​2016/​02/​MAP_​2016_​Considerations_​for_​Implementing_​Measures_​in_​Federal_​Programs_​-_​PAC-LTC.aspx. At the time, the risk-adjustment model was still under development. Following completion of that development work, we were able to test for measure validity and reliability as available in the measure specifications document provided above. Testing results are within range for similar outcome measures finalized in public reporting and value-based purchasing programs, including the SNFRM finalized for this program.

We invited public comment on our proposal to adopt this measure, the SNF 30-Day Potentially Preventable Readmission Measure (SNFPPR). The comments we received on this topic, with their responses, appear below.

Comment: One commenter called on us to establish a standardized process by which we could evaluate new measures for the Program, or alternatively a standard process to evaluate whether or not we should remove or retire a measure. The commenter suggested that we adopt the same methods under use in the Hospital IQR and Hospital VBP Programs.

Response: We do not believe that a standardized process is necessary for the SNF VBP Program because unlike the Hospital IQR and Hospital VBP Programs, we are statutorily limited in the SNF VBP Program to including only two measures (one at a time). Since we have not yet implemented the SNFPPR, we do not believe establishing a standardized process for replacing it is warranted at this time.

Comment: Some commenters supported our proposal to adopt the SNFPPR, including the measure's intent, and recognized that the measure will provide incentives for SNFs to coordinate care post-discharge. Some commenters specifically stated their support for the infectious conditions defined as potentially preventable, stating that many of these conditions are preventable using appropriate infection prevention interventions.

Response: We agree that the measure will provide strong incentives for care coordination and will appropriately capture preventable readmissions, including infection-related readmissions.

Comment: One commenter stated that SNFs should not be penalized for readmissions when the conditions that prompted them are unrelated to the reasons the patient was admitted to the SNF. The commenter also called on us to account for differences in each SNF's mix of low-income patients when calculating readmissions.

Response: We note that the SNF VBP Program's statute requires that the measures required under sections 1888(g)(1) and (2) of the Act must be “all-condition hospital readmission” measures, which we believe necessitates attributing readmissions to SNFs even in the case the commenter specifies.

We believe that the proposed risk adjustment methodology appropriately adjusts for SNFs' patient mix when calculating readmissions, particularly because the measure's risk adjustments were developed to harmonize with the Hospital Wide Readmission (HWR) measure (NQF #1789), and the SNFRM. We describe the risk adjustment variables in more detail in the draft SNF PPR technical report, which is available on our Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNFPPR-Technical-Report.pdf. We respond to commenter's point about sociodemographic or socioeconomic adjustments below in a subsequent response.

Comment: One commenter stated that we should develop additional criteria for SNFs that have implemented programs and policies to mitigate unplanned events. The commenter suggested that SNFs with standard fall precautions should not be penalized if a well-managed, low-risk dementia patient falls and sustains a fracture.

Response: We believe that SNFs with programs and policies that reduce the incidence of unplanned events may generally experience fewer readmissions over time. However, a potentially preventable readmission still presents the potential for harm to the patient and generates costs for the Medicare program. We wish to clarify that this is a measure of potentially preventable readmissions and that not all readmissions are preventable. The PPR rate is not expected to be 0. The focus of this measure is to identify excess PPR rates for the purposes of quality improvement. We believe the Program will encourage SNFs to take appropriate, effective steps to minimize this outcome for SNF patients.

Comment: One commenter suggested that we adopt a minimum denominator size for the SNFPPR measure of 25 stays, though they preferred 30, stating that 30 stays would produce more reliable results for low-volume SNFs. The commenter noted that observed variability increases substantially between 30 and 20 stays, and requested that we provide data on the variation in SNFPPR rates for SNFs with small denominator sizes.

Response: We wish to clarify that we did not propose a minimum denominator size for the SNFPPR measure. We acknowledge that increasing the denominator size for this measure may increase its reliability. However, doing so would exclude a substantial number of SNFs from the measure calculation and thus the SNF VBP Program However, as stated in the SNF PPR technical report available on our Web site (https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNFPPR-Technical-Report.pdf), we found 1 year of data to be sufficient to calculate this measure in a statistically reliable manner.

Comment: One commenter supported the proposed risk adjustment methodology for the SNFPPR, noting that the adjustments will provide a valid assessment of a facility's care quality in preventing unplanned, preventable hospital readmissions.

Response: We thank the commenter for their comment.

Comment: One commenter expressed concern about our proposal to use claims-based data for quality measurement. The commenter believes that claims-based data are not accurate compared to other types of quality measure data, and the commenter cautioned that having performance data is not the same as having highly reliable and accurate data. The commenter suggested that claims data may be better used as a supplement to traditional HAI surveillance after validation.

Response: With respect to the use of claims data to calculate this measure, multiple studies have been conducted to examine the validity of using Medicare hospital claims for several NQF-endorsed quality measures used in public reporting and value-based purchasing programs.^{[20 21 22]} These studies supported the use of claims data as a valid means for risk adjustment and assessing similar outcomes. Additionally, although assessment and other data sources may be valuable for risk adjustment, we are not aware of another data source aside from Medicare claims data that could be used to reliably assess the outcome of potentially preventable hospital readmissions during this readmission window.

Comment: One commenter expressed concerns about the use of readmissions measures for SNFs, stating that the sickest individuals are the most likely to be readmitted. The commenter also noted that the sickest individuals are the most likely to die, so facilities with excessive mortality rates may have lower readmission rates. Some commenters were concerned that facilities may be incentivized to delay needed care in order to improve their readmission scores and suggested that we include ER visits in the measure.

Response: We believe that the risk adjustment approach used in calculating the SNFPPR measure appropriately adjusts for patient case-mix even among patients that may be at end-of-life. We intend to conduct ongoing evaluation and monitoring to ensure that the measure does not result in unintended consequences for patients, such as increased mortality rates.

With respect to emergency room visits, we note while such visits can certainly be negative outcomes for patients, they are not readmissions within the definitions we have adopted for measures of readmissions. We agree with commenters that mortality is also an important clinical outcome, but in other settings where we assess both readmission and mortality rates, the two types of measures seem to correlate,^[23] which suggests that we do not see reductions in readmission rates as a consequence of increasing mortality rates.

Comment: One commenter suggested that we allow additional time between when we specify a quality measure for the Program and when we begin using the measures for payment purposes. The commenter stated that more lead time would better enable providers to understand new measures and address quality improvement issues.

Response: While we understand the commenter's concern, we must implement the Program in accordance with the deadlines specified in statute, and quality measure development is a lengthy process requiring significant time and testing to ensure that measures are clinically and statistically valid. We were required under section 1888(g)(1) of the Act to specify a skilled nursing facility all-cause, all-condition hospital readmission measure not later than October 1, 2015. Similarly, under section 1888(g)(2) of the Act, we are required to specify a measure of all-condition risk-adjusted potentially preventable hospital readmissions for skilled nursing facilities not later than October 1, 2016. Additionally, under section 1888(h)(1)(B) of the Act, we are required to begin making value-based incentive payments to SNFs on October 1, 2018 (the beginning of FY 2019). However, we intend to work with SNFs and other stakeholders to raise awareness and understanding of program requirements. For example, the confidential feedback reports required by PAMA are one mechanism through which we can educate SNFs about the measures and their performance on the measures prior to implementation.

Comment: One commenter was concerned that SNFs would not necessarily be able to verify the accuracy of the risk adjustment model, as they are unlikely to have access to complete information on sociodemographic characteristics, principal diagnosis during the proximal hospital stay, body system specific surgical indicators, comorbidities, length of stay during the proximal hospital stay, intensive care utilization, ESRD status, and the number of hospital stays during the prior year. The commenter suggested that we provide SNFs with verifiable prior hospitalization information used to calculate the risk adjustment.

Response: We thank the commenter for their concern over providers' ability to verify the accuracy of the data used for risk adjustment and to calculate this measure. We will take this comment under consideration as we determine which data elements would enable SNFs to verify their data and risk-standardized PPR rate. We refer readers to the review and correction subsection of this final rule for additional information.

Comment: One commenter recommended that we describe readmissions as “potentially preventable,” not “preventable,” stating that the literature on readmissions shows that they occur even when ideal care that conforms to all clinical guidelines is provided. The commenter noted that ambulatory care sensitive conditions and Patient Quality Indicators developed by AHRQ were intended to assess the availability of and access to ambulatory care services in a community, but have not been focused on individual hospitals and other providers. The commenter did not object to this focus, but requested that we modify our language and measure construction to account for the measure's use in tracking individual providers rather than the community. The commenter stated that our goal should not be zero readmissions, as SNFPPR rates of zero can only be achieved by denying hospital services to individuals.

Response: The readmissions to be measured in the SNFPPR are defined as those believed to be “potentially preventable,” as we understand that some SNF patients might be readmitted to the hospital even if they receive excellent care from the SNF. Both the SNFPPR and the SNFRM calculate facility-level risk-standardized readmission rates in order to provide quality of care information about individual providers rather than community-level characteristics. Given that the SNFPPR is capturing “potentially preventable” readmissions, the goal is not to reach zero readmissions, but is to identify excess rates of readmissions that could potentially have been avoided in order to assess the quality of care being furnished by individual SNFs.

Comment: Several commenters urged us to consider adjusting the SNFPPR for socioeconomic and/or sociodemographic factors. The commenter also urged us to conduct additional testing on the categories and codes used to identify PPRs.

Response: The categories and specific conditions used to identify potentially preventable readmissions were developed based on existing evidence and were vetted by a TEP, which included clinicians and post-acute care experts. We also conducted a comprehensive environmental scan to identify conditions for which readmissions may be considered potentially preventable. Results of this environmental scan and details of the TEP input received were made available on the CMS Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

Readmissions may be considered potentially preventable even if they may not appear to be clinically related to the patient's original reason for SNF admission. There is substantial evidence that the conditions included in the definition are preventable with sufficient medical monitoring and appropriate patient treatment during the SNF stay or adequately planned, explained, and implemented post-discharge instructions, including effective care coordination ensuring appropriate follow-up care after SNF discharge. Furthermore, this measure is based on Medicare claims data and it may not always be feasible to determine whether a subsequent readmission is or is not clinically related to the reason why the patient was admitted to the SNF.

With respect to socioeconomic or sociodemographic adjustment, we note that the NQF is currently undertaking a 2-year trial period in which new measures and measures undergoing maintenance review will be assessed to determine if risk-adjusting for sociodemographic factors is appropriate. This trial entails temporarily allowing inclusion of sociodemographic factors in the risk-adjustment approach for some performance measures. At the conclusion of the trial, NQF will issue recommendations on future permanent inclusion of sociodemographic factors. During the trial, measure developers are encouraged to submit information such as analyses and interpretations as well as performance scores with and without sociodemographic factors in the risk adjustment model. Several measures developed by CMS have been brought to NQF since the beginning of the trial. We, consistent with NQF's guidance to measure developers, have tested sociodemographic factors in the measures' risk models and made recommendations about whether or not to include these factors in the endorsed measure. We intend to continue engaging in the NQF process as we consider the appropriateness of adjusting for sociodemographic factors in our outcome measures.

Furthermore, the Office of the Assistant Secretary for Planning and Evaluation (ASPE) is conducting research to examine the impact of sociodemographic status on quality measures, resource use, and other measures under the Medicare program as directed by the IMPACT Act. We will closely examine the findings of the ASPE reports and related recommendations and consider how they apply to our quality programs at such time as they are available.

Comment: One commenter expressed concern that the SNFPPR proposed for the Program differs from the SNF QRP's readmission measure. The commenter noted that the VBP Program's measure assesses both post-discharge PPRs as well as those occurring during a SNF stay and includes an additional category of PPR of inadequate prevention of injury. The commenter urged us to consider a single measure for both programs.

Response: We made a policy decision to use two different measures for the SNF VBP and QRP Programs. Our rationale for this decision was that the readmission window associated with each measure assesses different aspects of SNF care. The readmission window for the SNFPPR measure was developed to align with the SNFRM which was previously adopted for the SNF VBP Program, and both of which are required by the SNF VBP Program's statute. Both the SNFRM and SNFPPR measure specifications, including the readmission window, were designed to harmonize with CMS's Hospital Wide All-Cause Unplanned Readmission (HWR) measure used in the Hospital IQR Program. The advantage of this window is that it assesses readmissions both during the SNF stay and post-SNF discharge for most SNF patients, depending on the SNF length of stay (LOS). For these measures, the focus is on transitions to the SNF from the prior proximal hospital stay, and we believe the alignment to be appropriate since the SNF VBP Program's statute specifically directs us to adopt measures of hospital readmissions.

The readmission window used for the SNF measure proposed for the SNF QRP to meet the IMPACT Act requirements was developed to align with other post-acute care readmission measures. The focus of this post-PAC discharge readmission window is on assessing potentially preventable hospital readmissions during the 30 days after discharge. We believe that assessing PPRs during each of these readmission windows provides valuable information for their respective programs.

Comment: One commenter was concerned about the measure's ability to pinpoint the SNF's care for a short-stay resident who is expected to move on to the community setting, and commenter noted that SNFs often do not have easy access to information needed to improve on the measure. The commenter called on CMS to provide claims data to SNFs so that facilities can verify the measure, determine whether or not they are receiving necessary patient information, and conduct quality improvement efforts.

Response: We appreciate the commenters' feedback. We are cognizant of providers' desire for more information on quality performance, and we are considering ways to provide the best information to SNFs. As required by statute and as discussed further below, we will provide quarterly confidential feedback reports to SNFs detailing their performance on measures specified for the Program, and we are interested in SNFs' feedback on the reports and on their contents once we provide them. We will take that feedback into account as we refine the quarterly reports to be most useful to SNFs for quality improvement efforts.

Comment: Commenter noted that the SNF QRP version of the SNFPPR counts unplanned readmissions to LTCHs and asked us to clarify why the SNF VBP version of the measure does not include readmissions to LTCHs.

Response: The SNFPPR was developed to harmonize with the SNFRM, previously adopted for the SNF VBP Program, and both measures do not count planned readmissions to LTCHs. However, the potentially preventable hospital readmission measure proposed for the SNF QRP to meet the requirements of the IMPACT Act does count readmissions to LTCHs in order to align with the other IMPACT Act measures. We intend to conduct analyses to determine the impact that including readmissions to LTCHs would have on the QRP measure performance; however, we expect that this will represent a relatively small number of readmissions and will have a minimal impact.

Comment: Commenter was concerned that SNFs would not necessarily be able to verify the accuracy of the risk adjustment model, as they are unlikely to have access to complete information on sociodemographic characteristics, principal diagnosis during the proximal hospital stay, body system specific surgical indicators, comorbidities, length of stay during the proximal hospital stay, intensive care utilization, ESRD status, and the number of hospital stays during the prior year. The commenter suggested that we provide SNFs with verifiable prior hospitalization information used to calculate the risk adjustment.

Response: We thank the commenter for their concern over providers' ability to verify the accuracy of the data used to calculate this measure. We will take this comment under consideration as we determine which data elements would enable SNFs to verify their data and risk-standardized PPR rate.

Comment: Commenter supported our proposal to adopt claims-based measures rather than measures based on self-reported data, stating that the latter are susceptible to gaming. The commenter also applauded our choice to count within-stay and post-discharge hospital readmissions in the measure. However, the commenter stated that we should extend the measured time period to 90 days, suggesting that the proposed 30-day time period is too short to capture poor care provided by a SNF. Another commenter supported the adoption of the SNFPPR and suggested that both the proposed and previously adopted measure (SNFRM) readmission measures could be improved by extending the readmission window. The commenter noted that about one-third of SNF stays are longer than the proposed 30-day window, and suggested that the current proposal could create incentives for SNFs to delay care until after the 30th day to avoid being penalized on the measure.

Response: We appreciate the commenter's support for the proposed measure, including the support for using claims data as the source for the measure's calculation. We are not aware of another data source aside from Medicare claims data that could be used to reliably assess the outcome of potentially preventable hospital readmissions for this specific readmission window.

The 30-day readmission window used in both the SNFRM (NQF #2510) and the proposed SNFPPR was developed to harmonize with measures used in the hospital setting, including the NQF-endorsed Hospital-Wide Risk-Adjusted All-Cause Unplanned Readmission Measure (NQF #1789). This readmission window was also vetted by technical expert panels. We appreciate the suggestion to consider a 90-day readmission window; however, we believe it would be difficult to ensure that potentially preventable hospital readmissions occurring up to 90 days after prior hospital discharge are attributable to the SNF care received. As we noted previously in this section, the advantage of this window is that it assesses readmissions both during the SNF stay and post-SNF discharge for most SNF patients, depending on the SNF length of stay. For these measures, the focus is on transitions to the SNF from the prior proximal hospital stay, and we believe the alignment to be appropriate since the SNF VBP Program's statute specifically directs us to adopt measures of hospital readmissions.

We intend to conduct ongoing evaluation and monitoring to assess for potential unintended consequences associated with the implementation of this measure. We will report results of our monitoring for potential unintended consequences—including the potential of SNFs to push needed care just past the 30-day window—in future SNF PPS rules.

Comment: Commenter expressed concern about our proposal to include the number of hospitalizations during the previous year as a factor in risk-adjustment. The commenter stated that this factor could result in adjusting a facility's rate for potentially preventable readmissions that occurred during the previous year. The commenter stated that a facility that did poorly preventing preventable readmissions during the prior year would receive a lower readmission target rate as a result.

Response: We agree with the comment that risk adjusting for the count of a beneficiary's prior year hospitalizations may include potentially preventable readmissions. However, we do not believe that the impact of risk adjusting for this will be driven by potentially preventable readmissions since this captures all hospital admissions as well as hospital readmissions. We have chosen to adjust for this factor at the patient-level because it is an indicator of several case-mix factors that we believe are important for risk adjustment. For example, a higher number of prior hospital stays may be indicative of a more complex or compromised clinical state. The number of prior hospital stays may also be related to otherwise unmeasured patient characteristics such as access, and patient compliance during the post-discharge period. Furthermore, we do not believe that including this as a risk adjuster will have a major impact on SNFs' performance on the measure.

Comment: Some commenters suggested that we adopt a measure that assesses the rate of readmissions of SNF beneficiaries to a hospital within 30 days of their discharge from the SNF to a lower level of care or the community.

Response: We agree that a 30-day post-discharge from SNF measure would also be valuable for assessing potentially preventable hospital readmissions; however, given the Program is limited to one measure at a time, we believe that the readmission window selected for the SNFPRR provides specific advantages for the reasons described in this section. We note that we are adopting the Potentially Preventable 30-Day Post-Discharge Readmission Measure for the SNF QRP. That measure assesses the rate of readmissions within 30 days of a SNF discharge.

Comment: Commenters stated that the SNFPPR needs additional risk adjustment in order to avoid establishing incentives for facilities to avoid admitting challenging patients. Commenters specifically called for risk adjustment for socioeconomic status, functional status, medical complexity, and cognitive impairment. Commenters specifically stated that functional and cognitive status are among the strongest predictors of future health care utilization.

Response: We developed a comprehensive claims-based risk-adjustment model that takes into account demographic and eligibility characteristics; principal diagnoses; types of surgery or procedure from the prior short-term hospital stay; comorbidities; length of stay and ICU/CCU utilization from the immediately prior short-term hospital stay; and number of admissions in the year preceding the SNF admission. We direct readers to the final measure specifications posted on the CMS Web site (https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNF-VBP.html), which includes results of the final risk adjustment model. This comprehensive risk-adjustment model is similar to those developed for other NQF-endorsed readmission measures. Results of our testing are within range for similar outcome measures finalized in public reporting and value-based purchasing programs, including the SNFRM.

We agree with the comment that functional and cognitive status are potentially important predictors of readmission outcomes. We intend to evaluate the feasibility of including functional and cognitive status in the future, including using standardized assessment data required by the IMPACT Act when they become available. We refer readers to our reply above on the topic of socioeconomic or sociodemographic adjustment.

Comment: One commenter questioned why we exclude SNF stays where the patient had one or more intervening PAC admissions between the prior proximal hospital discharge and SNF admission or after the SNF discharge, within the 30-day risk window. The commenter also questioned why we exclude SNF admissions where the patient had multiple SNF admissions after the prior proximal hospitalization, within the 30-day risk window. The commenter believed that our stated rationale for this exclusion could apply to any PAC setting and therefore disagreed with the exclusion.

Response: This measure was developed to align with the SNFRM previously adopted for the SNF VBP Program. Both measures exclude patients who have intervening IRF or LTCH admissions before their first SNF admission. In analyses conducted for the SNFRM (NQF #2510), we found that these patients started their SNF admission later in the 30-day readmission window and received services different from those received by patients admitted directly from the hospital to the SNF. As a result, we determined patients with intervening stays present a different risk for readmission than patients admitted directly to the SNF. SNF patients with intervening IRF/LTCH stays had the lowest rates of all-cause readmission (8.6 percent) as compared with those with no intervening IRF/LTCH stay. Additionally, we found that those with intervening IRF/LTCH admissions had longer hospital lengths of stay and more prior proximal hospitalizations involving surgical procedures compared to those without an intervening stay.

This issue also impacts a relatively small number of SNF stays; previous analyses showed that 6 percent of SNF stays had an intervening PAC stay (IRF, LTCH, or another SNF) or go home from their prior proximal hospitalization and are later admitted to a SNF within the 30-day readmission window. Combined, these analyses provide justification for excluding SNF admissions with intervening IRF or LTCH admissions, or with multiple SNF stays, by showing these exclusions will not have a substantial effect on the SNFPPR. Additionally, concerns about attribution, given the mix of providers these patients have received services from during the risk period, states for the appropriateness of excluding these patients. Lastly, patients with multiple PAC stays do not cluster in a small group of facilities, so no facilities are disproportionately impacted by these exclusions. We will continue to monitor, among other unintended consequences of introducing this measure, whether patients are being shifted to other PAC providers or being sent home before arriving at SNFs.

Comment: One commenter stated that we should not exclude SNF stays with a gap of greater than one day between discharge from the prior proximal hospitalization and admission to a SNF. The commenter stated that this exclusion criterion does not consider medically complex patients treated in IRFs and subsequently readmitted for issues that may be treated as comorbidities. The commenter stated that admissions to IRFs should be considered as proximal hospitalizations since IRFs are licensed as hospitals.

Response: This measure was developed to harmonize with our other hospital readmission measures, the SNFRM, and other potentially preventable readmission measures which do not consider post-acute care settings, like IRFs, as proximal hospitalizations. Although IRFs are licensed as hospitals, we include them in the PAC continuum of care and, as such, we have proposed potentially preventable hospital readmission measures for the IRF QRP.

Comment: Commenter stated that we should not finalize the SNFPPR because the measure specifications were not published for the Technical Expert Panel or the MAP to review prior to the proposed rule's display. The commenter also noted that the risk adjustment model is new, and stated that the measure should not be rushed to meet an artificial deadline.

Response: In order to be as transparent as possible with the public, we made the specifications we had completed available to the TEP and the MAP. We then continued developing the measure in order to meet the deadline under section 1888(g)(2) of the Act to specify the measure by October 1, 2016. We also wish to note that although we were not required to make the specifications available to the MAP prior to proposing to adopt it for the SNF VBP, we did make the final specifications available to the MAP for comments and feedback. The risk-adjustment model developed for the SNFPPR measure was also made available at the time of the proposed rule.

Comment: Commenter stated that we should not finalize the SNFPPR because the MAP only recommended the measure as “encourage further development,” and did not vote to “support” or “support with conditions.” The commenter suggested that we should submit the measure for NQF endorsement. The commenter also noted that the SNF VBP statute specifies that the measure should be adopted “as soon as practicable,” and stated their belief that measures that will affect beneficiary access and quality as well as providers should undergo consensus review.

Response: Although the measure is not currently NQF-endorsed, we did conduct additional testing subsequent to the December 2015 MAP meeting where this measure was discussed. Based on that testing, we were able to complete the risk adjustment model and evaluate facilities' PPR rates, and we made the results of our analyses available at the time of the proposed rule. We found that testing results were similar to the SNFRM (NQF #2510) and allowed us to conclude that the measure is sufficiently developed, valid and reliable for adoption in the SNF VBP Program.

Comment: One commenter also stated that we should await NQF endorsement of the SNFPPR before we adopt it for use in the SNF VBP Program and at a minimum, should wait until at least 2 years after the SNFRM has been used in the Program.

Response: We intend to submit the SNFPPR to NQF for consideration of endorsement. With regard to the waiting at least 2 years before we adopt the SNFPPR for use in the SNF VBP, we will take this comment under consideration.

Comment: Commenter stated that we should use an “actual readmission rate” to calculate SRRs rather than predicted readmissions, or we should show how predicted and actual readmissions result in significantly different rankings in order to justify their use in the methodology. The commenter understood the statistical rationale for using the risk-adjusted estimate instead of actual readmission rate in the SRR, but did not believe that this approach provides superior or more accurate information than the actual readmission rate, and will instead be more confusing. The commenter called on us to use a simpler method.

Response: The statistical approach for this measure, including the use of the predicted to expected PPR rate, is used in several other quality measures, including the NQF-endorsed all-cause unplanned readmission measures for post-acute care and the hospital-wide all-cause readmission measure (NQF #1789) and other hospital readmission measures used in the Hospital Inpatient Quality Reporting (IQR) Program. Our decision to use this approach was influenced by work we became aware of by an independent committee appointed by the Committee of Presidents of Statistical Societies. In its White Paper report, the committee approved CMS's approach as a valid modeling approach with preferred statistical characteristics.^[24] We believe that this approach makes providers with small numbers of eligible patient stays less vulnerable to reported rates driven by the influence of random variation in performance, and, thus, will maximize the value of assessing SNFs' performance in SNF VBP. We would also like to note that facilities will be given their observed or actual readmission rates in their reports.

Comment: Commenter stated that the SNFPPR should not exclude individuals who died during the SNF stay, noting that individuals who died could still have been hospitalized for a PPR prior to dying. Commenter stated that excluding these patients will overestimate readmission rates in SNFs with high rates of within-SNF stay mortality and could create incentives to let patients die in SNFs rather than sending them to the hospital.

Response: We wish to clarify that the SNFPPR measure does not exclude patients who die during the 30-day window. If an individual died and was hospitalized for a PPR prior to dying, this readmission would in fact be included in the numerator for the facility. For additional information on the SNFPPR's calculation and methodology, we refer readers to the final specification that we will post on the SNF VBP Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Value-Based-Programs/​Other-VBPs/​SNF-VBP.html.

Comment: Commenter called on us to harmonize the SNFPPR with other PAC PPR measures, noting that the SNFPPR is the only one of several measures that counts readmissions during a patient's stay and after discharge, depending on the SNF length of stay. The commenter stated that the MAP recommended that the measure track “within stay” readmissions in order to align with other measures and avoid duplication of efforts, and noted that readmissions will be counted in both the SNFPPR and the Potentially Preventable 30-Day Post-Discharge Readmission Measure for the SNF QRP measure. The commenter acknowledged our concern that not counting post-discharge hospitalizations may create incentives for SNFs to discharge patients prematurely, but stated that we have not presented any evidence that this will, in fact, occur and that we have numerous other programs available to monitor any such behavior by SNFs. This commenter stated that, if nothing else, we should reduce the readmission window to seven days post-discharge, suggesting that readmissions after seven days are more reflective of quality and access to ambulatory care.

Response: Our decision to develop the SNFPPR using this specific readmission window was intended to balance the relative advantages associated with assessing the outcome both during the SNF stay and potentially post-discharge with any possible incentives to discharge patients who represent the highest risk for readmission in order to avoid penalty. Given that this measure is the sole determinant of a value-based purchasing program for SNFs, we were limited to selecting one readmission window for the measure and believe that counting readmissions that may occur post-discharge but within the 30-day window would be most valuable, even though other quality programs outside the VBP may be available to monitor premature discharges in SNFs.

The 30-day window reflects a transitional time period wherein the acute care hospital and skilled nursing facility are responsible for coordinating the care of a patient moving from one setting to another and is consistent with readmission measures used in other value-based purchasing programs, such as the ESRD Quality Incentive Program and the Hospital Readmission Reduction Program, as well as readmission measures used in a number of quality reporting programs that apply to post-acute care providers. Furthermore, our analysis of readmission rates showed no patterns indicating that using a shorter or longer period would produce very different comparative results, though the overall rates would change. In addition, the NQF Standing Committee generally agreed that 30 days post-hospital discharge is an accepted standard for measuring readmissions. Longer windows may be subject to greater “noise” in the readmission rate. The measure as specified has the potential for this unintended consequence of delaying hospital care beyond the 30-day readmission window, but this is a danger that would be associated with any selected day threshold. In addition, we will continue to analyze whether there are changes in the number of days to hospital readmission over time in order to assess whether a change to the readmissions window is needed for this measure in the future.

After consideration of the public comments that we received, we are finalizing our proposal to adopt the SNF 30-Day Potentially Preventable Readmission Measure (SNFPPR) for the SNF VBP Program.

Section 1888(h)(2)(B) of the Act requires the Secretary to apply the all-condition risk-adjusted potentially preventable hospital readmission measure specified under paragraph (g)(2) instead of the measure specified under paragraph (g)(1) as soon as practicable. We will apply the measure specified under paragraph (g)(1) beginning in performance year CY 2017 for payment year FY 2019, and we will apply it until such a time as the measure specified under paragraph (g)(2) replaces the measure specified under paragraph (g)(1). We intend to propose the timing for the change to the paragraph (g)(2) measure in future rulemaking. We sought comment on when we should propose this change for the SNF VBP Program. The comments we received on this topic, with their responses, appear below.

Comment: One commenter stated that the SNFPPR should replace the SNFRM as soon as possible because the SNFPPR holds providers accountable for conditions that can be managed in the SNF. The commenter suggested that we could replace the SNFRM for scoring beginning in October 2019, after the first Program year. Still other commenters suggested that we transition the measure once it receives unconditional endorsement from NQF, or that we allow at least a full year for SNFs to receive and understand their SNFPPR data before we implement the measure. Another commenter suggested that we defer transitioning the Program from the SNFRM to the SNFPPR, citing the MAP's vote to recommend the measure to “encourage further development” and the commenter's belief that the measure should be subjected to additional public comments prior to its adoption.

Response: We thank commenters for these suggestions. We will consider these comments when we develop a future proposal to replace the SNFRM with the SNFPPR.

As noted previously in this section, we also intend to submit the SNFPPR to the NQF for consideration of endorsement as soon as possible.

c. Performance Standards

i. Background

Sections 1888(h)(3)(A) of the Act requires the Secretary to establish performance standards for the SNF VBP Program. Under paragraph (3)(B) of section 1888(h) of the Act, the performance standards must include levels of achievement and improvement, and under paragraph (3)(C) of such section, must be established and announced not later than 60 days prior to the beginning of the performance period for the FY involved.

In the FY 2016 SNF PPS final rule (80 FR 46419 through 46422), we summarized public comments we received on possible approaches to calculating performance standards under the SNF VBP Program. We specifically sought comment on the approaches that we have adopted for other Medicare VBP programs such as the Hospital VBP Program (Hospital VBP Program), the Hospital-Acquired Conditions Reduction Program (HAC Reduction Program), the Hospital Readmissions Reduction Program (HRRP), and the End-Stage Renal Disease Quality Incentive Program (ESRD QIP). We also sought comment on the best possible approach to measuring improvement, particularly given the SNF VBP Program's limitation to one measure for each program year.

ii. Proposed Performance Standards Calculation Methodology

We believe that an essential goal of the SNF VBP program is to provide incentives for all SNFs to improve the quality of care that they furnish to their residents. In determining what level of SNF performance would be appropriate to select as the performance standard for the quality measures specified under the SNF VBP program, we focused on selecting levels that would challenge SNFs to improve continuously or to maintain high levels of performance. To achieve this aim, we analyzed SNFRM data and examined how different achievement performance standards would impact SNFs' scores under the proposed scoring methodology described further below. As more data becomes available, we will continue to assess the appropriateness of these performance standards for the SNF VBP program and, if necessary, propose to refine these standards' definitions and calculation methodologies to better incentivize the provision of high-quality care.

(a) Proposed Achievement Performance Standard and Benchmark

Beginning with the FY 2019 SNF VBP program, we proposed to define the achievement performance standard (which we will refer to as the “achievement threshold”) for quality measures specified under the SNF VBP program as the 25th percentile of national SNF performance on the quality measure during the applicable baseline period. We believe this achievement threshold definition represents an achievable standard of excellence and will reward SNFs appropriately for their performance on the quality measures specified for the SNF VBP program. We further believe this achievement threshold definition will provide strong incentives for SNFs to improve their performance on the measures specified for the SNF VBP Program continuously and will result in a wide range of SNF measure scores that can be used in public reporting.

We further proposed to define the “benchmark” for quality measures specified under the SNF VBP program as the mean of the top decile of SNF performance on the quality measure during the applicable baseline period. We believe this definition represents demonstrably high but achievable standards of excellence; in other words, the benchmark will reflect observed scores for the group of highest-performing SNFs on a given measure. This proposed benchmark policy aligns with that used by the Hospital VBP Program. As stated in the FY 2016 SNF PPS final rule (80 FR 46419 through 46420), we believe the Hospital VBP Program's performance standards methodology is a well-understood methodology under which health care providers and suppliers can be rewarded both for providing high-quality care and for improving their performance over time. We therefore believe it is appropriate to align with the Hospital VBP Program in setting benchmarks for the SNF VBP Program.

We also proposed that SNFs would receive points along an achievement range, which is the scale between the achievement threshold and the benchmark. Under this proposal, SNFs would receive achievement points if they meet or exceed the achievement threshold for the specified measure, and could increase their achievement score based on higher levels of performance. (We described the proposed scoring methodology, including how we proposed to award points for both achievement and improvement, in the scoring methodology section of the proposed rule). This proposed achievement range policy aligns with that used by the Hospital VBP Program. We refer readers to the FY 2016 SNF PPS final rule (80 FR 46419 through 46420) for a discussion of the rationale behind aligning SNF VBP Program policies with the Hospital VBP Program. As stated in that rule, we believe that the Hospital VBP Program's performance standards methodology is well-understood and would allow us to reward SNFs both for providing high-quality care and for improving their performance over time. We stated our intent to publish the final performance standards using complete data from CY 2015 in the FY 2017 SNF PPS final rule, and we have updated the numerical values in Table.

The comments we received on this topic, with their responses, appear below.

Comment: Commenters supported our proposed performance standards calculations, including our proposal to define the achievement threshold as the 25th percentile of national SNF performance during the baseline period. Commenters also supported our proposal to define the benchmark as the mean of the top decile of all SNFs' performance on proposed measures. Some commenters requested that we establish and announce the achievement threshold and benchmark earlier in the year in order to give SNFs additional time to develop quality improvement strategies.

Response: We thank the commenters for their support. However, we do not believe we can establish and announce the achievement threshold and the benchmark earlier in the year given the time needed to compile claims data and compute the readmissions measures.

We also sought comment on whether we should consider adopting either the 50th or 15th percentiles of national SNFs' performance on the quality measure during the applicable baseline period. We sought comment on data or other analysis that we should consider regarding the impact on SNFs' financial viability and service delivery to beneficiaries at either the higher or lower alternative standard. For example, while the 50th percentile would represent a more challenging threshold for care quality improvement, that standard would align with the Hospital VBP Program and would likely result in higher value-based incentive payments to top-performing SNFs than other definitions, though the actual distribution of value-based incentive payments would depend on all SNFs' performance and on the statutory rules governing their distribution. Such a standard would likely result in lower value-based incentive payments to lower-performing SNFs, which could create substantial payment disparities among participating SNFs. Conversely, the 15th percentile would likely result in higher value-based incentive payments for lower-performing SNFs than other thresholds, with the corresponding result of lower value-based incentive-payments for top-performing SNFs compared to other thresholds. The comments we received on this topic, with their responses, appear below.

Comment: Commenter stated that we should not increase the proposed achievement threshold to 50 percent, noting that meeting such a standard may be difficult for small, rural, or frontier facilities with limited resources and low volume. The commenter also suggested that we should test the two-pronged process for performance standards for reliability and validity prior to payment and public reporting. Other commenters stated that the 2 percent withhold has a significant enough impact on providers that they need to take time to understand how to minimize payment penalties.

Response: As discussed further below, we are finalizing the definition of the achievement threshold as the 25th percentile of SNFs' performance during the applicable baseline period. We intend to monitor the effects of the performance standards' definition on SNFs' performance and on the provision of care to Medicare beneficiaries.

We are required by statute to implement the 2 percent withhold from Medicare payments for SNFs. We intend to monitor the Program's effects on the impact of care by SNFs. However, as explained more fully above, we do not believe we can allow SNFs more time than we have proposed in order to understand how to minimize payment penalties.

Comment: One commenter recommended that we adopt the 50th percentile for the achievement threshold, stating that we should maintain consistency across settings when calculating achievement scores.

Response: While we agree with the commenter in general that consistency across settings in our value-based purchasing programs is important, we also recognize that we must implement these programs differently where statutory language differs or where the different care setting necessitates a policy change from other programs. We remain concerned that adopting the 50th percentile for the definition of the achievement threshold would result in about half of SNFs receiving no points for achievement under the Program, which would mean that we are effectively unable to reward their performance, particularly in cases where they do not qualify for improvement points. Our intention with the SNF VBP Program is to provide strong incentives for SNFs to improve their performance on the Program's measures continuously, and we do not believe that effectively excluding about half of SNFs from receiving achievement points will further that objective. We balanced that intention with our desire to ensure that we award points under the Program for quality performance, and do not award substantial points for what we have measured as poor-quality care. Upon further consideration of the comments, we believe the 25th percentile appropriately balances those goals.

Comment: Commenter expressed concerns about the alternative levels of the achievement threshold presented in the rule, suggesting that the 25th percentile represents the best chance to balance incentive payments between low and high performers. The commenter urged us to test these alternatives prior to implementation and public reporting.

Response: We thank the commenter for their support, and as discussed further above, we share the commenter's concerns about the alternatives to the 25th percentile for the achievement threshold. Accordingly, we are finalizing the definition of the achievement threshold as the 25th percentile of SNFs' performance on the Program's measures during the applicable baseline period.

Comment: One commenter was concerned about the proposed definition for the benchmark under the Program, explaining their preference for additional testing of the benchmark prior to its public reporting and use in calculating incentive payments. The commenter was concerned about unintended consequences for nursing homes and medically-complex or otherwise high-risk patients.

Response: We intend to monitor the Program's effects on SNFs' provision of high-quality care to Medicare beneficiaries. However, as we stated in the proposed rule (81 FR 24246), we believe that the proposed definition of the benchmark represents a demonstrably high but achievable standard of excellence for all SNFs, including those SNFs that treat high-risk patients. We note further that the measures specified under the Program are risk adjusted for medically-complex or otherwise high-risk patients, and we believe that adjustment will mitigate the commenter's concerns about unintended consequences. We intend to monitor the effects of the measures' risk adjustment policy to ensure that SNFs serving those patients are scored appropriately and are not penalized for treating medically-complex or high-risk patients.

(b) Improvement Performance Standard

Beginning with the FY 2019 SNF VBP program, we proposed to define the improvement performance standard (which we will refer to as the “improvement threshold”) for quality measures specified under the SNF VBP program as each specific SNF's performance on the specified measure during the applicable baseline period. As discussed further below, we will measure SNFs' performance during both the proposed performance and baseline periods, and we will award improvement points by comparing SNFs' performance to the improvement threshold. We believe this improvement performance standard ensures that SNFs will be adequately incentivized to improve continuously their performance on the quality measures specified under the SNF VBP Program, and we believe it appropriately balances our view that we should both reward SNFs for high performance and encourage improved performance over time.

We invited public comment on this proposal. The comments we received on this topic, with their responses, appear below.

Comment: Some commenters expressed concern about the proposed improvement points formula, suggesting that the formula should not require unrealistic levels of improvement from providers that are already high achievers based on their baseline period scores. Other commenters noted that we have in other rules explained that measures should be dropped or changed when performance reaches a uniformly high level.

Response: SNFs that are already high achievers based on their baseline period scores will be able to score achievement points under the proposed scoring methodology. While the commenter is correct that it may be difficult for a SNF to score a substantial number of improvement points if that SNF has a high baseline period score, the proposed methodology allows SNFs to earn ten additional points for achievement than they are able to earn for improvement. We therefore believe that SNFs that are already high achievers are well-positioned to earn high scores under the Program so long as they maintain their high performance on the specified measures.

We thank commenters for the suggestion that we should adopt a policy to drop measures or change them when performance reaches a uniformly high level. In other contexts, we have described this as a “topped-out” measures policy. We have not considered adopting such a policy for the SNF VBP Program to date, but we will consider whether or not to do so in future rulemaking.

(c) Publication of Performance Standard Numerical Values

Section 1888(h)(3)(C) of the Act requires the Secretary to establish and announce the performance standards for a given SNF VBP program year not later than 60 days prior to the beginning of the performance period for the FY involved. Based on the proposed performance period of CY 2017 for the FY 2019 SNF VBP Program, we believe that we must establish and announce performance standards for the FY 2019 Program not later than November 1, 2016. We intend to establish and announce performance standards for the Program in the annual SNF PPS rule, which is effective on October 1 of each year.

However, finalizing numerical values of these performance standards is often logistically difficult because it requires the collection and analysis of large amounts of quality measure data in a short period of time. For example, the data file for a full year of SNF claims data is typically completed around May of the following year. To calculate a numerical value for a performance standard, we must perform multiple levels of analyses on the data to ensure that all appropriate SNFs and patients are included in measure calculations; perform the measure calculations themselves; and then use those calculations to determine the numerical value for the performance standards. If any individual step of this process is delayed, it may preclude us from publishing finalized numerical values for the finalized performance standards in the applicable SNF PPS final rule, which is typically displayed publicly by August 1 of each year.

To retain the flexibility needed to ensure that numerical values published for the finalized performance standards are accurate, we proposed to publish these numerical values no later than 60 days prior to the beginning of the performance period but, if necessary, outside of notice-and-comment rulemaking. As noted, we intend to publish numerical values for those performance standards in the final rule when practicable. However, in instances in which we cannot complete the necessary analyses in time to include them in the SNF PPS final rule, we proposed to publish the numerical values for the performance standards on the QualityNet Web site used by SNFs to receive VBP information as soon as practicable but in no event later than the statutorily required 60 days prior to the beginning of the performance period for the fiscal year involved. In this instance, we would notify SNFs and the public of the publication of the performance standards using a listserv email and posting on the QualityNet News portion of the Web site.

We welcomed public comment on this proposal. The comments we received on this topic, with their responses, appear below.

Comment: One commenter supported our proposed timing and method for publishing the numerical values of the performance standards and for payment adjustments. The commenter appreciated the complexity of calculating hospital readmission rates and understood that we may need to publish performance standards or payment information outside of rulemaking. The commenter believed this to be a reasonable trade-off in order to have the performance period occur as close to the payment adjustment as possible.

Response: We thank the commenter for their support.

After consideration of the public comments that we received, we are finalizing our performance standards policies as proposed. Specifically, we are finalizing our definition of the achievement performance standard, which we refer to as the “achievement threshold,” for quality measures specified under the SNF VBP Program as the 25th percentile of national SNF performance on the quality measure during the applicable baseline period. We are finalizing our proposal to define the “benchmark” for quality measures specified under the SNF VBP Program as the mean of the top decile of SNF performance on the applicable quality measure during the applicable baseline period. We are also finalizing our proposals that SNFs would receive points along an achievement range, which is the scale between the achievement threshold and the benchmark.

We are also finalizing our proposal to define the improvement performance standard (which we refer to as the “improvement threshold”) for quality measures specified under the SNF VBP Program as each specific SNF's performance on the specified measure during the applicable baseline period.

We are also finalizing our proposal to publish the numerical values of the achievement threshold and the benchmark no later than 60 days prior to the beginning of the performance period, but if necessary, outside of notice-and-comment rulemaking.

The final values for the achievement threshold and the benchmark for the FY 2019 Program are displayed below in Table 10. For clarity, and as discussed further above, we have inverted the SNFRM rate so that a higher rate represents better performance.

Table 10—Final FY 2019 SNF VBP Program Performance Standards *

Measure ID	Measure description	Achievement threshold	Benchmark
SNFRM	SNF 30-Day All-Cause Readmission Measure (NQF #2510)	0.79590	0.83601
* Note: Performance standards were calculated as of July 14, 2016 using CY 2015 data.

d. FY 2019 Performance Period and Baseline Period

i. Background

We refer readers to the FY 2016 SNF PPS final rule (80 FR 46422) for discussion of the considerations that we intend to take into account when specifying a performance period under the SNF VBP Program. We also explained our view that the SNF VBP Program necessitates adoption of a baseline period, similar to those adopted under the Hospital VBP Program and ESRD QIP, which we would use to establish performance standards and measure improvement.

We received public comments on this topic, and we refer readers to the FY 2016 SNF PPS final rule for a summary of those comments and our responses. We considered those comments when developing our performance and baseline period proposals for this proposed rule.

ii. Proposed FY 2019 Performance Period

In considering various performance periods that could apply for the FY 2019 SNF VBP Program, we recognized that we must balance the length of the performance period used to collect quality measure data and the amount of data needed to calculate reliable, valid measure rates with the need to finalize a performance period through notice and comment rulemaking. We therefore proposed to adopt CY 2017 (January 1, 2017 through December 31, 2017) as the performance period for the FY 2019 SNF VBP Program, with a 90-day run out period immediately thereafter for claims processing, based on the following considerations.

We strive to link performance furnished by SNFs as closely as possible to the payment year to ensure clear connections between quality measurement and value-based payment. We also strive to measure performance using a sufficiently reliable population of patients that broadly represent the total care provided by SNFs. As such, we anticipate that our annual performance period end date must provide sufficient time for SNFs to submit claims for the patients included in our measure population. Based on past experience with claims processing in other quality reporting and value-based purchasing programs, this time lag between care delivered to patients who are included in readmission measures and application of a payment consequence linked to reporting or performance on those measures has historically been close to 1 year. We also recognize that other factors contribute to the delay between data collection and payment impacts, including: The processing time needed to calculate measure rates using multiple sources of claims needed for statistical modeling; time for determining achievement and improvement scores; time for providers to review their measure rates and included patients; and processing time needed to determine whether a payment adjustment needs to be made to a provider's reimbursement rate under the applicable PPS based on its performance. Further, our preference is to adopt at least a 12-month period as the performance period, consistent with our view that using a full year's performance period provides sufficient levels of data accuracy and reliability for scoring SNF performance on the SNFRM and SNFPPR. We also believe that adopting a 12-month period for the performance period supports the direction provided of section 1888(g)(3) of the Act that the quality measures specified under the SNF VBP Program shall be designed to achieve a high level of reliability and validity. Specifically, we believe using a full year of claims data better ensures that the variation found among SNF performance on the measures is due to real differences between SNFs, and not within-facility variation due to issues such as seasonality. Additionally, we believe that adopting 12-month performance and baseline periods enables us to measure SNFs' performance on the specified measures in sequence, which we believe is necessary in order to measure SNFs on both achievement and improvement, as required by section 1888(h)(3)(B) of the Act.

Finally, we also considered the time necessary to calculate SNF-specific performance on the SNFRM after the conclusion of the performance period and to develop and provide SNF VBP scoring reports, including the requirement under section 1888(h)(7) of the Act that we inform each SNF of the adjustments to the SNF's payments as a result of the program not later than 60 days prior to the FY involved. Based on the requirements and concerns discussed above, we believe a 12-month time period is the only operationally feasible performance period for the SNF VBP Program.

We invited public comments on this proposal, and we respond to them in the next section.

iii. Proposed FY 2019 Baseline Period

As we have done in the Hospital VBP Program and the ESRD QIP, we proposed to adopt a baseline period for use in the SNF VBP Program.

We proposed to adopt calendar year 2015 claims (January 1, 2015 through December 31, 2015) as the baseline period for the FY 2019 SNF VBP Program and to use that baseline period as the basis for calculating performance standards. We stated that, as with the performance period, we will allow for a 90-day claims run out following the last date of discharge (December 31, 2015) before incorporating the 2015 claims in our database into the measure calculation.

We welcomed public comment on this proposal. The comments we received on this topic, as well as the comments that we received on the proposed performance period, with their responses, appear below.

Comment: One commenter supported our baseline and performance period proposals, stating their appreciation that we proposed a performance period as close to the payment period as possible.

Response: We thank the commenter for the support and agree. When developing these policies, we attempted to balance the length of the performance period with its proximity to the payment period, and we believe we have appropriately balanced those two factors.

Comment: One commenter was concerned about the delay between quality measurement and incentive payments or penalties, stating that providers need a clear link between practice and outcomes.

Response: As explained previously in this section, we believe that the proposed performance period is as close to the payment period as we can implement practically given the time necessary for claims submission and processing, as well as for scoring under the Program.

Comment: One commenter recommended that we expand the performance period for low-volume SNFs (which the commenter defined as SNFs having less than 25 stays) to 24 months, and that we exclude from the program SNFs that have fewer than 25 stays during the 2-year performance period. The commenter stated that this suggested exemption's effects would be insignificant on SNFs' scores in the aggregate, pointing to analysis that a similarly-structured 20-stay exclusion would only exempt about 7.4 percent of SNFs and just 1 percent of stays. The commenter noted that increasing the minimum stays count to 25 would increase the number of exempted SNFs to approximately 9.2 percent of all SNFs and about 1.6 percent of Medicare SNF stays, but also noted that expanding the performance period for low-volume SNFs would reduce the number of exempted SNFs and stays to 4.8 percent and 0.4 percent respectively. The commenter believed that these relatively low numbers of exempted SNFs and stays are justifiable since those SNFs are likely serving isolated areas or providing specialized care.

Response: We are sensitive to the effects the SNF VBP could have on beneficiaries' access to SNF care, and especially how the program might affect access to SNF care in rural and low-volume facilities.

However, while we appreciate the commenters' intent to ensure as broad participation as possible in the Program, we do not believe that a separate performance period for low-volume SNFs is feasible. Under section 1888(h)(3)(C) of the Act, we are required to establish and announce performance standards for a fiscal year not later than 60 days prior to the beginning of the performance period for that fiscal year. We do not believe we would comply with that requirement by establishing a longer performance period for certain SNFs. In addition, because we would not know which SNFs would have had fewer than 25 stays in their measure denominator until after the performance period concluded, it would be impossible for us to have provided the appropriate notice to those SNFs as required under section 1888(h)(3)(C) of the Act. Moreover, unless we established a separate baseline period for low-volume SNFs, we would be comparing performance and baseline periods of different durations, which raises questions about the validity of those performance comparisons over time. Further, we do not believe that a separate 24-month baseline period is appropriate, as it would require wholly separate calculations of measured performance using an additional year's claims data, which is both time-consuming and costly. Finally, we do not believe that low-volume SNFs are penalized by participating in the Program. The measures of readmissions adopted under the Program include an adjustment that reduces variability in low-volume SNFs' measured performance called “shrinkage estimation,” and we believe that this adjustment ensures that the measures are sufficiently reliable for the Program's purposes. However, we will continue to test and evaluate the Program's measures and will take this recommendation under consideration prior to transitioning from the SNFRM to the proposed SNFPPR measure in the SNF VBP Program.

After consideration of the public comments that we received, we are finalizing our proposals to adopt CY 2015 as the baseline period for the FY 2019 SNF VBP Program, and CY 2017 as the performance period for the same Program year.

e. SNF VBP Performance Scoring

i. Background

We refer readers to the FY 2016 SNF PPS final rule (80 FR 46422 through 46425) for a discussion of other Medicare VBP scoring methodologies, including the methodologies used by the Hospital VBP Program and HAC Reduction Program. We also discussed policy considerations related to the Hospital Readmission Reduction Program and the ESRD QIP in the performance standards section of that final rule (80 FR 46420 through 46421). We also discussed the potential application of an exchange function (80 FR 46424 through 46425) to translate SNF performance scores into value-based incentive payments under the SNF VBP Program.

We considered those issues, as well as comments we received on these issues, when developing our performance scoring policy below.

ii. SNF VBP Program Scoring Methodology

Section 1888(h)(4)(A) of the Act requires the Secretary develop a methodology for assessing the total performance of each SNF based on the performance standards established under section 1888(h)(3) of the Act for the measure applied under section 1888(h)(2) of the Act. Section 1888(h)(3)(B) of the Act further requires that these performance standards include levels of achievement and improvement and that, in calculating a facility's SNF performance score, the Secretary use the higher of either improvement or achievement.

After carefully reviewing and evaluating a number of scoring methodologies for the SNF VBP Program, we proposed to adopt a scoring model for the SNF VBP Program similar conceptually to that used by the Hospital VBP Program and the ESRD QIP, with certain modifications to allow us to better differentiate between SNFs' performance on the quality measures specified under the SNF VBP Program.^[25] We believe this hybrid appropriately accounts for the SNF VBP Program's statutory limitation to a single measure, will maintain consistency and alignment with other VBP programs already in place, and in doing so, will better enable SNFs to understand the SNF VBP Program. Specifically, we proposed to implement a 0 to 100-point scale for achievement scoring and a 0 to 90-point scale for improvement scoring. In addition, as discussed previously, we proposed to set the achievement threshold for the SNF VBP Program at the 25th percentile of SNF national performance on the quality measure during the baseline period rather than the 50th percentile achievement threshold used in the Hospital VBP Program, though as noted above, we also sought comment on whether or not we should consider adopting the 50th percentile or the 15th percentile.

We believe using wider scales of 0 to 100 points and 0 to 90 points instead of the 0 to 10 and 0 to 9 scales used in the Hospital VBP Program and ESRD QIP will allow us to calculate more granular performance scores for individual SNFs and provide greater differentiation between facilities' performance. We further believe that setting the achievement threshold for the SNF VBP Program at the 25th percentile of national SNF performance on the quality measure during the baseline period is preferable to the Hospital VBP Program's achievement threshold of the 50th percentile of national facility performance for this Program because it accounts for the statutory requirement that the SNF VBP Program include only one quality measure at a time. Unlike the Hospital VBP Program, which contains many measures across multiple domains, the SNF VBP Program is limited by statute to a single quality measure at a time. As a result, a hospital participating in the Hospital VBP Program could perform below the 50th percentile of national performance on one or more measures without experiencing a dramatic drop in its Total Performance Score because the hospital's performance on other measures would contribute to its total performance score. By contrast, if the SNF VBP Program used an achievement threshold of the 50th percentile of national SNF performance, approximately one-half of all SNFs nationwide would automatically receive 0 achievement points assuming no national improvement trends between baseline and performance periods. While these SNFs could still receive improvement points, we believe it is preferable to set a lower achievement threshold that would award the majority of SNFs at least some achievement points, thereby enabling us to differentiate performance among the lower-performing half of SNFs and enabling SNFs to continually increase their achievement score based on higher levels of performance. As stated above, as more data becomes available, we will continue to assess the appropriateness of this achievement threshold for the SNF VBP program and, if necessary, propose to refine these standards' definitions and calculation methodologies to better incentivize the provision of high-quality care.

For these reasons, we proposed to adopt the following scoring methodology beginning with the FY 2019 SNF VBP Program.

(a) Scoring of SNF Performance on the SNFRM

Because the SNF VBP Program uses only one measure to incentivize and assess facility performance and improvement, we believe it is important to ensure that SNFs and the public are able to understand these measure scores easily. SNFRM rates represent the percentage of qualifying patients at a facility that were readmitted within the risk window for the measure. As a result, lower SNFRM rates indicate lower rates of readmission, and are therefore an indicator of higher quality care. For example, a SNFRM rate of 0.14159 means that approximately 14.2 percent of qualifying patients discharged from that SNF were readmitted during the risk window.

We understand that the use of a “lower is better” rate could cause confusion among SNFs and the public. Therefore, we proposed to calculate scores under the Program by first inverting SNFRM rates using the following calculation:

SNFRM Inverted Rate = 1−Facility's SNFRM Rate

This calculation inverts SNFs' SNFRM rates such that higher SNFRM performance reflects better performance on the SNFRM. As a result, the same SNFRM rate presented above (0.14159) would result in a SNFRM inverted rate of 0.85841, which means that approximately 86 percent of qualifying patients discharged from that SNF were not readmitted during the risk window. We believe this inversion is important to incentivize improvement in a clear and understandable manner, and will also simplify public reporting of SNF performance for use in consumer, family, and caregiver decision-making. Further, under this proposal, all SNFRM inverted rates would be rounded to the fifth significant digit.

(b) Scoring SNFs' Performance Based on Achievement

We proposed that a SNF would earn an achievement score of 0 to 100 points based on where its performance on the specified measure fell relative to the achievement threshold (which we proposed above to define for the quality measures specified under the SNF VBP program as the 25th percentile of SNF performance on the quality measure during the applicable baseline period) and the benchmark (which we proposed to define as the mean of the top decile of SNF performance on the measure during the baseline period). As with the Hospital VBP Program, we proposed to award points to SNFs based on their performance as follows:

• If a SNF's SNFRM inverted rate was equal to or greater than the benchmark, the SNF would receive 100 points for achievement;
• If a SNF's SNFRM inverted rate was less than the achievement threshold (that is, the lower bound of the achievement range), the SNF would receive 0 points for achievement.
• If a SNF's SNFRM inverted rate was equal to or greater than the achievement threshold, but less than the benchmark, we would award between 0 and 100 points to the SNF according to the following formula:

The results of this formula would be rounded to the nearest whole number.

The SNF achievement score would therefore range between 0 and 100 points, with a higher achievement score indicating higher performance.

(c) Scoring SNF Performance Based on Improvement

We proposed that a SNF would earn an improvement score of 0 to 90 points based on how much its performance on the specified measure during the performance period improved from its performance on the measure during the baseline period. Under this proposal, a unique improvement range would be established for each SNF that defines the distance between the SNF's baseline period score and the national benchmark for the measure (which we propose to define as the mean of the top decile of SNF performance on the measure during the baseline period). We would then calculate a SNF improvement score for each SNF depending on its performance period score:

• If the SNF's performance period score was equal to or lower than its improvement threshold, the SNF would receive 0 points for improvement.
• If the SNF's performance period score was equal to or higher than the benchmark, the SNF would receive 90 points for improvement.
• If the SNF's performance period score was greater than its improvement threshold, but less than the benchmark, we would award between 0 and 90 points for improvement according to the following formula:

The results of this formula would be rounded to the nearest whole number.

(d) Establishing SNF Performance Scores

Consistent with sections 1888(h)(3)(B) and 1888(h)(4)(A) of the Act, we proposed to use the higher of a SNF's achievement and improvement scores to serve as the SNF's performance score for a given year of the SNF VBP Program. The resulting SNF performance score would be used as the basis for ranking SNF performance on the quality measures specified under the SNF VBP Program and establishing the value-based incentive payment percentage for each SNF for a given FY.

(e) Examples of the Proposed FY 2019 SNF VBP Program Scoring Methodology

In the proposed rule, we provided two examples to illustrate the proposed scoring methodology for the FY 2019 SNF VBP Program using hypothetical SNFs A, B, and C. The benchmark calculated for the SNFRM for all of these hypotheticals is 0.83915 (the mean of the top decile of SNF performance on the SNFRM in 2014), and the achievement threshold is 0.79551 (the 25th percentile of national SNF performance on the SNFRM in 2014). We noted that, as discussed previously, our proposal for scoring SNF performance on the SNFRM inverts the measure rates so that a higher rate represents better performance.

Figure AA shows the scoring for SNF A. SNF A's SNFRM rate of 0.15025 means that approximately 15 percent of qualifying patients discharged from SNF A were readmitted during the 30-day risk window. Under the proposed SNFRM scoring methodology, SNF A's SNFRM inverted rate would be calculated as follows:

Facility a SNFRM Inverted Rate = 1−0.15025

As a result of this calculation, Facility A's SNFRM inverted rate would be 0.84975 on the SNFRM for the performance period. This result indicates that approximately 85 percent of SNF A's qualifying patients were not readmitted during the 30-day risk window. Because SNF A's SNFRM inverted rate of 0.84975 exceeds the benchmark (that is, the mean of the top decile of facility performance, or 0.83915), SNF A would receive 100 points for achievement. Because SNF A has earned the maximum number of points possible for the SNFRM, its improvement score would not be calculated.

Figure BB shows the scoring for SNF B. As can be seen below, SNF B's performance on the SNFRM went from 0.21244, for a SNFRM inverted rate of 0.78756 (below the achievement threshold) in the baseline period to 0.18322, for a SNFRM inverted rate of 0.81668 (above the achievement threshold) in the performance period. Applying the achievement scoring methodology proposed above, SNF B would earn [49] achievement points for this measure, calculated as follows:

However, because SNF B's performance during the performance period is greater than its performance during the baseline period, but below the benchmark, we would calculate an improvement score as well. According to the improvement scale, based on SNF B's improved SNFRM inverted rate from 0.78756 to 0.81668, SNF B would receive 51 improvement points, calculated as follows:

In Figure CC, SNF C's performance on the SNFRM drops from 0.19487, for a SNFRM inverted rate of 0.80513, in the baseline period to 0.21148, for a SNFRM inverted rate 0.78852, in the performance period (a decline of 0.01661). Because this SNF's performance during the performance period is lower than the achievement threshold of 0.79551, it receives 0 points based on achievement. It would also receive 0 points for improvement, because its performance during the performance period is lower than its performance period during the baseline period. In this example, SNF C would receive 0 points for its SNF performance score.

The comments we received on this topic, with their responses, appear below.

Comment: One commenter supported the proposed scoring methodology, characterizing it as a reasonable approach that appropriately rewards achievement more than improvement.

Response: We thank the commenter for this feedback and agree. We believe the proposed scoring methodology complies with the Program's statutory requirement to score SNFs on both achievement and improvement while reserving the maximum scores for SNFs that are high achievers.

Comment: Some commenters appreciated our proposal to invert SNFs' performance rates on readmission measures to show that higher performance is better, particularly given the requirement to rank SNFs under the program.

Response: We thank the commenters for this feedback.

Comment: Some commenters supported the proposed 0 to 100 scoring approach, and called on us to monitor performance over time to ensure that the scores continue to reflect meaningful differences in care. Other commenters noted the proposed methodology's similarity to the HVBP program and expressed their support accordingly. Commenters also supported our proposed improvement scoring methodology, expressing appreciation that we intend to award fewer improvement points than achievement points. Commenters agreed that including the improvement score creates strong incentives for all SNFs to improve over time.

Response: We thank the commenters for their support.

Comment: One commenter suggested that we consider two additional factors for scoring adjustments, including the best ways to encourage palliative care without harming performance scores and how to adjust for individuals with specialized conditions that present increased risks of hospitalizations.

Response: We do not believe that the Program will discourage palliative care because the Program's measures do not hold SNFs accountable for admissions to hospice or other forms of palliative care, and we believe that the measures' risk adjustment appropriately controls for variations related to individuals' clinical status. However, we will monitor the Program's effects on access to care, and if necessary, will consider additional adjustments in the future.

After consideration of the public comments that we received, we are finalizing the scoring methodology for the SNF VBP Program as proposed.

f. SNF Value-Based Incentive Payments

i. Background

Paragraphs (5), (6), (7), and (8) of section 1888(h) of the Act outline several requirements for value-based incentive payments under the SNF VBP Program. Section 1888(h)(5)(A) of the Act requires that the Secretary increase the adjusted Federal per diem rate for skilled nursing facilities by the value-based incentive payment amount determined under section 1888(h)(5)(B) of the Act. That amount is to be determined by the product of the adjusted federal per diem rate and the value-based incentive payment percentage specified under section 1888(h)(5)(C) of the Act for each SNF for a FY.

Section 1888(h)(5)(C) of the Act requires that the value-based incentive payment percentage be based on the SNF performance score and must be appropriately distributed so that the highest-ranked SNFs receive the highest payments, the lowest-ranked SNFs receive the lowest payments, and that the payment rate for services furnished by SNFs in the lowest 40 percent of the rankings be less than would otherwise apply. Finally, the total amount of value-based incentive payments must be greater than or equal to 50 percent, but not greater than 70 percent, of the total amount of the reductions to payments for the FY specified under section 1888(h)(6) of the Act, as estimated by the Secretary. As discussed further below, we will propose to adopt in future rulemaking an exchange function to ensure that the total amount of value-based incentive payments made under the program each year meets those criteria.

Section 1888(h)(7) of the Act requires the Secretary, not later than 60 days prior to the fiscal year involved, to inform each SNF of the adjustments to its Medicare payments for services furnished by the SNF during the FY. Section 1888(h)(8) of the Act requires that the value-based incentive payment and payment reduction only apply for the FY involved, and not be taken into account in making payments to a SNF in a subsequent year.

We received a number of comments on incentive payments that will be made under the Program.

Comment: Several commenters recommended that we disburse the maximum 70 percent of payments withheld from SNFs as value-based incentive payments, stating that the larger the incentive, the greater the behavioral change. Commenters believed that making the largest amount of funds available would have the greatest impact on changing care practices.

Response: We thank commenters for this feedback. We will address the topic of value-based incentive payments under the Program in future rulemaking. We agree with commenters that the Program's incentive payments should be substantial enough to promote quality improvement through changing care practices.

Comment: One commenter stated that the SNF VBP Program should be budget-neutral, and suggested that we should reconsider the 50 to 70 percent payback to facilities under the Program.

Response: Section 1888(h)(5)(C)(ii)(III) of the Act requires that the total amount of value-based incentive payments available under the Program for a fiscal year range from between 50 percent and 70 percent of the total amount of the reductions to the adjusted Federal per diem rates otherwise applicable to skilled nursing facilities for that fiscal year, as estimated by the Secretary. As a result, we do not believe we have the authority to make the SNF VBP Program budget-neutral, or to vary the total amount that we will disburse in value-based incentive payments beyond the 50 to 70 percent range specified under the statute.

ii. Request for Comment on Exchange Function

As we discussed in the FY 2016 SNF PPS final rule (80 FR 46424 through 46425), we use a linear exchange function to translate a hospital's Total Performance Score under the Hospital VBP Program into the percentage multiplier to be applied to each Medicare discharge claim submitted by the hospital during the applicable FY. We intend to adopt a similar methodology to translate SNF performance scores into value-based incentive payment percentages under the SNF VBP Program. When considering that methodology, we sought public comments on the appropriate form and slope of the exchange function to determine how best to reward high performance and encourage SNFs to improve the quality of care provided to Medicare beneficiaries. As illustrated in Figure DD, we considered the following four mathematical exchange function options: Straight line (linear); concave curve (cube root function); convex curve (cube function); and S-shape (logistic function).

We received numerous public comments on the FY 2016 SNF PPS proposed rule, and we sought further public comments to inform our policies on this topic. We requested additional public comments on the specific form of the exchange function that we should propose in the future, including any additional forms beyond the four examples that we have illustrated above, and any considerations we should take into account when selecting an exchange function form that would best support quality improvement in SNFs.

Additionally, we will determine the precise slope of the exchange function after the performance period has concluded, because the distribution of SNFs' performance scores will form the basis for value-based incentive payments under the program. However, two additional considerations will affect the exchange function's slope. As required in section 1888(h)(5)(C)(ii)(II)(cc) of the Act, SNFs in the lowest 40 percent of the ranking determined under paragraph (4)(B) must receive a payment that is less than the payment rate for such services that would otherwise apply. Additionally, as described in this section, section 1888(h)(5)(C)(ii)(III) of the Act requires that the total amount of value-based incentive payments under the Program be greater than or equal to 50 percent, but not greater than 70 percent, of the total amount of reductions to SNFs' payments for the FY, as estimated by the Secretary. We intend to ensure that both of these requirements, as well as all other statutory requirements under the Program, are fulfilled when we specify the exchange function's slope.

We invited public comments on this topic. The comments we received on this topic, with their responses, appear below.

Comment: Commenter offered several principles for us to consider when developing our exchange function proposals in the future. The commenter suggested that top performing SNFs should receive an increase in their Medicare rates, that we should maximize the number of SNFs that do not receive a cut in their rates, that we should allow for continuous improvement, even for SNFs that are already high performers, and that differences in rehospitalization scores should be tied to meaningful differences in incentive payments. The commenter recommended that we adopt the logistic function and recommended against the cube root function, stating that the former balances incentives for low and high performers and that the latter creates very little incentive for performance improvement.

Response: We thank the commenter for this feedback, and we will take it into account as we develop proposals for the exchange function in the future.

g. SNF VBP Reporting

i. Confidential Feedback Reports

Section 1888(g)(5) of the Act requires that we provide quarterly confidential feedback reports to SNFs on their performance on the measures specified under sections 1888(g)(1) and (2) of the Act. Section 1888(g)(5) of the Act also requires that we begin providing those reports on October 1, 2016.

In order to meet the statutory deadline, we are developing the feedback reports, operational systems, and implementation guidance related to those reports. We intend to provide these reports to SNFs via the QIES system CASPER files currently used by SNFs to report quality performance.

We invited public comments on the appropriateness of the QIES system, and any considerations we should take into account when designing and providing these feedback reports. The comments we received on this topic, with their responses, appear below.

Comment: One commenter supported our proposal to use the QIES system to deliver feedback reports to SNFs. The commenter suggested that we provide these reports in a spreadsheet-based format to allow data aggregation within organizations.

Response: We thank the commenter for this feedback.

Comment: One commenter requested that trade organizations and other organizations that represent the interests of SNFs be provided access to SNFs' quarterly feedback reports. The commenter believed that these organizations can assure that SNF VBP data affecting each SNF will be protected and only shared with representatives for that particular SNF. The commenter noted that many SNFs are members of larger organizations, and that allowing further data distribution would enable these organizations to aggregate these reports rather than manually enter data voluntarily provided by each SNF. Commenter also requested that we provide a national data file with SNF VBP performance to these organizations that can help disseminate performance information to individual SNFs or their parent organizations.

Response: Section 1888(g)(5) of the Act requires us to provide confidential feedback reports to SNFs. We do not believe that we have the authority to share those confidential feedback reports with other entities.

Comment: One commenter requested that we consider using the QIES system to provide real-time data updates, or as close to real-time updates as possible. Commenter noted that we update our MDS data weekly to capture SNFs' most current measure rates in order to facilitate quality improvement efforts and suggested that we could do something similar with Part A claims and the Program's measures.

Response: Although we agree that SNFs would benefit from receiving the most up-to-date information as possible, it is not operationally feasible to provide SNFs with real-time data updates at this time. Unlike MDS data, claims-based measures require significant time to compute and are based on large pools of data. While we will, as described above, provide quarterly confidential feedback reports, we do not believe more frequent updates are possible at this time.

Comment: One commenter suggested several data elements that we could consider including in SNFs' quarterly reports, including readmission counts during and after the Part A stay, names of beneficiaries triggering readmissions, number of readmissions by PPR diagnosis, predicted and expected rates used to calculate the SSR for the prior rolling 12-month window, and national rates used to calculate achievement and improvement scores.

Response: We thank the commenter for this feedback. As we continue the Program's implementation, we will refine the quarterly reports in accordance with SNFs' feedback, and will take these suggestions into consideration.

ii. Proposed Two-Phase SNF VBP Data Review and Correction Process

(a) Background

Section 1888(g)(6) of the Act requires the Secretary to establish procedures to make public performance information on the measures specified under paragraphs (1) and (2) of such section. The procedures must ensure that a SNF has the opportunity to review and submit corrections to the information that will be made public for the facility prior to its being made public. This public reporting is also required by statute to begin no later than October 1, 2017. Additionally, section 1888(h)(9) of the Act requires the Secretary to make available to the public information regarding SNFs' performance under the SNF VBP Program, specifically including each SNF's performance score and the ranking of SNFs for each fiscal year.

Accordingly, we proposed to adopt a two-phase review and correction process for (1) SNFs' measure data that will be made public under section 1888(g)(6) of the Act, which will consist of each SNFs' performance on the measures specified under sections 1888(g)(1) and (2) of the Act, and (2) SNFs' performance information that will be made public under section 1888(h)(9) of the Act.

(b) Phase One: Review and Correction of SNFs' Quality Measure Information

We view the quarterly confidential feedback reports described previously in this section, as one possible means to provide SNFs an opportunity to review and provide corrections to their performance information. However, collecting SNF measure data and calculating measure performance scores takes a number of months following the end of a measurement period. Because it is not feasible to provide SNFs with an updated measure rate for each quarterly report or engage in review and corrections on a quarterly basis, we proposed to use one of the four reports each year to provide SNFs an opportunity to review their data slated for public reporting. In this specific quarterly report, we intend to provide SNFs: (1) A count of readmissions; (2) the number of eligible stays at the SNF; (3) the SNF's risk-standardized readmissions ratio; and (4) the national SNF measure performance rate. In addition, we intend to provide the patient-level information used in calculating the measure rate. However, we sought comment on what patient-level information would be most useful to SNFs and how we should make this information available if requested. We intend to address the topic of what specific information will be provided if requested in this specific quarterly report in future rulemaking, where we intend to propose a process for SNFs' requests for patient-level data. We intend to notify SNFs of this report's release via listserv email and posting on the QualityNet News portion of the Web site.

Therefore, we proposed to fulfill the statutory requirement that SNFs have an opportunity to review and correct information that is to be made public under section 1888(g)(6) of the Act by providing SNFs with an annual confidential feedback report that we intend to provide via the QIES system CASPER files. We further proposed that SNFs must, if they believe the report's contents to be in error, submit a correction request to [e-mail] with the following information:

• SNF's CMS Certification Number (CCN).
• SNF Name.
• The correction requested and the SNF's basis for requesting the correction. More specifically, the SNF must identify the error for which it is requesting correction, and explain its reason for requesting the correction. The SNF must also submit documentation or other evidence, if available, supporting the request. Additionally, any requests made during phase one of the proposed process will be limited to the quality measure information at issue.

We further proposed that SNFs must make any correction requests within 30 days of posting the feedback report via the QIES system CASPER files, not counting the posting date itself. For example, if we provide reports on October 1, 2017, SNFs must review those reports and submit any correction requests by October 31, 2017. We will not consider any requests for correction to quality measure data that are received after the close of the first phase of the proposed review and correction process. As discussed further in this section, any corrections sought during phase two of the proposed process will be limited to the SNF performance score calculation and the ranking.

We will review all timely phase one correction requests that we receive and will provide responses to SNFs that have requested corrections as soon as practicable.

(c) Phase Two: Review and Correction of SNF Performance Scores and Ranking

As required by section 1888(h)(7) of the Act, we intend to inform each SNF of its payment adjustments as a result of the SNF VBP Program not later than 60 days prior to the fiscal year involved. For the FY 2019 SNF VBP Program, we intend to notify SNFs of those payment adjustments via a SNF performance score report not later than 60 days prior to October 1, 2018. We intend to address the specific contents of that report in future rulemaking.

In that report, however, we also intend to provide SNFs with their SNF performance scores and ranking. By doing so, we intend to use the performance score report's provision to SNFs as the beginning of the second phase of the proposed review and correction process. By completing phase one, SNFs will have an opportunity to verify that their quality measure data are fully accurate and complete and as a result, phase two will be limited only to corrections to the SNF performance score's calculation and the SNF's ranking. Any requests to correct quality measure data that are received during phase two will be denied.

We intend to set out specific requirements for phase two of the proposed review and correction process in future rulemaking. To inform those proposals, we sought comments on what information would be most useful for us to provide to SNFs to facilitate their review of their SNF performance scores and ranking. As with the phase one process, we intend to adopt a 30-day time period for phase two review and corrections, beginning with the date on which we provide SNF performance score reports.

We invited public comments on this proposed two-phase review and correction process. The comments we received on this topic, with their responses, appear below.

Comment: One commenter only supported the 30-day deadline for correction requests if sufficient information is included in the quarterly reports. The commenter noted that SNFs may not be able to submit documentation or other evidence supporting a correction request within 30 days if they do not receive the names of the beneficiaries who were readmitted, when the readmission occurred, and the readmission diagnosis. Commenter appreciated that we may receive many correction requests, and suggested that we consider allowing corrections for missing data only annually, but corrections for when patients' admissions are listed incorrectly quarterly in order to streamline our reviews of correction requests. Another commenter requested that we provide SNF and hospital inpatient Part A claims to SNFs on a quarterly basis, both to facilitate quality improvement and correction requests. Commenter suggested that we could provide patient identifiable files to organizations that have a Business Associate Agreement with the SNF and allow the organizations to share data with the SNF. Commenter noted that many facilities do not have the capacity to analyze claims data, but many large organizations are working with SNFs to provide this service. Another commenter opposed the ability of SNFs to request data corrections in phase two of the proposed review and correction process unless all data in phase two is also included in the quarterly feedback reports in phase one, and the last quarterly report in phase one includes the final data used to calculate the rehospitalization score. Commenter explained that if SNFs will not be able to file correction requests based on phase two feedback reports, all of the data used to calculate the rehospitalization score needs to be in the phase one reports.

Response: We thank the commenters for this feedback. As we discuss further below in response to other comments, we are finalizing a policy whereby we will accept corrections on any quarterly report provided during a calendar year until the following March 31.

However, the feedback reports that we must provide to SNFs under the requirements at section 1888(g)(5) of the Act are specifically required to remain confidential. We do not believe that we have the authority to share those confidential feedback reports with other organizations than SNFs themselves. We note that SNFs are free to share their feedback reports with other organizations at their discretion.

We would like to clarify the distinction between the two phases of the proposed review and correction process. As we discussed in the proposed rule (81 FR 24255), the first phase is intended to allow SNFs to review and correct patient-level information that we used to calculate the measure rates. The second phase is intended to allow SNFs to review and correct only their performance scores and the ranking, not their measure rates. Although the two phases are separate, they will, taken together, provide SNFs with an opportunity to correct both the measure rates that are used to generate their performance scores and ranking, as well as their actual performance scores and ranking. We do not believe that we should conflate the two, or allow corrections to quality measure data (that is, phase one requests) during the phase two process, because the two phases are aimed at two separate purposes. We believe it to be necessary to finalize the claims data that SNFs will be able to correct in phase one so that those data may form the basis for performance calculations that SNFs will be able to review in phase two.

Comment: One commenter recommended that SNFs be provided access to the information used to calculate their rehospitalization scores and also information to estimate their adjustment factor based on the final exchange function. Commenter explained that SNFs will want to replicate their scores, so they will need their predicted rates, expected rates, national average, baseline period rates, and major “cut points” used to determine achievement and improvement points. The commenter also suggested that the ranking of achievement and improvement scores could be helpful to SNFs as well.

Response: We will take these comments into account as we develop the first quarterly feedback reports for SNFs, and look forward to additional feedback from SNFs after we provide them.

Comment: Commenter expressed support for the proposed review and corrections process

Response: We thank the commenter for their support.

Comment: Commenter supported our proposal to provide feedback reports to SNFs via the QIES system. However, the commenter did not support our plan to allow SNFs to seek corrections on an annual basis, and commenter recommended instead that we allow corrections on a quarterly basis with an annual deadline. The commenter suggested that the quarterly data that we provide should be sufficient to allow SNFs to verify the accuracy of their measured performance and suggested as a result that SNFs should be allowed to submit corrections quarterly.

Response: We understand the commenter's concern about the deadline following each quarterly confidential feedback report, and we will instead finalize a policy under which we will accept corrections to any quarterly report provided during a calendar year until the following March 31. We believe that this policy appropriately balances our desire to ensure that the measure data are sufficiently accurate with SNFs' need for sufficient information with which to evaluate the accuracy of those reports, and provides SNFs with more time to review each quarter's data than the 30 days that we initially proposed.

After consideration of the public comments that we received, we are finalizing the two-phase review and correction process as proposed, with the exception stated above that we will accept corrections to SNFs' quarterly confidential feedback reports during a calendar year until the following March 31.

iii. SNF VBP Public Reporting

Section 1888(h)(9)(A) of the Act requires that we make available to the public on the Nursing Home Compare Web site or its successor information regarding the performance of individual SNFs with respect to a FY, including the performance score for each SNF for the FY and each SNF's ranking, as determined under section 1888(h)(4)(B) of the Act. Additionally, section 1888(h)(9)(B) of the Act requires that we periodically post aggregate information on the SNF VBP Program on the Nursing Home Compare Web site or its successor, including the range of SNF performance scores, and the number of SNFs receiving value-based incentive payments and the range and total amount of those payments.

We intend to address this topic in future rulemaking. However, we invited public comments on the best means by which to display the SNF-specific and aggregate performance information for public consumption. The comments we received on this topic, with their responses, appear below.

Comment: Commenter supported public posting of SNFs performance scores, but not their rehospitalization rates, achievement or improvement scores. The commenter stated that achievement and improvement scores are not required to be posted publicly by statute and that they are not necessarily helpful to consumers. The commenter also stated against posting the risk adjusted SNFRM or SNFPPR rates, noting that these measures differ from other rehospitalization measures publicly posted by CMS.

Response: We thank the commenter for this feedback. We will propose details on public reporting of SNF VBP Program performance information in the future and will take these comments into account at that time.

Comment: Commenter supported posting of the aggregate value-based incentive payments, as well as the range of those payments and the number of SNFs receiving payment adjustments, but did not support posting individual SNF payments. The commenter noted that individual SNF payments are the product of rehospitalization scores, volume of admissions and patient case mix RUG payments, so actual payment adjustments could be confusing to the public.

Response: We thank the commenter for this feedback and agree that we will need to communicate clearly with the public about the information that we post publicly. We will take these comments into account when we propose details on public posting of SNF VBP payments information in the future.

iv. Ranking SNF Performance

Section 1888(h)(4)(B) of the Act requires ranking the SNF performance scores determined under paragraph (A) of such section from low to high. Additionally, and as discussed in this section, we are required to publish the ranking of SNF performance scores for a FY on Nursing Home Compare or a successor Web site.

To meet these requirements, we proposed to order SNF performance scores from low to high and publish those rankings on both the Nursing Home Compare and QualityNet Web sites. However, because SNF performance scores will not be calculated until after the performance period concludes after CY 2017 (that is, during CY 2018), and because SNFs must be provided their value-based incentive payment adjustments not later than 60 days prior to the FY involved, we intend to publish the ranking for FY 2019 SNF VBP payment implications after August 1, 2018.

We invited public comments on the most appropriate format and Web site for the ranking's publication. The comments we received on this topic, with their responses, appear below.

Comment: Commenter stated that any public posting of SNFs' ranking under the Program must be clearly indicated, and suggested that rank number 1 should be reserved for the SNF with the best rehospitalization score, not the worst score. Commenter explained that the public may be confused about the ranking unless clear and easy to understand information on the ranking's direction is posted. Commenter also supported our plan to post the ranking on the Nursing Home Compare Web site.

Response: We thank the commenter for this feedback and will take it into account as we develop the ranking that will be publicly posted. We agree with the commenter that we will need to be clear about what the ranking means when it is posted. We note that section 1888(h)(4)(B) of the Act directs that the ranking of SNF performance scores (not SNF rehospitalization rates) under the Program be ordered from low to high, and we intend to be as clear as possible about SNFs' placements on the ranking.

We will address this topic further in future rulemaking. We note that, because we will compute FY 2019 SNF performance scores after the completion of the performance period (finalized above as CY 2017), we will not publish the ranking or other SNF-specific performance information for the FY 2019 Program until at least the summer of CY 2018.

2. Skilled Nursing Facility (SNF) Quality Reporting Program (QRP)

a. Background and Statutory Authority

We seek to promote higher quality and more efficient health care for Medicare beneficiaries, and our efforts are furthered by QRPs coupled with public reporting of that information.

The Improving Medicare Post-Acute Care Transformation Act of 2014 (IMPACT Act) added section 1899B to the Act that imposed new data reporting requirements for certain PAC providers, including SNFs, and required that the Secretary implement a SNF quality reporting program (SNF QRP). Section 1888(e)(6)(B)(i)(II) of the Act requires that each SNF submit, for FYs beginning on or after the specified application date (as defined in section 1899B(a)(2)(E) of the Act), data on quality measures specified under section 1899B(c)(1) of the Act and data on resource use and other measures specified under section 1899B(d)(1) of the Act in a manner and within the time frames specified by the Secretary. In addition, section 1888(e)(6)(B)(i)(III) of the Act requires, for FYs beginning on or after October 1, 2018, that each SNF submit standardized patient assessment data required under section 1899B(b)(1) of the Act in a manner and within the time frames specified by the Secretary. Section 1888(e)(6)(A)(i) of the Act requires that, for FYs beginning with FY 2018, if a SNF does not submit data, as applicable, on quality and resource use and other measures in accordance with section 1888(e)(6)(B)(i)(II) of the Act and on standardized patient assessment in accordance with section 1888(e)(6)(B)(i)(III) of the Act for such FY, the Secretary must reduce the market basket percentage described in section 1888(e)(5)(B)(ii) of the Act by 2 percentage points. The SNF QRP applies to freestanding SNFs, SNFs affiliated with acute care facilities, and all non-CAH swing-bed rural hospitals.

We refer readers to the FY 2016 SNF PPS final rule (80 FR 46427 through 46429) for information on the requirements of the IMPACT Act

In the FY 2016 SNF PPS final rule, we finalized the general timeline and sequencing of activities under the SNF QRP. Please refer to the FY 2016 SNF PPS final rule (80 FR 46427 through 46429) for more information on these topics.

In addition, in implementing the SNF QRP and IMPACT Act requirements in the FY 2016 SNF PPS final rule, we established our approach for identifying cross-setting measures and processes for the adoption of measures including the application and purpose of the Measure Application Partnership (MAP) and the notice and comment rulemaking process. For more information on these topics, please refer to the FY 2016 SNF PPS final rule (80 FR 46427 through 46429).

b. General Considerations Used for Selection of Measures for the SNF QRP

We refer readers to the FY 2016 SNF PPS final rule (80 FR 46429 through 46431) for a detailed discussion of the considerations we apply in measure selection for the SNF QRP, such as alignment with the CMS Quality Strategy,^[26] which incorporates the three broad aims of the National Quality Strategy.^[27] Overall, we strive to promote high quality and efficiency in the delivery of health care to the beneficiaries we serve. Performance improvement leading to the highest quality health care requires continuous evaluation to identify and address performance gaps and reduce the unintended consequences that may arise in treating a large, vulnerable, and aging population. QRPs, coupled with public reporting of quality information, are critical to the advancement of health care quality improvement efforts. Valid, reliable, and relevant quality measures are fundamental to the effectiveness of our QRPs. Therefore, selection of quality measures is a priority for CMS in all of its QRPs.

In the FY 2017 SNF PPS proposed rule, we proposed to adopt for the SNF QRP one measure that we are specifying under section 1899B(c)(1)(C) of the Act to meet the Medication Reconciliation domain: (1) Drug Regimen Review Conducted with Follow-Up for Identified Issues—Post-Acute Care Skilled Nursing Facility Quality Reporting Program. Further, we proposed to adopt for the SNF QRP three measures to meet the resource use and other measure domains identified in section 1899B(d)(1) of the Act: (1) Medicare Spending per Beneficiary—Post-Acute Care Skilled Nursing Facility Quality Reporting Program; (2) Discharge to Community—Post Acute Care Skilled Nursing Facility Quality Reporting Program; and (3) Potentially Preventable 30-Day Post-Discharge Readmission Measure for Skilled Nursing Facility Quality Reporting Program.

In our development and specification of measures, we employ a transparent process in which we seek input from stakeholders and national experts and engage in a process that allows for pre-rulemaking input on each measure, as required by section 1890A of the Act.

To meet this requirement, we provided the following opportunities for stakeholder input. Our measure development contractor convened technical expert panels (TEPs) that included stakeholder experts and patient representatives on July 29, 2015 for the Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, on August 25, 2015, September 25, 2015, and October 5, 2015 for the Discharge to Community—PAC SNF QRP, on August 12 and 13, 2015 and October 14, 2015 for the Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP, and on October 29 and 30, 2015 for the Medicare Spending per Beneficiary measures. In addition, we released draft quality measure specifications for public comment on the Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP from September 18, 2015 to October 6, 2015, for the Discharge to Community—PAC SNF QRP from November 9, 2015 to December 8, 2015, for the Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP from November 2, 2015 to December 1, 2015, and for the Medicare Spending per Beneficiary measures from January 13, 2016 to February 5, 2016. Further, we implemented a public mailbox, [e-mail], for the submission of public comments. This PAC mailbox is accessible on our post-acute care quality initiatives Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-of-2014-Data-Standardization-and-Cross-Setting-MeasuresMeasures.html.

Additionally, we sought public input from the MAP PAC, Long-Term Care Workgroup during the annual in-person meeting held December 14 and 15, 2015. The final MAP report is available at http://www.qualityforum.org/​Publications/​2016/​02/​MAP_​2016_​Considerations_​for_​Implementing_​Measures_​in_​Federal_​Programs_​-_​PAC-LTC.aspx. The MAP is composed of multi-stakeholder groups convened by the NQF, our current contractor under section 1890(a) of the Act, tasked to provide input on the selection of quality and efficiency measures described in section 1890(b)(7)(B) of the Act.

The MAP reviewed each measure that we proposed in the proposed rule for use in the SNF QRP. For more information on the MAP, we refer readers to the FY 2016 SNF PPS final rule (80 FR 46430 through 46431). Further, for more information on the MAP's recommendations, we refer readers to the MAP 2015-2016 Considerations for Implementing Measures in Federal Programs public report at http://www.qualityforum.org/​Publications/​2016/​02/​MAP_​2016_​Considerations_​for_​Implementing_​Measures_​in_​Federal_​Programs_​-_​PAC-LTC.aspx.

We received a number of general comments on our measure selection process.

Comment: Many commenters supported the goals of the IMPACT Act, including the implementation of cross-setting measures across PAC settings. One of these commenters stated that the use of standardized and interoperable patient assessment data will allow for better cross-setting comparisons of quality and will support the development of better quality measures with uniform risk standardization. The commenter also recognized that the standardization of data collected across PAC settings is an ongoing process and will require continued refinement.

Response: We appreciate the commenters' support for the implementation of cross-setting measures across PAC settings as required by the IMPACT Act. We believe that standardizing patient assessment data will allow for the exchange of data among PAC providers in order to facilitate care coordination and improve patient outcomes.

Comment: Several commenters expressed concern with the compressed timeline in which CMS is adopting measures for the SNF QRP. Additionally, one commenter believes the “hurried pace” of the development process may lead to negative unintended consequences and may preclude stakeholder input. The commenter suggested that a less compressed comment period and implementation timeline provided would be less disruptive to measure development. Several commenters suggested that the measures be refined further prior to their implementation in the SNF QRP.

Response: We recognize the timeline and pace to implement the requirements of the IMPACT Act is ambitious. However, we have taken steps to ensure the scientific rigor of measure development, including testing measures under development and soliciting stakeholder feedback during both the measure development and rulemaking process. We have also worked to be responsive to stakeholder concerns about the length of various comment periods, and in response to those concerns, we have extended our public comment periods for measures under development on several occasions. We also encourage feedback through our IMPACT Act PAC Quality Initiative resource and feedback mailbox at [e-mail] or at the SNF QRP resource and feedback mailbox at [e-mail]. We intend to continually monitor, refine, and update all measures if necessary to ensure that they do not result in unintended consequences. With regard to refining measures prior to their implementation, we interpret this to refer to further refinement of the measures prior to adoption. We understand and agree that measures should be developed prior to adoption and have engaged in several activities to ensure further refinement which are described in the specifc measure sections below.

Comment: One commenter expressed concern that SNFs will be held responsible for outcomes of care when other care coordination arrangements such as Accountable Care Organizations, Medicare bundled payments, and Medicaid managed care arrangements for dual eligibles are available. The commenter believes that overlapping care coordination initiatives and SNF QRP measures will cause confusion and diffuse accountability for the outcomes of care. One commenter suggested streamlining measures to reduce the redundancy of reporting. Another commenter was concerned that SNFs would be confused by the various measures, and thought that there would be unintended consequences as a result.

Response: Although we recognize that there might be some overlap along the lines suggested by the commenters, the SNF QRP is being designed to assess the quality care specific furnished by SNFs to Medicare beneficiaires. We believe that this information will be important for quality improvement purposes. We will continue to provide outreach and education to SNFs including trainings and National Provider Calls to help them understand the requirements and measures adopted for the SNF QRP. We also appreciate the concern that SNF QRP measures be aligned to minimize reporting requirements when possible. We will nonetheless seek, where feasible, to align the SNF QRP with existing reporting requirements.

Comment: We received several comments regarding NQF endorsement of the proposed measures. One commenter voiced support of the measures and encouraged submission of the measures for NQF endorsement. Several commenters expressed concern about the lack of NQF endorsement for measures and suggested additional measure testing and development. One commenter requested that CMS provide a timeline for submission of the measures to NQF. Additionally, commenters recommended NQF endorsement prior to public reporting.

Response: We recognize the importance of consensus endorsement and, where possible, seek to adopt measures for the SNF QRP that are endorsed by the NQF. To the extent that we adopt measures under our exception authority, we intend to seek NQF-endorsement of those measures and will do so as soon as is feasible. Regardless of whether the measures are or are not NQF-endorsed at the time we adopt them, they have all been tested for reliability and validity, and we believe that the results of that testing support our conclusion that they are sufficiently reliable and valid to warrant their adoption in the SNF QRP. The results of our reliability and validity testing for these measures may be found in Measure Specifications for Measures Adopted in the FY 2017 SNF QRP Final Rule, posted on the CMS SNF QRP Web page at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

Comment: Several commenters stated that the NQF MAP committee did not support the proposed measures; instead, they recommended that we delay measure implementation until the measures are fully developed and tested and brought back to the MAP for further consideration. One commenter suggested that TEP members and other stakeholders who provided feedback in the measure development process did not support the measures moving forward without further testing.

Response: We interpret this comment to address the activities of the Measures Application Partnership, a multi-stakeholder partnership convened by NQF that provides input to the U.S. Department of Health and Human Services (HHS) on its selection of measures for certain Medicare programs. We would like to clarify that the MAP provided the recommendation of “encourage continued development” for the proposed measures. According to the MAP, the term “encourage continued development,” is applied when a measure addresses a critical program objective or promotes alignment but is in an earlier stage of development. In contrast, the MAP uses the phrase “do not support” when it does not support a measure at all.

Since the MAP recommendation of “encourage continued development” for the proposed measures during the December 2015 NQF-convened PAC LTC MAP meeting, we have further refined the measure specifications based on additional validity and reliability testing. Our efforts included: A pilot test in 12 post-acute care settings, including SNFs, to determine the feasibility of assessment items for use in calculation of the Drug Regimen Review Conducted with Follow-Up for Identified Issues measure and further development of risk-adjusted models for the Discharge to Community, Medicare Spending per Beneficiary and Potentially Preventable Readmissions measures. Additional information regarding testing that was performed since the MAP Meeting, TEP meetings, and public comment periods is further described below in our responses to comments on individual proposed measures.

For these reasons, we believe that the measures have been fully and robustly developed, and believe they are appropriate for implementation and should not be delayed.

Comment: One commenter expressed concern about a lack of consistency and comparability of measures across PAC settings and believed it inappropriate to compare performance across provider types due to the lack of appropriate risk adjustment. We also received comments from MedPAC conveying that findings from their work on a unified PAC payment system suggest overlap in where Medicare beneficiaries are treated for similar care in PAC settings. As a result of this work, MedPAC recommended that the IMPACT Act measures use a uniform definition, specification, and risk adjustment method to facilitate quality comparison across PAC settings to inform Medicare beneficiary choice, and so that Medicare can evaluate the value of services it pays for. MedPAC further noted that differences in rates should reflect differences in quality of care rather than differences in the way rates are constructed.

Response: For each of the proposed measures, we applied consistent models where feasible in order to develop their definitions, other technical specifications and approach to risk-adjustment.

However, there are nuances among the four PAC provider types which must be taken into account in order to address issues such as patient acuity and medical complexity. As a result, we have risk-adjusted measures and included provider-specific refinements. For example, for the Discharge to Community measure, risk adjustment for ventilator use is included in LTCH and SNF settings, but not IRF settings. We investigated the need for risk adjustment for ventilator use in IRFs, but found that less than 0.01 percent of the IRF population had ventilator use in the IRF. Given the low frequency of ventilator use in IRFs, any associated estimates would not be reliable; thus, ventilator use is not included as a risk adjuster in the IRF setting measure. We believe that the measures proposed for the SNF QRP will inform beneficiaries on the differences in quality rather than differences in measure construction because we have taken into account the factors necessary to ensure meaningful comparability within the SNF providers and as able, across the post-acute providers.

Comment: A number of commenters expressed concerns regarding the validity and reliability of IMPACT Act measures and encouraged us to analyze data to ensure comparability across post-acute care settings, prior to implementation.

Response: We have tested for validity and reliability all of the IMPACT Act measures, and the results of that testing is available in Measure Specifications for Measures Adopted in the FY 2017 SNF QRP Final Rule, posted on the CMS SNF QRP Web page at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html .

We intend to continue to monitor the reliability and validity of the SNF QRP measures, including whether the measures are reliable and valid for cross-setting purposes.

Comment: One commenter expressed concern that the proposed measures could adversely affect low-volume or rural SNFs. Another commenter expressed concerns about the ability to compare measure rates across facilities due to varying patient volumes, recommending the use of patient days as the denominator for SNF quality measures.

Response: We do not believe the proposed measures will adversely affect low-volume or rural SNFs. We wish to clarify that our measures and/or our proposals to implement these measures were designed to mitigate any potential impact that may be caused by low volume. For example, the statistical approach used for two of the claims-based measures incorporates a shrinkage estimator intended to ensure that smaller facilities are not vulnerable to rates driven by the influence of random variation in their raw rates. Additionally, for some of the measures, public reporting requirements exclude reporting of facilities with fewer than 25 resident stays during the reporting period. We would like to clarify that the quality, resource use and other measures in the SNF QRP are based on stay-level outcomes, not day-level outcomes. The measures examine events occurring at SNF discharge or after SNF discharge; therefore, the measures are based on number of discharges. For example, the proposed quality measure Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP would not be appropriate for data calculation on a daily basis. The data collected for this measure is at admission and discharge and reflects data recorded throughout the entire patient stay.

Comment: One commenter expressed concern that the proposed measures will incentivize SNFs to avoid admitting medically complex residents, which would result in unintended consequences.

Response: To mitigate the risk of creating incentives for SNFs to avoid admitting medically complex residents, who may be at higher risk for poor outcomes and higher costs, we have included factors related to medical complexity in the risk adjustment methodology used in our measures. We also intend to conduct ongoing monitoring to assess for potential unintended consequences associated with the implementation of these measures.

c. Policy for Retaining SNF QRP Measures Adopted for Future Payment Determinations

In the FY 2016 SNF PPS final rule (80 FR 46431 through 46432), we finalized our policy for measure removal and also finalized that when we adopt a measure for the SNF QRP for a payment determination, this measure will be automatically retained in the SNF QRP for all subsequent payment determinations unless we propose to remove, suspend, or replace the measure. We did not propose any new policies related to measure retention or removal in the FY 2017 SNF PPS proposed rule. For further information on how measures are considered for removal, suspension, or replacement, please refer to the FY 2016 SNF PPS final rule (80 FR 46431 through 46432).

d. Process for Adoption of Changes to SNF QRP Measures

In the FY 2016 SNF PPS final rule (80 FR 46432), we finalized our policy pertaining to the process for adoption of non-substantive and substantive changes to SNF QRP measures. We did not propose to make any changes to this policy.

e. Quality Measures Previously Finalized for Use in the SNF QRP

The SNF QRP quality measures for the FY 2018 payment determinations and subsequent years are presented in Table 11. Measure specifications for the previously adopted measures adapted from non-SNF settings are available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html under the downloads section at the bottom of the page.

Table 11—Quality Measures Previously Finalized for Use in the SNF QRP

Measure title and NQF #	SNF PPS final rule	Data collection start date	Annual payment determination: Initial and subsequent APU years
Percent of Residents or Patients with Pressure Ulcers That Are New or Worsened (Short Stay) (NQF #0678)	Adopted in the FY 2016 SNF PPS Final Rule (80 FR 46433 through 46440)	October 1, 2016	FY 2018 and subsequent years.
Application of the NQF-endorsed Percent of Residents Experiencing One or More Falls with Major Injury (Long Stay) (NQF #0674)	Adopted in the FY 2016 SNF PPS Final Rule (80 FR 46440 through 46444)	October 1, 2016	FY 2018 and subsequent years.
Application of Percent of Long-Term Care Hospital Patients with an Admission and Discharge Functional Assessment and a Care Plan That Addresses Function (NQF #2631)	Adopted in the FY 2016 SNF PPS Final Rule (80 FR 46444 through 46453)	October 1, 2016	FY 2018 and subsequent years.

f. SNF QRP Quality, Resource Use and Other Measures for FY 2018 Payment Determinations and Subsequent Years

For the FY 2018 payment determination and subsequent years, in addition to the quality measures identified in Table 11 that we are retaining under our policy described in section V.B.3., we proposed to adopt three new measures for the SNF QRP. These three measures were developed to meet the requirements of the IMPACT Act. They are: (1) Medicare Spending per Beneficiary—PAC SNF QRP; (2) Discharge to Community—PAC SNF QRP; and (3) Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP. Through the use of standardized quality measures and standardized data, the intent of the Act, among other obligations, is to enable interoperability and access to longitudinal information for such providers to facilitate coordinated care, improved outcomes, and overall quality comparisons. The measures are described in more detail below.

For the risk adjustment of the resource use and other measures, we understand the important role that sociodemographic status plays in the care of patients. However, we continue to have concerns about holding providers to different standards for the outcomes of their patients of diverse sociodemographic status because we do not want to mask potential disparities or minimize incentives to improve the outcomes of disadvantaged populations. We routinely monitor the impact of sociodemographic status on providers' results on our measures.

The NQF is currently undertaking a 2-year trial period in which new measures and measures undergoing maintenance review will be assessed to determine if risk-adjusting for sociodemographic factors is appropriate. For 2-years, NQF will conduct a trial of temporarily allowing inclusion of sociodemographic factors in the risk-adjustment approach for some performance measures. At the conclusion of the trial, NQF will issue recommendations on future permanent inclusion of sociodemographic factors. During the trial, measure developers are expected to submit information such as analyses and interpretations as well as performance scores with and without sociodemographic factors in the risk adjustment model.

Furthermore, the Office of the Assistant Secretary for Planning and Evaluation (ASPE) is conducting research to examine the impact of sociodemographic status on quality measures, resource use, and other measures under the Medicare program as directed by the IMPACT Act. We will closely examine the findings of the ASPE reports and related Secretarial recommendations and consider how they apply to our quality programs at such time as they are available.

We invited public comment on how socioeconomic and demographic factors should be used in risk adjustment for the resource use and other measures. The comments we received on this topic, with their responses, appear below.

Comment: Several commenters supported the inclusion of sociodemographic status adjustment in quality measures, resource use, and other measures. Commenters suggested that failure to account for these patient characteristics could penalize SNFs for providing care to a more medically-complex and socioeconomically disadvantaged patient population and affect provider performance. Some commenters expressed concerns about standardization and interoperability of the measures as it pertains to risk-adjusting, particularly for SDS characteristics. Many commenters recommended incorporating socioeconomic factors as risk-adjustors for the measures and several commenters suggested conducting additional testing and/or NQF endorsement prior to implementation of these measures. In addition, many commenters recommended including functionality as an additional risk-adjustment factor, and several commenters suggested risk-adjustment for cognitive impairment. One commenter recommended varied standards for patient outcomes with individuals of diverse SDS statuses.

A few commenters, including MedPAC, did not support risk-adjustment of measures by SES or SDS status. One commenter did not support risk-adjustment because it can hide disparities and create different standards of care for SNFs based on the demographics in the facility. MedPAC stated that risk adjustment can hide disparities in care and suggested that risk-adjustment reduces pressure on providers to improve quality of care for low-income Medicare beneficiaries. Instead, MedPAC supported peer provider group comparisons with providers of similar low-income beneficiary populations. Another commenter stated that SDS factors should not be included in measures that assess the resident outcome during a SNF stay, but should only be considered for measures evaluating care after the SNF discharge.

Response: We appreciate the considerations and suggestions conveyed in relation to the measures and the importance in balancing appropriate risk adjustment along with ensuring access to high quality care. We note that in the measures that are risk adjusted we do take into account characteristics associated with medical complexity, as well as factors such as age where appropriate to do so. For those cross-setting post-acute measures such as those intended to satisfy the IMPACT Act domains that use the patient assessment-based data elements for risk adjustment, we have either made such items standardized, or intend to do so as feasible. With regard to the incorporation of additional factors, such as cognitive impairment and function, we have and will continue to take such factors into account, which would include further testing as part of our ongoing measure development monitoring activities. As discussed previously, we intend to seek NQF endorsement for our measures.

We also received suggestions pertaining to the incorporation of socioeconomic factors as risk-adjustors for the measures, including in those measures that pertain to after the resident was discharged from the SNF, additional testing and/or NQF endorsement prior to implementation of these measures, and comments that pertain to potential consequences associated with such risk adjustors and alternative approaches to grouping comparative data. We wish to reiterate that as previously discussed, NQF is currently undertaking a 2-year trial period in which new measures and measures undergoing maintenance review will be assessed to determine if risk-adjusting for sociodemographic factors is appropriate. This trial entails temporarily allowing inclusion of sociodemographic factors in the risk-adjustment approach for some performance measures. At the conclusion of the trial, NQF will issue recommendations on future permanent inclusion of sociodemographic factors. During the trial, measure developers are encouraged to submit information such as analyses and interpretations as well as performance scores with and without sociodemographic factors in the risk adjustment model. Several measures developed by CMS have been brought to NQF since the beginning of the trial. CMS, in compliance with NQF's guidance, has tested sociodemographic factors in the measures' risk models and made recommendations about whether or not to include these factors in the endorsed measure. We intend to continue engaging in the NQF process as we consider the appropriateness of adjusting for sociodemographic factors in our outcome measures.

Furthermore, the Office of the Assistant Secretary for Planning and Evaluation (ASPE) is conducting research to examine the impact of sociodemographic status on quality measures, resource use, and other measures under the Medicare program as directed by the IMPACT Act. We will closely examine the findings of the ASPE reports and related Secretarial recommendations and consider how they apply to our quality programs at such time as they are available.

i. Measure That Addresses the IMPACT Act Domain of Resource Use and Other Measures: Total Estimated MSPB-PAC SNF QRP

We proposed an MSPB-PAC SNF QRP measure for inclusion in the SNF QRP for the FY 2018 payment determination and subsequent years. Section 1899B(d)(1)(A) of the Act requires the Secretary to specify resource use measures, including total estimated Medicare spending per beneficiary, on which PAC providers consisting of SNFs, Inpatient Rehabilitation Facilities (IRFs), Long-Term Care Hospitals (LTCHs), and Home Health Agencies (HHAs) are required to submit necessary data specified by the Secretary.

Rising Medicare expenditures for post-acute care as well as wide variation in spending for these services underlines the importance of measuring resource use for providers rendering these services. Between 2001 and 2013, Medicare PAC spending grew at an annual rate of 6.1 percent and doubled to $59.4 billion, while payments to inpatient hospitals grew at an annual rate of 1.7 percent over this same period.^[28] A study commissioned by the Institute of Medicine found that variation in PAC spending explains 73 percent of variation in total Medicare spending across the United States.^[29]

We reviewed the NQF's consensus-endorsed measures and were unable to identify any NQF-endorsed resource use measures for PAC settings. As such, we proposed this MSPB-PAC SNF QRP measure under the Secretary's authority to specify non—NQF-endorsed measures under section 1899B(e)(2)(B) of the Act. Given the current lack of resource use measures for PAC settings, our MSPB-PAC SNF QRP measure would provide valuable information to SNF providers on their relative Medicare spending in delivering services to approximately 1.7 million Medicare beneficiaries.^[30]

The MSPB-PAC SNF QRP episode-based measure would provide actionable and transparent information to support SNF providers' efforts to promote care coordination and deliver high quality care at a lower cost to Medicare. The MSPB-PAC SNF QRP measure holds SNF providers accountable for the Medicare payments within an “episode of care” (episode), which includes the period during which a patient is directly under the SNF's care, as well as a defined period after the end of the SNF treatment, which may be reflective of and influenced by the services furnished by the SNF. MSPB-PAC SNF QRP episodes, constructed according to the methodology described below, have high levels of Medicare spending with substantial variation. In FY 2014, Medicare FFS beneficiaries experienced 1,534,773 MSPB-PAC SNF QRP episodes. The mean payment-standardized, risk-adjusted episode spending for these episodes is $26,279. There is substantial variation in the Medicare payments for these MSPB-PAC SNF QRP episodes—ranging from approximately $6,090 at the 5th percentile to approximately $60,050 at the 95th percentile. This variation is partially driven by variation in payments occurring after SNF treatment.

Evaluating Medicare payments during an episode creates a continuum of accountability between providers that should improve post-treatment care planning and coordination. While some stakeholders throughout the measure development process supported the MSPB-PAC measures and felt that measuring Medicare spending was critical for improving efficiency, others believed that resource use measures did not reflect quality of care in that they do not take into account patient outcomes or experience beyond those observable in claims data. However, SNFs involved in the provision of high-quality PAC care as well as appropriate discharge planning and post-discharge care coordination would be expected to perform well on this measure since beneficiaries would likely experience fewer costly adverse events (for example, avoidable hospitalizations, infections, and emergency room usage). Further, it is important that the cost of care be explicitly measured so that, in conjunction with other quality measures, we can publicly report which SNFs provide high quality care at lower cost.

We developed a MSPB-PAC measure for each of the four PAC settings. We proposed an LTCH-specific MSPB-PAC measure in the FY 2017 IPPS/LTCH proposed rule (81 FR 25216 through 25220), an IRF-specific MSBP-PAC measure in the FY 2017 IRF proposed rule (81 FR 24197 through 24201), a SNF-specific MSPB-PAC measure in the FY 2017 SNF proposed rule (81 FR 24258 through 24262), and a HHA-specific MSBP-PAC measure in the CY 2017 HH proposed rule (81 FR 43760 through 43764). The four setting-specific MSPB-PAC measures are closely aligned in terms of episode construction and measure calculation. Each MSPB-PAC measure assesses Medicare Part A and Part B spending within an episode, and the numerator and denominator are defined similarly. However, setting-specific measures allow us to account for differences between settings in payment policy, the types of data available, and the underlying health characteristics of beneficiaries.

The MSPB-PAC measures mirror the general construction of the inpatient prospective payment system (IPPS) hospital MSPB measure, which was adopted for the Hospital IQR Program beginning with the FY 2014 program, and was implemented in the Hospital VBP Program beginning with the FY 2015 program. The measure was endorsed by the NQF on December 6, 2013 (NQF #2158).^[31] The hospital MSPB measure evaluates hospitals' Medicare spending relative to the Medicare spending for the national median hospital during a hospital MSPB episode. It assesses Medicare Part A and Part B payments for services performed by hospitals and other healthcare providers within a hospital MSPB episode, which is comprised of the periods immediately prior to, during, and following a patient's hospital stay.^[32 33] Similarly, the MSPB-PAC measures assess all Medicare Part A and Part B payments for fee-for-service (FFS) claims with a start date during the episode window (which, as discussed in this section, is the time period during which Medicare FFS Part A and Part B services are counted towards the MSPB-PAC SNF QRP episode). There are differences between the MSPB-PAC measures and the hospital MSPB measure to reflect differences in payment policies and the nature of care provided in each PAC setting. For example, the MSPB-PAC measures exclude a limited set of services (for example, for clinically unrelated services) provided to a beneficiary during the episode window, while the hospital MSPB measure does not exclude any services.

MSPB-PAC episodes may begin within 30 days of discharge from an inpatient hospital as part of a patient's trajectory from an acute to a PAC setting. A SNF stay beginning within 30 days of discharge from an inpatient hospital would therefore be included once in the hospital's MSPB measure, and once in the SNF provider's MSPB-PAC measure. Aligning the hospital MSPB and MSPB-PAC measures in this way creates continuous accountability and aligns incentives to improve care planning and coordination across inpatient and PAC settings.

We sought and considered the input of stakeholders throughout the measure development process for the MSPB-PAC measures. We convened a TEP consisting of 12 panelists with combined expertise in all of the PAC settings on October 29 and 30, 2015 in Baltimore, Maryland. A follow-up email survey was sent to TEP members on November 18, 2015 to which seven responses were received by December 8, 2015. The MSPB-PAC TEP Summary Report is available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​Downloads/​Technical-Expert-Panel-on-Medicare-Spending-Per-Beneficiary.pdf. The measures were also presented to the MAP Post-Acute Care/Long-Term Care (PAC/LTC) Workgroup on December 15, 2015. As the MSPB-PAC measures were under development, there were three voting options for members: Encourage continued development, do not encourage further consideration, and insufficient information.^[34] The MAP PAC/LTC workgroup voted to “encourage continued development” for each of the MSPB-PAC measures.^[35] The MAP PAC/LTC workgroup's vote of “encourage continued development” was affirmed by the MAP Coordinating Committee on January 26, 2016.^[36] The MAP's concerns about the MSPB-PAC measures, as outlined in their final report “MAP 2016 Considerations for Implementing Measures in Federal Programs: Post-Acute Care and Long-Term Care” and Spreadsheet of Final Recommendations, were taken into consideration during the measure development process and are discussed as part of our responses to public comments, described below.^[37 38]

Since the MAP's review and recommendation of continued development, CMS continued to refine risk adjustment models and conduct measure testing for the IMPACT Act measures consistent with the MAP's recommendations. The IMPACT Act measures are consistent with the information submitted to the MAP and support the scientific acceptability of these measures for use in quality reporting programs.

In addition, a public comment period, accompanied by draft measures specifications, was open from January 13 to 27, 2016 and extended to February 5. A total of 45 comments on the MSPB-PAC measures were received during this 3.5 week period. The comments received also covered each of the MAP's concerns as outlined in their Final Recommendations.^[39] The MSPB-PAC Public Comment Summary Report is available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​Downloads/​2016_​03_​24_​mspb_​pac_​public_​comment_​summary_​report.pdf and the MSPB-PAC Public Comment Supplementary Materials are available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​Downloads/​2016_​03_​24_​mspb_​pac_​public_​comment_​summary_​report_​supplementary_​materials.pdf: These documents contain the public comments, along with our responses including statistical analyses. The MSPB-PAC SNF QRP measure, along with the other MSPB-PAC measures, as applicable, will be submitted for NQF endorsement when feasible.

To calculate the MSPB-PAC SNF QRP measure for each SNF provider, we first define the construction of the MSPB-PAC SNF QRP episode, including the length of the episode window as well as the services included in the episode. Next, we apply the methodology for the measure calculation. The specifications are discussed further in this section. More detailed specifications for the MSPB-PAC measures, including the MSPB-PAC SNF QRP measure, are available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

The comments we received on this topic, with their responses, appear below.

Comment: Several commenters expressed concern about the lack of NQF endorsement for proposed measures; some believed that the measure should not be finalized until NQF endorsement is obtained.

Response: We thank the commenters for their concern regarding the lack of NQF endorsement and refer readers to section III.D.2.b. where we also discuss this topic.

Comment: Several commenters noted the NQF MAP committee did not endorse the proposed measure, believing that the measure should not be finalized until the support of the MAP is obtained.

Response: We appreciate the comments about the NQF MAP committee, and direct readers to section III.D.2.b. where we also discuss this topic.

Comment: Some commenters recommended the use of uniform single MSPB-PAC measure that could be used to compare providers' resource use across settings, but they also recognized that we do not have a uniform PPS for all the PAC settings currently. In the absence of a single PAC PPS, they recommend a single MSPB-PAC measure for each setting that could be used to compare providers within a setting. Under a single measure, the episode definitions, service inclusions/exclusions, and risk adjustment methods would be the same across all PAC settings.

Response: We thank the commenters. The four separate MSPB-PAC measures reflect the unique characteristics of each PAC setting and the population it serves. The four setting specific MSPB-PAC measures are defined as consistently as possible across settings given the differences in the payment systems for each setting, and types of patients served in each setting. We have taken into consideration these differences and aligned the specifications, such as episode definitions, service inclusions/exclusions and risk adjustment methods for each setting, to the extent possible while ensuring the accuracy of the measures in each PAC setting.

Each of the measures assess Medicare Part A and Part B spending during the episode window which begins upon admission to the provider's care and ends 30 days after the end of the treatment period. The service-level exclusions are harmonized across settings. The definition of the numerator and denominator is the same across settings. However, specifications differ between settings when necessary to ensure that the measures accurately reflect patient care and align with each setting's payment system. For example, Medicare pays LTCHs and IRFs a stay-level payment based on the assigned MS-LTC-DRG and CMG, respectively, while SNFs are paid a daily rate based on the RUG level, and HHA providers are reimbursed based on a fixed 60-day period for standard home health claims. While the definition of the episode window is consistent across settings and is based on the period of time that a beneficiary is under a given provider's care, the duration of the treatment period varies to reflect how providers are reimbursed under the PPS that applies to each setting. The length of the post-treatment period is consistent between settings. There are also differences in the services covered under the PPS that applies to each setting: For example, durable medical equipment, prosthetics, orthotics, and supplies (DMEPOS) claims are covered LTCH, IRF, and SNF services but are not covered HHA services. This affects the way certain first-day service exclusions are defined for each measure.

We recognize that beneficiaries may receive similar services as part of their overall treatment plan in different PAC settings, but believe that there are some important differences in beneficiaries' care profiles that are difficult to capture in a single measure that compares resource use across settings.

Also, the risk adjustment models for the MSPB-PAC measures share the same covariates to the greatest extent possible to account for patient case mix. However, the measures also incorporate additional setting-specific information where available to increase the predictive power of the risk adjustment models. For example, the MSPB-PAC LTCH QRP risk adjustment model uses MS-LTC-DRGs and Major Diagnostic Categories (MDCs) and the MSPB-PAC IRF QRP model includes Rehabilitation Impairment Categories (RICs). The HH and SNF settings do not have analogous variables that directly reflect a patient's clinical profile.

We will continue to work towards a more uniform measure across settings as we gain experience with these measures, and we plan to conduct further research and analyses about comparability of resource use measures across settings for clinically similar patients, different treatment periods and windows, risk adjustment, service exclusions, and other factors.

Comment: A few commenters noted that the MSPB-PAC measures are resource use measures that are not a standalone indicator of quality.

Response: We appreciate the comment regarding the proposed MSPB-PAC measures as resource use measures. The MSPB-PAC SNF QRP measure is one of four QRP measures that were proposed in the FY 2017 SNF PPS proposed rule for inclusion in the SNF QRP: In addition to the MSPB-PAC SNF QRP measure, these proposed measures were the Discharge to Community—PAC SNF QRP measure (81 FR 24262 through 24264), the Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP (81 FR 24264 through 24267), and the Drug Regimen Review Conducted With Follow-Up for Identified Issues—PAC SNF QRP measure (81 FR 24267 through 24269). As part of the SNF QRP, the MSPB-PAC SNF QRP measure will be paired with quality measures; we direct readers to section III.D.2.e. for a discussion of quality measures previously finalized for use in the SNF QRP. We believe it is important that the cost of care be explicitly measured so that, in conjunction with other quality measures, we can publicly report which SNF providers are involved in the provision of high quality care at lower cost.

Comment: One commenter expressed concern over the short timeframe available for stakeholder input.

Response: We appreciate the feedback regarding the timing issues related to IMPACT Act implementation. It is our intent to move forward with IMPACT Act implementation in a manner in which the measure development process continues to be transparent, and includes input and collaboration from experts, the PAC provider community, and the public at large. It is of the utmost importance to us to continue to engage stakeholders, including providers as well as residents and their families, throughout the measure development lifecycle through their participation in our measure development public comment periods, the pre-rulemaking process, TEPs convened by our measure development contractors, open door forums and other opportunities. We have provided multiple opportunities for stakeholder input on the MSPB-PAC measures, including the TEP, NQF MAP public comment period and in-person meeting, pre-rulemaking public comment period, and 60-day public comment period on the proposed SNF QRP rule. A summary of TEP proceedings, the MSPB-PAC Public Comment Summary Report and MSPB-PAC Public Comment Supplementary Materials are available at the links provided above. We thank all stakeholders for their thoughtful feedback on and engagement with the measure development and rulemaking process.

(a) Episode Construction

An MSPB-PAC SNF QRP episode begins at the episode trigger, which is defined as the patient's admission to a SNF. The admitting facility is the attributed provider, for whom the MSPB-PAC SNF QRP measure is calculated. The episode window is the time period during which Medicare FFS Part A and Part B services are counted towards the MSPB-PAC SNF QRP episode. Because Medicare FFS claims are already reported to the Medicare program for payment purposes, SNF providers would not be required to report any additional data to CMS for calculation of this measure. Thus, there would be no additional data collection burden from the implementation of this measure.

The episode window is comprised of a treatment period and an associated services period. The treatment period begins at the trigger (that is, on the day of admission to the SNF) and ends on the day of discharge from that SNF. Readmissions to the same facility occurring within 7 or fewer days do not trigger a new episode, and instead are included in the treatment period of the original episode. When two sequential stays at the same SNF occur within 7 or fewer days of one another, the treatment period ends on the day of discharge for the latest SNF stay. The treatment period includes those services that are provided directly or reasonably managed by the SNF provider that are directly related to the beneficiary's care plan. The associated services period is the time during which Medicare Part A and Part B services (with certain exclusions) are counted towards the episode. The associated services period begins at the episode trigger and ends 30 days after the end of the treatment period. The distinction between the treatment period and the associated services period is important because clinical exclusions of services may differ for each period. Certain services are excluded from the MSPB-PAC SNF QRP episodes because they are clinically unrelated to SNF care, and/or because SNF providers may have limited influence over certain Medicare services delivered by other providers during the episode window. These limited service-level exclusions are not counted towards a given SNF provider's Medicare spending to ensure that beneficiaries with certain conditions and complex care needs receive the necessary care. Certain services that are determined to be outside of the control of a SNF provider include planned hospital admissions, management of certain preexisting chronic conditions (for example, dialysis for end-stage renal disease (ESRD), and enzyme treatments for genetic conditions), treatment for preexisting cancers, organ transplants, and preventive screenings (for example, colonoscopy and mammograms). Exclusion of such services from the MSPB-PAC SNF QRP episode ensures that facilities do not have disincentives to treat patients with certain conditions or complex care needs.

An MSPB-PAC episode may begin during the associated services period of an MSPB-PAC SNF QRP episode in the 30 days post-treatment. One possible scenario occurs where a SNF provider discharges a beneficiary who is then admitted to an IRF within 30 days. The IRF claim would be included once as an associated service for the attributed provider of the first MSPB-PAC SNF QRP episode and once as a treatment service for the attributed provider of the second MSPB-PAC IRF QRP episode. As in the case of overlap between hospital and PAC episodes discussed earlier, this overlap is necessary to ensure continuous accountability between providers throughout a beneficiary's trajectory of care, as both providers share incentives to deliver high quality care at a lower cost to Medicare. Even within the SNF setting, one MSPB-PAC SNF QRP episode may begin in the associated services period of another MSPB-PAC SNF QRP episode in the 30 days post-treatment. The second SNF claim would be included once as an associated service for the attributed SNF provider of the first MSPB-PAC SNF QRP episode and once as a treatment service for the attributed SNF provider of the second MSPB-PAC SNF QRP episode. Again, this ensures that SNF providers have the same incentives throughout both MSPB-PAC SNF QRP episodes to deliver quality care and engage in patient-focused care planning and coordination. If the second MSPB-PAC SNF QRP episode were excluded from the second SNF provider's MSPB-PAC SNF QRP measure, that provider would not share the same incentives as the first SNF provider of first MSPB-PAC SNF QRP episode. The MSPB-PAC SNF QRP measure was designed to benchmark the resource use of each attributed provider against what its spending is expected to be as predicted through risk adjustment. As discussed further in this section, the measure takes the ratio of observed spending to expected spending for each episode and then takes the average of those ratios across all of the attributed provider's episodes. The measure is not a simple sum of all costs across a provider's episodes, thus mitigating concerns about double counting.

The comments we received on this topic, with their responses, appear below.

Comment: One commenter expressed concern about how claims are counted and attributed to providers.

Response: We appreciate the commenter's concern, but note that there were no further specifics detailing the nature of this concern. We designed the attribution process to hold SNF providers accountable for the Medicare payments within an “episode of care” (episode), which includes the period during which a patient is directly under the SNF's care, as well as a defined period after the end of the SNF treatment. An MSPB-PAC SNF QRP episode begins at the episode trigger, which is defined as the patient's admission to a SNF. The admitting facility is the attributed provider, for whom the MSPB-PAC SNF QRP measure is calculated. The episode window is the time period during which Medicare FFS Part A and Part B services are counted towards the MSPB-PAC SNF QRP episode. The standardized allowed amounts on the claims for those services are summed to calculate observed episode spending. Further details on episode construction and attribution, as they relate to how claims are counted are in the MSPB-PAC Measure Specifications, a link for which has been provided above.

(b) Measure Calculation

Medicare payments for Part A and Part B claims for services included in MSPB-PAC SNF QRP episodes, defined according to the methodology above, are used to calculate the MSPB-PAC SNF QRP measure. Measure calculation involves determination of the episode exclusions, the approach for standardizing payments for geographic payment differences, the methodology for risk adjustment of episode spending to account for differences in patient case mix, and the specifications for the measure numerator and denominator.

(i) Exclusion Criteria

In addition to service-level exclusions that remove some payments from individual episodes, we exclude certain episodes in their entirety from the MSPB-PAC SNF QRP measure to ensure that the MSPB-PAC SNF QRP measure accurately reflects resource use and facilitates fair and meaningful comparisons between SNF providers. The episode-level exclusions are as follows:

• Any episode that is triggered by a SNF claim outside the 50 states, DC, Puerto Rico, and U.S. Territories.
• Any episode where the claim(s) constituting the attributed SNF provider's treatment have a standard allowed amount of zero or where the standard allowed amount cannot be calculated.
• Any episode in which a beneficiary is not enrolled in Medicare FFS for the entirety of a 90-day lookback period (that is, a 90-day period prior to the episode trigger) plus episode window (including where the beneficiary dies), or is enrolled in Part C for any part of the lookback period plus episode window.
• Any episode in which a beneficiary has a primary payer other than Medicare for any part of the 90-day lookback period plus episode window.
• Any episode where the claim(s) constituting the attributed SNF provider's treatment include at least one related condition code indicating that it is not a prospective payment system bill.

The comments we received on this topic, with their responses, appear below.

Comment: One commenter expressed general support for the list of episode-level exclusions proposed for the MSPB-PAC SNF QRP measure.

Response: We thank the commenter for its support.

(ii) Standardization and Risk Adjustment

Section 1899B(d)(2)(C) of the Act requires that the MSPB-PAC measures are adjusted for the factors described under section 1886(o)(2)(B)(ii) of the Act, which include adjustment for factors such as age, sex, race, severity of illness, and other factors that the Secretary determines appropriate. Medicare payments included in the MSPB-PAC SNF QRP measure are payment standardized and risk-adjusted. Payment standardization removes sources of payment variation not directly related to clinical decisions and facilitates comparisons of resource use across geographic areas. We proposed to use the same payment standardization methodology that was used in the NQF-endorsed hospital MSPB measure. This methodology removes geographic payment differences, such as wage index and geographic practice cost index (GPCI), incentive payment adjustments, and other add-on payments that support broader Medicare program goals including indirect graduate medical education (IME) and hospitals serving a disproportionate share of uninsured patients (DSH).^[40]

Risk adjustment uses patient claims history to account for case-mix variation and other factors that affect resource use but are beyond the influence of the attributed SNF provider. To assist with risk adjustment, we create mutually exclusive and exhaustive clinical case mix categories using the most recent institutional claim in the 60 days prior to the start of the MSPB-PAC SNF QRP episode. The beneficiaries in these clinical case mix categories have a greater degree of clinical similarity than the overall SNF patient population, and allow us to more accurately estimate Medicare spending. Our MSPB-PAC SNF QRP measure, adapted for the SNF setting from the NQF-endorsed hospital MSPB measure uses a regression framework with a 90-day hierarchical condition category (HCC) lookback period and covariates including the clinical case mix categories, HCC indicators, age brackets, indicators for originally disabled, ESRD enrollment, and long-term care status, and selected interactions of these covariates where sample size and predictive ability make them appropriate. We sought and considered public comment regarding the treatment of hospice services occurring within the MSPB-PAC SNF QRP episode window. Given the comments received, we proposed to include the Medicare spending for hospice services but risk adjust for them, such that MSPB-PAC SNF QRP episodes with hospice services are compared to a benchmark reflecting other MSPB-PAC SNF QRP episodes with hospice services. We believe this strikes a balance between the measure's intent of evaluating Medicare spending and ensuring that providers do not have incentives against the appropriate use of hospice services in a patient-centered continuum of care.

We understand the important role that sociodemographic factors, beyond age, play in the care of patients. However, we continue to have concerns about holding providers to different standards for the outcomes of their patients of diverse sociodemographic status because we do not want to mask potential disparities or minimize incentives to improve the outcomes of disadvantaged populations. We will monitor the impact of sociodemographic status on providers' results on our measures.

The NQF is currently undertaking a 2-year trial period in which new measures and measures undergoing maintenance review will be assessed to determine if risk-adjusting for sociodemographic factors is appropriate. For 2 years, NQF will conduct a trial of temporarily allowing inclusion of sociodemographic factors in the risk-adjustment approach for some performance measures. At the conclusion of the trial, NQF will issue recommendations on future permanent inclusion of sociodemographic factors. During the trial, measure developers are expected to submit information such as analyses and interpretations as well as performance scores with and without sociodemographic factors in the risk adjustment model.

Furthermore, the Office of the Assistant Secretary for Planning and Evaluation (ASPE) is conducting research to examine the impact of sociodemographic status on quality measures, resource use, and other measures under the Medicare program as required by the IMPACT Act. We will closely examine the findings of the ASPE reports and related Secretarial recommendations and consider how they apply to our quality programs at such time as they are available.

While we conducted analyses on the impact of age by sex on the performance of the MSPB-PAC SNF QRP risk-adjustment model, we did not propose to adjust the MSPB-PAC SNF QRP measure for socioeconomic factors. As this MSPB-PAC SNF QRP measure would be submitted for NQF endorsement, we prefer to await the results of this trial and study before deciding whether to risk adjust for socioeconomic factors. We will monitor the results of the trial, studies, and recommendations. We invited public comment on how socioeconomic and demographic factors should be used in risk adjustment for the MSPB-PAC SNF QRP measure. The comments we received on this topic, with their responses, appear below.

Comment: Several commenters recommended that the risk adjustment model for the MSPB-PAC SNF QRP measure include variables for SES/SDS factors. A commenter recommended that a “fairer” approach than using SES/SDS factors as risk adjustment variables would be to compare resource use levels that have not been adjusted for SES/SDS factors across peer providers (that is, providers with similar shares of beneficiaries with similar SES characteristics).

Response: With regard to the suggestions that the model include sociodemographic factors and the suggestion pertaining to an approach with which to convey data comparisons, we refer readers to section III.D.2.f. where we also discuss these topics.

Comment: Some commenters recommended that additional variables be included in risk adjustment to better capture clinical complexity. A few commenters suggested the inclusion of functional and cognitive status and other patient assessment data. Commenters recommended that additional variables should include obesity, amputations, CVAs (hemiplegia/paresis), and ventilator status.

Response: We thank the commenters for their suggestions. The HCC indicators that are already included in the risk adjustment model account for amputations, hemiplegia, and paresis. We believe that the other risk adjustment variables adequately adjust for ventilator dependency and obesity by accounting for HCCs, clinical case mix categories, and prior inpatient and ICU length of stay.

We recognize the importance of accounting for beneficiaries' functional and cognitive status in the calculation of predicted episode spending. We considered the potential use of functional status information in the risk adjustment models for the MSPB-PAC measures. However, we decided to not include this information derived from current setting-specific assessment instruments given the move towards standardized data as mandated by the IMPACT Act. We will revisit the inclusion of functional status in these measures' risk adjustment models in the future when the standardized functional status data mandated by the IMPACT Act-mandated become available. Once they are available, we will take a gradual and systematic approach in evaluating how they might be incorporated. We intend to implement any changes if appropriate based on testing.

Comment: One commenter expressed concern that the measures will give incentive to SNFs to avoid admitting medically complex residents, which would result in unintended consequences.

Response: To mitigate the risk of creating incentives for SNFs to avoid admitting medically complex residents, who may be at higher risk for poor outcomes and higher costs, we have included factors related to medical complexity in the risk adjustment methodology for the MSPB-PAC SNF QRP measure. We also intend to conduct ongoing monitoring to assess for potential unintended consequences associated with the implementation of this measure.

Comment: One commenter recommended that SNFs providing palliative care should be treated the same way as SNFs providing hospice care.

Response: We thank the commenter for their concern and note that the risk adjustment model used in the MSPB-PAC SNF QRP measure does not adjust for the type of care provided in the SNF, such as hospice-type or palliative care services. However, the episode spending for beneficiaries who receive hospice care within the episode window is benchmarked only against the expected episode-level spending of similar beneficiaries. This is achieved through the inclusion of a risk adjustment indicator for beneficiaries for whom Medicare pays hospice claims during the episode window. We adjust for beneficiaries with hospice claims as these patients have different characteristics from those who are not receiving hospice care services; one requirement of eligibility for hospice services under Part A is that beneficiaries must be terminally ill with a life expectancy of 6 months or less. In contrast, palliative care services can encompass any comfort care services (such as pain medication) at any stage of treatment of illness or condition. Given the challenges of identifying the range of services that could indicate palliative care and the wide variety of patients receiving this type of care, we believe that adjusting for the presence of hospice claims and not palliative care services supports the goal of providing fair comparisons between providers.

(iii) Measure Numerator and Denominator

The MPSB-PAC SNF QRP measure is a payment-standardized, risk-adjusted ratio that compares a given SNF provider's Medicare spending against the Medicare spending of other SNF providers within a performance period. Similar to the hospital MSPB measure, the ratio allows for ease of comparison over time as it obviates the need to adjust for inflation or policy changes.

The MSPB-PAC SNF QRP measure is calculated as the ratio of the MSPB-PAC Amount for each SNF provider divided by the episode-weighted median MSPB-PAC Amount across all SNF providers. To calculate the MSPB-PAC Amount for each SNF provider, one calculates the average of the ratio of the standardized episode spending over the expected episode spending (as predicted in risk adjustment), and then multiplies this quantity by the average episode spending level across all SNF providers nationally. The denominator for a SNF provider's MSPB-PAC SNF QRP measure is the episode-weighted national median of the MSPB-PAC Amounts across all SNF providers. An MSPB-PAC SNF QRP measure of less than 1 indicates that a given SNF provider's resource use is less than that of the national median SNF provider during a performance period. Mathematically, this is represented in equation (A) below:

Where

• Y_ij = attributed standardized spending for episode i and provider j
• Y_ij = expected standardized spending for episode i and provider j, as predicted from risk adjustment
• n_j = number of episodes for provider j
• n = total number of episodes nationally
• i ε {I_j} = all episodes i in the set of episodes attributed to provider j.

The comments we received on this topic, with their responses, appear below.

Comment: A few commenters expressed concern about comparing mean to median values leading to inaccurate measure calculation. Commenters requested clarification on proposed values to ensure fairness.

Response: We appreciate the commenters' concerns. As noted in the MSPB-PAC Public Comment Summary Report for which a link has been provided above, we clarify that a provider's MSPB-PAC Amount is the average of observed over expected spending across a provider's episodes. Comparing a provider's MSPB-PAC Amount to the national median MSPB-PAC Amount does not affect the rank ordering of providers, and will therefore not lead to inaccurate measure calculations because the attributed provider's rank relative to the median will not change.

Comment: One commenter recommended including payments made by the SNF to non-Medicare payers so that providers cannot simply shift costs to other payers.

Response: We thank the commenter for the input and note that this measure only includes beneficiaries who are continuously enrolled in Medicare FFS for the entirety of a 90-day lookback period (that is, a 90-day period prior to the episode trigger) plus episode window. We do not have the ability to assess payments made by private payers or track beneficiary coinsurance or deductibles paid for plans outside of Medicare. CMS will monitor this issue using administrative claims data from Medicare as a part of ongoing measure monitoring and evaluation.

Comment: One commenter recommended that a geographic-specific (for example, state or regional) median should be used instead of the national median, citing differences in cost, patient population, and regulation.

Response: We appreciate the commenter's input. As noted in the proposed rule, (81 FR 24260), we proposed to use the same payment standardization methodology as that used in the NQF-endorsed hospital MSPB measure to account for variation in Medicare spending. This methodology removes geographic payment differences, such as wage index and geographic practice cost index (GPCI), incentive payment adjustments, and other add-on payments that support broader Medicare program goals including indirect graduate medical education (IME) and hospitals serving a disproportionate share of uninsured patients (DSH). We believe that this approach accounts for the differences that the commenter raises while also maintaining consistency with the NQF-endorsed hospital MSPB measure's methodology for addressing regional variation through payment standardization.

(c) Data Sources

The MSPB-PAC SNF QRP resource use measure is an administrative claims-based measure. It uses Medicare Part A and Part B claims from FFS beneficiaries and Medicare eligibility files.

(d) Cohort

The measure cohort includes Medicare FFS beneficiaries with a SNF treatment period ending during the data collection period.

(e) Reporting

We intend to provide initial confidential feedback to providers, prior to public reporting of this measure, based on Medicare FFS claims data from discharges in CY 2016. We intend to publicly report this measure using claims data from discharges in CY 2017.

We proposed to use a minimum of 20 episodes for reporting and inclusion in the SNF QRP. For the reliability calculation, as described in the measure specifications, a link for which has been provided above, we used data from FY 2014. The reliability results support the 20 episode case minimum, and 100 percent of SNF providers had moderate or high reliability (above 0.4).

The comments we received on this topic, with their responses, appear below.

Comment: Several commenters supported a period during which providers would be able to preview and correct measure and quality data.

Response: We appreciate the comments, and direct readers to section III.D.2.n. where we discuss this topic in detail.

Comment: Some commenters recommended an initial confidential data preview period for providers, prior to public reporting.

Response: Providers will receive a confidential preview report with 30 days for review in advance of their data and information being publicly displayed.

Comment: Some commenters recommended that the MSPB-PAC SNF QRP measure be tested for reliability and validity prior to finalization.

Response: The MSPB-PAC SNF QRP measure has been tested for reliability using FY 2014 data. The reliability results support the 20 episode case minimum, and 100 percent of SNF providers had moderate or high reliability (above 0.4). Further details on the reliability calculation are provided in the MSPB-PAC Measure Specifications, a link for which has been provided above.

Comment: One commenter suggested that descriptive statistics on the measure score by provider-level characteristics (for example, rural/urban status and bed size) would be useful to evaluate measure design decisions.

Response: We thank the commenter for their input. The following table 12 shows the MSPB-PAC SNF provider scores by provider characteristics, calculated using FY 2014 data.

Table 12—MSPB-PAC SNF Scores by Provider Characteristics

Provider characteristic	Number of providers	Mean score	Score percentile
			1st	10th	25th	50th	75th	90th	99th
All Providers	15,446	1.01	0.38	0.66	0.84	1.01	1.18	1.35	1.69
Urban/Rural:
Urban	10,656	1.03	0.46	0.73	0.87	1.02	1.18	1.35	1.68
Rural	4,786	0.96	0.29	0.56	0.74	0.96	1.16	1.35	1.71
Unknown	4	1.12	0.89	0.89	0.90	1.05	1.34	1.51	1.51
Ownership Type:
For profit	10,705	1.07	0.47	0.77	0.92	1.06	1.22	1.39	1.72
Non-profit	3,693	0.87	0.32	0.56	0.70	0.86	1.03	1.18	1.56
Government	1,008	0.89	0.20	0.49	0.66	0.87	1.12	1.31	1.66
Unknown	40	0.52	0.18	0.31	0.38	0.52	0.62	0.79	0.89
Census Division:
New England	943	0.91	0.44	0.68	0.79	0.91	1.04	1.14	1.40
Middle Atlantic	1,708	1.00	0.46	0.69	0.84	1.00	1.16	1.30	1.59
East North Central	3,009	1.07	0.50	0.76	0.92	1.06	1.21	1.39	1.69
West North Central	1,989	0.82	0.27	0.52	0.67	0.82	0.97	1.12	1.43
South Atlantic	2,369	1.03	0.41	0.75	0.90	1.03	1.17	1.31	1.60
East South Central	1,083	1.07	0.34	0.64	0.88	1.08	1.28	1.44	1.72
West South Central	2,076	1.13	0.40	0.75	0.96	1.13	1.31	1.49	1.79
Mountain	732	0.90	0.23	0.61	0.78	0.92	1.05	1.15	1.46
Pacific	1,529	1.03	0.43	0.68	0.84	1.01	1.20	1.40	1.75
Other	8	0.51	0.39	0.39	0.43	0.53	0.56	0.68	0.68
Bed Count:
0-49	1,877	0.82	0.24	0.49	0.61	0.79	1.00	1.20	1.70
50-99	5,799	1.00	0.36	0.64	0.82	0.99	1.17	1.36	1.70
100-199	6,846	1.06	0.52	0.78	0.91	1.05	1.20	1.36	1.67
200-299	726	1.08	0.55	0.78	0.91	1.06	1.23	1.42	1.69
300 +	198	1.03	0.45	0.75	0.87	1.01	1.16	1.35	1.62
No. of Episodes:
0-99	10,048	1.01	0.33	0.63	0.82	1.01	1.20	1.40	1.73
100-249	4,298	1.01	0.52	0.75	0.88	1.01	1.15	1.28	1.53
250-499	960	0.96	0.52	0.69	0.83	0.97	1.08	1.20	1.45
500-1000	136	0.96	0.57	0.74	0.88	0.96	1.08	1.19	1.35
1000 +	4	0.86	0.73	0.73	0.80	0.87	0.92	0.98	0.98

In summary, after consideration of the public comments we received, we are finalizing the specifications of the MSPB-PAC SNF QRP resource use measure, as proposed. A link for the measure specifications has been provided above.

Specifically, we are finalizing the definition of an MSPB-PAC SNF QRP episode, beginning from episode trigger. An episode window comprises a treatment period beginning at the trigger and ending upon discharge, and an associated services period beginning at the trigger and ending 30 days after the end of the treatment period. Readmissions to the same SNF within 7 or fewer days do not trigger a new episode and are instead included in the treatment period of the first episode.

We exclude certain services that are clinically unrelated to SNF care and/or because SNF providers may have limited influence over certain Medicare services delivered by other providers during the episode window. We also exclude certain episodes in their entirety from the MSPB-PAC SNF QRP measure, such as where a beneficiary is not enrolled in Medicare FFS for the entirety of the lookback period plus episode window.

We finalize the inclusion of Medicare payments for Part A and Part B claims for services included in the MSPB-PAC SNF QRP episodes to calculate the MSPB-PAC SNF QRP measure.

We are finalizing our proposal to risk adjust using covariates including age brackets, HCC indicators, prior inpatient stay length, ICU stay length, clinical case mix categories, and indicators for originally disabled, ESRD enrollment, long-term care status, and hospice claim in episode window. The measure also adjusts for geographic payment differences such as wage index and GPCI, and adjusts for Medicare payment differences resulting from IME and DSH.

We calculate the individual providers' MSPB-PAC Amount which is inclusive of MSPB-PAC SNF QRP observed episode spending over the expected episode spending as predicted through risk adjustment. Individual SNF providers' scores are calculated as their individual MSPB-PAC Amount divided by the median MSPB-PAC amount across all SNFs.

ii. Measure to Address the IMPACT Act Domain of Resource Use and Other Measures: Discharge to Community—Post Acute Care (PAC) Skilled Nursing Facility (SNF) Quality Reporting Program (QRP)

Sections 1899B(d)(1)(B) and 1899B(a)(2)(E)(ii) of the Act require the Secretary to specify a measure to address the domain of discharge to community by SNFs, LTCHs, and IRFs by October 1, 2016, and HHAs by January 1, 2017. We proposed to adopt the measure, Discharge to Community—PAC SNF QRP, for the SNF QRP for the FY 2018 payment determination and subsequent years as a Medicare FFS claims-based measure to meet this requirement.

This measure assesses successful discharge to the community from a SNF setting, with successful discharge to the community including no unplanned rehospitalizations and no death in the 31 days following discharge from the SNF. Specifically, this measure reports a SNF's risk-standardized rate of Medicare FFS residents who are discharged to the community following a SNF stay, and do not have an unplanned readmission to an acute care hospital or LTCH in the 31 days following discharge to community, and who remain alive during the 31 days following discharge to community. The term “community”, for this measure, is defined as home or self care, with or without home health services, based on Patient Discharge Status Codes 01, 06, 81, and 86 on the Medicare FFS claim.^[41 42] This measure is 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.

Discharge to a community setting is an important health care outcome for many residents for whom the overall goals of post-acute care 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 residents who are not expected to make functional improvement during their SNF stay, and for residents who may be expected to decline functionally due to their medical condition. The discharge to community outcome offers a multi-dimensional view of preparation for community life, including the cognitive, physical, and psychosocial elements involved in a discharge to the community.^[43 44]

In addition to being an important outcome from a resident and family perspective, patients and residents discharged to community settings, on average, incur lower costs over the recovery episode, compared with those discharged to institutional settings.^[45 46] Given the high costs of care in institutional settings, encouraging SNFs to prepare residents for discharge to community, when clinically appropriate, may have cost-saving implications for the Medicare program.^[47] Also, providers have discovered that successful discharge to community was a major driver of their ability to achieve savings, where capitated payments for post-acute care were in place.^[48] For 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 residents' out-of-pocket expenditures.^[49]

Analyses conducted for 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.^[50] 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.^[51]

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 resident groups. Variation in discharge to community rates has been reported within and across post-acute settings; across a variety of facility-level characteristics, such as geographic location (for example, regional location, urban or rural location), ownership (for example, for-profit or nonprofit), and freestanding or hospital-based units; and across patient-level characteristics, such as race and gender.^{[52 53 54 55 56 57]} Discharge to community rates in the IRF setting have been reported to range from about 60 to 80 percent.^{[58 59 60 61 62 63]} Longer-term studies show that rates of discharge to community from IRFs have decreased over time as IRF length of stay has decreased.^[64 65] Greater variation in discharge to community rates is seen in the SNF setting, with rates ranging from 31 to 65 percent.^{[66 67 68 69]} In the SNF Medicare FFS population, using CY 2013 national claims data, we found that approximately 44 percent of residents were discharged to the community. A multi-center study of 23 LTCHs demonstrated that 28.8 percent of 1,061 patients who were ventilator-dependent on admission were discharged to home.^[70] A single-center study revealed that 31 percent of LTCH hemodialysis patients were discharged to home.^[71] One study noted that 64 percent of beneficiaries who were discharged from the home health episode did not use any other acute or post-acute services paid by Medicare in the 30 days after discharge.^[72] However, significant numbers of patients were admitted to hospitals (29 percent) and lesser numbers to SNFs (7.6 percent), IRFs (1.5 percent), home health (7.2 percent) or hospice (3.3 percent).^[73]

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.^{[74 75 76 77]} Many of these interventions involve discharge planning or specific rehabilitation strategies, such as addressing discharge barriers and improving medical and functional status.^{[78 79 80 81]} The effectiveness of these interventions suggests that improvement in discharge to community rates among post-acute care residents is possible through modifying provider-led processes and interventions.

A TEP convened by our measure development contractor was strongly supportive of the importance of measuring discharge to community outcomes, and implementing the measure, Discharge to Community—PAC SNF QRP in the SNF QRP. The panel provided input on the technical specifications of this measure, including the feasibility of implementing the measure, as well as the overall measure reliability and validity. A summary of the TEP proceedings is available on the PAC Quality Initiatives Downloads and Videos Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

We also solicited stakeholder feedback on the development of this measure through a public comment period held from November 9, 2015, through December 8, 2015. Several stakeholders and organizations, including the MedPAC, among others, supported this measure for implementation. The public comment summary report for the measure is available on the CMS Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

The NQF-convened MAP met on December 14 and 15, 2015, and provided input on the use of this Discharge to Community—PAC SNF QRP measure in the SNF QRP. The MAP encouraged continued development of the measure to meet the mandate of the IMPACT Act. The MAP supported the alignment of this measure across PAC settings, using standardized claims data. More information about the MAP's recommendations for this measure is available at http://www.qualityforum.org/​Publications/​2016/​02/​MAP_​2016_​Considerations_​for_​Implementing_​Measures_​in_​Federal_​Programs_​-_​PAC-LTC.aspx.

Since the MAP's review and recommendation of continued development, we have continued to refine risk-adjustment models and conduct measure testing for this measure, as recommended by the MAP. This measure is consistent with the information submitted to the MAP, and the original MAP submission and our continued refinements support its scientific acceptability for use in quality reporting programs. As discussed with the MAP, we fully anticipate that additional analyses will continue as we submit this measure to the ongoing measure maintenance process.

We reviewed the NQF's consensus-endorsed measures and were unable to identify any NQF-endorsed resource use or other measures for post-acute care focused on discharge to community. In addition, we are unaware of any other post-acute care measures for discharge to community that have been endorsed or adopted by other consensus organizations. Therefore, we proposed the measure, Discharge to Community—PAC SNF QRP, under the Secretary's authority to specify non—NQF-endorsed measures under section 1899B(e)(2)(B) of the Act.

We proposed to use data from the Medicare FFS claims and Medicare eligibility files to calculate this measure. We proposed to use data from the “Patient Discharge Status Code” on Medicare FFS claims to determine whether a resident was discharged to a community setting for calculation of this measure. In all PAC settings, we tested the accuracy of determining discharge to a community setting using the “Patient Discharge Status Code” on the PAC claim by examining whether discharge to community coding based on PAC claim data agreed with discharge to community coding based on PAC assessment data. We found agreement between the two data sources in all PAC settings, ranging from 94.6 percent to 98.8 percent. Specifically, in the SNF setting, using 2013 data, we found 94.6 percent agreement in discharge to community codes when comparing discharge status codes on claims and the Discharge Status (A2100) on the Minimum Data Set (MDS) 3.0 discharge assessment, when the claims and MDS assessment had the same discharge date. We further examined the accuracy of the “Patient Discharge Status Code” on the PAC claim by assessing how frequently discharges to an acute care hospital were confirmed by follow-up acute care claims. We discovered that 88 percent to 91 percent of IRF, LTCH, and SNF claims with acute care discharge status codes were followed by an acute care claim on the day of, or day after, PAC discharge. We believed these data support the use of the claims “Patient Discharge Status Code” for determining discharge to a community setting for this measure. In addition, this measure can feasibly be implemented in the SNF QRP because all data used for measure calculation are derived from Medicare FFS claims and eligibility files, which are already available to CMS.

Based on the evidence discussed above, we proposed to adopt the measure, Discharge to Community—PAC SNF QRP, for the SNF QRP for FY 2018 payment determination and subsequent years. This measure is calculated using 1 year of data. We proposed a minimum of 25 eligible stays in a given SNF for public reporting of the measure for that SNF. Since Medicare FFS claims data are already reported to the Medicare program for payment purposes, and Medicare eligibility files are also available, SNFs will not be required to report any additional data to CMS for calculation of this measure. The measure denominator is the risk-adjusted expected number of discharges to community. The measure numerator is the risk-adjusted estimate of the number of residents who are discharged to the community, do not have an unplanned readmission to an acute care hospital or LTCH in the 31-day post-discharge observation window, and who remain alive during the post-discharge observation window. The measure is risk-adjusted for variables such as age and sex, principal diagnosis, comorbidities, ventilator status, ESRD status, and dialysis, among other variables. For technical information about the proposed measure, including information about the measure calculation, risk adjustment, and denominator exclusions, we referred readers to the document titled, Proposed Measure Specifications for Measures Proposed in the FY 2017 SNF QRP Proposed Rule available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

We stated in the proposed rule that we intend to provide initial confidential feedback to SNFs, prior to public reporting of this measure, based on Medicare FFS claims data from discharges in CY 2016. We intend to publicly report this measure using claims data from discharges in CY 2017. We plan to submit this measure to the NQF for consideration for endorsement.

We invited public comment on our proposal to adopt the measure, Discharge to Community—PAC SNF QRP, for the SNF QRP. The comments we received on this topic, with our responses, appear below.

Comment: Several commenters, including MedPAC, supported the Discharge to Community—PAC SNF QRP measure, noting that it is a critical measure assessing the ability of a PAC provider to rehabilitate patients and enable them to return to the home and community-based setting. One commenter noted that measuring the rate that the various PAC settings discharge patients to the community, without an admission (or readmission) to an acute care hospital within 30 days, is one of the most relevant patient-centered measures that exists in the post-acute care area. Commenters noted that most older adults want to live independently in their homes and communities, that returning home following care was an important concern of Medicare beneficiaries, and that successful transitions to community would decrease potentially preventable readmissions. Two commenters supported CMS's efforts to develop aligned yet distinctive risk-adjusted discharge to community measures for IRFs, SNFs and LTCHs, given the inherent variability in patient/resident profiles across these settings. Commenters agreed that discharge to community was an important outcome not just for patients expected to make functional improvement and return to their previous level of independence, but also for patients not expected to make functional improvement, or those who may be expected to decline functionally due to their medical condition. One commenter stated that achieving a standardized and interoperable patient assessment data set and stable quality measures as quickly as possible would allow for better cross-setting comparisons and the evolution of better quality measures with uniform risk standardization. One commenter expressed support for the use of claims data over assessment data in calculating the Discharge to Community—PAC SNF QRP measure, stating that assessment data could be susceptible to gaming by providers.

Response: We thank the commenters for their support of the Discharge to Community—PAC SNF QRP measure, and their recognition of its patient-centeredness, its relevance for patients with a range of functional abilities and prognosis, and its potential to reduce post-discharge readmissions. We also thank commenters for their support of use of claims data, and their support of standardized and interoperable patient assessment data and quality measures. As mandated by the IMPACT Act, we are moving toward the goal of standardized patient assessment data and quality measures across PAC settings.

Comment: One commenter interpreted our measure proposal language as suggesting that functional improvement is not a requirement, and encouraged that Medicare coverage for maintenance nursing and therapy be ensured and reflected by the measure.

Response: Our intent in the measure proposal was to acknowledge that discharge to community can be an important goal even for patients who may not be able to make functional improvement. This measure does not impact Medicare coverage rules for maintenance nursing and therapy.

Comment: Several commenters requested that “home” be defined broadly to reflect the place an individual calls “home”, including assisted living facilities, residential care settings, or other congregate community housing.

Response: We agree with the commenters that “home” should be defined broadly for the discharge to community measure. In addition to home, our definition of community includes settings such as group home, foster care, and independent living and other residential care arrangements.^[82] For further details on measure specifications, including the definition of community, we refer readers to the Measure Specifications for Measures Adopted in the FY 2017 SNF QRP Final Rule, posted on the CMS SNF QRP Web page at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

Comment: Several commenters expressed concerns regarding the use of the Patient Discharge Status Code variable to define community discharges. Commenters emphasized that it was important to ensure that only home and community based settings were included in the definition of community, and were concerned that Code 01 (Discharge to home or self care), which is included in the definition of community, included institutional settings such as jail or law enforcement. One commenter expressed that many settings included under Code 01 do not satisfy the home and community based settings rule, and may be inconsistent with the integration mandate of the Americans with Disabilities Act. Commenters strongly recommended that we either revise discharge status code 01 to exclude non community-based settings, or use alternative variables to capture discharge to community.

Response: We agree with the commenters that the discharge to community measure should only capture discharges to home and community based settings. We believe that the comment referring to the “home and community based settings rule” refers to Medicaid regulations applicable to services authorized under sections 1915(c), 1915(i) and 1915(k) of the Act, which are provided through waivers or state plans amendments approved by CMS. We would like to clarify that this measure only captures discharges to home and community based settings, not to institutional settings, and is consistent with both Medicaid regulations requiring home and community based settings to support integration, and also with the Americans with Disabilities Act (ADA), based on Patient Discharge Status Codes 01, 06, 81, and 86 on the Medicare FFS PAC claim.^[83] Discharges to jail or law enforcement are not included under Code 01 of the Patient Discharge Status Code; rather these are included under Code 21 (Discharged/transferred to Court/Law Enforcement).

We also note that Title II of the ADA regulations requires public entities to administer services, programs, and activities in the most integrated setting appropriate to the needs of qualified individuals with disabilities (28 CFR 35.130(d)). The preamble discussion of the “integration regulation” explains that “the most integrated setting” is one that enables individuals with disabilities to interact with nondisabled persons to the fullest extent possible. Integrated settings are those that provide individuals with disabilities opportunities to live, work, and receive services in the greater community, like individuals without disabilities (28 CFR part 35, app. A (2010) (addressing § 35.130)).

Comment: Several commenters stated that PAC patients/residents discharged to a nursing facility as long-term care residents should not be considered discharges to community, particularly if they were discharged to the nursing facility from the Medicare-certified skilled nursing part of the same nursing home, and even if they resided in a long-term nursing facility at baseline. Commenters emphasized that a nursing home does not represent an individual's own home in their own community. These commenters interpreted the proposed measure specifications as allowing these discharges to a nursing facility to be coded as “group home”, “foster care”, or “other residential care arrangement” under discharge status code 01. Commenters expressed concern that coding discharges from the SNF to residential/long-term care facility within the same nursing home as discharges to community would unfairly advantage SNFs and artificially inflate their discharge to community rates, would disadvantage other PAC providers, would negate the value of the measure, and would miscommunicate facility's actual discharge to community performance to the average Medicare beneficiary. Commenters also noted that including nursing facility discharges as community discharges could incentivize SNFs to not do the hard work that actual, meaningful discharge planning to the community requires.

Response: We agree with the commenters that discharges to long-term care nursing facilities, or any other institutional settings, should not be coded as discharges to community. We also recognize the differences in required discharge planning processes and resources for discharging a patient/resident to the community compared with discharging to a long-term nursing facility. The discharge to community measure only captures discharges to home and community based settings as discharges to community, based on Patient Discharge Status Codes 01, 06, 81, and 86 on the Medicare FFS PAC claim.^[84] These codes do not include discharges to long-term care nursing facilities or any other institutional setting that may violate the integration mandate of title II of the ADA. Instead, depending on the nature of the facility to which patients/residents are discharged, such discharges may be coded on the Medicare FFS claim as 04, 64, 84, 92, or another appropriate code for an institutional discharge.

In response to the commenters' concerns that SNFs may be unfairly advantaged by this measure as compared with other PAC providers, we would like to note that, in our measure development samples, the national discharge to community rate for SNFs was 47.26 percent, while this rate for IRFs was considerably higher (69.51 percent). Further, using an MDS-claims linked longitudinal file, we found that, of SNF stays that had a pre-hospitalization non-PPS MDS assessment suggesting prior nursing facility residence, two-thirds had a discharge status code of 30 (still patient), and approximately 18 percent had a discharge status code of 02 (acute hospital); less than 5 percent of these patients had a discharge status code of 01 (discharge to home or self care).

Comment: Several commenters recommended that the discharge to community measure should entirely exclude baseline long-stay nursing facility residents, as they could not be reasonably expected to discharge to the community after their PAC stay. One commenter noted that the measure fails to consider when a patient's “home” is a custodial nursing facility and the patient's post-acute episode involves a discharge back to his or her “home.” Another commenter noted that baseline nursing facility residents have a very different discharge process back to the nursing facility compared with patients discharged to the community. This commenter recommended that different measures be developed for the baseline nursing facility resident population, such as return to prior level of function, improvement in function, prevention of further functional decline, development of pressure ulcers, or accidental falls. This commenter also recognized our current efforts in monitoring transitions of care and quality requirements in long-term care facilities. One commenter suggested that we use the Minimum Data Set to identify and exclude baseline nursing facility residents.

Response: We appreciate the commenters' concerns and their recommendation to exclude baseline nursing facility residents from the discharge to community measure, and to distinguish baseline custodial nursing facility residents who are discharged back to the nursing facility after their SNF stay. We recognize that patients/residents who permanently lived in a nursing facility at baseline may not be expected to discharge back to a home and community based setting after their PAC stay. We also recognize that, for baseline nursing facility residents, a discharge back to their nursing facility represents a discharge to their baseline residence. We agree with the commenter about the differences in discharge planning processes when discharging a patient/resident to the community compared with discharging to a long-term nursing facility. However, using Medicare FFS claims alone, we are unable to accurately identify baseline nursing facility residents. Potential future modifications of the measure could include the assessment of the feasibility and impact of excluding baseline nursing facility residents from the measure through the addition of patient assessment-based data. However, we note that, currently, the IRF-PAI is the only PAC assessment that contains an item related to pre-hospital baseline living setting.

Comment: One commenter raised concerns that the measure does not exclude individuals admitted to a SNF for Part A services, but who have an expressed goal to remain in the facility for long-term care and never be discharged back to community. The commenter specifically noted that there appears to be a relationship between SNF turnover rate and discharge to community rates. They noted that SNFs with low turnover, which they offered as a marker for being a primarily long-term care facility, had low discharge to community rates compared with SNFs with high turnover.

Response: This measure risk adjusts for several case-mix variables that may be related to preferences for facility-based long-term care such as age, diagnoses from the prior acute stay, comorbidities in the year preceding PAC admission, length of prior acute stay, number of prior hospitalizations in the past year, and ventilator use. Further, by excluding patients on hospice and those whose prior acute stay was for medical treatment of cancer, we are excluding SNF residents who may be more likely to transfer to a nursing facility at the end of their SNF stay. There are no claims data we could currently use to identify residents with an expressed goal to remain in the nursing home for long-term care. As we agree this is an important aspect of this measure work, we will consider assessing the ability to identify residents with an expressed goal to remain in the nursing home for long-term care, and the impact of such an exclusion on the measure performance.

Comment: MedPAC recommended that we confirm discharge to a community setting with the absence of a subsequent claim to a hospital, IRF, SNF, or LTCH, to ensure that discharge to community rates reflect actual facility performance. Other commenters also recommended that we assess the reliability and validity of the Patient Discharge Status code on PAC claims, expressing concerns about the accuracy of these data without further definition and validation. Commenters cited MedPAC and other studies, noting that Patient Discharge Status Codes often have low reliability, and this could impact accurate portrayal of measure performance.

Response: We are committed to developing measures based on reliable and valid data. This measure does confirm the absence of hospital or LTCH claims following discharge to a community setting. Unplanned hospital and LTCH readmissions following the discharge to community, including those on the day of SNF discharge, are considered an unfavorable outcome. We will consider verifying the absence of IRF and SNF claims following discharge to a community setting, as we continue to refine this measure. Nonetheless, we would like to note that an ASPE report on post-acute care relationships found that, following discharge to community settings from IRFs, LTCHs, or SNFs in a 5 percent Medicare sample, IRFs or SNFs were very infrequently reported as the next site of post-acute care.^[85]

Because the discharge to community measure is a measure of discharge destination from the PAC setting, we have chosen to use the PAC-reported discharge destination (from the Medicare FFS claims) to determine whether a patient/resident was discharged to the community (based on discharge status codes 01, 06, 81, 86). We assessed the reliability of the claims discharge status code by examining agreement between discharge status on claims and assessment instruments for the same stay in all four PAC settings. We found between 94 and 99 percent agreement in coding of community discharges on matched claims and assessments in each of the PAC settings. We also assessed how frequently discharges to acute care, as indicated on the PAC claim, were confirmed by follow-up acute care claims, and found that 88 percent to 91 percent of IRF, LTCH, and SNF claims indicating acute care discharge were followed by an acute care claim on the day of, or day after, PAC discharge. We believe that these data support the use of the “Patient Discharge Status Code” from the PAC claim for determining discharge to a community setting for this measure.

The use of the claims discharge status code to identify discharges to the community was discussed at length with the TEP convened by our measure development contractor. TEP members did not express significant concerns regarding the accuracy of the claims discharge status code in coding community discharges, nor about our use of the discharge status code for defining this quality measure. A summary of the TEP proceedings is available on the PAC Quality Initiatives Downloads and Videos Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

Comment: One commenter recommended that, in all PAC settings, patients who are discharged home and then admitted to a SNF or nursing facility during the 31-day post-discharge window not be counted as successful discharges to the community. The commenter suggested that MDS data could be used to identify individuals admitted to nursing homes.

Response: We agree that it is important to track whether patients remain in the community in the post-discharge observation window in order to ensure that facilities are appropriately discharging patients to the community. In the measure, we examine post-discharge unplanned acute care or LTCH readmissions, thereby accounting for more serious, acute readmissions in the post-discharge window. In future versions of the measure, we will consider looking for IRF, SNF, and nursing facility admissions and readmissions in the 31-day post-discharge window when examining discharge to community outcomes.

Comment: A few commenters requested clarification on the calculation of the discharge to community measure rates. One commenter questioned why estimates were used rather than observed rates.

Response: A successful discharge to community outcome includes patients discharged to the community who remain alive for 31 days post-discharge with no unplanned readmission. The method used requires the use of estimates because the observed rates are statistically adjusted to account for patient mix in each facility. The statistical model also estimates facility-level effects. In brief, we first calculate the sum of the probabilities of discharge to community of all patients/residents in the facility, including both the impact of patient/resident characteristics and the impact of the facility; this equals the “predicted number” of discharges to community after adjusting for the facility's case mix. We then calculate the “expected number” of discharges to community for the same patients/residents 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 at the average facility. This ratio is 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.

Details on the risk adjustment methodology and measure calculation algorithm for the discharge to community measure are available in the Measure Specifications for Measures Adopted in the FY 2017 SNF QRP Final Rule, posted on the CMS SNF QRP Web page at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html. Specifically, we refer readers to Sections 2.1.8—Statistical Risk Model and Risk Adjustment Covariates, and 2.1.9—Measure Calculation Algorithm.

Comment: One commenter had concerns that there was overlap between the potentially preventable readmission measure and the discharge to community measure under the SNF QRP. The commenter noted that using two separate measures may be confusing to consumers and providers, making it challenging for SNFs to track and improve performance on these metrics.

Response: There are distinct differences between the discharge to community and potentially preventable readmission measures under the SNF QRP. Although there may be some overlap in the outcomes captured across the two measures (for example, residents who have a potentially preventable readmission also have an unsuccessful discharge to community) each measure has a distinct purpose, outcome definition, and measure population. For example, the discharge to community measure assesses the rate of successful discharges to the community, defined as discharge to a community setting without post-discharge unplanned readmissions or death, while the potentially preventable readmission measure assesses the rate of readmissions that may be potentially prevented for patients/residents discharged to lower levels of care from the SNF.

Our goal is to develop measures that are meaningful to patients and consumers, and assist them in making informed choices when selecting post-acute providers. Since the goal of PAC for most patients and family members is to be discharged to the community and remain in the community, from a patient/consumer perspective, it is important to assess whether a patient remained in the community after discharge and to separately report discharge to community rates. In addition to assessing the success of community discharges, the inclusion of post-discharge readmission and death outcomes is intended to avoid the potential unintended consequence of inappropriate discharges to the community.

Analysis on our measure development sample has shown that, of SNF patients discharged to the community, approximately 15 percent had an unplanned readmission in the post-discharge observation window. The mean number of days from SNF discharge to readmission was 12.2 with a standard deviation of 9.7; 25 percent of readmissions occurred within 3 days of SNF discharge, and 50 percent within 10 days. Ignoring these post-discharge readmissions occurring soon after discharge to community would fail to reflect our intent with this measure.

Comment: One commenter suggested that the discharge to community measure examine emergency room visits in the post-discharge observation window, in addition to unplanned readmissions. The commenter noted that this addition would impose no additional data collection burden on SNFs or hospitals, since these data are already collected by us.

Response: The discharge to community measure captures patients that are discharged to the community and remain in the community post-discharge. An emergency room visit that does not result in hospitalization would not be considered a failure to remain in the community. Nevertheless, we will assess emergency room visit rates in the post-discharge observation window to monitor for increasing rates, and potential indication of poor quality of care or inappropriate community discharges.

Comment: Some commenters had questions regarding death in the post-discharge window. One commenter requested clarification as to why an unexpected death, such as an accidental death, in the post-discharge observation window would count against a SNF's measure rate on the discharge to community measure. Another commenter recommended that the measure exclude patients who have been discharged to the community and expire within the post-discharge observation window. The commenter stated that the types of patients treated in SNFs varied greatly, and including post-discharge death in the measure could lead to an inaccurate reflection of the quality of care furnished by the SNF.

Response: Including 31-day post-discharge mortality outcomes is intended to identify successful discharges to community, and to avoid the potential unintended consequence of inappropriate community discharges. We have found, through our analyses on our measure development sample, that death in the 31 days following discharge to community is an infrequent event, with only 2.0 percent of SNF Medicare FFS beneficiaries discharged to community dying during that period. In addition, accidental or unrelated deaths in the post-discharge window are expected to be rare and randomly distributed. We do not expect such deaths to disproportionately affect measure rates for specific facilities. Finally, we do not expect facilities to achieve a 0 percent death rate in the measure's post-discharge observation window; however, one focus of the measure is to identify facilities with unexpectedly high rates of death for quality monitoring purposes.

Comment: A few commenters requested clarification on whether patients who are discharged to home under hospice care qualify as a discharge to community for the purposes of the measure. One commenter also requested clarification on how a patient who elects hospice care after SNF discharge but within the post-discharge observation window would be counted in the measure. Two commenters suggested that patients who die on hospice within the post-discharge observation window not be excluded from the discharge to community measures, but instead be considered successful discharges to the community. One commenter noted that dying at home is the preference of the majority of Americans, and nursing homes should not be penalized for helping a person choose where they want their life to end. The other commenter believed that excluding patients on hospice could create an incentive to keep dying individuals in a SNF or discharge them to the hospital.

Response: The discharge to community measure excludes patients discharged to home- or facility-based hospice care. Thus, discharges to hospice are not considered discharges to community, but rather are excluded from the measure calculation. We are are adding an exclusion of patients/residents with a hospice benefit in the post-discharge observation window to the proposed Discharge to Community—PAC SNF QRP measure, in response to public comments received on this measure proposal, comments received during measure development, and our ongoing analysis and testing.

In response to commenters' concerns about the exclusion of hospice patients/residents, we would like to note that we that we reached the decision to exclude patients/residents discharged to hospice after discussion with our TEP members and hospice clinical experts, comparison of post-discharge death rates for hospice and non-hospice patients/residents, and comparison of discharge planning and goals of care for hospice and non-hospice patients/residents. We concluded that it would be conceptually confusing to include in the discharge to community outcome both patients/residents who are successfully rehabilitated to live in the community for whom death is an undesirable outcome, and patients/residents who are terminally ill, and wish to die in the comfort of their home. The rationale for the added exclusion of patients/residents with a post-discharge hospice benefit aligns with the rationale for exclusion of discharges to hospice.

Comment: One commenter suggested that the measure does not appropriately account for patients who seek other end-of-life care in the community, beyond hospice.

Response: There are no current data sources available that would enable us to identify patients seeking end-of-life care that is separate from hospice services.

Comment: One commenter suggested that we revise the measure name to reflect that it only applies to the Medicare FFS population. The commenter was concerned that, in many states, a large proportion of Medicare beneficiaries served by SNFs are not enrolled in Medicare FFS; thus, the measure may not reflect a SNF's overall discharge to community rate, but rather the discharge to community rate among FFS beneficiaries only.

Response: We will take the commenter's suggestion into consideration.

Comment: Several commenters had concerns that the risk adjustment methodology does not include adjustment for sociodemographic or socioeconomic status. Commenters noted the importance of home and community supports such as caregiver availability, willingness, and ability to support the person in the community, and availability of an established home in determining a beneficiary's ability to be discharged to community and remain in their home or community setting. Commenters believed that sociodemographic and socioeconomic factors were strong predictors of return to the community, and since they were outside a provider's control, they should be accounted for in risk adjustment. One commenter expressed concern that the measure does not adjust for regional differences in community-based needs and supports that result from factors such as geographic variance in availability of affordable housing. Another commenter suggested that the measure account for rurality, since limited alternative services may be available in rural areas, making discharge to community less feasible.

Response: We understand the importance of home and community supports and availability of housing for ensuring a successful discharge to community outcome. The discharge to community measure is a claims-based measure and, currently, there are no standardized data on variables such as living status, family and caregiver supports, or housing availability across across the four PAC settings. We appreciate and will consider the commenter's suggestion to account for potential challenges of discharging patients to the community in rural areas. As we refine the measure in the future, we will consider testing and adding additional relevant data sources and standardized items for risk adjustment of this measure. With regard to the suggestions regarding risk adjustment pertaining to sociodemographic and socioeconomic factors, we refer the readers to section III.D.2.f. for a more detailed discussion of the role of SES/SDS factors in risk adjustment of our measures.

Comment: One commenter raised concerns that the measure does not adjust for factors that are unique to certain specific provider types, such as providers offering dedicated services to specialty residents, for example, those with HIV/AIDS. The commenter noted that providers caring for these populations may encounter greater challenges in discharging patients to the community due to special needs such as affordable and safe housing, mental health and substance abuse counseling, and medication management and supports.

Response: We appreciate the commenters' suggestion that the discharge to community measure should adjust for providers primarily caring for specialty populations that may encounter greater challenges with discharge to community settings. Our risk adjustment model accounts for a comprehensive list of diagnoses and comorbidities, including HIV/AIDS. We will consider testing for an association between providers primarily caring for specialty populations and discharge to community outcomes as we refine this measure.

Comment: One commenter emphasized the relationship between functional gains made by patients during their SNF stay and their ability to discharge to the community. The commenter stated that return to one's previous home represents part of the goal of care; additionally, it is also important that the patient is able to function to the greatest possible extent in the home and community setting, and achieve the highest quality of life possible. The commenter recommended that we delay adopting this measure until it incorporates metrics that assess whether patients achieved their functional and independence goals based on their plan of care and their specific condition.

Many other commenters suggested that we include functional status in the risk adjustment for the discharge to community measure. Commenters noted that the literature demonstrates evidence that higher functional and cognitive status are strong predictors of individuals' ability to live independently, whereas lower functional status was a strong predictor of requiring long-term nursing home placement. Another commenter noted that functional status is associated with increased risk of 30-day all-cause hospital readmissions, and since readmissions and discharge to community are closely related, functional status risk adjustment is also important for this measure. One commenter suggested that the SNF and LTCH measures include risk adjustment that is similar to the risk adjustment for Case-Mix Groups (CMGs) in the IRF setting and Activities of Daily Living in the HHA setting. One commenter interpreted the measure proposal as stating that we will not adjust the quality measures, including the discharge to community measure, to account for functional status of beneficiaries until such data are collected under the IMPACT Act.

Response: We agree that it is important to assess various aspects of patient outcomes that are indicative of successful discharge from the SNF setting. We also agree that functional status may be related to discharge to community outcomes, and that it is important to test functional status risk adjustment when assessing discharge to community outcomes. The discharge to community measure does include functional status risk adjustment in the IRF setting using CMGs from claims, and in the home health setting using Activities of Daily Living from claims. As mandated by the IMPACT Act, we are moving toward the goal of collecting standardized patient assessment data for functional status across PAC settings. Currently, the SNF Quality Reporting Program includes a process measure related to functional status assessment: Application of Percent of Long-Term Care Hospital Patients with an Admission and Discharge Functional Assessment and a Care Plan That Addresses Function (NQF #2631). Once standardized functional status data become available across settings, it is our intent to use these data to assess patients' functional gains during their PAC stay, and to examine the relationship between functional status, discharge destination, and patients' ability to discharge to community. As we examine these relationships between functional outcomes and discharge to community outcomes in the future, we will assess the feasibility of leveraging these standardized patient assessment data to incorporate functional outcomes into the discharge to community measure. Standardized cross-setting patient assessment data will also allow us to examine interrelationships between the quality and resource use measures in each PAC setting, to understand how these measures are correlated.

Comment: One commenter stated that ventilator use is included as a risk adjuster in the LTCH setting only, but should be used across all settings. This commenter also requested information on the hierarchical logistic regression modeling and variables that will be used for risk adjustment.

Response: We would like to clarify that risk adjustment for ventilator use is included in both LTCH and SNF settings. We investigated the need for risk adjustment for ventilator use in IRFs, but found that less than 0.01 percent of the IRF population (19 patient stays in 2012, and 9 patient stays in 2013) had ventilator use in the IRF. Given the low frequency of ventilator use in IRFs, any associated estimates would not be reliable; thus, ventilator use is not included as a risk adjuster in the IRF setting measure. However, we will continue to assess this risk adjuster for inclusion in the IRF model for this measure.

For details on measure specifications, modeling, and calculations, we refer readers to the Measure Specifications for Measures Adopted in the FY 2017 SNF QRP Final Rule, posted on the CMS SNF QRP Web page at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

Comment: Two commenters conveyed concerns about unintended consequences of the discharge to community measure. One commenter was concerned about increased costs to the health care system in instances where patients have difficult transitions to community, have subsequent difficulty accessing SNF care, and experience costlier inpatient care as a consequence. Another commenter had concerns that the discharge to community measure may limit access to specialty services, limit access to care for low-income populations; create perverse incentives for providers; or impact the finances of post-acute care providers based on factors beyond their control. One commenter stated that effective risk adjustment would be important to avoid unintended consequences of decreased access for patients who may need a longer SNF stay.

Response: We appreciate the commenter's concerns regarding potential unintended consequences of the discharge to community measure. We expect that, on average, discharges to community settings rather than institutional settings will result in lower healthcare costs. To avoid potential unintended consequences of inappropriate discharges to the community, this measure examines acute care and LTCH readmissions and death in the 31-day post-discharge observation window; the measure thus incentivizes providers to ensure safe transitions to the community without post-discharge unplanned readmissions. In future modifications of the measure, we will consider looking for IRF, SNF, and nursing facility admissions and readmissions in the 31-day post-discharge window when examining discharge to community outcomes. With regard to the commenter's concern that the measure may result in decreased access for patients who may need a longer SNF stay, we would like to clarify that the measure does not examine the length of a SNF stay and does not incentivize facilities to avoid patients/residents who may need a longer stay in the facility. The measure examines discharge destination from the SNF, irrespective of their length of stay.

As with all our measures, we will monitor for unintended consequences as part of measure monitoring and evaluation to ensure that measures do not reduce quality of care or access for patients, result in disparities for certain patient sub-groups, or adversely affect healthcare spending.

Comment: One commenter conveyed appreciation that the measure would be revised using an ICD-9 to ICD-10 crosswalk.

Response: We thank the commenter for their appreciation of proposed measure updates using the ICD-9 to ICD-10 crosswalk, as stated in the Proposed Measure Specifications for Measures Proposed in the FY 2017 SNF QRP Proposed Rule.

Comment: One commenter encouraged us to provide PAC settings with access to measure performance data as early as possible so providers have time to adequately review these data, and implement strategies to decrease readmissions where necessary.

Response: We intend to provide initial confidential feedback to PAC providers, prior to public reporting of this measure, based on Medicare FFS claims data from discharges in CY 2016.

Comment: Several commenters expressed concern about the lack of NQF endorsement for the measure, and suggested additional measure testing and development. One commenter requested that we provide a timeline for submission of the proposed measures to NQF. Additionally, commenters recommended NQF endorsement prior to implementation or public reporting.

Response: We thank the commenter for their comments regarding NQF endorsement. We would like to clarify that the discharge to community measure has been fully developed and tested. We plan to submit the Discharge to Community—PAC SNF QRP measure to the NQF for consideration for endorsement.

Final Decision: After careful consideration of the public comments, we are finalizing our proposal to adopt the measure, Discharge to Community—PAC SNF QRP as a Medicare FFS claims-based measure for the FY 2018 payment determination and subsequent years, with the added exclusion of residents with a hospice benefit in the 31-day post-discharge observation window. For measure specifications, we refer readers to the Measure Specifications for Measures Adopted in the FY 2017 SNF QRP Final Rule, posted on the CMS SNF QRP Web page at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

iii. Measure To Address the IMPACT Act Domain of Resource Use and Other Measures: Potentially Preventable 30-Day Post-Discharge Readmission Measure for Skilled Nursing Facility Quality Reporting Program

Sections 1899B(a)(2)(E)(ii) and 1899B(d)(1)(C) of the Act require the Secretary to specify measures to address the domain of all-condition risk-adjusted potentially preventable hospital readmission rates by SNFs, LTCHs, and IRFs by October 1, 2016, and HHAs by January 1, 2017. We proposed the measure Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP as a Medicare FFS claims-based measure to meet this requirement for the FY 2018 payment determination and subsequent years.

The measure assesses the facility-level risk-standardized rate of unplanned, potentially preventable hospital readmissions for Medicare FFS beneficiaries in the 30 days post-SNF discharge. The SNF admission must have occurred within up to 30 days of discharge from a prior proximal hospital stay which is defined as an inpatient admission to an acute care hospital (including IPPS, CAH, or a psychiatric hospital). Hospital readmissions include readmissions to a short-stay acute care hospitals or an LTCH, with a diagnosis considered to be unplanned and potentially preventable. This measure is claims-based, requiring no additional data collection or submission burden for SNFs. Because the measure denominator is based on SNF admissions, each Medicare beneficiary may be included in the measure multiple times within the measurement period. Readmissions counted in this measure are identified by examining Medicare FFS claims data for readmissions to either acute care hospitals (IPPS or CAH) or LTCHs that occur during a 30-day window beginning two days after SNF discharge. This measure is conceptualized uniformly across the PAC settings, in terms of the measure definition, the approach to risk adjustment, and the measure calculation. Our approach for defining potentially preventable hospital readmissions is described in more detail below.

Hospital readmissions among the Medicare population, including beneficiaries that utilize PAC, are common, costly, and often preventable.^[86 87] MedPAC and a study by Jencks et al. estimated that 17 to 20 percent of Medicare beneficiaries discharged from the hospital were readmitted within 30 days. MedPAC found that more than 75 percent of 30-day and 15-day readmissions and 84 percent of 7-day readmissions were considered “potentially preventable.” ^[88] In addition, MedPAC calculated that annual Medicare spending on potentially preventable readmissions would be $12 billion for 30-day, $8 billion for 15-day, and $5 billion for 7-day readmissions in 2005.^[89] For hospital readmissions from SNFs, MedPAC deemed 76 percent of readmissions as “potentially avoidable”—associated with $12 billion in Medicare expenditures.^[90] Mor et al. analyzed 2006 Medicare claims and SNF assessment data (Minimum Data Set), and reported a 23.5 percent readmission rate from SNFs, associated with $4.3 billion in expenditures.^[91] Fewer studies have investigated potentially preventable readmission rates from the remaining post-acute care settings.

We have addressed the high rates of hospital readmissions in the acute care setting, as well as in PAC. For example, we developed the following measure: Skilled Nursing Facility 30-Day All-Cause Readmission Measure (SNFRM) (NQF #2510), as well as similar measures for other PAC providers (NQF #2502 for IRFs and NQF #2512 for LTCHs).^[92] These measures are endorsed by the NQF, and the NQF endorsed SNF measure (NQF #2510) was adopted into the SNF VBP Program in the FY 2016 SNF final rule (80 FR 46411 through 46419). Note that these NQF endorsed measures assess all-cause unplanned readmissions.

Several general methods and algorithms have been developed to assess potentially avoidable or preventable hospitalizations and readmissions for the Medicare population. These include the Agency for Healthcare Research and Quality's (AHRQ's) Prevention Quality Indicators, approaches developed by MedPAC, and proprietary approaches, such as the 3M ^[TM] algorithm for Potentially Preventable Readmissions.^{[93 94 95]} Recent work led by Kramer et al. for MedPAC identified 13 conditions for which readmissions were deemed as potentially preventable among SNF and IRF populations.^[96 97] Although much of the existing literature addresses hospital readmissions more broadly and potentially avoidable hospitalizations for specific settings like long-term care, these findings are relevant to the development of potentially preventable readmission measures for PAC.^{[98 99 100]}

Potentially Preventable Readmission Measure Definition: We conducted a comprehensive environmental scan, analyzed claims data, and obtained input from a TEP to develop a definition and list of conditions for which hospital readmissions are potentially preventable. The Ambulatory Care Sensitive Conditions and Prevention Quality Indicators, developed by AHRQ, served as the starting point in this work. For patients in the 30-day post-PAC discharge period, a potentially preventable readmission (PRR) refers to a readmission for which the probability of occurrence could be minimized with adequately planned, explained, and implemented post discharge instructions, including the establishment of appropriate follow-up ambulatory care. Our list of PPR conditions is categorized by 3 clinical rationale groupings:

• Inadequate management of chronic conditions;
• Inadequate management of infections; and
• Inadequate management of other unplanned events.

Additional details regarding the definition for potentially preventable readmissions are available in the document titled, Proposed Measure Specifications for Measures Proposed in the FY 2017 SNF QRP Proposed Rule, available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

This measure focuses on readmissions that are potentially preventable and also unplanned. Similar to the SNF 30-Day All-Cause Readmission Measure (NQF #2510), this measure uses the current version of the CMS Planned Readmission Algorithm as the main component for identifying planned readmissions. A complete description of the CMS Planned Readmission Algorithm, which includes lists of planned diagnoses and procedures, can be found on the CMS Web site at http://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​HospitalQualityInits/​Measure-Methodology.html. In addition to the CMS Planned Readmission Algorithm, this measure incorporates procedures that are considered planned in post-acute care settings, as identified in consultation with TEPs. Full details on the planned readmissions criteria used, including the CMS Planned Readmission Algorithm and additional procedures considered planned for post-acute care, can be found in the document titled, Proposed Measure Specifications for Measures Proposed in the FY 2017 SNF QRP Proposed Rule at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

This measure, Potentially Preventable 30-Day Post-Discharge Readmission Measure for Skilled Nursing Facility Quality Reporting Program, assesses potentially preventable readmission rates while accounting for patient demographics, principal diagnosis in the prior hospital stay, comorbidities, and other patient factors. While estimating the predictive power of patient characteristics, the model also estimates a facility-specific effect, common to patients treated in each facility. This measure is calculated for each SNF based on the ratio of the predicted number of risk-adjusted, unplanned, potentially preventable hospital readmissions that occur within 30 days after a SNF discharge, including the estimated facility effect, to the estimated predicted number of risk-adjusted, unplanned inpatient hospital readmissions for the same patients treated at the average SNF. A ratio above 1.0 indicates a higher than expected readmission rate (worse) while a ratio below 1.0 indicates a lower than expected readmission rate (better). This ratio is referred to as the standardized risk ratio (SRR). The SRR is then multiplied by the overall national raw rate of potentially preventable readmissions for all SNF stays. The resulting rate is the risk-standardized readmission rate (RSRR) of potentially preventable readmissions.

An eligible SNF stay is followed until: (1) The 30-day post-discharge period ends; or (2) the patient is readmitted to an acute care hospital (IPPS or CAH) or LTCH. If the readmission is unplanned and potentially preventable, it is counted as a readmission in the measure calculation. If the readmission is planned, the readmission is not counted in the measure rate.

This measure is risk adjusted. The risk adjustment modeling estimates the effects of patient characteristics, comorbidities, and select health care variables on the probability of readmission. More specifically, the risk-adjustment model for SNFs accounts for demographic characteristics (age, sex, original reason for Medicare entitlement), principal diagnosis during the prior proximal hospital stay, body system specific surgical indicators, comorbidities, length of stay during the patient's prior proximal hospital stay, intensive care unit (ICU) utilization, end-stage renal disease status, and number of acute care hospitalizations in the preceding 365 days.

This measure is calculated using 1 calendar year of FFS claims data, to ensure the statistical reliability of this measure for facilities. In addition, we proposed a minimum of 25 eligible stays for public reporting of the measure.

A TEP convened by our measure development contractor provided recommendations on the technical specifications of this measure, including the development of an approach to define potentially preventable hospital readmission for PAC. Details from the TEP meetings, including TEP members' ratings of conditions proposed as being potentially preventable, are available in the TEP Summary Report available on the CMS Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html. We also solicited stakeholder feedback on the development of this measure through a public comment period held from November 2 through December 1, 2015. Comments on the measure varied, with some commenters supportive of the measure, while others either were not in favor of the measure, or suggested potential modifications to the measure specifications, such as including standardized function data. A summary of the public comments is also available on the CMS Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

The MAP encouraged continued development of the measure. Specifically, the MAP stressed the need to promote shared accountability and ensure effective care transitions. More information about the MAP's recommendations for this measure is available at http://www.qualityforum.org/​Publications/​2016/​02/​MAP_​2016_​Considerations_​for_​Implementing_​Measures_​in_​Federal_​Programs_​-_​PAC-LTC.aspx. At the time, the risk-adjustment model was still under development. Following completion of that development work, we were able to test for measure validity and reliability as identified in the measure specifications document provided above. Testing results are within range for similar outcome measures finalized in public reporting and value-based purchasing programs, including the SNFRM (NQF #2510) adopted into the SNF VBP Program in the FY 2016 SNF final rule (80 FR 46411 through 46419).

We reviewed the NQF's consensus endorsed measures and were unable to identify any NQF endorsed measures focused on potentially preventable hospital readmissions. We are unaware of any other measures for this IMPACT Act domain that have been endorsed or adopted by other consensus organizations. Therefore, we proposed the Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP, under the Secretary's authority to specify non-NQF-endorsed measures under section 1899B(e)(2)(B) of the Act, for the SNF QRP for the FY 2018 payment determination and subsequent years given the evidence previously discussed above.

We plan to submit the measure to the NQF for consideration of endorsement. We stated in the proposed rule that we intended to provide initial confidential feedback to SNFs, prior to public reporting of this measure, based on 1 calendar year of claims data from discharges in CY 2016. We also stated that we intended to publicly report this measure using claims data from CY 2017.

We invited public comment on our proposal to adopt the measure, Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP. We received several comments, which are summarized with our responses below.

Comment: MedPAC and several other commenters expressed general support for the proposed Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP. One commenter noted that the PPR measure would supplement the all-cause readmission measure by creating an incentive for SNFs to focus attention on managing SNF residents that are chronically ill as well as to manage or avoid infections. Some commenters specifically supported the post-PAC discharge readmission window, noting that SNFs should be accountable for safe transitions to the community or next care setting.

Response: We thank commenters for their support of this measure.

Comment: One commenter specifically supported the inclusion of infectious conditions in the “inadequate management of infections” and “inadequate management of other unplanned events” categories in the measure's definition of potentially preventable hospital readmissions. Another commenter expressed support for the inclusion of chronic conditions and infections as conditions for which readmissions would be considered potentially preventable. Another commenter expressed appreciation for the focus on preventable readmissions, but urged us to continue evaluating and testing the measure to ensure that the codes used for the PPR definition are clinically relevant. One commenter expressed concern over being “penalized” for readmissions that are clinically unrelated to a patient's original reason for SNF admission.

Response: We thank commenters for their support of this measure domain and the list of PPR conditions developed for this measure. Though readmissions may be considered potentially preventable even if they may not appear to be clinically related to the patient's original reason for SNF admission, there is substantial evidence that the conditions included in the definition may be preventable with adequately planned, explained, and implemented post-discharge instructions, including the establishment of appropriate follow-up ambulatory care. Furthermore, this measure is based on Medicare FFS claims data, and it may not always be feasible to determine whether a subsequent readmission is or is not clinically related to the reason why the patient was receiving SNF care. We intend to conduct ongoing evaluation and monitoring of this measure.

Comment: Several commenters expressed concern over the cross-setting alignment of the proposed PPR measures. One commenter encouraged us to assess readmission measures across the agency's programs to ensure that they promote collaboration and support readmission reduction efforts. MedPAC commented that the measure definition and risk adjustment should be identical across PAC settings so that potentially preventable readmission rates can be compared across settings. Another commenter expressed concern specifically over the “nonalignment” between the IRF and SNF versions of the measure, adding that this may lead to confusion.

Response: The PPR definition (that is, list of conditions for which readmissions would be considered potentially preventable) is aligned for measures with the same readmission window, regardless of PAC setting. Specifically, the post-PAC discharge PPR measures that were developed for each of the PAC settings contain the same list of PPR conditions. Although there are some minor differences in the specifications across the measures (for example, years of data used to calculate the measures to ensure reliability and some of the measure exclusions necessary to attribute responsibility to the individual settings), the IMPACT Act PPR measures are standardized. As described for all IMPACT Act measures in section III.D.2.f., the statistical approach for risk adjustment is also aligned across the measures; however, there is variation in the exact risk adjusters. The risk-adjustment models are empirically driven and differ between measures as a consequence of case mix differences, which is necessary to ensure that the estimates are valid.

Comment: One commenter expressed concern that the post-discharge readmission window provides an opportunity for patient health to decline following discharge due to factors beyond providers' control, including patient behavior, noting these factors vary considerably among patients. The commenter suggested the measure reflect the shared responsibility of all parties involved in a patient's care, such as caregivers and the patients themselves. The commenter also suggested we clarify how patients that expire within the readmission window are handled in the measure.

Response: The focus of the PPR measure is to identify excess PPR rates for the purposes of quality improvement. There is substantial evidence that certain readmissions can be prevented with adequately planned, explained, and implemented post-discharge instructions, including the establishment of appropriate follow-up ambulatory care. We are aware that there are certain patient characteristics that may increase the risk of readmission, and a number of these conditions are accounted for in the risk-adjustment model. We would also like to clarify that patients who expire during the SNF stay are excluded because there is no post-SNF discharge window to observe the outcome. However, we do include patients that expire during the post-SNF discharge readmission window to assess the outcome as it is relevant for all patients discharged from SNFs. This is also consistent with other NQF-endorsed readmission measures.

Comment: Several commenters raised concerns over the risk-adjustment approach for the PPR measures, urging us to incorporate factors such as cognitive and functional status, supply variables, and SES/SDS factors into the measure's risk adjustment. One commenter noted that assessment instruments, such as the MDS, provide data sources for various patient clinical characteristics. Furthermore, the commenter expressed that because the IMPACT Act mandates the standardization of assessment instruments, the IMPACT Act measures should incorporate standardized items as risk adjusters.

Another commenter supported the proposed risk-adjustment methodology commenting that it will provide a valid assessment of quality of care in preventing unplanned, preventable hospital readmissions.

Response: The risk-adjustment model takes into account medical complexity, as patients with multiple risk factors will rate as having higher risk of readmission. For those cross-setting post-acute measures such as those intended to satisfy the IMPACT Act domains that use the patient assessment-based data elements for risk adjustment, we have either made such items standardized, or intend to do so as feasible. We wish to note that we intend to evaluate the feasibility of including functional and cognitive status when standardized assessment data become available. With regard to the suggestions pertaining to risk adjustment methodologies pertaining to sociodemographic factors we refer the readers to section section III.D.2.f. where we also discuss these topics.

Comment: Some commenters cautioned against potential unintended consequences of the measure, in particular, noting that the measure could incentivize SNFs to delay necessary readmission to the hospital or prolong the SNF stay. One commenter noted that the measure could cause SNFs to be selective about the patients they admit (that is, “cherry pick” their patients), and suggested that an appropriate risk adjustment could prevent this.

Response: We intend to conduct ongoing monitoring to assess for potential unintended consequences associated with the implementation of this measure, and we will take these suggestions into account. A major goal of risk adjustment is to ensure that patient case mix is taken into account in order to allow for fair comparisons of facilities. The risk of readmission for patients in poor health is taken into account by the risk-adjustment model used in the calculation of this measure. Given this is a post-SNF discharge measure, SNFs would have no incentive to delay hospital readmissions.

Comment: One commenter suggested that the PPR measure incorporate both inpatient and emergency room (ER) visits because a measure that captures both would be more understandable to consumers. Another expressed concern regarding overlap between the proposed PPR measure and the discharge to community measure, and the implications for quality improvement.

Response: We appreciate the comment suggesting that the measure include inpatient as well as ER visits. However, we wish to clarify that the PPR measure was developed to fulfill the IMPACT Act's statutory requirement for a measure to address the domain of potentially preventable hospital readmissions. We agree that ER or emergency department visits are also an important outcome, but they are not hospital readmissions.

We discuss above the similarities and differences between the PPR and discharge to community measure. Although there are conceptual similarities between the measures, we believe that each measure provides important information for quality improvement purposes and will enable SNFs to target different aspects of care provided.

Comment: One commenter provided comments on the statistical approach used to calculate the measure, recommending that we use the actual readmission rate (that is, observed) as the numerator of the SRR rather than the predicted number of readmissions, or provide evidence to justify this more complicated methodology. The commenter acknowledged the aims of the risk-adjustment model but suggested using the actual instead of the predicted number of readmissions so that the numerator of the SRR is clearer and more actionable for facilities, and is not likely to result in substantial changes to the relative ranking of facilities. The same commenter also indicated support for the current minimum denominator size—25 patients—for public reporting but suggested that a minimum size of 30 would improve the reliability of the measurement.

Response: The statistical approach for this measure, including the use of the predicted to expected readmission rate, is used in several other readmission measures, including the SNFRM (NQF #2510) and other NQF-endorsed readmission measures. Not using this approach would render providers with small numbers of eligible patient stays excessively vulnerable to reported rates driven by the influence of random variation in performance, limiting the value of the public reporting their measure performance. We would also like to note that facilities will be given their observed rates in their reports.

We acknowledge that increasing the minimum denominator size for public reporting of this measure may increase the reliability of the measure, but doing so would prevent a substantial number of facilities from reporting this measure.

Comment: One commenter commented that we should not finalize this measure because the measure was still under development and the MAP did not vote to support it, but instead encouraged continued development. In addition, this commenter said we should submit the measure for NQF endorsement and only propose NQF endorsed measures. Another commenter encouraged additional testing and evaluation of the measure prior to implementation.

Response: We intend to submit this measure to NQF for consideration of endorsement. Although the measure is not currently endorsed, we did conduct additional testing subsequent to the MAP meeting. Based on that testing, we were able to complete the risk adjustment model and evaluate facilities' PPR rates, and we made the results of our analyses available at the time of the proposed rule. We found that testing results were similar to the SNFRM (NQF #2510) and allowed us to conclude that the measure is sufficiently developed, valid and reliable for adoption in the SNF QRP.

Comment: One commenter expressed concern that we used language that suggested all readmissions are preventable and recommends the use of the term “may be avoidable” in place of “should be avoidable” in describing readmissions. The commenter was concerned that the language used would imply that the goal of the measure is for providers to reach zero percent PPR.

Another commenter expressed concern about the accuracy of claims-based data, but supported the effort to limit the data collection burden placed on providers.

Response: We agree with the commenter that this is a measure of potentially preventable readmissions and that not all readmissions are preventable. We wish to clarify that the PPR rate is not expected to be 0. The goal of the measure is to identify excess PPR rates for the purposes of quality improvement.

With respect to the use of claims data to calculate this measure, multiple studies have been conducted to examine the validity of using Medicare hospital claims to calculate several NQF endorsed quality measures for public reporting.^{[101 102 103]} These studies supported the use of claims data as a valid means for risk adjustment and assessing similar outcomes. Additionally, although assessment and other data sources may be valuable for risk adjustment, we are not aware of another data source aside from Medicare claims data that could be used to reliably assess the outcome of potentially preventable hospital readmissions post-SNF discharge.

Final Decision: After careful consideration of the public comments, we are finalizing our proposal to adopt the measure, Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP beginning with the FY 2018 payment determination. Measure Specifications for Measures Adopted in the FY 2017 SNF QRP Final Rule are available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

g. SNF QRP Quality Measure Finalized for the FY 2020 Payment Determination and Subsequent Years

We proposed to adopt one new quality measure to meet the requirements of the IMPACT Act for the FY 2020 payment determination and subsequent years. The measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, addresses the IMPACT Act quality domain of Medication Reconciliation.

1. Quality Measure Addressing the IMPACT Act Domain of Medication Reconciliation: Drug Regimen Review Conducted With Follow-Up for Identified Issues—Post Acute Care (PAC) Skilled Nursing Facility Quality Reporting Program

Sections 1899B (a)(2)(E)(i)(III) and 1899B(c)(1)(C) of the Act require the Secretary to specify a quality measure to address the domain of medication reconciliation by October 1, 2018 for IRFs, LTCHs and SNFs; and by January 1, 2017 for HHAs. We proposed to adopt the quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PPAC SNF QRP, for the SNF QRP as a resident-assessment based, cross-setting quality measure to meet the IMPACT Act requirements with data collection beginning October 1, 2018 for the FY 2020 payment determinations and subsequent years.

This measure assesses whether PAC providers were responsive to potential or actual clinically significant medication issue(s) when such issues were identified. Specifically, the proposed quality measure reports the percentage of resident stays in which a drug regimen review was conducted at the time of admission and timely follow-up with a physician occurred each time potential clinically significant medication issues were identified throughout that stay. For this proposed quality measure, a drug regimen review is defined as the review of all medications or drugs the patient is taking to identify any potential clinically significant medication issues. This proposed quality measure utilizes both the processes of medication reconciliation and a drug regimen review, in the event an actual or potential medication issue occurred. The measure informs whether the PAC facility identified and addressed each clinically significant medication issue and if the facility responded or addressed the medication issue in a timely manner. Of note, drug regimen review in PAC settings is generally considered to include medication reconciliation and review of the patient's drug regimen to identify potential clinically significant medication issues.^[104] (Please note: In the proposed rule, footnote 94 was inadvertently labeled ibid, which attributed the reference to the American Geriatric Society. In this final rule, we have corrected the reference and replaced it with the intended one, Institute of Medicine. Preventing Medication Errors. Washington, DC: National Academies Press; 2006.) This measure is applied uniformly across the PAC settings.

Medication reconciliation is a process of reviewing an individual's complete and current medication list. Medication reconciliation is a recognized process for reducing the occurrence of medication discrepancies that may lead to Adverse Drug Events (ADEs).^[105] Medication discrepancies occur when there is conflicting information documented in the medical records. The World Health Organization regards medication reconciliation as a standard operating protocol necessary to reduce the potential for ADEs that cause harm to patients. Medication reconciliation is an important patient safety process that addresses medication accuracy during transitions in resident care and in identifying preventable ADEs.^[106] The Joint Commission added medication reconciliation to its list of National Patient Safety Goals (2005), suggesting that medication reconciliation is an integral component of medication safety.^[107] The Society of Hospital Medicine published a statement in agreement of the Joint Commission's emphasis and value of medication reconciliation as a patient safety goal.^[108] There is universal agreement that medication reconciliation directly addresses resident safety issues that can result from medication miscommunication and unavailable or incorrect information.^{[109]  [110] [111]}

The performance of timely medication reconciliation is valuable to the process of drug regimen review. Preventing and responding to ADEs is of critical importance as ADEs account for significant increases in health services utilization and costs ^{[112]  [113]  [114]} including subsequent emergency room visits and re-hospitalizations.^[115] Annual health care costs from ADEs in the United States are estimated at $3.5 billion, resulting in 7,000 deaths annually.^[116]

Medication errors include the duplication of medications, delivery of an incorrect drug, inappropriate drug omissions, or errors in the dosage, route, frequency, and duration of medications. Medication errors are one of the most common types of medical errors and can occur at any point in the process of ordering and delivering a medication. Medication errors have the potential to result in an ADE.^{[117]  [118]  [119]  [120]  [121]  [122]} Inappropriately prescribed medications are also considered a major healthcare concern in the United States for the elderly population, with costs of roughly $7.2 billion annually.^[123]

There is strong evidence that medication discrepancies occur during transfers from acute care facilities to post-acute care facilities. Discrepancies occur when there is conflicting information documented in the medical records. Almost one-third of medication discrepancies have the potential to cause patient harm.^[124] Medication discrepancies upon admission to SNFs have been reported as occurring at a rate of more than 21 percent. It has been found that at least one medication discrepancy occurred in more than 71 percent of all the SNF admissions.^[125] An estimated fifty percent of patients experienced a clinically important medication error after hospital discharge in an analysis of two tertiary care academic hospitals.^[126]

Medication reconciliation has been identified as an area for improvement during transfer from the acute care facility to the receiving post-acute care facility. Post-acute care facilities report gaps in medication information between the acute care hospital and the receiving post-acute care setting when performing medication reconciliation.^[127 128] Hospital discharge has been identified as a particularly high risk point in time, with evidence that medication reconciliation identifies high levels of discrepancy.^{[129]  [130]  [131]  [132]  [133]  [134]} Also, there is evidence that medication reconciliation discrepancies occur throughout the patient stay.^{[135]  [136]} For older patients who may have multiple comorbid conditions and thus multiple medications, transitions between acute and post-acute care settings can be further complicated,^[137] and medication reconciliation and patient knowledge (medication literacy) can be inadequate post-discharge.^[138] The proposed quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, provides an important component of care coordination for PAC settings and would affect a large proportion of the Medicare population who transfer from hospitals into PAC services each year. For example, in 2013, 1.7 million Medicare FFS beneficiaries had SNF stays, 338,000 beneficiaries had IRF stays, and 122,000 beneficiaries had LTCH stays.^[139]

A TEP convened by our measure development contractor provided input on the technical specifications of this proposed quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, including components of reliability, validity and the feasibility of implementing the measure across PAC settings. The TEP supported the measure's implementation across PAC settings and was supportive of our plans to standardize this measure for cross-setting development. A summary of the TEP proceedings is available on the PAC Quality Initiatives Downloads and Video Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

We solicited stakeholder feedback on the development of this measure by means of a public comment period held from September 18 through October 6, 2015. Through public comments submitted by several stakeholders and organizations, we received support for implementation of this measure. The public comment summary report for the measure is available on the CMS Public Comment Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

The NQF-convened MAP met on December 14 and 15, 2015 and provided input on the use of this proposed quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP. The MAP encouraged continued development of the proposed quality measure to meet the mandate added by the IMPACT Act. The MAP agreed with the measure gaps identified by us including medication reconciliation, and stressed that medication reconciliation be present as an ongoing process. More information about the MAPs recommendations for this measure is available at http://www.qualityforum.org/​Publications/​2016/​02/​

MAP_2016_Considerations_for_Implementing_Measures_in_Federal_Programs_-_PAC-LTC.aspx.

Since the MAP's review and recommendation of continued development, we have continued to refine this measure consistent with the MAP's recommendations. The measure is consistent with the information submitted to the MAP and support its scientific acceptability for use in quality reporting programs. Therefore, we proposed this measure for implementation in the SNF QRP as required by the IMPACT Act.

We reviewed the NQF's endorsed measures and identified one NQF-endorsed cross-setting quality measure related to medication reconciliation, which applies to the SNF, LTCH, IRF, and HHA settings of care: Care for Older Adults (COA) (NQF #0553). The quality measure, Care for Older Adults (COA) (NQF #0553) assesses the percentage of adults 66 years and older who had a medication review. The Care for Older Adults (COA) (NQF #0553) measure requires at least one medication review conducted by a prescribing practitioner or clinical pharmacist during the measurement year and the presence of a medication list in the medical record. This is in contrast to the proposed quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, which reports the percentage of resident stays in which a drug regimen review was conducted at the time of admission and that timely follow-up with a physician occurred each time one or more potential clinically significant medication issues were identified throughout that stay.

After careful review of both quality measures, we decided to propose the quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP for the following reasons:

• The IMPACT Act requires the implementation of quality measures using patient assessment data that are standardized and interoperable across PAC settings. The quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, employs three standardized resident-assessment data elements for each of the four PAC settings so that data are standardized, interoperable, and comparable; whereas, the Care for Older Adults (COA), (NQF #0553) quality measure does not contain data elements that are standardized across all four PAC settings.
• The quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, requires the identification of potential clinically significant medication issues at the beginning, during and at the end of the resident's stay to capture data on each resident's complete PAC stay; whereas, the Care for Older Adults (COA), (NQF #0553) quality measure only requires annual documentation in the form of a medication list in the medical record of the target population.
• The quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, includes identification of the potential clinically significant medication issues and communication with the physician (or physician designee), as well as resolution of the issue(s) within a rapid timeframe (by midnight of the next calendar day); whereas, the Care for Older Adults (COA), (NQF #0553) quality measure does not include any follow-up or timeframe in which the follow-up would need to occur.
• The quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, does not have age exclusions; whereas, the Care for Older Adults (COA), (NQF #0553) quality measure limits the measure's population to patients aged 66 and older.
• The quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, will be reported to SNFs quarterly to facilitate internal quality monitoring and quality improvement in areas such as resident safety, care coordination and resident satisfaction; whereas, the Care for Older Adults (COA), (NQF #0553) quality measure would not enable quarterly quality updates, and thus data comparisons within and across PAC providers would be difficult due to the limited data and scope of the data collected.

Therefore, based on the evidence discussed above, we proposed to adopt the quality measure entitled, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, for the SNF QRP for FY 2020 payment determination and subsequent years. We plan to submit the quality measure to the NQF for consideration for endorsement.

The calculation of the proposed quality measure would be based on the data collection of three standardized items to be included in the MDS. The collection of data by means of the standardized items would be obtained at admission and discharge. For more information about the data submission required for this measure, please see section V.B.9. of the FY 2017 SNF PPS proposed rule (81 FR 24270 through 24273).

The standardized items used to calculate this proposed quality measure do not duplicate existing items currently used for data collection within the MDS. The measure denominator is the number of resident stays with a discharge or expired assessment during the reporting period. The measure numerator is the number of stays in the denominator where the medical record contains documentation of a drug regimen review conducted at: (1) Admission; and (2) discharge with a look back through the entire resident stay, with all potential clinically significant medication issues identified during the course of care and followed-up with a physician or physician designee by midnight of the next calendar day. This measure is not risk adjusted. For technical information about this measure including information about the measure calculation and discussion pertaining to the standardized items used to calculate this measure, refer to the document titled, Proposed Measure Specifications for Measures Proposed in the FY 2017 SNF QRP Proposed Rule available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

Data for the proposed quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, would be collected using the MDS with submission through the Quality Improvement Evaluation System (QIES) Assessment Submission and Processing (ASAP) system.

We invited public comment on our proposal to adopt the quality measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, for the SNF QRP. The comments we received on this topic, with their responses, appear below.

Comment: Several commenters, including MedPAC, expressed support for the quality measure. Further, several commenters expressed appreciation to us for proposing a quality measure to address the IMPACT Act domain, Medication Reconciliation, acknowledging the importance of medication reconciliation for addressing resident safety issues. Several commenters emphasized the importance of preventing and responding to Adverse Drug Events (ADEs) to reduce health services utilization and associated healthcare costs and emphasized that medication reconciliation is fundamental to resident safety during care transitions.

Response: We appreciate the commenters' support for the quality measure and the recognition of the importance of medication reconciliation as addressed in the measure. We agree that medication reconciliation is an important patient safety process for addressing medication accuracy during transitions in patient care and identifying preventable Adverse Drug Events (ADEs), which may lead to reduced health services utilization and associated costs.

Comment: We received several comments regarding concerns about whether the measure has continued to be refined since the NQF-convened MAP meeting in December 2015. Many commenters noted that the MAP recommended “continued development” for the measure and requested evidence of robust testing of the measure to support measure validity. Several commenters requested that we test this measure prior to implementing it as part of the quality reporting system. One commenter further expressed that testing would enable us to more fully understand the benefits and limitations of the measure and its implication for providers and patients. Several commenters expressed concern that the measure was not NQF endorsed.

Response: Since the time of the NQF-convened MAP, with our measure contractor, we tested this measure in a pilot test involving twelve post-acute care facilities (IRF, SNF, LTCH), representing variation across geographic location, size, profit status, and clinical records system. Two clinicians in each facility collected data on a sample of 10 to 20 patients for a total of 298 records (147 qualifying pairs). Analysis of agreement between coders within each participating facility indicated a 71 percent agreement for item DRR-01 ^[140] Drug Regimen Review (admission); 69 percent agreement for item DRR-02 ^[141] Medication Follow-up (admission); and 61 percent agreement for DRR-03 ^[142] Medication Intervention (During Stay and Discharge). Overall, pilot testing enabled us to verify feasibility of the measure. Furthermore, measure development included convening a technical expert panel (TEP) to provide input on the technical specifications of this proposed quality measure, including components of reliability, validity and the feasibility of implementing the measure across PAC settings. The TEP included SNF stakeholders and supported the measure's implementation across PAC settings and was supportive of our plans to standardize this measure for cross-setting development. A summary of the TEP proceedings is available on the PAC Quality Initiatives Downloads and Videos Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​Post-Acute-Care-Quality-Initiatives/​IMPACT-Act-of-2014/​IMPACT-Act-Downloads-and-Videos.html.

As noted above, we plan to conduct further testing on this measure once we have started collecting data from the PAC settings. Analysis of this data will allow us to evaluate whether the measure satisfies NQF endorsement criteria (for example, measure performance). Once we have completed this additional measure performance testing, we plan to submit the measure to NQF for endorsement.

Comment: We received several comments about the lack of a specific definition of clinically significant medication issues for the measure. Several commenters were concerned that the phrase could be interpreted differently by the many providers involved in a resident's treatment, and that this could result in a challenge to collect reliable and accurate data for this quality measure. Several commenters requested that we provide additional guidance regarding this definition. One commenter suggested that it was premature for us to provide clarifying language because a related proposed rule regarding Discharge Planning (Reform of Requirements for Long-Term Care Facilities, 80 FR 42168) has not been finalized. One commenter further conveyed that, without further guidance on the definition of clinically significant, there are likely to be variations in measure performance that are not based on differences in care, but rather on differences in data collection.

Response: For this measure, potential clinically significant medication issues are defined as those issues that, in the clinician's professional judgment, warrant interventions, such as alerting the physician and/or others, and the timely completion of any recommended actions (by midnight of the next calendar day) so as to avoid and mitigate any untoward or adverse outcomes. The definition of “clinically significant” in this measure was conceptualized during the measure development process. For purposes of the measure, the decision regarding whether or not a medication issue is “clinically significant” will need to be made on a case-by-case basis, but we also intend to provide additional guidance and training on this issue.

Comment: We received several comments related to the State Operations Manual (SOM) § 483.60(c). One commenter requested that we provide further guidance on how the measure relates to the “medication regimen review” within the SOM. Many commenters recommended that the definitions of potentially clinically significant medication issues and drug regimen review align with similar definitions in the SOM. One commenter further requested that we allow the existing SNF SOM required reviews to fulfill the requirements of the measure. One commenter further noted that the definitions contained in the measure are not as clinically detailed (as the SOM), are not PAC setting inclusive, and do not acknowledge the need for a multiple disciplinary team. The commenter also noted that the SOM uses the term “medication” rather than “drug” and offers that “medication” is a more appropriate title to the measure. One commenter conveyed a need for clarification in how the measure will interface with the current SNF requirements for drug regimen review. One commenter expressed concern that the requirements of the measure potentially conflict with the requirements CMS SNF State Operations Manual.

Response: We acknowledge the commenters' request to align other regulatory requirements involving medication regimen review with the measure such as the State Operations Manual § 483.60(c). We would like to note that during the development of this measure, the definitions as detailed in the SOM were taken into consideration. We do not believe that the measure's use of terminology of “clinically significant” overrides the guidance as outlined in the SOM. Further, we wish to clarify that the specification of the measure does not preclude the activities of drug regimen reviews that are consistent with the SOM. We would like to reiterate that this measure was developed to assess whether PAC providers were responsive to potential or actual clinically significant medication issue(s) when such issues were identified and was not developed for regulatory purposes for Skilled Nursing Facilities to be in compliance with the requirements of the 42 CFR part 483. In particular, the SOM Appendix PP—Guidance to Surveyors for Long Term Care Facilities, under § 483.60(c) Drug Regimen Review, references pharmacy services requirements where: (1) The drug regimen of each resident must be reviewed at least once a month by a licensed pharmacist; and (2) The pharmacist must report any irregularities to the attending physician, and the director of nursing, and these reports must be acted upon. The measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP reports the percentage of resident stays in which a drug regimen review was conducted at the time of admission, and timely follow-up with a physician occurred each time potential clinically significant medication issues were identified throughout that stay.

Comment: Several commenters were concerned that the measure does not meet the medication reconciliation domain of the IMPACT Act. In particular, these commenters believe that the proposed quality measure goes beyond the statutory mandate by incorporating drug regimen (medication) review into the measure. Commenters supported measure development related to the concepts of drug regimen review and medication reconciliation in reducing unnecessary rehospitalizations, preventable adverse events, and improving health care outcomes, but maintained that the services provided as part of drug regimen review are distinctly different from the services provided as part of medication reconciliation, and that they are completed by different members of the care team. One commenter conveyed that the measure has not been proven to be relevant to medication reconciliation.

Response: We disagree with the commenters' suggestion that the measure does not meet the requirements of the IMPACT Act. Medication reconciliation and drug regimen review are interrelated activities; while medication reconciliation is a process that identifies the most accurate and current list of medications, particularly during transitions of care, it also includes the evaluation of the name, dosage, frequency, and route. Drug regimen review is a process that necessitates and includes the review of all medications for additional purposes such as the identification of potential adverse effects. The process of drug regimen review includes medication reconciliation at the time of resident transitions and throughout the resident's stay. Therefore, we believe that medication reconciliation and drug regimen review are processes that are appropriate to combine in a single measure for purposes of the SNF QRP.

Comment: We received several comments regarding the time frame for the measure and resulting burden. Several commenters noted that requiring SNFs to notify the physician within one day was unreasonable. One commenter was concerned that the requirement that a physician be contacted within a day was too prescriptive, given that it may take more than a day for a physician to return a call, and suggested that we adopt a more reasonable standard. Further, another commenter suggested that this timeline created a mandate that many SNFs simply won't be able to meet. One commenter acknowledged that medication issues need to be resolved with urgency, but conveyed that the timeframe requirements of the measure are not feasible, citing limitations with the prescriber's and the hospitalist's availability to respond to issues and limited access to information technology that supports the prompt resolution of issues. Another commenter also noted that while clinically significant medical issues are required to be reported in a timely process, the word timely has not been adequately defined. One commenter suggested that we abandon the measure and instead verify that medication reconciliation is provided upon admission. Another commenter suggested that we clarify whether physician follow up is only required for clinically significant issues, rather than each time the drug regimen review is conducted.

Several commenters conveyed concern that the time frame of the measure (for example, following up by midnight of the next calendar day) will create challenges for rural SNFs without an in-house pharmacy or physicians, and that the measure will increase operational and financial challenges for long-term care providers. A few commenters asked us to consider reforms to mitigate the burden for providers located in rural areas. Another commenter conveyed that additional questions on the MDS would result in additional staff cost and effort. One commenter noted that many SNFs have not implemented electronic medical records, which will increase the burden associated with collecting this information. One commenter recommended that we work with stakeholders to develop a policy that aligns with the resident's best interest and accounts for the complex post-acute care setting.

Response: We appreciate the challenges that SNFs face when they have to coordinate resident care with a treatment team that may include physicians, non-physician practitioners, pharmacists and others, and also appreciate that some of these treatment team members might not work full-time at the SNF. However, we chose to set the intervention timeline as midnight of the next calendar day because we believe this timeline is consistent with current standard clinical practice where a clinically significant medication issue arises. We believe that high quality care should be provided wherever resident services are administered, including small and rural facilities, and that these activities, in addition to any regulatory requirements, ensure such high quality care is provided and patient harm avoided.

Comment: We received several comments related to the role of pharmacists in drug regimen review. One commenter expressed concern that the measure would require frequent consultant pharmacist visits to the SNF without providing more funding to cover additional expenses. Many commenters suggested that we redefine the measure to allow the SNF to determine which licensed professional provides the medication reconciliation. These commenters recommended that we recognize the essential role that pharmacists play in providing services to beneficiaries. One commenter submitted a study that noted the monetary savings that drug regimen review by pharmacists have provided to post-acute care residential facilities. Several commenters expressed that pharmacists should receive compensation for service they provide around this measure. One commenter encouraged us to consider ways in which to provide incentives to LTC pharmacies for the savings and improved care.

Response: We recognize the essential role that pharmacists, as well as other members of the SNF treatment team, play in furnishing services to Medicare beneficiaries. This measure does not supersede or conflict with current CMS guidance or regulations related to drug regimen review. The measure also does not specify what clinical professional is required to perform these activities.

Comment: We received several comments pertaining to the scope of the measure. One commenter conveyed that the CMS definition of Medication Reconciliation in a measure for hospitals differs from the definition for purposes of the proposed SNF QRP measure. One commenter conveyed opposition to the measure, expressing that the measure calculation proposes to capture a number of action steps within this single measure. Many commenters expressed concerns that the measure may not accurately capture SNF performance, given all the work that the SNF and pharmacy undertake to ensure that medication-related issues are addressed prior to dispensing medication.

Response: The Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP measure evaluates medication reconciliation in conjunction with drug regimen review in the post acute care setting, which distinguishes it from solely medication reconciliation that is conducted in the hospital which we believe the commenter is referring to. We believe it is appropriate that the measure captures multiple action steps in a single measure as drug regimen review is a multifaceted process that should take place throughout the resident's stay.

Comment: We received a comment suggesting that we inaccurately represented that an article by American Geriatric Society suggests (and therefore aides our position) that drug regimen review includes a medication reconciliation and review of the patient's drug regimen to identify potential issues.

Response: The commenter is correct regarding an inaccurate reference. We inadvertently attributed reference to the American Geriatric Society in our discussion. Therefore, we have corrected the reference and replaced it with the intended one (Institute of Medicine. Preventing Medication Errors. Washington, DC: National Academies Press; 2006).

Comment: One commenter supported the need for medication reconciliation, but had concerns about factors outside the facility's control. The commenter conveyed the challenge of medication reconciliation across the continuum, conveying the importance of a discharge summary from the prior care setting that includes a thorough medication list, by indication, in avoiding therapeutic duplication. The commenter suggested that we consider the need for increased collaboration with hospitals to address this issue. Other commenters, including MedPAC, suggested that we develop a measure that evaluates whether PAC providers are sending medication lists home or to the next level of care. These commenters suggested that requiring providers to transfer medication lists may improve monitoring of the patient's condition, which may help prevent readmissions and unintended medical harm. Another commenter recommended that we add a medication management measure to fully address patients' medication management routine needs in order to prepare patients for discharge to PAC settings or the community.

Response: We appreciate the comments about the importance of collaboration across the continuum of care, as well as the value of a detailed discharge summary from the prior level of care. We believe that all providers should strive to ensure accurate, sufficient, and efficient patient-centered care during their care transitions across the continuum, including medication oversight. Thus while we may implement quality measures that address gaps in quality, such as information exchange during care transitions, ultimately providers must act to ensure that such coordination is taking place.

We appreciate the commenter's comment and interest in future quality measure development, including measures related to sending a medication list at discharge and adding a medication management measure. As a requirement of this measure and as with common clinical practice, PAC facilities are expected to document information pertaining to the process of drug regimen review, which includes medication reconciliation, in the resident's discharge medical record. However, we will take the commenters recommendations into consideration as we continue to develop additional quality measures under the domain of Medication Reconciliation.

Comment: One commenter encouraged us to make the reporting of the measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP, available to SNFs in real time through the CASPER Quality Measures report in QIES ASAP system.

Response: We thank the commenter for their suggestion. We anticipate making this measure information available to SNFs in the CASPER Quality Measures reports beginning approximately in October, 2020. Confidential SNF feedback on this measure will be made available to SNFs in October, 2019.

Comment: We received a comment about the role of registered nurses in the medication reconciliation process. The commenter recognized the critical importance of medication reconciliation and cited research demonstrating that registered nurses (RNs) are more likely to identify medication discrepancies in nursing facilities than licensed practical nurses (LPNs); the commenter encouraged us, in the Conditions of Participation for Skilled Nursing Facilities (SNFs) and Nursing Facilities (NFs), to require that facilities employ RNs 24 hours per day.

Response: We thank the commenter for recognizing the importance of medication reconciliation and the role of registered nurses in the medication reconciliation process.

Comment: We received a comment about materials that were posted on the CMS Public Comment Web site for a public comment period held from September 18 through October 6, 2015. The comment specifically included specific questions regarding the language used in the “Importance” section of the Measure Justification Form, which requests the measure developer quote verbatim currently published clinical practice guidelines. The commenter noted the absence of an “Outcome 1,” which is defined as functional status, in the quoted material. Additionally, the commenter expressed concern about specific targets within the goal of reducing polypharmacy and about guidelines for calculating creatinine clearance levels and about the Cockcroft Gault Score. Finally, the commenter noted that it is clinically unrealistic to have an expected outcome of “No adverse drug reactions, no drugs ordered to treat side effects or adverse reaction.”

Response: We thank the commenter for their comments but wish to clarify that the document they reference, the Measure Justification Form, was posted for a prior public comment period that was not part of the proposed rule. We also wish to clarify that language that was commented on was derived directly from published clinical practice guidelines and not by CMS.

Final Decision: After consideration of the public comments, we are finalizing our proposal to adopt the measure, Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP measure for the SNF QRP for the FY 2020 payment determination and subsequent years, as described in the Measure Specifications for Measures Adopted in the FY 2017 SNF QRP final rule, available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​Skilled-Nursing-Facility-Quality-Reporting-Program/​SNF-Quality-Reporting-Program-Measures-and-Technical-Information.html.

h. SNF QRP Quality Measures and Measure Concepts Under Consideration for Future Years

We invited comment on the importance, relevance, appropriateness, and applicability for each of the quality measures in Table 13 for future years in the SNF QRP. We are developing a measure related to the IMPACT Act domain, accurately communicating the existence of and providing for the transfer of health information and care preferences of an individual to the individual, family caregiver of the individual, and providers of services furnishing items and services to the individual, when the individual transitions. We are considering the possibility of adding quality measures that rely on the patient's perspective; that is, measures that include patient-reported experience of care and health status data. For this purpose, we are considering a measure focused on pain and four measures focused on function that rely on the collection of patient-reported data. Finally, we are considering a measure related to health and well-being, Percent of Residents or Patients Who Were Assessed and Appropriately Given the Seasonal Influenza Vaccine, and a measure related to patient safety, Percent of SNF Residents Who Newly Received an Antipsychotic Medication.

Table 13—SNF QRP Quality Measures Under Consideration for Future Years

IMPACT Act Domain	Accurately communicating the existence of and providing for the transfer of health information and care preferences of an individual to the individual, family caregiver of the individual, and providers of services furnishing items and services to the individual, when the individual transitions.
IMPACT Act Measure	• Transfer of health information and care preferences when an individual transitions.
NQS Priority	Patient- and Caregiver-Centered Care.
Measures	• Percent of Residents Who Self-Report Moderate to Severe Pain • Application of the Change in Self-Care Score for Medical Rehabilitation Patients (NQF #2633) • Application of the Change in Mobility Score for Medical Rehabilitation Patients (NQF #2634) • Application of the Discharge Self-Care Score for Medical Rehabilitation Patients (NQF #2635) • Application of the Discharge Mobility Score for Medical Rehabilitation Patients (NQF #2636).
NQS Priority	Health and Well-Being.
Measure	• Percent of Residents or Patients Who Were Assessed and Appropriately Given the Seasonal Influenza Vaccine.
NQS Priority	Patient Safety.
Measure	• Percent of SNF Residents Who Newly Received an Antipsychotic Medication.
The comments we received on this topic, with their responses, appear below.

Comment: We received several comments supporting the inclusion of measures regarding the transfer of health information and care preferences. One commenter encouraged the inclusion of measures that capture the role of family caregivers in supporting care transitions, quality outcomes, and individual care preferences. Another commenter recommended pilot testing measures regarding transfer of health information and preferences; while another suggested a measure that would incentivize the adoption of health IT around the domain requirement to support the electronic transmission of health information and care preferences.

Response: We thank the commenters for their comments and agree that the transfer of health information across PAC settings is important to capture. As we move through the development of this measure concept, we will consider the inclusion of the role of family caregivers in supporting care transitions, quality outcomes, and individual care preferences. In addition, we will take into consideration the commenters' recommendations pertaining to the pilot testing for these measure concepts.

Comment: We received comments that were broadly supportive of patient- and caregiver-reported measures and agreed that they are meaningful to patients and their families.

Response: We thank the commenters for their support of patient-reported measures under consideration for future implementation in the SNF QRP and agree with the importance of patient- and caregiver-centered measures such as these.

Comment: Several commenters supported the potential future use of the four self-reported function measures. One commenter supported risk adjustment of these measures and the focus on patient-centered outcomes. Another supported the use of the four self-reported function measures applied from the IRF setting and emphasized the importance of alignment across PAC settings and encouraged measure testing in the SNF setting prior to implementation. Another commenter recommended that SNF residents should be excluded from measures related to change in function if there is no expectation of functional improvement.

Several commenters suggested the development of function measures addressing cognition. One commenter remarked on the limited number of items in the MDS related to communication, cognition, and swallowing and noted that these three domains stand as major obstacles to validly determine the status, needs, and outcomes of individuals with neurological disorders. The commenter encouraged us to adopt a specific screening tool, the Montreal Cognitive Assessment (MoCA), or similar screening tools and assessment tools (that is, CARE-C) to best meet the needs of Medicare beneficiaries and the intent of the IMPACT Act.

Another commenter recommended that we consider community-based measures of function, examining patient outcomes after they are discharged from a PAC setting. One commenter encouraged the development of an outcome measure to meet the IMPACT Act domain of functional status, suggesting the NH Compare measure, Percent of Residents Whose Need for Help with Activities of Daily Living has Increased (Long Stay).

Response: We thank the commenters for their support of the four self-reported function measures under consideration for future implementation in the SNF QRP. We also appreciate commenters' suggestions regarding the development and specification of these measures as well as additional measure concepts or areas related to function that we should consider. We agree that the implementation of outcome measures of function in the SNF QRP is a priority. We also agree that future measure development should include other areas of function, such as communication, cognition, and swallowing. We will continue to engage stakeholders in future measure development. We will take these suggested quality measure concepts and recommendations regarding measure specifications into consideration in our ongoing measure development and testing efforts.

Comment: We received several comments regarding pain management and prevention. One commenter suggested that we consider HCAHPS measures related to pain control, while another commenter suggested such a measure should reflect a patient-centered approach to pain management instead of level and frequency of pain symptoms. We also received a comment encouraging the use of the CAHPS NH survey to examine resident and family members' experience of care.

Response: We will take these suggested quality measure concepts and recommendations regarding measure specifications into consideration in our ongoing measure development and testing efforts.

Comment: We received several comments supporting a future seasonal influenza vaccination measure. Several commenters encouraged us to consider other immunization measures for the SNF QRP, including a pneumococcal vaccine measure. One commenter encouraged consideration of the cost of delivering these services as they may have financial implications for SNFs.

Response: We thank the commenters for their support of a future seasonal influenza vaccination measure. Cost burden for providers is always a consideration as we develop and implement new measures. We appreciate the commenters' feedback on potential measure development areas related to immunization. We will take their recommendations into consideration in our measure development and testing efforts, as well as in our ongoing efforts to identify and propose appropriate measures for the SNF QRP in the future.

Comment: We received several comments supporting the inclusion of the antipsychotic quality measure (listed on the Nursing Home Compare Web site) in the SNF QRP. One commenter supported the measure but cautioned against adapting the pre-existing, non-NQF-endorsed antipsychotic measures currently used in nursing homes, indicating that these process measures do not provide a linkage to clinical outcomes or intermediate outcomes. Commenters also emphasized the need for the measures to account for situations where continued or newly prescribed antipsychotics would be clinically appropriate.

Response: We appreciate commenters' feedback on this potential measure development area. We will take their recommendations into consideration in our measure development and testing efforts, as well as in our ongoing efforts to identify and propose appropriate measures for the SNF QRP in the future.

Comment: Commenters suggested additional measures and measure concepts for us to consider for future implementation in the SNF QRP, including workforce-related measures and measures assessing resident experience of care, engagement, and shared decision-making. Several commenters recommended that CMS consider incorporating various Nursing Home Compare measures into the SNF QRP.

Response: We thank commenters for their suggestions regarding areas for potential future measure development. We will take their recommendations into consideration in our measure development and testing efforts, as well as in our ongoing efforts to identify and propose appropriate measures for the SNF QRP in the future.

i. Form, Manner, and Timing of Quality Data Submission

i. Participation/Timing for New SNFs

In the FY 2016 SNF PPS final rule (80 FR 46455), we established the requirements associated with the timing of data submission, beginning with the submission of data required for the FY 2018 payment determination, for new SNFs. We finalized that a new SNF would be required to begin reporting data on any quality measures finalized for that program year by no later than the first day of the calendar quarter subsequent to 30 days after the date on its CMS Certification Number (CCN) notification letter. For example, for the FY 2018 payment determinations, if a SNF received its CCN on August 28, 2016, and 30 days are added (August 28 + 30 days = September 27), the SNF would be required to submit data for residents who are admitted beginning on October 1, 2016. We did not propose any new policies related to the participation and timing for new SNFs.

ii. Finalized Data Collection Timelines and Requirements for the FY 2018 Payment Determination and Subsequent Years

In the FY 2016 SNF PPS final rule (80 FR 46457), for the FY 2018 payment determination, we finalized that SNFs submit data on the three finalized quality measures for residents who are admitted to the SNF on and after October 1, 2016, and discharged from the SNF up to and including December 31, 2016, using the data submission method and schedule that we proposed in this section. We also finalized that we would collect that single quarter of data for FY 2018 to remain consistent with the usual October release schedule for the MDS, to give SNFs a sufficient amount of time to update their systems so that they can comply with the new data reporting requirements, and to give CMS a sufficient amount of time to determine compliance for the FY 2018 program. The proposed use of one quarter of data for the initial year of quality reporting is consistent with the approach we used to implement a number of other QRPs, including the LTCH, IRF, and Hospice QRPs.

We also finalized that, following the close of the reporting quarter, October 1, 2016, through December 31, 2016, for the FY 2018 payment determination, SNFs would have an additional 5.5 months to correct and/or submit their quality data and we finalized that the final deadline for submitting data for the FY 2018 payment determination would be May 15, 2017 (80 FR 46457). The statement that SNFs would have an additional 5.5 months was incorrect in that the time between the close of the quarter on December 31, 2016 and May 15, 2017 is 4.5 months, not 5.5 months. Therefore, we proposed that SNFs will have 4.5 months, from January 1, 2017 through May 15, 2017, following the data submission period of October 1, 2016 through December 31, 2016, in which to complete their data submissions and make corrections to their data where necessary.

Table 14—Finalized Measures, Data Collection Source, Data Collection Period and Data Submission Deadlines Affecting the FY 2018 Payment Determination

Quality measure	Data collection source	Data collection period	Data submission deadline for FY 2018 payment determination
NQF # 0678: Percent of Patients or Residents with Pressure Ulcers that are New or Worsened	MDS	10/01/16-12/31/16	May 15, 2017.
NQF # 0674: Application of Percent of Residents Experiencing One or More Falls with Major Injury (Long Stay)	MDS	10/01/16-12/31/16	May 15, 2017.
NQF # 2631: Application of Percent of Long-Term Care Hospital Patients with an Admission and Discharge Functional Assessment and a Care Plan that Addresses Function	MDS	10/01/16-12/31/16	May 15, 2017.

We invited public comments on our proposal to correct the time frame for SNFs to correct and/or submit their quality data used for the FY 2018 payment determination to consist of 4.5 months rather than the 5.5 months stated in the FY 2016 SNF PPS final rule (80 FR 46457). We received no comments on this proposed correction.

Final decision: We are finalizing as proposed that for the FY 2018 payment determination, SNFs will have 4.5 months following the end of the reporting quarter to complete their data submissions and make corrections to their data where necessary.

iii. Data Collection Timelines and Requirements for the FY 2019 Payment Determinations and Subsequent Years

In the FY 2016 SNF PPS final rule (80 FR 46457), we finalized that, for the FY 2019 payment determination, we would collect data from the 2nd through 4th quarters of FY 2017 (that is, data for residents who are admitted from January 1st and discharged up to and including September 30th) to determine whether a SNF has met its quality reporting requirements for that FY. In the FY 2016 SNF PPS final rule we also finalized that beginning with the FY 2020 payment determination, we would move to a full year of fiscal year (FY) data collection. We intend to propose the FY 2019 payment determination quality reporting data submission deadlines in future rulemaking.

In the FY 2016 SNF PPS final rule (80 FR 46457), we also finalized that we would collect FY 2018 data in a manner that would remain consistent with the usual October release schedule for the MDS. However, to align with the data reporting cycles in other quality reporting programs, in contrast to fiscal year data collection that we finalized last year, we are now proposing to move to calendar year (CY) reporting following the initial reporting of data from October 1, 2016, through December 31, 2016, as finalized in the FY 2016 SNF PPS final rule (80 FR 46457), for the FY 2018 payment determination.

More specifically, we proposed to follow a CY schedule for measure and data submission requirements that includes quarterly deadlines following each quarter of data submission, beginning with data reporting for the FY 2019 payment determinations. Each quarterly deadline will occur approximately 4.5 months after the end of a given calendar quarter as outlined below in Table 15. This timeframe will give SNFs enough time to submit corrections to the assessment data, as discussed below. Thus, if finalized, the FY 2019 payment determination would be based on 12 calendar months of data reporting beginning on January 1, 2017, and ending on December 31, 2017 (that is, data from January 1, 2017, up to and including December 31, 2017.) This approach would enable CMS to move to a full 12 months of data reporting immediately following the first 3 months of reporting (October 1, 2016 through December 31, 2016 for the FY 2018 payment determination) rather than an interim year which uses only 9 months of data, and a subsequent 12 months of FY data reporting following the initial reporting for the FY 2018 payment determination.

Our proposal to implement, for the FY 2019 payment determination and all subsequent years for assessment-based data submitted via the MDS, calendar year, quarterly data collection periods followed by data submission deadlines is consistent with the approach taken by the LTCH QRP and the IRF QRP, which are based on CY data and for which each data collection quarterly period is followed by a 4.5 month time frame that allows for the continued submission and correction of data until a deadline has been reached for that quarter of data. At that point, the data submitted becomes a frozen “snapshot” of data for both public reporting purposes and for the purposes of determining compliance in meeting the data reporting thresholds.

Table 15—Proposed Data Collection Period and Data Submission Deadlines Affecting the FY 2019 Payment Determination and Subsequent Years

Quality measure	Data collection source	Data collection/submission quarterly reporting period *	Quarterly review and correction periods and data submission quarterly deadlines for FY 2019 payment determination **
NQF # 0678: Percent of Patients or Residents with Pressure Ulcers that are New or Worsened	MDS	CY 2017 Q1—1/1/2017-3/31/2017	CY 2017 Q1 Deadline: August 15, 2017.
NQF # 0674: Application of Percent of Residents Experiencing One or More Falls with Major Injury (Long Stay)		CY 2017 Q2—4/1/2017-6/30/17	CY 2017 Q2 Deadline: November 15, 2017.
NQF #2631: Application of Percent of Long-Term Care Hospital Patients with an Admission and Discharge Functional Assessment and a Care Plan that Addresses Function		CY 2017 Q3—7/1/2017-9/30/2017 CY 2017 Q4—10/1/2017-12/31/2017	CY 2017 Q3 Deadline: February 15, 2018. CY 2017 Q4 Deadline: May 15, 2018.
* Data collection/submission will follow a similar quarterly reporting period schedule for subsequent CYs.
** Data review and correction periods and data submission deadlines will follow a similar quarterly schedule for subsequent CYs.

We invited public comments on our proposal to adopt calendar year data collection time frames, following the initial 3-month reporting period from October 1, 2016, to December 31, 2016, for all measures finalized for adoption into the SNF QRP. The comments we received on this topic, with their responses, appear below.

Comment: We received several comments supporting our proposal to move to a CY reporting schedule to align with the LTCH and IRF QRPs.

Response: We appreciate the commenters' support of our proposal to move to a calendar year reporting schedule, which is consistent with the approach we also use for the LTCH and IRF QRPs. We seek to align requirements across QRPs whenever possible.

Comment: We received one comment supporting the continuation of the October release schedule for updates to the MDS and the alignment of data collection with that October release schedule.

Response: We appreciate the commenters' support of our alignment of the beginning of the initial data collection period for new measures with the October release schedule for the MDS and moving to CY reporting following the initial data collection period.

Further, we proposed that beginning with FY 2019 payment determination, assessment-based measures finalized for adoption into the SNF QRP will follow a CY schedule of data reporting, quarterly review and correction periods, and data submission deadlines as provided in Tables 15 and 16 for all subsequent payment determination years unless otherwise specified:

Table 16—Proposed Data Collection Period and Data Submission Deadlines Affecting the FY 2019 Payment Determination and Subsequent Years

CY data collection quarter	Data collection/submission quarterly reporting period	Quarterly review and correction periods and data submission deadlines for payment determination
Quarter 1	January 1-March 31	April 1-August 15.
Quarter 2	April 1-June 30	July 1-November 15.
Quarter 3	July 1-September 30	October 1-February 15.
Quarter 4	October 1-December 31	January 1-May 15.

We invited public comments on the proposed data collection period and data submission deadlines for all assessment-based measures finalized for adoption into the SNF QRP beginning with the FY 2019 payment determination, specifically, on our use of CY reporting with data submission deadlines following a period of approximately 4.5 months after each quarterly data collection period to enable the correction of such data, as outlined in Table 16. We received no additional comments on this proposed general schedule.

Final decision: We are finalizing our proposed data collection period and data submission deadlines for all assessment-based measures finalized for adoption into the SNF QRP beginning with FY 2019 payment determination, as outlined in Tables 15 and 16.

iv. Timeline and Data Submission Mechanisms for Claims-Based Measures for the FY 2018 Payment Determination and Subsequent Years

The Medicare Spending per Beneficiary—PAC SNF QRP, Discharge to Community—PAC SNF QRP, and Potentially Preventable Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP measures are Medicare FFS claims-based measures. Because claims-based measures can be calculated based on data that are already reported to the Medicare program for payment purposes, no additional information collection will be required from SNFs. As discussed in section V.B.6. of the FY 2017 SNF PPS proposed rule (81 FR 24257 through 24267), for the Medicare Spending per Beneficiary—PAC SNF QRP Measure, the Discharge to Community—PAC SNF QRP measure and the Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP, we proposed to use 1 year of claims data beginning with CY 2016 claims data to inform confidential feedback reports for SNFs, and CY 2017 claims data for public reporting.

We invited public comments on this proposal. We did not receive any comments specifically related to this proposal.

Final Decision: We are finalizing the timeline and data submission mechanisms for claims-based measures proposed for the FY 2018 payment determination and subsequent years as proposed in Tables 15 and 16.

v. Timeline and Data Submission Mechanisms for the FY 2020 Payment Determination and Subsequent Years for New SNF QRP Assessment-Based Quality Measure

We proposed that SNFs would submit data on the Drug Regimen Review measure by completing data elements to be included in the MDS and then submitting the MDS to CMS through the Quality Improvement and Evaluation System (QIES), Assessment Submission and Processing System (ASAP) system beginning October 1, 2018. For more information on SNF QRP reporting through the QIES ASAP system, refer to the “Related Links” section at the bottom of https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​index.html?​redirect=​/​NursingHomeQualityInits/​30_​NHQIMDS30TechnicalInformation.asp#TopOfPage.

We invited public comments on our proposed SNF QRP data collection requirements for the Drug Regimen Review measure for the FY 2020 payment determination and subsequent years. We did not receive any comments related to this topic.

For the FY 2020 payment determination, we proposed that SNFs submit data on the proposed assessment-based quality measure for residents who are admitted to the SNF on and after October 1, 2018, and discharged from SNF Part A covered stays (that is, both residents discharged from Part A covered stays and physically discharged) up to and including December 31, 2018, using the data submission schedule that we proposed in this section.

We proposed to collect a single quarter of data for the FY 2020 payment determination to remain consistent with the usual October release schedule for the MDS, to give SNFs a sufficient amount of time to update their systems so that they can comply with the new data reporting requirements, and to give CMS a sufficient amount of time to determine compliance for the FY 2020 program. The proposed use of one quarter of data for the initial year of assessment data reporting in the SNF QRP is consistent with the approach we used previously for the SNF QRP and in other QRPs, including the LTCH, IRF, and Hospice QRPs in which we have finalized the use of fewer than 12 months of data.

We also proposed that following the close of the reporting quarter, October 1, 2018, through December 31, 2018, for the FY 2020 payment determination, SNFs would have an additional 4.5 months to correct and/or submit their quality data and that the final deadline for submitting data for the FY 2020 payment determination would be May 15, 2019. We further proposed that for the FY 2021 payment determination and subsequent years, we will collect data using the CY reporting cycle as previously proposed in section V.B.9.c. of the FY 2017 SNF PPS proposed rule (81 FR 24271 through 24272).

Table 17—Proposed New SNF QRP Assessment-Based Quality Measures Data Collection Period and Data Submission Deadlines Affecting the FY 2020 Payment Determination

Quality measure	Data collection source	Data collection/ submission reporting period	Data submission deadline for FY 2020 payment determination
Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP	MDS	10/01/18-12/31/18	May 15, 2019.

We invited public comment on the proposed new SNF QRP assessment-based quality measure data collection period and data submission deadline affecting the FY 2020 payment determination. We did not receive comments related to this topic.

Final Decision: We are finalizing as proposed the timeline and data submission mechanism for the FY 2020 payment determination for the new assessment-based quality as provided in Table 17.

For this measure, we also proposed to follow a CY schedule for measure and data submission requirements that includes quarterly deadlines following each quarter of data submission, beginning with data reporting for the FY 2021 payment determinations. As previously discussed, each quarterly deadline will occur approximately 4.5 months after the end of a given calendar quarter as outlined in Table 18. Thus, if finalized, the FY 2021 payment determination would be based on 12 calendar months of data reporting beginning January 1, 2019, and ending December 31, 2019. Table 18 provides the data submission and collection method, data collection period and data submission timelines for the assessment-based quality measure affecting the FY 2021 payment determination and subsequent years.

Table 18—Proposed New SNF QRP Assessment-Based Quality Measure Data Collection Period and Data Submission Deadline Affecting FY 2021 Payment Determination and Subsequent Years

Quality measure	Data collection source	Data collection/ submission quarterly reporting period *	Data submission quarterly deadlines for FY 2021 payment determination **
Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP	MDS	CY 19 Q1, 1/1/2019-3/31/2019	CY 2019 Q1 Deadline: August 15, 2019.
		CY 19 Q2, 4/1/2019-6/30/19	CY 2019 Q2 Deadline: November 15, 2019.
		CY 19 Q3, 7/1/2019-9/30/2019	CY 2019 Q3 Deadline: February 15, 2020.
		CY 19 Q4, 10/1/2019-12/31/2019	CY 2019 Q4 Deadline: May 15, 2020.
* Data collection/submission will follow a similar quarterly reporting period schedule for subsequent CYs.
** Data review and correction periods and data submission deadlines will follow a similar quarterly schedule for subsequent CYs.

We invited public comment on the SNF QRP assessment-based quality measure data collection period and data submission deadline affecting the FY 2021 payment determination and subsequent years for the new assessment-based measure. We did not receive comments related to this topic.

Final Decision: We are finalizing as proposed the timeline and data submission mechanism for the FY 2021 payment determination and subsequent years for the new SNF QRP assessment-based quality measure as outlined in Table 18.

j. SNF QRP Data Completion Thresholds for the FY 2018 Payment Determination and Subsequent Years

We refer readers to the FY 2016 SNF PPS final rule (80 FR 46458) for our finalized policies regarding data completion thresholds for the FY 2018 payment determination and subsequent years. We finalized that, beginning with the FY 2018 payment determination, SNFs must report all of the data necessary to calculate the proposed quality measures on at least 80 percent of the MDS assessments that they submit. We also finalized that, for the FY 2018 SNF QRP, any SNF that does not meet the proposed requirement that 80 percent of all MDS assessments submitted contain 100 percent of all data items necessary to calculate the SNF QRP measures would be subject to a reduction of 2 percentage points to its FY 2018 market basket percentage. We finalized that a SNF has reported all of the data necessary to calculate the measures if the data actually can be used for purposes of calculating the quality measures, as opposed to, for example, the use of a dash [-], to indicate that the SNF was unable to perform a pressure ulcer assessment. We wish to clarify that the provision we finalized will affect FY 2018 payment determinations and subsequent years and is dependent upon the successful achievement of the completion threshold of the data used to calculate the measures we finalize. We did not propose any changes to these policies. While we did not solicit comments specifically regarding the data completion threshold for the SNF QRP, we did receive one comment related to this topic.

Comment: One commenter suggested that the 80 percent data completion threshold finalized the SNF PPS FY 2016 final rule is set too low and requested that, for the FY 2018 payment determination, the data completion threshold be increased to at least ninety percent.

Response: We intend to reevaluate this threshold over time and will propose to modify it, if warranted, based on our analysis.

k. SNF QRP Data Validation Requirements for the FY 2018 Payment Determination and Subsequent Years

We refer readers to the FY 2016 SNF PPS final rule (80 FR 46458 through 46459) for a summary of our approach to the development of data validation process for the SNF QRP. At this time, we are continuing to explore data validation methodology that will limit the amount of burden and cost to SNFs, while allowing us to establish estimations of the accuracy of SNF QRP data. We did not propose any further details pertaining to the data validation process for the SNF QRP, but we plan to do so in future rulemaking cycles. While we did not solicit comments specifically regarding data validation requirements for the SNF QRP, we received several comments related to this topic.

Comment: Several commenters agreed that validation of quality measure data is important in IMPACT Act implementation. One commenter recommended that we utilize pure data checks to identify both inconsistencies between QRP measures and MDS items and that data from these audits should be provided as part of SNF feedback reports to improve data accuracy. This commenter also suggested that we audit suspicious data patterns using trained MDS experts and present a list of validation checks to providers and MDS vendors to help improve data accuracy and expedite the process. Another commenter suggested revising and testing revisions to the survey protocol to review resident assessments and instituting penalties for violating resident assessment requirements.

Response: We thank the commenters for their input on policies that we should consider pertaining to data validation and accuracy analysis. We appreciate the commenters' suggestions to ensure data accuracy such as a combination of pure data checks to identify inconsistencies. We encourage providers to engage in available opportunities to improve the accuracy of their data. These suggestions will be taken into consideration as we develop the data validation methodologies for the SNF QRP.

l. SNF QRP Submission Exception and Extension Requirements for the FY 2018 Payment Determination and Subsequent Years

We refer readers to the FY 2016 SNF PPS final rule (80 FR 46459 through 46460) for our finalized policies regarding submission exception and extension requirements for the FY 2018 payment determination and subsequent years. We did not propose any changes to these policies.

m. SNF QRP Reconsideration and Appeals Procedures for the FY 2018 Payment Determination and Subsequent Years

We refer the reader to the FY 2016 SNF PPS final rule (80 FR 46460 through 46461) for a summary of our finalized reconsideration and appeals procedures for the SNF QRP for FY 2018 payment determination and subsequent years. We did not propose any changes to these procedures.

n. Public Display of Quality Measure Data for the SNF QRP & Procedures for the Opportunity To Review and Correct Data and Information

Section 1899B(g) of the Act requires the Secretary to establish procedures for public reporting of SNFs' performance, including the performance of individual SNFs, on quality measures specified under paragraph (c)(1) and resource use and other measures specified under paragraph (d)(1) of the Act (collectively, IMPACT Act measures) beginning not later than 2 years after the applicable specified application date under section 1899B(a)(2)(E) of the Act. Under section 1899B(g)(2) of the Act, the procedures must ensure, including through a process consistent with the process applied under section 1886(b)(3)(B)(viii)(VII) of the Act, which refers to public display and review requirements in the Hospital Inpatient Quality Reporting Program (HIQR), that each SNF has the opportunity to review and submit corrections to its data and information that are to be made public prior to the information being made public. In future rulemaking, we intend to propose a policy to publicly display performance information for individual SNFs on IMPACT Act measures, as required under the Act.

We proposed in the FY 2017 SNF PPS proposed rule to implement procedures that would allow individual SNFs to review and correct their data and information on IMPACT Act measures that are to be made public before those measure data are made public.

For assessment-based measures, we proposed a process by which we would provide each SNF with a confidential feedback report that would allow the SNF to review its performance on such measures and, during a review and correction period, to review and correct the data the SNF submitted to CMS via the CMS Quality Improvement and Evaluation System (QIES) Assessment Submission and Processing (ASAP) system for each such measure. In addition, during the review and correction period, the SNF would be able to request correction of any errors in the assessment-based measure rate calculations.

We proposed that these confidential feedback reports would be available to each SNF using the Certification and Survey Provider Enhanced Reporting (CASPER) System. We refer to these reports as the SNF Quality Measure (QM) Reports. We proposed to provide monthly updates to the data contained in these reports that pertain to assessment-based data, as the data become available. We proposed to provide the reports so that providers would be able to view their data and information at both the facility- and resident-level for quality measures. The CASPER facility-level QM Reports may contain information such as the numerator, denominator, facility rate, and national rate. The CASPER patient-level QM Reports may contain individual patient information which will provide information related to which patients were included in the quality measures to identify any potential errors. In addition, we would make other reports available in the CASPER System, such as MDS data submission reports and provider validation reports, which would disclose SNFs' data submission status, providing details on all items submitted for a selected assessment and the status of records submitted. Additional information regarding the content and availability of these confidential feedback reports would be provided on an ongoing basis at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​SNF-Quality-Reporting.html.

As proposed in section III.D.2.i.ii. of the FY 2017 SNF PPS Proposed Rule (81 FR 24270), SNFs would have approximately 4.5 months after the reporting quarter to correct any errors that appear on the CASPER-generated QM reports pertaining to their assessment-based data used to calculate the assessment-based measures. During the time of data submission for a given quarterly reporting period and up until the quarterly submission deadline, SNFs could review and perform corrections to errors in the assessment data used to calculate the measures and could request correction of measure calculations. However, once the quarterly submission deadline occurs, the data is “frozen” and calculated for public reporting; providers can no longer submit any corrections. We would encourage SNFs to submit timely assessment data during a given quarterly reporting period and review their data and information early during the review and correction period so that they can identify errors and resubmit data before the data submission deadline.

As noted in this section, the data would be populated into the confidential feedback reports, and we intend to update the reports monthly with all data that have been submitted and are available. We believe that a proposed data submission and review period, consisting of the reporting quarter plus approximately 4.5 months, is sufficient time for SNFs to submit, review and, where necessary, correct their data and information. These proposed time frames and deadlines for review and correction of assessment-based measures and data satisfy the statutory requirement that SNFs be provided the opportunity to review and correct their data and information that is to be made public and are consistent with the informal process hospitals follow in the Hospital Inpatient Quality Reporting (IQR) Program.

We proposed that, in addition to the data collection/submission quarterly reporting periods that are followed by data review and correction periods and submission deadlines, we would give SNFs a 30-day preview period prior to public display during which SNFs may preview the performance information on their measures that will be made public. We proposed to provide a preview report also using the CASPER System with which SNFs are familiar. The CASPER preview reports would inform providers of their performance on each measure which will be publicly reported. The CASPER preview reports for the reporting quarter will be available after the 4.5-month review and correction period and its data submission deadline, and the reports are refreshed on a quarterly basis for those measures publicly reported quarterly and annually for those measures publicly reported annually. We proposed to give SNFs 30 days to review this information, beginning from the date on which they can access the preview report. Corrections to the underlying data would not be permitted during this time; however, SNFs may contest incorrect measure calculations during the 30-day preview period. We proposed that if CMS determines that the measure, as it is displayed in the preview report, contains a calculation error, CMS could suppress the data on the public reporting Web site, recalculate the measure and publish it at the time of the next scheduled public display date. This process would be consistent with that followed in the Hospital IQR Program. If finalized, we intend to utilize a subregulatory mechanism, such as our SNF QRP Web site, to explain the process for how and when providers may ask for a correction to their measure calculations.

We invited public comment on these proposals. The comments we received on this topic, with their responses, appear below.

Comment: Several commenters, including MedPAC, supported public reporting of the cross-setting quality measures.

Response: We appreciate the support from MedPAC and several other commenters for public reporting of quality measures across post-acute care settings. We will continue to move forward with cross-setting measure development and public reporting of these measures to meet the mandate of the IMPACT Act.

Comment: One commenter was concerned about measure methodology associated with public reporting. The commenter stated that a year or more between the report date and penalties would not be meaningful or effective in changing behaviors.

Response: We appreciate the concern raised regarding the measure methodology associated with public reporting and the time delay between the performance period and public display of the quality measure results. We assume commenter's use of the term “measure methodology” to refer to how the quality measure is calculated. We first want to clarify that there are no penalties associated with quality measure performance. The quality measures for public display reflect basic fundamental processes or outcomes of providing good quality care. SNFs should have internal processes established to monitor and improve their care. Additionally, through the Certification and Survey Provider Enhanced Reports (CASPER) system, providers are able to review their data and performance results via reports that are available to them well in advance of public display of the quality measures for the purposes of ongoing quality improvement. We discuss such reports in greater detail below and such reports will enable providers to review their data on an ongoing basis so that they can utilize this information to improve their quality of care.

Comment: One commenter was concerned that the review and correction process may not provide SNFs enough information to validate measure values.

Response: We appreciate the commenter's concern regarding the review and correct process. In addition to the CASPER QM and Review and Correct Reports as described earlier in the proposed rule, SNFs have opportunities to review their information and validate their data for measure calculation using other reports such as data submission reports available through CASPER which gives providers information on fatal errors and warning messages related to data submission. For example, various data submission reports provide details regarding assessment items submitted for a selected MDS 3.0 assessment and others summarize errors encountered in assessments submitted during a specified period. We believe these CASPER reports will provide SNFs with sufficient information to validate measure values.

In addition to assessment-based measures, we have also proposed claims-based measures for the SNF QRP. Section 1899B(g)(2) of the Act requires prepublication provider review and correction procedures that are consistent with those followed in the Hospital IQR Program. For claims-based measures used in the Hospital IQR Program, we provide hospitals 30 days to preview their claims-based measures and data in a preview report containing aggregate hospital-level data. We proposed to adopt a similar process for the SNF QRP.

Prior to the public display of our claims-based measures, in alignment with the Hospital IQR, HAC and Hospital VBP Programs, we proposed to make available through the CASPER system a confidential preview report that will contain information pertaining to claims-based measure rate calculations, for example, facility and national rates. Such data and information would be for feedback purposes only and could not be corrected. This information would be accompanied by additional confidential information based on the most recent administrative data available at the time we extract the claims data for purposes of calculating the rates. Because the claims-based measures are calculated on an annual basis, these confidential CASPER QM reports for claims-based measures would be refreshed annually. SNFs would have 30 days from the date the preview report is made available in which to review this information. The 30-day preview period is the only time when SNFs would be able to see claims-based measures before they are publicly displayed. SNFs will not be able to make corrections to underlying claims data during this preview period, nor will they be able to add new claims to the data extract. However, SNFs may request that we correct our measure calculation if the SNF believes it is incorrect during the 30 day preview period. We proposed that if we agree that the measure, as it is displayed in the preview report, contains a calculation error, we would suppress the data on the public reporting Web site, recalculate the measure, and publish it at the time of the next scheduled public display date. This process would be consistent with that followed in the Hospital IQR Program. If finalized, we intend to utilize a subregulatory mechanism, such as our SNF QRP Web site, to explain the process for how and when providers may contest their measure calculations.

The proposed claims-based measures—Medicare Spending per Beneficiary—PAC SNF QRP Measure; Discharge to Community—PAC SNF QRP and Potentially Preventable 30 Day Post-Discharge Readmission Measure for SNF QRP—use Medicare administrative data from hospitalizations for Medicare FFS beneficiaries. Public reporting of data will be based on one CY of data. We proposed to create data extracts using claims data for these claims based measures, at least 90 days after the last discharge date in the applicable period (12 calendar months preceding), which we will use for the calculations. For example, if the last discharge date in the applicable period for a measure is December 31, 2017, for data collection January 1, 2017, through December 31, 2017, we would create the data extract on approximately March 31, 2018, at the earliest, and use that data to calculate the claims-based measures for that applicable period. Since SNFs would not be able to submit corrections to the underlying claims snapshot or add claims (for those measures that use SNF claims) to this data set at the conclusion of the at least 90-day period following the last date of discharge used in the applicable period, at that time we would consider SNF claims data to be complete for purposes of calculating the claims-based measures.

We proposed that beginning with data that will be publicly displayed in 2018, claims-based measures will be calculated using claims data with at least a 90 day run off period after the last discharge date in the applicable period, at which time we would create a data extract or snapshot of the available claims data to use for the measure calculations. This timeframe allows us to balance the need to provide timely program information to SNFs with the need to calculate the claims-based measures using as complete a data set as possible. As noted, under this proposed procedure, during the 30-day preview period, SNFs would not be able to submit corrections to the underlying claims data or add new claims to the data extract. This is for two reasons. First, for certain measures, the claims data used to calculate the measure is derived not from the SNF's claims, but from the claims of another provider. For example, the measure Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP uses claims data submitted by the hospital to which the patient was readmitted. The claims are not those of the SNF and, therefore, the SNF could not make corrections to them. Second, even where the claims used to calculate the measures are those of the SNF, it would not be not possible to correct the data after it is extracted for the measures calculation. This is because it is necessary to take a static “snapshot” of the claims to perform the necessary measure calculations.

We seek to have as complete a data set as possible. We recognize that the proposed at least 90-day “run-out” period when we would take the data extract to calculate the claims-based measures is less than the Medicare program's current timely claims filing policy, under which providers have up to one year from the date of discharge to submit claims. We considered a number of factors in determining that the proposed at least 90-day run-out period is appropriate to calculate the claims-based measures. After the data extract is created, it takes several months to incorporate other data needed for the calculations (particularly in the case of risk-adjusted or episode-based measures). We then need to generate and check the calculations. Because several months lead time is necessary after acquiring the data to generate the claims-based calculations, if we were to delay our data extraction point to 12 months after the last date of the last discharge in the applicable period, we would not be able to deliver the calculations to SNFs sooner than 18 to 24 months after the last discharge. We believe this would create an unacceptably long delay, both for SNFs and for us to deliver timely calculations to SNFs for quality improvement.

We invited public comment on these proposals. The comments we received on this topic, with their responses, appear below.

Comment: Several commenters recommended we provide real time reporting for assessment-based measures and every six months reporting for claims-based measures.

Response: SNFs will have an opportunity to review and utilize their data using confidential reports provided through the Certification and Survey Provider Enhanced Reports (CASPER) system as close to real time as is feasible. We intend to provide SNF Review and Correct reports that will allow providers to review information on assessment-based measures and anticipate the reports will be updated at least monthly. The decision to update claims-based measures on an annual basis was to ensure that the amount of data received during the reporting period was sufficient to generate reliable measure rates. However, we will look into the feasibility of providing SNFs with information more frequently.

Comment: One commenter was concerned with the 90-day run-out period for the claims-based measures because claims not filed within this period may negatively impact measure rates.

Response: We wish to clarify that we proposed for the claims-based measures to be calculated using claims data with at least a 90 day run off period after the last discharge date in the applicable period. We established this as the minimum run off period so as to use the most recently available data when calculating the claims-based measures. We developed this proposal to balance the need to provide timely program information to SNFs with the need to calculate the claims-based measures using as complete a data set as possible.

Final Decision: After careful consideration of public comments, we are finalizing these proposals as proposed.

o. Mechanism for Providing Feedback Reports to SNFs

Section 1899B(f) of the Act requires the Secretary to provide confidential feedback reports to post-acute care providers on their performance for the measures specified under paragraphs (c)(1) and (d)(1), beginning 1 year after the specified application date that applies to such measures and PAC providers. As discussed earlier, the reports we proposed to provide to SNFs to review their data and information would be confidential feedback reports that would enable SNFs to review their performance on the measures required under the SNF QRP. We proposed that these confidential feedback reports would be available to each SNF using the CASPER System. Data contained within these CASPER reports would be updated, as previously described, on a monthly basis as the data become available except for claims-based measures which can only be previewed on an annual basis.

We intend to provide detailed procedures to SNFs on how to obtain their confidential feedback CASPER reports on the SNF QRP Web site at http://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​SNF-Quality-Reporting.html. We proposed to use the CMS Quality Improvement and Evaluation System (QIES) Assessment Submission and Processing (ASAP) system to provide quality measure reports in a manner consistent with how providers obtain such reports to date. The QIES ASAP system is a confidential and secure system with access granted to providers, or their designees.

We sought public comment on this proposal to satisfy the requirement to provide confidential feedback reports to SNFs. The comments we received on this topic, with their responses, appear below.

Comment: One commenter supported our plan to make the feedback reports available in QIES ASAP through CASPER.

Response: We appreciate the commenter's support for providing feedback reports through CASPER.

Comment: Several commenters recommended that we conduct a “dry run” in which providers receive confidential preview reports prior to publicly reporting new SNF QRP measures so that providers can become familiar with the methodology, understand the measure results, know how well they are performing, and have an opportunity to give us feedback on potential technical issues with the measures.

Response: We appreciate that implementation activities such as dry runs are valuable prior to measure implementation to ensure the usability of a measure and educate providers. We intend to offer SNFs information and outreach training related to their measures so that they become familiar with the measure's methodology and understand how to interpret the confidential preview reports, which they will receive prior to the public reporting of new SNF QRP measures. SNFs will also receive additional confidential reports such as the SNF facility and resident level QM Reports and Review and Correct reports which we are developing. The Review and Correct Report will display all of the reporting quarters so that SNFs can identify errors in their data prior to and up until the submission deadline (freeze date) of a given quarter. The Review and Correct Report will provide updates regarding our data with a cumulative rate that will reflect publicly reported performance. We believe that these various reports will provide an indication on how well the SNF is performing as well as opportunities to provide us feedback on technical issues with the measures. The SNF Review and Correct Reports will be available beginning in the spring of 2017 and will be issued prior to the public reporting of SNF QRP measures. We refer readers to the SNF QRP Web site at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​NursingHomeQualityInits/​SNF-Quality-Reporting.html for further information, where we will address the process of accessing reports. We will continue to engage stakeholders and ask for recommendations to take into consideration for future public reporting development for the SNF QRP.

Final Decision: After careful consideration of public comments, we are finalizing our policies for providing confidential feedback reports to SNFs as proposed.

3. SNF Payment Models Research

In the FY 2017 SNF PPS proposed rule (81 FR 24275 through 24276), we provided an update on the progress we have made in the SNF Payment Models Research project. Specifically, we discussed the two prior Technical Expert Panels (TEPs) hosted by Acumen, LLC, the contractor conducting this research. On June 15, 2016, during the comment period associated with the FY 2017 SNF PPS proposed rule, Acumen hosted a third TEP which brought together many of the concepts and developments from the prior TEPs and analysis. We received a great deal of support from TEP panelists, as well as some excellent feedback on ways to improve the research going forward. As noted in the FY 2017 SNF PPS proposed rule, materials associated with these TEPs are available on the CMS Web site at https://www.cms.gov/​Medicare/​Medicare-Fee-for-Service-Payment/​SNFPPS/​therapyresearch.html.

In the FY 2017 SNF PPS proposed rule, we requested comments on the SNF PMR project. The comments we received on this topic, with responses, appear below.

Comment: Many commenters supported the goals of the research effort, specifically to develop a replacement for the existing SNF PPS that reimburses providers based on resident characteristics and not service provision. Some commenters stated that we should consider adding certain elements into the new payment system, such as a high cost outlier payment, separate payment for non-therapy ancillaries, and shifting from a per diem payment to a stay-based or episode-based payment schedule. One commenter stated that we should consider incorporating an episode-based payment model specifically for speech-language pathology services. A few commenters stated that the reformed payment system should consider a resident's socioeconomic status. Finally, some of these commenters asked that we try to align the new PPS model with other existing or future post-acute care payment models.

Response: We appreciate the support for this project, and will consider the suggestions made by commenters. However, we would note that, in order to develop a revised payment model that is implementable without requiring additional statutory authority, we have decided to only pursue those options which would be authorized within existing statutory constraints. Among other things, we believe this precludes the possibility of an outlier policy or non-per diem payment.

Comment: A few commenters expressed concern regarding the timeline for reform of the existing SNF PPS, with one commenter expressing frustration that we have not yet implemented a revised SNF PPS. These commenters stated that we should implement reform as soon as possible.

Response: We appreciate these commenters' concerns regarding the timing for implementing reform, but would note that reform of a system which covers such a wide range of services and such a diverse population of beneficiaries requires time to be completed correctly. We are moving as expeditiously as possible, ensuring that we allow sufficient time for requesting and considering public comments.

Comment: A few commenters expressed concerns regarding the data being used for the research. One commenter stated that we should not use any data from the Staff Time and Resource Intensity Verification, or STRIVE, project. A few commenters stated that SNF cost report data may not represent a viable source of data upon which to base a revised SNF PPS. One commenter expressed concern regarding the potential use of ADL information collected on the MDS as a source of nursing resource information, as the number of medications a resident is taking would not be taken into account. Finally, a few commenters stated that we should refrain from implementing a revised SNF PPS until new resident data, such as that required by the IMPACT Act, is available for analysis.

Response: We appreciate the concerns raised by these commenters and will pass along these concerns to our contractor performing the research so that it can take them into account as the research continues to evolve.

Comment: One commenter provided comments on information the commenter received participating in a TEP associated with the research project. Specifically, the commenter expressed concern regarding the possibility of combining physical and occupational therapy together under a single rate component. The commenter also made reference to the possibility of an additional TEP in Fall 2016.

Response: We appreciate this commenter's thoughts on the TEP materials, as well as their participation on the panel itself. We will pass these comments on to our contractor performing the research to ensure that this, and other comments made by the commenter during the panel, are taken into account. With regard to the possibility of another TEP in Fall 2016, we have discussed plans with the contractor to host an additional TEP in Fall 2016.

We appreciate all of the comments received on this topic and look forward to providing additional details on the CMS Web site and in future rulemaking. We invite the public to provide comments outside of the rulemaking process by contacting us at [e-mail].

IV. Collection of Information Requirements

Section III.D.2.f. of this preamble sets out three claims-based measures that we are adopting for the SNF QRP beginning with the FY 2018 payment year: (1) Medicare Spending per Beneficiary—PAC SNF QRP; (2) Discharge to Community—PAC SNF QRP; and (3) Potentially Preventable 30-Day Post-Discharge Readmission Measure for SNF QRP. Because they are claims-based, the measures can be calculated using data that are already reported to the Medicare program for payment purposes. Consequently, we believe there will be no additional burden on SNFs in connection with the the reporting of data needed to calculate these measures.

We did not receive any public comments on this topic in response to the FY 2017 SNF PPS proposed rule.

For the FY 2020 payment determination and subsequent years, we are adopting for the SNF QRP an assessment-based measure entitled Drug Regimen Review Conducted with Follow-Up for Identified Issues—PAC SNF QRP. The data for this measure will be collected and reported using the MDS (version effective October 1, 2018). While the reporting of data on quality measures is an information collection, we believe that the burden associated with modifications to the MDS fall under the PRA exception (provided in section 1899B(m) of the IMPACT Act of 2014) because they are required to achieve the standardization of patient assessment data. The requirement and burden will, however, be submitted to OMB for review and approval when the modifications to the MDS or other applicable PAC assessment instruments have achieved standardization and are no longer exempt from the requirements under section 1899B(m).

We estimate the additional elements for the new assessment measure will take 7.5 minutes of nursing/clinical staff time to report data on admission and 2.5 minutes of nursing/clinical staff time to report data on discharge, for a total of 10 minutes. We estimate that the additional MDS-RAI items will be completed by Registered Nurses (RN) for approximately 75 percent of the time required and Pharmacists for approximately 25 percent of the time required. Individual providers determine the staffing resources necessary. We estimate 2,101,370 discharges from 16,484 SNFs annually, with an additional burden of 10 minutes. This would equate to 350,228 total hours or 21.25 hours per SNF. We believe this work will be completed by RNs (75 percent) and Pharmacists (25 percent). We obtained mean hourly wages for these staff from the U.S. Bureau of Labor Statistics' (BLS) May 2015 National Occupational Employment and Wage Estimates (http://www.bls.gov/​oes/​current/​oes_​nat.htm), to account for overhead and fringe benefits, we have doubled the mean hourly wage. Per the National Occupational Employment and Wage Estimates, the mean hourly wage for a RN (BLS occupation code: 29-1141) is $34.14/hr. However, to account for overhead and fringe benefits, we have double the mean hourly wage, making it $68.28/hr for an RN. The mean hourly wage for a pharmacist (BLS occupation code: 29-1051) is $57.34/hr. To account for overhead and fringe benefits, we have double the mean hourly wage, making it $114.68/hr for a pharmacist. Given these wages and time estimates, the total cost related to the four measures is estimated at $1,697.17 per SNF annually, or $27,976,212.64 [(262,671 hr × $68.28/hr) + (87,557 hr × $114.68/hr)] for all SNFs annually. These values have been updated from the FY 2017 SNF PPS proposed rule to reflect the more recent 2015 wage estimates. While we are setting out burden, the requirements and associated estimates will not be submitted to OMB for approval under Paperwork Reduction Act of 1995 (44 U.S.C. 3501 et seq.) since the burden estimates are either claims-based or associated with the exemption under section 1899B(m) of the IMPACT Act of 2014. We are setting out the burden as a courtesy to advise interested parties of the time and costs. These figures are not in the RIA section of this rule.

We received the following comment in response to the FY 2017 SNF PPS proposed rule.

Comment: One commenter agreed that standardization and associated collection of this MDS-based measure is PRA exempt. However, the commenter suggested that the estimate provided by CMS in the proposed rule is insufficient.

Response: For burden associated with this FY 2017 SNF PPS final rule, we considered the comment while planning to implement new items on the MDS. The comment was general in that it did not identify the estimate of concern nor did it identify what the correct estimate should be. While considering the comment, we revised our hourly wage estimate to account for more recent BLS wage data. Otherwise, our final estimate is unchanged from what was proposed.

As described in further detail in section III.D.1.b. of this final rule, we are adopting the SNFPPR measure for the SNF VBP Program. Like the SNFRM (NQF #2510), which was adopted for the SNF VBP Program in the FY 2016 SNF PPS final rule (80 FR 46419), the SNFPPR measure is also claims-based. Because claims-based measures are calculated based on claims that are already submitted to the Medicare program for payment purposes, there is no additional burden associated with data collection or submission for the SNFPPR measure. Thus there is no additional reporting burden associated with the SNFPPR measure.

We did not receive any public comments on this topic in response to the FY 2017 SNF PPS proposed rule.

Comments on any of the aforementioned collection of information claims must be received by the OMB desk officer by August 29, 2016.

To be assured consideration, comments and recommendations must be received via one of the following transmissions: OMB, Office of Information and Regulatory Affairs, Attention: CMS Desk Officer, Fax Number: [phone] OR, Email: [e-mail].

V. Economic Analyses

A. Regulatory Impact Analysis

1. Introduction

We have examined the impacts of this final rule as required by Executive Order 12866 on Regulatory Planning and Review (September 30, 1993), Executive Order 13563 on Improving Regulation and Regulatory Review (January 18, 2011), the Regulatory Flexibility Act (RFA, September 19, 1980, Pub. L. 96-354), section 1102(b) of the Act, section 202 of the Unfunded Mandates Reform Act of 1995 (UMRA, March 22, 1995; Pub. L. 104-4), Executive Order 13132 on Federalism (August 4, 1999), and the Congressional Review Act (5 U.S.C. 804(2)).

Executive Orders 12866 and 13563 direct agencies to assess all costs and benefits of available regulatory alternatives and, if regulation is necessary, to select regulatory approaches that maximize net benefits (including potential economic, environmental, public health and safety effects, distributive impacts, and equity). Executive Order 13563 emphasizes the importance of quantifying both costs and benefits, of reducing costs, of harmonizing rules, and of promoting flexibility. This rule has been designated an economically significant rule, under section 3(f)(1) of Executive Order 12866. Accordingly, we have prepared a regulatory impact analysis (RIA) as further discussed below, and the rule has been reviewed by OMB.

2. Statement of Need

This final rule updates the SNF prospective payment rates for FY 2017 as required under section 1888(e)(4)(E) of the Act. It also responds to section 1888(e)(4)(H) of the Act, which requires the Secretary to provide for publication in the Federal Register before the August 1 that precedes the start of each FY, the unadjusted federal per diem rates, the case-mix classification system, and the factors to be applied in making the area wage adjustment. As these statutory provisions prescribe a detailed methodology for calculating and disseminating payment rates under the SNF PPS, we do not have the discretion to adopt an alternative approach.

3. Overall Impacts

This final rule sets forth updates of the SNF PPS rates contained in the SNF PPS final rule for FY 2016 (80 FR 46390). Based on the above, we estimate that the aggregate impact would be an increase of $920 million in payments to SNFs, resulting from the SNF market basket update to the payment rates, as adjusted by the MFP adjustment. The impact analysis of this final rule represents the projected effects of the changes in the SNF PPS from FY 2016 to FY 2017. Although the best data available are utilized, there is no attempt to predict behavioral responses to these changes or to make adjustments for future changes in such variables as days or case-mix.

Certain events may occur to limit the scope or accuracy of our impact analysis, as this analysis is future-oriented and, thus, very susceptible to forecasting errors due to certain events that may occur within the assessed impact time period. Some examples of possible events may include newly-legislated general Medicare program funding changes by the Congress or changes specifically related to SNFs. In addition, changes to the Medicare program may continue to be made as a result of previously-enacted legislation or new statutory provisions. Although these changes may not be specific to the SNF PPS, the nature of the Medicare program is such that the changes may interact and, thus, the complexity of the interaction of these changes could make it difficult to predict accurately the full scope of the impact upon SNFs.

In accordance with sections 1888(e)(4)(E) and 1888(e)(5) of the Act, we update the FY 2016 payment rates by a factor equal to the market basket percentage change adjusted by the MFP adjustment to determine the payment rates for FY 2017. As discussed previously, for FY 2012 and each subsequent FY, as required by section 1888(e)(5)(B) of the Act, as amended by section 3401(b) of the Affordable Care Act, the market basket percentage is reduced by the MFP adjustment. The special AIDS add-on established by section 511 of the MMA remains in effect until such date as the Secretary certifies that there is an appropriate adjustment in the case mix. We have not provided a separate impact analysis for the MMA provision. Our latest estimates indicate that there are fewer than 4,800 beneficiaries who qualify for the add-on payment for residents with AIDS. The impact to Medicare is included in the total column of Table 19. In updating the SNF PPS rates for FY 2017, we made a number of standard annual revisions and clarifications mentioned elsewhere in this final rule (for example, the update to the wage and market basket indexes used for adjusting the federal rates).

The annual update set forth in this final rule applies to SNF PPS payments in FY 2017. Accordingly, the analysis that follows only describes the impact of this single year. In accordance with the requirements of the Act, we will publish a notice or rule for each subsequent FY that will provide for an update to the SNF PPS payment rates and include an associated impact analysis.

4. Detailed Economic Analysis

The FY 2017 SNF PPS payment impacts appear in Table 19. Using the most recently available data, in this case FY 2015, we apply the current FY 2016 wage index and labor-related share value to the number of payment days to simulate FY 2016 payments. Then, using the same FY 2015 data, we apply the FY 2017 wage index and labor-related share value to simulate FY 2017 payments. We tabulate the resulting payments according to the classifications in Table 19 (for example, facility type, geographic region, facility ownership), and compare the simulated FY 2016 payments to the simulated FY 2017 payments to determine the overall impact. In Section III.B.2 and III.B.4 of this final rule, we discussed an error in calculating the FY 2017 wage index budget neutrality factor in the FY 2017 SNF PPS proposed rule and how this error affected the impact table in the FY 2017 SNF PPS proposed rule (81 FR 24278). Specifically, we stated that in calculating the proposed wage index budget neutrality factor, we inadvertently neglected to update the wage index data used in the calculation with the most recently available FY 2017 data. As we discussed in section III.B.2. and III.B.4. of this final rule, this same error (the use of non-updated wage index data) which resulted in an incorrect calculation of the proposed wage index budget neutrality factor also resulted in inaccurate wage index impacts in Table 19 of the FY 2017 SNF PPS proposed rule. We have corrected this error, and Table 19 of this final rule includes corrected impact values based on updated FY 2017 wage index data. The breakdown of the various categories of data in the table follows:

• The first column shows the breakdown of all SNFs by urban or rural status, hospital-based or freestanding status, census region, and ownership.
• The first row of figures describes the estimated effects of the various changes on all facilities. The next six rows show the effects on facilities split by hospital-based, freestanding, urban, and rural categories. The next nineteen rows show the effects on facilities by urban versus rural status by census region. The last three rows show the effects on facilities by ownership (that is, government, profit, and non-profit status).
• The second column shows the number of facilities in the impact database.
• The third column shows the effect of the annual update to the wage index. This represents the effect of using the most recent wage data available. The total impact of this change is zero percent; however, there are distributional effects of the change.
• The fourth column shows the effect of all of the changes on the FY 2017 payments. The update of 2.4 percent (consisting of the market basket increase of 2.7 percentage points, reduced by the 0.3 percentage point MFP adjustment) is constant for all providers and, though not shown individually, is included in the total column. It is projected that aggregate payments will increase by 2.4 percent, assuming facilities do not change their care delivery and billing practices in response.

As illustrated in Table 19, the combined effects of all of the changes vary by specific types of providers and by location. For example, due to changes finalized in this rule, providers in the urban Outlying region would experience a 1.7 percent increase in FY 2017 total payments.

Table 19—Projected Impact to the SNF PPS for FY 2017

	Number of facilities FY 2017	Update wage data (percent)	Total change (percent)
Group:
Total	15,445	0.0	2.4
Urban	10,946	0.0	2.4
Rural	4,499	0.3	2.7
Hospital based urban	467	−0.2	2.2
Freestanding urban	10,479	0.0	2.4
Hospital based rural	320	0.5	2.9
Freestanding rural	4,179	0.3	2.7
Urban by region:
New England	797	−0.8	1.6
Middle Atlantic	1,481	−0.1	2.3
South Atlantic	1,862	−0.2	2.2
East North Central	2,095	−0.1	2.3
East South Central	547	−0.1	2.3
West North Central	907	−0.2	2.2
West South Central	1,323	0.3	2.7
Mountain	509	−0.1	2.3
Pacific	1,420	0.6	3.0
Outlying	5	−0.6	1.7
Rural by region:
New England	139	0.1	2.5
Middle Atlantic	221	0.4	2.8
South Atlantic	507	−0.2	2.2
East North Central	933	0.2	2.6
East South Central	530	0.4	2.8
West North Central	1,087	0.5	2.9
West South Central	745	0.6	3.0
Mountain	233	0.7	3.2
Pacific	104	−0.4	2.0
Ownership:
Government	1,051	0.1	2.5
Profit	10,766	0.0	2.4
Non-profit	3,628	−0.1	2.3
Note: The Total column includes the 2.7 percent market basket increase, reduced by the 0.3 percentage point MFP adjustment. Additionally, we found no SNFs in rural outlying areas.

5. Alternatives Considered

As described in this section, we estimate that the aggregate impact for FY 2017 under the SNF PPS would be an increase of $920 million in payments to SNFs, resulting from the SNF market basket update to the payment rates, as adjusted by the MFP adjustment.

Section 1888(e) of the Act establishes the SNF PPS for the payment of Medicare SNF services for cost reporting periods beginning on or after July 1, 1998. This section of the statute prescribes a detailed formula for calculating payment rates under the SNF PPS and does not provide for the use of any alternative methodology. It specifies that the base year cost data to be used for computing the SNF PPS payment rates must be from FY 1995 (October 1, 1994, through September 30, 1995). In accordance with the statute, we also incorporated a number of elements into the SNF PPS (for example, case-mix classification methodology, a market basket index, a wage index, and the urban and rural distinction used in the development or adjustment of the federal rates). Further, section 1888(e)(4)(H) of the Act specifically requires us to disseminate the payment rates for each new FY through the Federal Register and to do so before the August 1 that precedes the start of the new FY. Accordingly, we are not pursuing alternatives for the payment methodology as discussed previously.

6. Accounting Statement

As required by OMB Circular A-4 (available online at www.whitehouse.gov/​sites/​default/​files/​omb/​assets/​regulatory_​matters_​pdf/​a-4.pdf), in Table 20, we have prepared an accounting statement showing the classification of the expenditures associated with the provisions of this final rule. Table 20 provides our best estimate of the possible changes in Medicare payments under the SNF PPS as a result of the policies in this final rule, based on the data for 15,427 SNFs in our database. All expenditures are classified as transfers to Medicare providers (that is, SNFs).

Table 20—Accounting Statement: Classification of Estimated Expenditures, From the 2016 SNF PPS Fiscal Year to the 2017 SNF PPS Fiscal Year

Category	Transfers
Annualized Monetized Transfers	$920 million.*
From Whom To Whom?	Federal Government to SNF Medicare Providers.
* The net increase of $920 million in transfer payments is a result of the MFP-adjusted market basket increase of $920 million.

7. Conclusion

This final rule sets forth updates of the SNF PPS rates contained in the SNF PPS final rule for FY 2016 (80 FR 46390). Based on the above, we estimate the overall estimated payments for SNFs in FY 2017 are projected to increase by $920 million, or 2.4 percent, compared with those in FY 2016. We estimate that in FY 2017 under RUG-IV, SNFs in urban and rural areas would experience, on average, a 2.4 and 2.7 percent increase, respectively, in estimated payments compared with FY 2016. Providers in the rural Mountain region would experience the largest estimated increase in payments of approximately 3.2 percent. Providers in the urban New England region would experience the smallest estimated increase in payments of 1.6 percent.

8. Effects of the Requirements for the SNF VBP and SNF QRP Program

The requirements set forth for the SNF VBP and SNF QRP Program in this final rule would not impact SNFs in FY 2017; therefore, we are not including a regulatory impact analysis for the SNF VBP and SNF QRP Program in this final rule.

B. Regulatory Flexibility Act Analysis

The RFA requires agencies to analyze options for regulatory relief of small entities, if a rule has a significant impact on a substantial number of small entities. For purposes of the RFA, small entities include small businesses, non-profit organizations, and small governmental jurisdictions. Most SNFs and most other providers and suppliers are small entities, either by reason of their non-profit status or by having revenues of $27.5 million or less in any 1 year. We utilized the revenues of individual SNF providers (from recent Medicare Cost Reports) to classify a small business, and not the revenue of a larger firm with which they may be affiliated. As a result, we estimate approximately 91 percent of SNFs are considered small businesses according to the Small Business Administration's latest size standards (NAICS 623110), with total revenues of $27.5 million or less in any 1 year. (For details, see the Small Business Administration's Web site at http://www.sba.gov/​category/​navigation-structure/​contracting/​contracting-officials/​eligibility-size-standards). In addition, approximately 25 percent of SNFs classified as small entities are non-profit organizations. Finally, individuals and states are not included in the definition of a small entity.

This final rule sets forth updates of the SNF PPS rates contained in the SNF PPS final rule for FY 2016 (80 FR 46390). Based on the above, we estimate that the aggregate impact would be an increase of $920 million in payments to SNFs, resulting from the SNF market basket update to the payment rates, as adjusted by the MFP adjustment. While it is projected in Table 19 that most providers would experience a net increase in payments, we note that some individual providers within the same region or group may experience different impacts on payments than others due to the distributional impact of the FY 2017 wage indexes and the degree of Medicare utilization.

Guidance issued by the Department of Health and Human Services on the proper assessment of the impact on small entities in rulemakings utilizes a cost or revenue impact of 3 to 5 percent as a significance threshold under the RFA. According to MedPAC, Medicare covers approximately 12 percent of total patient days in freestanding facilities and 21 percent of facility revenue (Report to the Congress: Medicare Payment Policy, March 2016, available at http://medpac.gov/​documents/​reports/​chapter-7-skilled-nursing-facility-services-(march-2016-report).pdf). As a result, for most facilities, when all payers are included in the revenue stream, the overall impact on total revenues should be substantially less than those impacts presented in Table 19. As indicated in Table 19, the effect on facilities is projected to be an aggregate positive impact of 2.4 percent. As the overall impact on the industry as a whole, and thus on small entities specifically, is less than the 3 to 5 percent threshold discussed previously, the Secretary has determined that this final rule would not have a significant impact on a substantial number of small entities.

In addition, section 1102(b) of the Act requires us to prepare a regulatory impact analysis if a rule may have a significant impact on the operations of a substantial number of small rural hospitals. This analysis must conform to the provisions of section 604 of the RFA. For purposes of section 1102(b) of the Act, we define a small rural hospital as a hospital that is located outside of an MSA and has fewer than 100 beds. This final rule would affect small rural hospitals that (1) furnish SNF services under a swing-bed agreement or (2) have a hospital-based SNF. We anticipate that the impact on small rural hospitals would be similar to the impact on SNF providers overall. Moreover, as noted in previous SNF PPS final rules (most recently the one for FY 2016 (80 FR 46476)), the category of small rural hospitals would be included within the analysis of the impact of this final rule on small entities in general. As indicated in Table 19, the effect on facilities is projected to be an aggregate positive impact of 2.4 percent. As the overall impact on the industry as a whole is less than the 3 to 5 percent threshold discussed above, the Secretary has determined that this final rule would not have a significant impact on a substantial number of small rural hospitals.

C. Unfunded Mandates Reform Act Analysis

Section 202 of the Unfunded Mandates Reform Act of 1995 (UMRA) also requires that agencies assess anticipated costs and benefits before issuing any rule whose mandates require spending in any 1 year of $100 million in 1995 dollars, updated annually for inflation. In 2016, that threshold is approximately $146 million. This final rule does not include any mandate on state, local, or tribal governments in the aggregate, or by the private sector, of $146 million.

D. Federalism Analysis

Executive Order 13132 establishes certain requirements that an agency must meet when it issues a proposed rule (and subsequent final rule) that imposes substantial direct requirement costs on state and local governments, preempts state law, or otherwise has federalism implications. This final rule would have no substantial direct effect on state and local governments, preempt state law, or otherwise have federalism implications.

E. Congressional Review Act

This final rule is subject to the Congressional Review Act provisions of the Small Business Regulatory Enforcement Fairness Act of 1996 (5 U.S.C. 801 et seq.) and has been transmitted to the Congress and the Comptroller General for review.

In accordance with the provisions of Executive Order 12866, this final rule was reviewed by the Office of Management and Budget.

Dated: July 18, 2016.

Andrew M. Slavitt,

Acting Administrator, Centers for Medicare & Medicaid Services.

Dated: July 25, 2016.

Sylvia M. Burwell,

Secretary, Department of Health and Human Services.

Footnotes

1.  Friedman, B., and Basu, J.: The rate and cost of hospital readmissions for preventable conditions. Med. Care Res. Rev. 61(2):225-240, 2004. doi:10.1177/1077558704263799.

2.  Jencks, S.F., Williams, M.V., and Coleman, E.A.: Rehospitalizations among patients in the Medicare Fee-for-Service Program. N. Engl. J. Med. 360(14):1418-1428, 2009. doi:10.1016/j.jvs.2009.05.045.

3.  MedPAC: Payment policy for inpatient readmissions, in Report to the Congress: Promoting Greater Efficiency in Medicare. Washington, DC, pp. 103-120, 2007. Available from http://www.medpac.gov/​documents/​reports/​Jun07_​EntireReport.pdf.

4.  Ibid.

5.  Ibid.

6.  Mor, V., Intrator, O., Feng, Z., et al.: The revolving door of rehospitalization from SNFs. Health Aff. 29(1):57-64, 2010. doi:10.1377/hlthaff.2009.0629.

7.  National Quality Forum: All-Cause Admissions and Readmissions Measures. pp. 1-319, April 2015. Available from http://www.qualityforum.org/​Publications/​2015/​04/​All-Cause_​Admissions_​and_​Readmissions_​Measures_​-_​Final_​Report.aspx.

8.  Goldfield, N.I., McCullough, E.C., Hughes, J.S., et al.: Identifying potentially preventable readmissions. Health Care Finan. Rev. 30(1):75-91, 2008. Available from http://www.ncbi.nlm.nih.gov/​pmc/​articles/​PMC4195042/​ .

9.  Agency for Healthcare Research and Quality: Prevention Quality Indicators Overview. 2008.

10.  MedPAC: Online Appendix C: Medicare Ambulatory Care Indicators for the Elderly. pp. 1-12, prepared for Chapter 4, 2011. Available from http://www.medpac.gov/​documents/​reports/​Mar11_​Ch04_​APPENDIX.pdf?​sfvrsn=​0 .

11.  Kramer, A., Lin, M., Fish, R., et al.: Development of Inpatient Rehabilitation Facility Quality Measures: Potentially Avoidable Readmissions, Community Discharge, and Functional Improvement. pp. 1-42, 2015. Available from http://www.medpac.gov/​documents/​contractor-reports/​development-of-inpatient-rehabilitation-facility-quality-measures-potentially-avoidable-readmissions-community-discharge-and-functional-improvement.pdf?​sfvrsn=​0 .

12.  Kramer, A., Lin, M., Fish, R., et al.: Development of Potentially Avoidable Readmission and Functional Outcome SNF Quality Measures. pp. 1-75, 2014. Available from http://www.medpac.gov/​documents/​contractor-reports/​mar14_​snfqualitymeasures_​contractor.pdf?​sfvrsn=​0 .

13.  Allaudeen, N., Vidyarthi, A., Maselli, J., et al.: Redefining readmission risk factors for general medicine patients. J. Hosp. Med. 6(2):54-60, 2011. doi:10.1002/jhm.805.

14.  Gao, J., Moran, E., Li, Y.-F., et al.: Predicting potentially avoidable hospitalizations. Med. Care 52(2):164-171, 2014. doi:10.1097/MLR.0000000000000041.

15.  Walsh, E.G., Wiener, J.M., Haber, S., et al.: Potentially avoidable hospitalizations of dually eligible Medicare and Medicaid beneficiaries from nursing facility and home‐and community‐based services waiver programs. J. Am. Geriatr. Soc. 60(5):821-829, 2012. doi:10.1111/j.1532-5415.2012.03920.

16.  National Quality Forum: All-Cause Admissions and Readmissions Measures. pp. 1-319, April 2015. National Quality Forum: All-Cause Admissions and Readmissions Measures. pp. 1-319, April 2015. Available from http://www.qualityforum.org/​Publications/​2015/​04/​All-Cause_​Admissions_​and_​Readmissions_​Measures_​-_​Final_​Report.aspx .

17.  Available by searching for “1789” at http://www.qualityforum.org/​QPS/​QPSTool.aspx.

18.  National Quality Forum: Measure Applications Partnership Pre-Rulemaking Report: 2013 Recommendations of Measures Under Consideration by HHS. pp. 1-394, February 2013. Available from https://www.qualityforum.org/​Publications/​2013/​02/​MAP_​Pre-Rulemaking_​Report_​-_​February_​2013.aspx .

19.  https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​QualityMeasures/​Downloads/​2015-Measures-Under-Consideration-List.pdf.

20.  Bratzler DW, Normand SL, Wang Y, et al. An administrative claims model for profiling hospital 30-day mortality rates for pneumonia patients. PLoS One 2011;6(4):e17401.

21.  Keenan PS, Normand SL, Lin Z, et al. An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure. Circulation 2008;1(1):29-37.

22.  Krumholz HM, Wang Y, Mattera JA, et al. An administrative claims model suitable for profiling hospital performance based on 30-day mortality rates among patients with heart failure. Circulation 2006;113:1693-1701.

23.  See Medicare Hospital Quality Chartbook 2010, p. 12, https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​HospitalQualityInits/​downloads/​HospitalChartBook.pdf.

24.  The COPSS-CMS White Paper Committee. Statistical Issues in Assessing Hospital Performance. January 2012. Available at https://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​HospitalQualityInits/​Downloads/​Statistical-Issues-in-Assessing-Hospital-Performance.pdf.

25.  We refer readers to the FY 2013 IPPS final rule for a discussion of the Hospital VBP Program scoring methodology (76 FR 2466 through 2470).

26.  http://www.cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​QualityInitiativesGenInfo/​CMS-Quality-Strategy.html.

27.  http://www.ahrq.gov/​workingforquality/​nqs/​nqs2011annlrpt.htm.

28.  MedPAC, “A Data Book: Health Care Spending and the Medicare Program,” (2015). 114.

29.  Institute of Medicine, “Variation in Health Care Spending: Target Decision Making, Not Geography,” (Washington, DC: National Academies 2013). 2.

30.  2013 figures. MedPAC, “Medicare Payment Policy,” Report to the Congress (2015). xvii-xviii.

31.  QualityNet, “Measure Methodology Reports: Medicare Spending per Beneficiary (MSPB) Measure,” (2015). http://www.qualitynet.org/​dcs/​ContentServer?​pagename=​QnetPublic%2FPage%2FQnetTier3&​cid=​1228772053996.

32.  QualityNet, “Measure Methodology Reports: Medicare Spending per Beneficiary (MSPB) Measure,” (2015). http://www.qualitynet.org/​dcs/​ContentServer?​pagename=​QnetPublic%2FPage%2FQnetTier3&​cid=​1228772053996.

33.  FY 2012 IPPS/LTCH PPS Final Rule (76 FR 51619).

34.  National Quality Forum, Measure Applications Partnership, “Process and Approach for MAP Pre-Rulemaking Deliberations, 2015-2016” (February 2016) http://www.qualityforum.org/​WorkArea/​linkit.aspx?​LinkIdentifier=​id&​ItemID=​81693.

35.  National Quality Forum, Measure Applications Partnership Post-Acute Care/Long-Term Care Workgroup, “Meeting Transcript—Day 2 of 2” (December 15, 2015) 104-106 http://www.qualityforum.org/​WorkArea/​linkit.aspx?​LinkIdentifier=​id&​ItemID=​81470.

36.  National Quality Forum, Measure Applications Partnership, “Meeting Transcript—Day 1 of 2” (January 26, 2016) 231-232 http://www.qualityforum.org/​WorkArea/​linkit.aspx?​LinkIdentifier=​id&​ItemID=​81637.

37.  National Quality Forum, Measure Applications Partnership, “MAP 2016 Considerations for Implementing Measures in Federal Programs: Post-Acute Care and Long-Term Care” Final Report, (February 2016) http://www.qualityforum.org/​Publications/​2016/​02/​MAP_​2016_​Considerations_​for_​Implementing_​Measures_​in_​Federal_​Programs_​-_​PAC-LTC.aspx.

38.  National Quality Forum, Measure Applications Partnership, “Spreadsheet of MAP 2016 Final Recommendations” (February 1, 2016) http://www.qualityforum.org/​WorkArea/​linkit.aspx?​LinkIdentifier=​id&​ItemID=​81593.

39.  National Quality Forum, Measure Applications Partnership, “Spreadsheet of MAP 2016 Final Recommendations” (February 1, 2016) http://www.qualityforum.org/​WorkArea/​linkit.aspx?​LinkIdentifier=​id&​ItemID=​81593 .

40.  QualityNet, “CMS Price (Payment) Standardization—Detailed Methods” (Revised May 2015) https://qualitynet.org/​dcs/​ContentServer?​c=​Page&​pagename=​QnetPublic%2FPage%2FQnetTier4&​cid=​1228772057350.

41.  National Uniform Billing Committee Official UB-04 Data Specifications Manual 2017, Version 11, July 2016, Copyright 2016, American Hospital Association.

42.  This definition is not intended to suggest that board and care homes, assisted living facilities, or other 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.

43.  El-Solh AA, Saltzman SK, Ramadan FH, Naughton BJ. Validity of an artificial neural network in predicting discharge destination from a postacute geriatric rehabilitation unit. Archives of physical medicine and rehabilitation. 2000;81(10):1388-1393.

44.  Tanwir S, Montgomery K, Chari V, Nesathurai S. Stroke rehabilitation: Availability of a family member as caregiver and discharge destination. European journal of physical and rehabilitation medicine. 2014;50(3):355-362.

45.  Dobrez D, Heinemann AW, Deutsch A, Manheim L, Mallinson T. Impact of Medicare's prospective payment system for inpatient rehabilitation facilities on stroke patient outcomes. American journal of physical medicine & rehabilitation/Association of Academic Physiatrists. 2010;89(3):198-204.

46.  Gage B, Morley M, Spain P, Ingber M. Examining Post Acute Care Relationships in an Integrated Hospital System. Final Report. RTI International;2009.

47.  Ibid.

48.  Doran JP, Zabinski SJ. Bundled payment initiatives for Medicare and non-Medicare total joint arthroplasty patients at a community hospital: Bundles in the real world. The journal of arthroplasty. 2015;30(3):353-355.

49.  Newcomer RJ, Ko M, Kang T, Harrington C, Hulett D, Bindman AB. Health Care Expenditures After Initiating Long-term Services and Supports in the Community Versus in a Nursing Facility. Medical Care. 2016; 54(3):221-228.

50.  Gage B, Morley M, Spain P, Ingber M. Examining Post Acute Care Relationships in an Integrated Hospital System. Final Report. RTI International;2009.

51.  Ibid.

52.  Reistetter TA, Karmarkar AM, Graham JE, et al. Regional variation in stroke rehabilitation outcomes. Archives of physical medicine and rehabilitation. 2014; 95(1):29-38.

53.  El-Solh AA, Saltzman SK, Ramadan FH, Naughton BJ. Validity of an artificial neural network in predicting discharge destination from a postacute geriatric rehabilitation unit. Archives of physical medicine and rehabilitation. 2000; 81(10):1388-1393.

54.  March 2015 Report to the Congress: Medicare Payment Policy. Medicare Payment Advisory Commission; 2015.

55.  Bhandari VK, Kushel M, Price L, Schillinger D. Racial disparities in outcomes of inpatient stroke rehabilitation. Archives of physical medicine and rehabilitation. 2005; 86(11):2081-2086.

56.  Chang PF, Ostir GV, Kuo YF, Granger CV, Ottenbacher KJ. Ethnic differences in discharge destination among older patients with traumatic brain injury. Archives of physical medicine and rehabilitation. 2008; 89(2):231-236.

57.  Berges IM, Kuo YF, Ostir GV, Granger CV, Graham JE, Ottenbacher KJ. Gender and ethnic differences in rehabilitation outcomes after hip-replacement surgery. American journal of physical medicine & rehabilitation/Association of Academic Physiatrists. 2008; 87(7):567-572.

58.  Galloway RV, Granger CV, Karmarkar AM, et al. The Uniform Data System for Medical Rehabilitation: Report of patients with debility discharged from inpatient rehabilitation programs in 2000-2010. American journal of physical medicine & rehabilitation/Association of Academic Physiatrists. 2013; 92(1):14-27.

59.  Morley MA, Coots LA, Forgues AL, Gage BJ. Inpatient rehabilitation utilization for Medicare beneficiaries with multiple sclerosis. Archives of physical medicine and rehabilitation. 2012; 93(8):1377-1383.

60.  Reistetter TA, Graham JE, Deutsch A, Granger CV, Markello S, Ottenbacher KJ. Utility of functional status for classifying community versus institutional discharges after inpatient rehabilitation for stroke. Archives of physical medicine and rehabilitation. 2010; 91(3):345-350.

61.  Gagnon D, Nadeau S, Tam V. Clinical and administrative outcomes during publicly-funded inpatient stroke rehabilitation based on a case-mix group classification model. Journal of rehabilitation medicine. 2005; 37(1):45-52.

62.  DaVanzo J, El-Gamil A, Li J, Shimer M, Manolov N, Dobson A. Assessment of patient outcomes of rehabilitative care provided in inpatient rehabilitation facilities (IRFs) and after discharge. Vienna, VA: Dobson DaVanzo & Associates, LLC; 2014.

63.  Kushner DS, Peters KM, Johnson-Greene D. Evaluating Siebens Domain Management Model for Inpatient Rehabilitation to Increase Functional Independence and Discharge Rate to Home in Geriatric Patients. Archives of physical medicine and rehabilitation. 2015; 96(7):1310-1318.

64.  Galloway RV, Granger CV, Karmarkar AM, et al. The Uniform Data System for Medical Rehabilitation: Report of patients with debility discharged from inpatient rehabilitation programs in 2000-2010. American journal of physical medicine & rehabilitation/Association of Academic Physiatrists. 2013; 92(1):14-27.

65.  Mallinson T, Deutsch A, Bateman J, et al. Comparison of discharge functional status after rehabilitation in skilled nursing, home health, and medical rehabilitation settings for patients after hip fracture repair. Archives of physical medicine and rehabilitation. 2014; 95(2):209-217.

66.  El-Solh AA, Saltzman SK, Ramadan FH, Naughton BJ. Validity of an artificial neural network in predicting discharge destination from a postacute geriatric rehabilitation unit. Archives of physical medicine and rehabilitation. 2000; 81(10):1388-1393.

67.  Hall RK, Toles M, Massing M, et al. Utilization of acute care among patients with ESRD discharged home from skilled nursing facilities. Clinical journal of the American Society of Nephrology: CJASN. 2015; 10(3):428-434.

68.  Stearns SC, Dalton K, Holmes GM, Seagrave SM. Using propensity stratification to compare patient outcomes in hospital-based versus freestanding skilled-nursing facilities. Medical care research and review: MCRR. 2006; 63(5):599-622.

69.  Wodchis WP, Teare GF, Naglie G, et al. Skilled nursing facility rehabilitation and discharge to home after stroke. Archives of physical medicine and rehabilitation. 2005; 86(3):442-448.

70.  Scheinhorn DJ, Hassenpflug MS, Votto JJ, et al. Post-ICU mechanical ventilation at 23 long-term care hospitals: A multicenter outcomes study. Chest. 2007;131(1):85-93.

71.  Thakar CV, Quate-Operacz M, Leonard AC, Eckman MH. Outcomes of hemodialysis patients in a long-term care hospital setting: A single-center study. American journal of kidney diseases: The official journal of the National Kidney Foundation. 2010;55(2):300-306.

72.  Wolff JL, Meadow A, Weiss CO, Boyd CM, Leff B. Medicare home health patients' transitions through acute and post-acute care settings. Medical care. 2008;46(11):1188-1193.

73.  Ibid.

74.  Kushner DS, Peters KM, Johnson-Greene D. Evaluating Siebens Domain Management Model for Inpatient Rehabilitation to Increase Functional Independence and Discharge Rate to Home in Geriatric Patients. Archives of physical medicine and rehabilitation. 2015;96(7):1310-1318.

75.  Wodchis WP, Teare GF, Naglie G, et al. Skilled nursing facility rehabilitation and discharge to home after stroke. Archives of physical medicine and rehabilitation. 2005;86(3):442-448.

76.  Berkowitz RE, Jones RN, Rieder R, et al. Improving disposition outcomes for patients in a geriatric skilled nursing facility. Journal of the American Geriatrics Society. 2011;59(6):1130-1136.

77.  Kushner DS, Peters KM, Johnson-Greene D. Evaluating use of the Siebens Domain Management Model during inpatient rehabilitation to increase functional independence and discharge rate to home in stroke patients. PM & R: The journal of injury, function, and rehabilitation. 2015;7(4):354-364.

78.  Kushner DS, Peters KM, Johnson-Greene D. Evaluating Siebens Domain Management Model for Inpatient Rehabilitation to Increase Functional Independence and Discharge Rate to Home in Geriatric Patients. Archives of physical medicine and rehabilitation. 2015;96(7):1310-1318.

79.  Wodchis WP, Teare GF, Naglie G, et al. Skilled nursing facility rehabilitation and discharge to home after stroke. Archives of physical medicine and rehabilitation. 2005;86(3):442-448.

80.  Berkowitz RE, Jones RN, Rieder R, et al. Improving disposition outcomes for patients in a geriatric skilled nursing facility. Journal of the American Geriatrics Society. 2011;59(6):1130-1136.

81.  Kushner DS, Peters KM, Johnson-Greene D. Evaluating use of the Siebens Domain Management Model during inpatient rehabilitation to increase functional independence and discharge rate to home in stroke patients. PM & R: The journal of injury, function, and rehabilitation. 2015;7(4):354-364.

82.  National Uniform Billing Committee Official UB-04 Data Specifications Manual 2017, Version 11, July 2016, Copyright 2016, American Hospital Association.

83.  Ibid.

84.  Ibid.

85.  Gage B, Morley M, Spain P, Ingber M. Examining Post Acute Care Relationships in an Integrated Hospital System Final Report. RTI International; 2009.

86.  Friedman, B., and Basu, J.: The rate and cost of hospital readmissions for preventable conditions. Med. Care Res. Rev. 61(2):225-240, 2004. doi:10.1177/1077558704263799.

87.  Jencks, S.F., Williams, M.V., and Coleman, E.A.: Rehospitalizations among patients in the Medicare Fee-for-Service Program. N. Engl. J. Med. 360(14):1418-1428, 2009. doi:10.1016/j.jvs.2009.05.045.

88.  MedPAC: Payment policy for inpatient readmissions, in Report to the Congress: Promoting Greater Efficiency in Medicare. Washington, DC, pp. 103-120, 2007. Available from http://www.medpac.gov/​documents/​reports/​Jun07_​EntireReport.pdf.

89.  ibid.

90.  ibid.

91.  Mor, V., Intrator, O., Feng, Z., et al.: The revolving door of rehospitalization from skilled nursing facilities. Health Aff. 29(1):57-64, 2010. doi:10.1377/hlthaff.2009.0629.

92.  National Quality Forum: All-Cause Admissions and Readmissions Measures. pp. 1-319, April 2015. Available from http://www.qualityforum.org/​Publications/​2015/​04/​All-Cause_​Admissions_​and_​Readmissions_​Measures_​-_​Final_​Report.aspx.

93.  Goldfield, N.I., McCullough, E.C., Hughes, J.S., et al.: Identifying potentially preventable readmissions. Health Care Finan. Rev. 30(1):75-91, 2008. Available from http://www.ncbi.nlm.nih.gov/​pmc/​articles/​PMC4195042/​.

94.  Agency for Healthcare Quality and Research: Prevention Quality Indicators Overview. 2008.

95.  MedPAC: Online Appendix C: Medicare Ambulatory Care Indicators for the Elderly. pp. 1-12, prepared for Chapter 4, 2011. Available from http://www.medpac.gov/​documents/​reports/​Mar11_​Ch04_​APPENDIX.pdf?​sfvrsn=​0 .

96.  Kramer, A., Lin, M., Fish, R., et al.: Development of Inpatient Rehabilitation Facility Quality Measures: Potentially Avoidable Readmissions, Community Discharge, and Functional Improvement. pp. 1-42, 2015. Available from http://www.medpac.gov/​documents/​contractor-reports/​development-of-inpatient-rehabilitation-facility-quality-measures-potentially-avoidable-readmissions-community-discharge-and-functional-improvement.pdf?​sfvrsn=​0.

97.  Kramer, A., Lin, M., Fish, R., et al.: Development of Potentially Avoidable Readmission and Functional Outcome SNF Quality Measures. pp. 1-75, 2014. Available from http://www.medpac.gov/​documents/​contractor-reports/​mar14_​snfqualitymeasures_​contractor.pdf?​sfvrsn=​0 .

98.  Allaudeen, N., Vidyarthi, A., Maselli, J., et al.: Redefining readmission risk factors for general medicine patients. J. Hosp. Med. 6(2):54-60, 2011. doi:10.1002/jhm.805.

99.  Gao, J., Moran, E., Li, Y.-F., et al.: Predicting potentially avoidable hospitalizations. Med. Care 52(2):164-171, 2014. doi:10.1097/MLR.0000000000000041.

100.  Walsh, E.G., Wiener, J.M., Haber, S., et al.: Potentially avoidable hospitalizations of dually eligible Medicare and Medicaid beneficiaries from nursing facility and home‐and community‐based services waiver programs. J. Am. Geriatr. Soc. 60(5):821-829, 2012. doi:10.1111/j.1532-5415.2012.03920.x.

101.  Bratzler DW, Normand SL, Wang Y, et al. An administrative claims model for profiling hospital 30-day mortality rates for pneumonia patients. PLoS One 2011;6(4):e17401.

102.  Keenan PS, Normand SL, Lin Z, et al. An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure. Circulation 2008;1(1):29-37.

103.  Krumholz HM, Wang Y, Mattera JA, et al. An administrative claims model suitable for profiling hospital performance based on 30-day mortality rates among patients with heart failure. Circulation 2006;113:1693-1701.

104.  Institute of Medicine. Preventing Medication Errors. Washington, DC: National Academies Press; 2006.

105.  Ibid.

106.  Leotsakos A., et al. Standardization in patient safety: The WHO High 5s project. Int J Qual Health Care. 2014:26(2):109-116.

107.  The Joint Commission. 2016 Long Term Care: National Patient Safety Goals Medicare/Medicaid Certification-based Option. (NPSG.03.06.01).

108.  Greenwald, J.L., Halasyamani, L., Greene, J., LaCivita, C., et al. (2010). Making inpatient medication reconciliation patient centered, clinically relevant and implementable: A consensus statement on key principles and necessary first steps. Journal of Hospital Medicine, 5(8), 477-485.

109.  Leotsakos A., et al. Standardization in patient safety: The WHO High 5s project. Int J Qual Health Care. 2014:26(2):109-116.

110.  The Joint Commission. 2016 Long Term Care: National Patient Safety Goals Medicare/Medicaid Certification-based Option. (NPSG.03.06.01).

111.  IHI. Medication Reconciliation to Prevent Adverse Drug Events [Internet]. Cambridge, MA: Institute for Healthcare Improvement; [cited 2016 Jan 11]. Available from: http://www.ihi.org/​topics/​adesmedicationreconciliation/​Pages/​default.aspx.

112.  Institute of Medicine. Preventing Medication Errors. Washington, DC: National Academies Press; 2006.

113.  Jha AK, Kuperman GJ, Rittenberg E, et al. Identifying hospital admissions due to adverse drug events using a computer-based monitor. Pharmacoepidemiol Drug Saf. 2001;10(2):113-119.

114.  Hohl CM, Nosyk B, Kuramoto L, et al. Outcomes of emergency department patients presenting with adverse drug events. Ann Emerg Med. 2011;58:270-279.

115.  Kohn LT, Corrigan JM, Donaldson MS. To Err Is Human: Building a Safer Health System Washington, DC: National Academies Press; 1999.

116.  Greenwald, J.L., Halasyamani, L., Greene, J., LaCivita, C., et al. (2010). Making inpatient medication reconciliation patient centered, clinically relevant and implementable: A consensus statement on key principles and necessary first steps. Journal of Hospital Medicine, 5(8), 477-485.

117.  Institute of Medicine. To err is human: Building a safer health system. Washington, DC: National Academies Press; 2000.

118.  Lesar TS, Briceland L, Stein DS. Factors related to errors in medication prescribing. JAMA. 1997:277(4): 312-317.

119.  Bond CA, Raehl CL, & Franke T. Clinical pharmacy services, hospital pharmacy staffing, and medication errors in United States hospitals. Pharmacotherapy. 2002:22(2): 134-147.

120.  Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, et al. Incidence of adverse drug events and potential adverse drug events. Implications for prevention. JAMA. 1995:274(1): 29-34.

121.  Barker KN, Flynn EA, Pepper GA, Bates DW, & Mikeal RL. Medication errors observed in 36 health care facilities. JAMA. 2002: 162(16):1897-1903.

122.  Bates DW, Boyle DL, Vander Vliet MB, Schneider J, & Leape L. Relationship between medication errors and adverse drug events. J Gen Intern Med. 1995:10(4): 199-205.

123.  Fu, Alex Z., et al. “Potentially inappropriate medication use and healthcare expenditures in the US community-dwelling elderly.” Medical care 45.5 (2007): 472-476.

124.  Wong, Jacqueline D., et al. “Medication reconciliation at hospital discharge: Evaluating discrepancies.” Annals of Pharmacotherapy 42.10 (2008): 1373-1379.

125.  Tjia, J., Bonner, A., Briesacher, B.A., McGee, S., Terrill, E., & Miller, K. (2009). Medication discrepancies upon hospital to skilled nursing facility transitions. Journal of general internal medicine, 24(5), 630-635.

126.  Kripalani S, Roumie CL, Dalal AK, et al. Effect of a pharmacist intervention on clinically important medication errors after hospital discharge: A randomized controlled trial. Ann Intern Med. 2012:157(1):1-10.

127.  Gandara, Esteban, et al. “Communication and information deficits in patients discharged to rehabilitation facilities: An evaluation of five acute care hospitals.” Journal of Hospital Medicine 4.8 (2009): E28-E33.

128.  Gandara, Esteban, et al. “Deficits in discharge documentation in patients transferred to rehabilitation facilities on anticoagulation: Results of a system wide evaluation.” Joint Commission Journal on Quality and Patient Safety 34.8 (2008): 460-463.

129.  Coleman EA, Smith JD, Raha D, Min SJ. Post hospital medication discrepancies: Prevalence and contributing factors. Arch Intern Med. 2005 165(16):1842-1847.

130.  Wong JD, Bajcar JM, Wong GG, et al. Medication reconciliation at hospital discharge: Evaluating discrepancies. Ann Pharmacother. 2008 42(10):1373-1379.

131.  Hawes EM, Maxwell WD, White SF, Mangun J, Lin FC. Impact of an outpatient pharmacist intervention on medication discrepancies and health care resource utilization in post hospitalization care transitions. Journal of Primary Care & Community Health. 2014; 5(1):14-18.

132.  Foust JB, Naylor MD, Bixby MB, Ratcliffe SJ. Medication problems occurring at hospital discharge among older adults with heart failure. Research in Gerontological Nursing. 2012, 5(1): 25-33.

133.  Pherson EC, Shermock KM, Efird LE, et al. Development and implementation of a post discharge home-based medication management service. Am J Health Syst Pharm. 2014; 71(18): 1576-1583.

134.  Pronovosta P, Weasta B, Scwarza M, et al. Medication reconciliation: A practical tool to reduce the risk of medication errors. J Crit Care. 2003; 18(4): 201-205.

135.  Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, et al. Incidence of adverse drug events and potential adverse drug events. Implications for prevention. JAMA. 1995:274(1): 29-34.

136.  Himmel, W., M. Tabache, and M. M. Kochen. “What happens to long-term medication when general practice patients are referred to hospital?.”European journal of clinical pharmacology 50.4 (1996): 253-257.

137.  Chhabra, P.T., et al. (2012). “Medication reconciliation during the transition to and from LTC settings: A systematic review.” Res Social Adm Pharm 8(1): 60-75.

138.  Kripalani S, Roumie CL, Dalal AK, et al. Effect of a pharmacist intervention on clinically important medication errors after hospital discharge: A randomized controlled trial. Ann Intern Med. 2012:157(1):1-10.

139.  March 2015 Report to the Congress: Medicare Payment Policy. Medicare Payment Advisory Commission; 2015.

140.  DRR pilot items DRR-01, DRR-02 and DRR-03 are equivalent to the proposed rule DRR PAC instrument items N. 2001, N. 2003 and N. 2005.

141.  DRR pilot items DRR-01, DRR-02 and DRR-03 are equivalent to the proposed rule DRR PAC instrument items N. 2001, N. 2003 and N. 2005.

142.  DRR pilot items DRR-01, DRR-02 and DRR-03 are equivalent to the proposed rule DRR PAC instrument items N. 2001, N. 2003 and N. 2005.