Skip to Content

Proposed Rule

Medicare Program; Advancing Care Coordination Through Episode Payment Models (EPMs); Cardiac Rehabilitation Incentive Payment Model; and Changes to the Comprehensive Care for Joint Replacement Model (CJR)

Document Details

Information about this document as published in the Federal Register.

Enhanced Content

Relevant information about this document from Regulations.gov provides additional context. This information is not part of the official Federal Register document.

Published Document

This document has been published in the Federal Register. Use the PDF linked in the document sidebar for the official electronic format.

Start Preamble Start Printed Page 50794

AGENCY:

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

ACTION:

Proposed rule.

SUMMARY:

This proposed rule proposes to implement three new Medicare Parts A and B episode payment models under section 1115A of the Social Security Act. Acute care hospitals in certain selected geographic areas will participate in retrospective episode payment models targeting care for Medicare fee-for-service beneficiaries receiving services during acute myocardial infarction, coronary artery bypass graft, and surgical hip/femur fracture treatment episodes. All related care within 90 days of hospital discharge will be included in the episode of care. We believe this model will further our goals of improving the efficiency and quality of care for Medicare beneficiaries receiving care for these common clinical conditions and procedures. This proposed rule also includes several proposed modifications to the Comprehensive Care for Joint Replacement model.

DATES:

Comment period: To be assured consideration, comments on this proposed rule must be received at one of the addresses provided in the ADDRESSES section no later than 5 p.m. EDT on October 3, 2016.

ADDRESSES:

In commenting, please refer to file code CMS-5519-P. Because of staff and resource limitations, we cannot accept comments by facsimile (FAX) transmission.

You may submit comments in one of four ways (please choose only one of the ways listed):

1. Electronically. You may submit electronic comments on this regulation to http://www.regulations.gov. Follow the “Submit a comment” instructions.

2. By regular mail. You may mail written comments to the following address ONLY: Centers for Medicare & Medicaid Services, Department of Health and Human Services, Attention: CMS-5519-P, P.O. Box 8013, Baltimore, MD 21244-1850.

Please allow sufficient time for mailed comments to be received before the close of the comment period.

3. By express or overnight mail. You may send written comments to the following address ONLY: Centers for Medicare & Medicaid Services, Department of Health and Human Services, Attention: CMS-5519-P, Mail Stop C4-26-05, 7500 Security Boulevard, Baltimore, MD 21244-1850.

4. By hand or courier. Alternatively, you may deliver (by hand or courier) your written comments ONLY to the following addresses prior to the close of the comment period:

a. For delivery in Washington, DC— Centers for Medicare & Medicaid Services, Department of Health and Human Services, Room 445-G, Hubert H. Humphrey Building, 200 Independence Avenue SW., Washington, DC 20201

(Because access to the interior of the Hubert H. Humphrey Building is not readily available to persons without federal government identification, commenters are encouraged to leave their comments in the CMS drop slots located in the main lobby of the building. A stamp-in clock is available for persons wishing to retain a proof of filing by stamping in and retaining an extra copy of the comments being filed.)

b. For delivery in Baltimore, MD— Centers for Medicare & Medicaid Services, Department of Health and Human Services, 7500 Security Boulevard, Baltimore, MD 21244-1850.

If you intend to deliver your comments to the Baltimore address, call telephone number (410) 786-7195 in advance to schedule your arrival with one of our staff members.

Comments erroneously mailed to the addresses indicated as appropriate for hand or courier delivery may be delayed and received after the comment period.

For information on viewing public comments, see the beginning of the SUPPLEMENTARY INFORMATION section.

Start Further Info

FOR FURTHER INFORMATION CONTACT:

For questions related to the proposed EPMs: NEPMRULE@cms.hhs.gov.

For questions related to the CJR model: CJR@cms.hhs.gov.

End Further Info End Preamble Start Supplemental Information

SUPPLEMENTARY INFORMATION:

Inspection of Public Comments: All comments received before the close of the comment period are available for viewing by the public, including any personally identifiable or confidential business information that is included in a comment. We post all comments received before the close of the comment period on the following Web site as soon as possible after they have been received: http://www.regulations.gov. Follow the search instructions on that Web site to view public comments.

Comments received timely will also be available for public inspection as they are received, generally beginning approximately 3 weeks after publication of a document, at the headquarters of the Centers for Medicare & Medicaid Services, 7500 Security Boulevard, Baltimore, Maryland 21244, Monday through Friday of each week from 8:30 a.m. to 4 p.m. To schedule an appointment to view public comments, phone 1-800-743-3951.

Electronic Access

This Federal Register document is also available from the Federal Register online database through Federal Digital System (FDsys), a service of the U.S. Government Printing Office. This database can be accessed via the internet at http://www.gpo.gov/​fdsys/​.

Alphabetical List of Acronyms

Because of the many terms to which we refer by acronym, abbreviation, or short form in this proposed rule, we are listing the acronyms, abbreviations and short forms used and their corresponding terms in alphabetical order.

ACE Acute-care episode

ACO Accountable Care Organization

ALOS Average length of stay

AMA American Medical Association

AMI Acute Myocardial Infarction

APM Alternative Payment Model

ASC QRP Ambulatory Surgical Center Quality Reporting Program

ASC Ambulatory Surgical Center

ASPE Assistant Secretary for Planning and Evaluation

BPCI Bundled Payments for Care Improvement

CABG Coronary Artery Bypass Graft

CAD Coronary artery disease

CAH Critical access hospital

CBSA Core-Based Statistical Area

CC Complication or comorbidity

CCDA Consolidated clinical document architecture

CCDE Core clinical data elements

CCN CMS Certification Number

CEC Comprehensive ESRD Care Initiative

CEHRT Certified Electronic Health Record Technology

CFR Code of Federal Regulations

CJR Comprehensive Care for Joint Replacement

CMHC Community Mental Health CenterStart Printed Page 50795

CMI Case Mix Index

CMMI Center for Medicare and Medicaid Innovation

CMP Civil monetary penalty

CMS Centers for Medicare & Medicaid Services

CoP Condition of Participation

CPC Comprehensive Primary Care Initiative

CPT Current Procedural Terminology

CR Cardiac rehabilitation

CSA Combined Statistical Area

CVICU Cardiovascular intensive care units

CY Calendar year

DME Durable medical equipment

DMEPOS Durable medical equipment, prosthetics, orthotics, and supplies

DSH Disproportionate Share Hospital

ECQM Electronic Clinical Quality Measures

EFT Electronic funds transfer

EHR Electronic health record

E/M Evaluation and management

EPM Episode payment model

ESCO ESRD Seamless Care Organization

ESRD End-Stage Renal Disease

FFS Fee-for-service

GAAP Generally-Accepted Accounting Principles

GEM General Equivalence Mapping

GPCI Geographic Practice Cost Index

HAC Hospital-Acquired Condition

HACRP Hospital-Acquired Condition Reduction Program

HCAHPS Hospital Consumer Assessment of Healthcare Providers and Systems

HCC Hierarchical Condition Category

HCPCS Healthcare Common Procedure Coding System

HHA Home health agency

HHPPS Home Health Prospective Payment System

HHRG Home Health Resource Group

HHS U.S. Department of Health and Human Services

HH QRP Home Health Quality Reporting Program

HICN Health Insurance Claim Number

HIPPA Health Insurance Portability and Accountability Act

HIQR Hospital Inpatient Quality Reporting

Health IT Health Information Technology

HLMR HCAHPS Linear Mean Roll Up

HOOS Hip Dysfunction and Osteoarthritis Outcome Score

HOPD Hospital outpatient department

HRRP Hospital Readmissions Reductions Program

HRR Hospital Referral Region

HVBP Hospital Value-Based Purchasing Program

HIV Human Immunodeficiency Virus

ICD-9-CM International Classification of Diseases, 9th Revision, Clinical Modification

IRFQRICD-10-CM International Classification of Diseases, 10th Revision, Clinical Modification

ICR Intensive Cardiac Rehabilitation

IME Indirect medical education

IPPS Inpatient Prospective Payment System

IPF Inpatient psychiatric facility

IRF QRP Inpatient Rehabilitation Facility Quality Reporting Program

IPF QRP Inpatient Psychiatric Facility Quality Reporting Program

IRF Inpatient rehabilitation facility

KOOS Knee Injury and Osteoarthritis Outcome Score

LEJR Lower-extremity joint replacement

LIP Low-income percentage

LOS Length-of-stay

LTCH QRP Long-Term Care Hospital Quality Reporting Program

LTCH Long-term care hospital

LUPA Low-utilization payment adjustment

MAC Medicare Administrative Contractor

MACRA Medicare Access and CHIP Reauthorization Act of 2015

MAPCP Multi-Payer Advanced Primary Care Practice

MAT Measure Authoring Tool

MCC Major complications or comorbidities

MCCM Medicare Care Choices Model

MDC Major diagnostic category

MDH Medicare-Dependent Hospital

MedPAC Medicare Payment Advisory Commission

MIPS Merit-based Incentive Payment System

MP Malpractice

MSA Metropolitan Statistical Area

MS-DRG Medical Severity Diagnosis-Related Group

NPI National Provider Identifier

NPRA Net Payment Reconciliation Amount

NQF National Quality Forum

OCM Oncology Care Model

OIG Department of Health and Human Services' Office of the Inspector General

OPPS Outpatient Prospective Payment System

OQR Outpatient Quality Reporting

PBPM Per-beneficiary per-month

PCI Percutaneous Coronary Intervention

PCMH Primary Care Medical Homes

PE Practice Expense

PFS Physician Fee Schedule

PGP Physician group practice

PQRS Physician Quality Reporting System

PHA Partial hip arthroplasty

PPS Prospective Payment System

PRO Patient-Reported Outcome

PROMIS Patient-Reported Outcomes Measurement Information Systems

PRO-PM Patient-Reported Outcome Performance Measure

PTCA Percutaneous transluminal coronary angioplasty

PY Performance year

QIO Quality Improvement Organization

RAC Recovery Audit Contractor

RRC Rural Referral Center

RSCR Risk-Standardized Complication Rate

RSRR Risk-Standardized Readmission Rate

RSMR Risk-Standardized Mortality Rate

RVU Relative Value Unit

SCH Sole Community Hospital

SHFFT Surgical hip/femur fracture treatment

SILS2 Single Item Health Literacy Screening

SNF QRP Skilled Nursing Facility Quality Reporting Program

SNF Skilled nursing facility

THA Total hip arthroplasty

TIN Taxpayer identification number

TKA Total knee arthroplasty

TP Target price

UHDDS Uniform Hospital Discharge Data Set

VR-12 Veterans Rand 12 Item Health Survey

Table of Contents

I. Executive Summary

A. Purpose

B. Summary of the Major Provisions

1. Model Overview—EPM episodes of care

2. Model Scope

3. Payment

4. Similar, Previous, and Concurrent Models

5. Overlap with Ongoing CMS Efforts

6. Quality Measures and Reporting Requirements

7. Beneficiary Protections

8. Financial Arrangements

9. Data Sharing

10. Program Waivers

C. Summary of Economic Effects

II. Background

III. Provisions of the Proposed Regulations

A. Selection of Episodes for Episode Payment Models in this Rulemaking and Potential Future Directions

1. Selection of Episodes for Episode Payment Models in this Rulemaking

a. Overview

b. SHFFT Model

c. AMI and CABG Models

2. Advanced Alternative Payment Model Considerations

a. Overview for the EPMs

b. EPM Participant Tracks

c. Clinician Financial Arrangements Lists under the EPMs

d. Documentation Requirements

3. Future Directions for Episode Payment Models

a. Refinements to the BPCI Initiative Models

b. Potential Future Condition-Specific Episode Payment ModelsStart Printed Page 50796

c. Potential Future Event-Based Episode Payment Models for Procedures and Medical Conditions

d. Health Information Technology Readiness for Potential Future Episode Payment Models

B. Proposed Definition of the Episode Initiator and Selected Geographic Areas

1. Background

2. Proposed definition of episode initiator

3. Financial responsibility for episode of care

4. Proposed Geographic Unit of Selection and Exclusion of Selected Hospitals

5. Overview and Options for Geographic Area Selection for AMI and CABG Episodes

a. Exclusion of Certain MSAs

b. Proposed Selection Approach

(1) Factors Considered but Not Used

(2) Sample Size Calculations and the Number of Selected MSAs

(3) Method of Selecting MSAs

C. Episode Definition for EPMs

1. Background

2. Overview of Proposed Three New Episode Payment Models

3. Clinical Dimensions of AMI, CABG, and SHFFT Model Episodes

a. Definition of the Clinical Conditions Included in AMI, CABG, and SHFFT Model Episodes

(1) AMI (Medical Management and PCI) Model

(2) CABG Model

(3) SHFFT (Excludes Lower Extremity Joint Replacement) Model

b. Definition of the Related Services Included in EPM Episodes

4. EPM Episodes

a. Beneficiary Care Inclusion Criteria and Beginning of EPM Episodes

(1) General Beneficiary Care Inclusion Criteria

(2) Beginning AMI Model Episodes

(3) Beginning CABG Model Episodes

(4) Beginning SHFFT Episodes

(5) Special Policies for Hospital Transfers of Beneficiaries with AMI

b. Middle of EPM Episodes

c. End of EPM Episodes

(1) AMI and CABG Models

(2) SHFFT Model

D. Methodology for Setting EPM Episode Prices and Paying EPM Participants in the AMI, CABG, and SHFFT Models

1. Background

a. Overview

b. Key Terms for EPM Episode Pricing and Payment

2. Performance Years, Retrospective Episode Payments, and Two-Sided Risk EPMs

a. Performance Period

b. Retrospective Payment Methodology

c. Two-Sided Risk EPMs

3. Adjustments to Actual EPM Episode Payments and to Historical Episode Payments used to Set Episode Prices

a. Overview

b. Special Payment Provisions

c. Services that Straddle Episodes

d. High-Payment EPM Episodes

e. Treatment of Reconciliation Payments and Medicare Repayments when Calculating Historical EPM-Episode Payments to Update EPM Benchmark and Quality-Adjusted Target Prices

4. EPM-Episode Price-Setting Methodologies

a. Overview

(1) AMI model

(2) CABG model

(3) SHFFT model

b. EPM-Episode Benchmark and Quality-Adjusted Target Price Features

(1) Risk-Stratifying EPM-Episode Benchmark Prices based on MS-DRG and Diagnosis

(2) Adjustments to Account for EPM-Episode Price Variation

(a) Adjustments for Certain AMI Model Episodes with Chained Anchor Hospitalizations

(b) Adjustments for CABG Model Episodes

(c) Adjustments for Certain AMI Model Episodes with CABG Readmissions

(d) Potential Future Approaches to setting Target Prices for AMI and Hip Fracture Episodes

(e) Summary of Pricing Methodologies for AMI, CABG, and SHFFT Model Episode Scenarios

(3) 3 Years of Historical Data

(4) Trending Historical Data to the Most Recent Year

(5) Update Historical EPM-Episode Payments for Ongoing Payment System Updates

(6) Blend Hospital-Specific and Regional Historical Data

(7) Define Regions as U.S. Census Divisions

(8) Normalize for Provider-Specific Wage Adjustment Variations

(9) Combining Episodes to Set Stable Benchmark and Quality-Adjusted Target Prices

(10) Effective Discount Factors

c. Approach to Combine Pricing Features for all SHFFT Model Episodes and AMI Model Episodes without CABG readmissions

d. Approach to Combine Pricing Features for CABG Model Episodes

(1) Anchor Hospitalization Portion of CABG Model Episodes

(2) Approach to Combine Pricing Features for Post-Anchor Hospitalization Portion of CABG Model Episodes

(3) Combine CABG Anchor Hospitalization Benchmark Price and CABG Post-Anchor Hospitalization Benchmark Price

e. Approach to Combine Pricing Features for AMI Model episodes with CABG Readmissions

5. Process for Reconciliation

a. Net Payment Reconciliation Amount (NPRA)

b. Payment Reconciliation

c. Reconciliation Report

6. Adjustments for Overlaps with Other Innovation Center Models and CMS Programs

a. Overview

b. Provider Overlap

(1) BPCI Participant Hospitals in Geographic Areas Selected for EPMs

(2) BPCI Physician Group Practice (PGP) Episode Initiators in Hospitals Participating in EPMs

c. Beneficiary Overlap

(1) Beneficiary Overlap with BPCI

(2) Beneficiary Overlap with the CJR Model and other EPMs

(3) Beneficiary Overlap with Shared Savings Models and Programs

d. Payment Reconciliation of Overlap with non-ACO CMS Models and Programs

7. Limits or Adjustments to EPM Participants' Financial Responsibility

a. Overview

b. Limit on Actual EPM-Episode Payment Contribution to Repayment Amounts and Reconciliation Payments

(1) Limit on Actual EPM-Episode Payment Contribution to Repayment Amounts

(2) Limitation on Reconciliation Payments

c. Additional Protections for Certain EPM Participants

(1) Proposed Policies for Certain EPM Participants to Further Limit Repayment Responsibility

(2) Considerations for Hospitals Serving a High Percentage of Potentially Vulnerable Populations

d. Application of Stop-Gain and Stop-Loss Limits

e. EPM Participant Responsibility for Increased Post-Episode Payments

8. Appeals Process

a. Overview

b. Notice of calculation error (first level appeal)

c. Dispute Resolution Process (second level of appeal)

d. Exception to the Notice of Calculation Error Process and Notice of Termination

e. Limitations on review

E. EPM quality measures, public display, and use of quality measures in the EPM payment methodology

1. Background

2. Selection of Proposed Quality Measures for the EPMs

a. Overview of Quality Measure Selection

b. AMI Model Quality Measures

c. CABG Model Quality Measures

d. SHFFT Model Quality Measures

3. Proposed Use of Quality Measures in the EPM Payment Methodologies

a. Overview of EPM Composite Quality Score Methodology

b. Determining Quality Measure Performance

c. Determining Quality Measure Improvement

d. Determining Successful Submission of Voluntary Data for AMI and SHFFT Models

(1) Hybrid AMI Mortality (NQF #2473) Voluntary Data

(2) Patient-Reported Outcomes and Limited Risk Variable Voluntary Data Following Elective Primary THA/TKA

e. Calculation of the EPM-Specific Composite Quality Score

(1) AMI Model Composite Quality Score

(2) CABG Model Composite Quality Score

(3) SHFFT Model Composite Quality Score

f. EPM Pay-for-Performance Methodologies to Link Quality and Payment

(1) Overview of Pay-for-Performance Proposals Applicable to the EPMs

(2) AMI and CABG Model Pay-for-Performance Methodology

(a) AMI Model Pay-for-Performance Methodology

(b) CABG Model Pay-for-Performance MethodologyStart Printed Page 50797

(c) Interface Considerations for the AMI and CABG Model Methodologies

(3) SHFFT Model Pay-for-Performance Methodology

4. Details on Quality Measures for the EPMs

a. AMI Model-Specific Measures

(1) Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate Following Acute Myocardial Infarction (AMI) Hospitalization (NQF #0230) (MORT-30-AMI)

(a) Background

(b) Data Sources

(c) Cohort

(d) Inclusion and Exclusion Criteria

(e) Risk-Adjustment

(f) Calculating the Risk-Standardized Mortality Ratio (RSMR) and Performance Period

(2) Excess Days in Acute Care after Hospitalization for Acute Myocardial Infarction (AMI Excess Days)

(a) Background

(b) Data Sources

(c) Cohort

(d) Inclusion and Exclusion Criteria

(e) Risk-Adjustment

(f) Calculating the Rate and Performance Period

(3) Hybrid Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate Following Acute Myocardial Infarction (AMI) Hospitalization (NQF# 2473)(Hybrid AMI Mortality)

(a) Background

(b) Data Sources

(c) Cohort

(d) Inclusion and Exclusion Criteria

(e) Risk-Adjustment

(f) Calculating the Risk-Standardized Mortality Ratio (RSMR) and Performance Period

(g) Requirements for Successful Submission of AMI Voluntary Data

b. CABG Model-Specific Measure

(1) Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) Following Coronary Artery Bypass Graft (CABG) Surgery (NQF# 2558)(MORT-30-CABG)

(a) Background

(b) Data Source

(c) Cohort

(d) Inclusion and Exclusion Criteria

(e) Risk-Adjustment

(f) Calculating the Risk-Standardized Mortality Ratio (RSMR) and Performance Period

c. SHFFT Model-Specific Measures

(1) Hospital Level Risk Standardized Complication Rate (RSCR) Following Elective Primary Total Hip Arthroplasty (THA) and/or Total Knee Arthroplasty (TKA) (NQF #1550) (Hip/Knee Complications)

(a) Background

(b) Data Sources

(c) Cohort

(d) Inclusion and Exclusion Criteria

(e) Risk Adjustment

(f) Calculating the Risk Standardized Complication Rate and Performance Period

(2) Hospital-Level Performance Measure(s) of Patient-Reported Outcomes Following Elective Primary Total Hip and/or Total Knee Arthroplasty

(a) Background

(b) Data Sources

(c) Cohort

(d) Inclusion and Exclusion Criteria

(e) Outcome

(f) Risk Adjustment (if applicable)

(g) Calculating the Risk Standardized Rate

(h) Requirements for Successful Submission of THA/TKA Patient-Reported Outcome-Based Voluntary Data

d. Measure Used for All EPMs

(1) Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) Survey (NQF #0166)

(a) Background

(b) Data Sources

(c) Cohort

(d) Inclusion and Exclusion Criteria

(e) Case-Mix Adjustment

(f) HCAHPS Scoring

(g) Calculating the Rate and Performance Period

e. Potential Future Measures

5. Form, Manner, and Timing of Quality Measure Data Submission

6. Display of Quality Measures and Availability of Information for the Public from the AMI, CABG, and SHFFT Models

F. Compliance Enforcement and Termination of an Episode Payment Model

1. Overview and Background

2. Proposed Compliance Enforcement for EPMs

3. Proposed Termination of an Episode Payment Model

G. Monitoring and Beneficiary Protection

1. Introduction and Summary

2. Beneficiary Choice

3. Beneficiary Notification

4. Monitoring for Access to Care

5. Monitoring for Quality of Care

6. Monitoring for Delayed Care

H. Access to Records and Record Retention

I. Financial Arrangements under EPM

1. Background

2. Overview of the EPM Financial Arrangements

3. EPM Collaborators

4. Sharing Arrangements under EPM

a. General

b. Requirements

c. Gainsharing Payment, Alignment Payment, and Internal Cost Savings Conditions and Restrictions

d. Documentation Requirements

5. Distribution Arrangements under the EPM

a. General

b. Requirements

6. Downstream Distribution Arrangements under the EPM

a. General

b. Requirements

7. Summary of Proposals for Sharing, Distribution, and Downstream Distribution Arrangements under the EPM

8. Enforcement Authority

9. Beneficiary Engagement Incentives under the EPM

a. General

b. Technology Provided to an EPM Beneficiary

c. Clinical Goals of the EPM

d. Documentation of Beneficiary Incentives

10. Compliance with Fraud and Abuse Laws

J. Proposed Waivers of Medicare Program Requirements

1. Overview

2. Summary of Waivers Adopted Under the CJR Model

3. Analysis of Current Model Data

a. Analysis of Waiver Usage

b. Analysis of Discharge Destination—Post-Acute Care Usage

c. Analysis of Hospital Mean Length of Stay Data

4. Post-Discharge Home Visits

a. AMI Model

b. CABG Model

c. SHFFT Model

5. Billing and Payment for Telehealth Services

6. SNF 3-Day Rule

a. Waiver of SNF 3-Day Rule

b. Additional Beneficiary Protections under the SNF 3-Day Stay Rule Waiver

7. Waivers of Medicare Program Rules to Allow Reconciliation Payment or Repayment Actions Resulting from the Net Payment Reconciliation Amount

8. New Waiver for Providers and Suppliers of Cardiac Rehabilitation and Intensive Cardiac Rehabilitation Services Furnished to EPM Beneficiaries During an AMI or CABG Episode

K. Data Sharing

1. Overview

2. Beneficiary Claims Data

3. Aggregate Regional Data

4. Timing and Period of Baseline Data

5. Frequency and Period of Claims Data Updates for Sharing Beneficiary-Identifiable Claims Data During the Performance Period

6. Legal Permission to Share Beneficiary-Identifiable Data

7. Data Considerations with Respect to EPM and CJR Collaborators

L. Coordination with other agencies

IV. Evaluation Approach

A. Background

B. Design and Evaluation Methods

C. Data Collection Methods

D. Key Evaluation Research Questions

E. Evaluation Period and Anticipated Reports

V. Comprehensive Care for Joint Replacement Model

A. Participant Hospitals in the CJR Model

B. Inclusion of Reconciliation and Repayment Amounts when Updating Data for Target Prices

C. Quality-Adjusted Target Price

D. Reconciliation

1. Hospital Responsibility for Increased Post-Episode Payments

2. ACO Overlap and Subsequent Reconciliation Calculation

3. Stop-Loss and Stop-Gain Limits

4. Proposed Modifications to Reconciliation Process

E. Use of Quality Measures and the Composite Quality Score

1. Hospitals Included in Quality Performance Distribution

2. Quality Improvement Points

3 Relationship of composite quality score to quality categoriesStart Printed Page 50798

4. Maximum Composite Quality Score

5e. Acknowledgement of Voluntary Data Submission

6. Calculation of the HCAHPS Linear Mean Roll-up (HLMR) Score

F. Accounting for Overlap with CMS ACO Models and the Medicare Shared Savings Program

G. Appeals Process

H. Beneficiary Notification

1. Physician and PGP Provision of Notice

2. Other CJR collaborators provision of notice

3. Beneficiary Notification Compliance and Records

4. Compliance with § 510.110

I. Compliance Enforcement

1. Failure to comply.

J. Financial Arrangements under the CJR model

1. Definitions related to Financial Arrangements

a. Addition to the definition of CJR collaborators

b. Deleting the term collaborator agreements

c. Addition of CJR activities

2. Sharing arrangements

a. General

b. Requirements

c. Gainsharing Payment, Alignment Payment, and Internal Cost Savings Conditions and Restrictions.

d. Documentation

3. Distribution arrangements

a. General

b. Requirements

4. Downstream Distribution Arrangements under the CJR model

a. General

b. Requirements

5. Summary of Proposals for Sharing, Distribution, and Downstream Distribution

K. Beneficiary Incentives under the CJR model

L. Access to Records and Record Retention

M. Waivers of Medicare Program Rules to Allow Reconciliation Payment or Repayment Actions Resulting From the Net Payment Reconciliation Amount

N. SNF 3-day Waiver Beneficiary Protections

O. Advanced Alternative Payment Model Requirements

1. Overview for CJR

2. CJR Participant Hospital Track

3. Clinician Financial Arrangements Lists under the CJR Model

4. Documentation Requirements

VI. Cardiac Rehabilitation Incentive Payment Model

A. Background

B. Overview of the CR Incentive Payment Model

1. Rationale for the CR Incentive Payment Model

2. General Design of the CR Incentive Payment Model

C. CR Incentive Payment Model Participants

D. CR/ICR Services that Count Towards CR Incentive Payments

E. Determination of CR Incentive Payments

1. Determination of CR Amounts that Sum to Determine a CR Incentive Payment

2. Relation of CR Incentive Payments to EPM Pricing and Payment Policies and Sharing Arrangements for EPM-CR participants

3. CR Incentive Payment Report

4. Proposed Timing for Making CR Incentive Payments

F. Provisions for FFS-CR Participants

1. Access to Records and Retention for FFS-CR participants

2. Appeals Process for FFS-CR Participants

a. Overview

b. Notice of Calculation Error (first level appeal).

c. Dispute Resolution Process (second level of appeal)

d. Exception to the Notice of Calculation Error Process and Notice of Termination.

e. Limitations on review.

3. Data Sharing for FFS-CR Participants

a. Overview

b. Data Sharing with CR participants

4. Compliance Enforcement for FFS-CR Participants and Termination of the CR Incentive Payment Model

5. Enforcement Authority for FFS-CR Participants

6. Beneficiary Engagement Incentives for FFS-CR Participants

7. Waiver of Physician Definition for Providers and Suppliers of CR/ICR Services Furnished to FFS-CR Beneficiaries During an AMI Care Period or CABG Care Period

a. Overview of Program Rule Waivers Under an EPM

b. General Physician Requirements for Furnishing CR/ICR Services

c. Proposed Waiver of Physician Definition For Providers and Suppliers of CR/ICR Services Furnished to EPM Beneficiaries During AMI or CABG Model Episodes

d. Proposed Waiver of Physician Definition For Providers or Suppliers of CR/ICR Services Furnished to FFS-CR Beneficiaries During AMI Care Periods or CABG Care Periods

G. Considerations Regarding Financial Arrangements Under the CR Incentive Payment Model

VII. Collection of Information Requirements

VIII. Response to Comments

IX. Regulatory Impact Analysis

A. Statement of Need

1. Need for EPM Proposed Rule

2. Need for CJR Modifications

3. Need for CR Incentive Payment Model

4. Aggregate Impact of EPMs, CJR, and CR Incentive Payment Model

B. Overall Impact

C. Anticipated Effects

1. Overall Magnitude of the Model and its Effects on the Market

a. EPMs

b. CJR

c. CR Incentive Payment Model

d. Aggregate Effects on the Market

2. Effects on the Medicare Program

a. EPMs

(1) Assumptions

(2) Analyses

(3) Uncertainties

b. CJR

(1) Assumptions and Uncertainties

(2) Analyses

c. CR Incentive Payment Model

(1) Assumptions and Uncertainties

(2) Analysis

3. Effects on Beneficiaries

4. Effects on Small Rural Hospitals

5. Effects on Small Entities

6. Effects on Collection of Information

7. Unfunded Mandates

D. Alternatives Considered

E. Accounting Statement and Table

F. Conclusion

Regulations Text

I. Executive Summary

A. Purpose

The purpose of this proposed rule—Advancing Care Coordination through Episode Payment Models, is to propose the creation and testing of three new episode payment models (EPMs) and a Cardiac Rehabilitation (CR) incentive payment model under the authority of the Center for Medicare and Medicaid Innovation (CMMI or “the Innovation Center”). Section 1115A of the Social Security Act (“the Act”) authorizes the Innovation Center to test innovative payment and service-delivery models to reduce Medicare, Medicaid, and Children's Health Insurance Program expenditures while preserving or enhancing the quality of care furnished to such programs' beneficiaries. Under the fee-for-service (FFS) program, Medicare makes separate payments to providers and suppliers for the items and services furnished to a beneficiary over the course of treatment (an episode of care). With the amount of payments dependent on the volume of services delivered, providers may not have incentives to invest in quality-improvement and care-coordination activities. As a result, care may be fragmented, unnecessary, or duplicative. The goal for the proposed EPMs is to improve the quality of care provided to beneficiaries in an applicable episode while reducing episode spending through financial accountability.[1] The proposed EPMs would include models for episodes of care surrounding an acute myocardial infarction (AMI), coronary artery bypass graft (CABG), and surgical hip/femur fracture treatment excluding lower extremity joint replacement (SHFFT). Under the proposed rule, the Centers for Medicare & Medicaid Services (CMS) will test whether an EPM for AMI, CABG, and SHFFT episodes of care will reduce Medicare expenditures while preserving or enhancing the quality of care for Medicare beneficiaries. We anticipate the proposed models would benefit Medicare beneficiaries by improving the Start Printed Page 50799coordination and transition of care, improving the coordination of items and services paid for through FFS Medicare, encouraging more provider investment in infrastructure and redesigned care processes for higher-quality and more efficient service delivery, and incentivizing higher-value care across the inpatient and post-acute care spectrum. We propose to test the proposed EPMs for 5 performance years, beginning July 1, 2017, and ending December 31, 2021.

Within this proposed rule, we propose three distinct EPMs focused on episodes of care for AMI, CABG, and SHFFT episodes. We chose these episodes for the proposed models because, as discussed in depth in section III.A. of this proposed rule, we believe hospitals would have significant opportunity to redesign care and improve quality of care furnished during the applicable episode. In addition, significant variation in spending occurs during these high-expenditure, common episodes. The proposed EPMs would enable hospitals to consider the most appropriate strategies for care redesign, including: (1) increasing post-hospitalization follow-up and medical management for patients; (2) coordinating across the inpatient and post-acute care spectrum; (3) conducting appropriate discharge planning; (4) improving adherence to treatment or drug regimens; (5) reducing readmissions and complications during the post-discharge period; (6) managing chronic diseases and conditions that may be related to the proposed EPMs' episodes; (7) choosing the most appropriate post-acute care setting; and (8) coordinating between providers and suppliers such as hospitals, physicians, and post-acute care providers. The proposed EPMs would offer hospitals the opportunity to examine and better understand their own care processes and patterns with regard to patients in AMI, CABG, and SHFFT episodes, as well as the processes of post-acute care providers and physicians.

We previously have used our statutory authority under section 1115A of the Act to test other episode payment models such as the Bundled Payments for Care Improvement (BPCI) initiative and Comprehensive Care for Joint Replacement (CJR) model. Bundled payments for multiple services in an episode of care hold participating organizations financially accountable for that episode of care. Such models also allow participants to receive payments based in part on the reduction in Medicare expenditures that arise from such participants' care redesign efforts. This payment can be used for investments in care redesign strategies and infrastructure, as well as to incentivize collaboration with other providers and suppliers furnishing services to beneficiaries included in the models.

We believe the proposed EPMs would further the Innovation Center's mission and the Administration's goal of increasingly paying for value and outcomes, rather than for volume alone,[2] by promoting the alignment of financial and other incentives for all health care providers caring for beneficiaries during SHFFT, CABG, or AMI episodes. The acute care hospital where an eligible beneficiary has an initial hospitalization for one of the procedures or clinical conditions included in these proposed EPMs would be held accountable for spending during the episode of care. EPM participants could earn reconciliation payments by appropriately reducing expenditures and meeting certain quality metrics. EPM participants also would gain access to data and educational resources to better understand care patterns during the inpatient hospitalization and post-acute periods, as well as associated spending. Payment approaches that reward providers for assuming financial and performance accountability for a particular episode of care create incentives for the implementation and coordination of care redesign between participants and other providers and suppliers such as physicians and post-acute care providers.

The proposal for the AMI, CABG, and SHFFT models would require the participation of hospitals in multiple geographic areas that might not otherwise participate in testing episode payment for the proposed episodes of care. CMS is testing other episode payment models with the BPCI initiative and the CJR model. The BPCI initiative is voluntary; providers applied to participate and chose from 48 clinical episodes. BPCI participants entered the at-risk phase between 2013 and 2015 and have the option to continue participating in the initiative through FY 2018. In the CJR model, acute care hospitals in selected geographic areas are required to participate in the CJR model for all eligible lower-extremity joint replacement (LEJR) episodes that initiate at a CJR participant hospital. The CJR model began its first of 5 performance years on April 1, 2016. Realizing the full potential of new EPMs will require the engagement of an even broader set of providers than have participated to date in our episode payment models such as the BPCI initiative and the CJR model. As such, we are interested in testing and evaluating the impact of episode payment for the three proposed EPMs in a variety of circumstances, including those hospitals that may not otherwise participate in such a test.

While we note that testing of the CJR model that began in April 2016 will allow CMS to gain experience with requiring hospitals to participate in an episode payment model, the clinical circumstances of the episodes we are proposing (AMI, CABG, and SHFFT) differ in important ways from the LEJR episodes included in the CJR model. LEJR procedures are common among the Medicare population, and the majority of such procedures are elective. In contrast, under the three proposed EPMs, CMS would test episode payment for certain cardiac conditions and procedures, as well as SHFFT. We expect the patient population included in these episodes would be substantially different from the patient population in CJR episodes, due to the clinical nature of the cardiac and SHFFT episodes. Beneficiaries in these episodes commonly have chronic conditions that contribute to the initiation of the episodes, and need both planned and unplanned care throughout the EPM episode following discharge from the initial hospitalization that begins the episode. Both AMI and CABG model episodes primarily include beneficiaries with cardiovascular disease, a chronic condition which likely contributed to the acute events or procedures that initiate the episodes. About half the average AMI model historical episode spending was for the initial hospitalization, with the majority of spending following discharge from the initial hospitalization due to hospital readmissions, while there was relatively less spending on SNF services, Part B professional services, and hospital outpatient services. In CABG model historical episodes, about three-quarters of episode spending was for the initial hospitalization, with the remaining episode spending relatively evenly divided between Part B professional services and hospital readmissions, and a lesser percentage on SNF services. Similar to AMI episodes, post-acute care provider use was relatively uncommon in CABG model historical episodes, while hospital readmissions during CABG model historical episodes were relatively common. SHFFT model historical episodes also were accompanied by substantial spending Start Printed Page 50800for hospital readmissions, and post-acute care provider use in these episodes also was high. The number of affected beneficiaries and potential impact of the models on quality and Medicare spending present an important opportunity to further the Administration's goal of shifting health care payments to support the quality of care over the quantity of services by promoting better coordination among health care providers and suppliers and greater efficiency in the care of beneficiaries in these models, while reducing Medicare expenditures.[3] Pay-for-performance episode payment models such as the three EPMs proposed in this rulemaking financially incentivize improved quality of care and reduced cost by aligning the financial incentives of all providers and suppliers caring for model beneficiaries with these goals. This alignment leads to a heightened focus on care coordination and management throughout the episode that prioritizes the provision of those items and services which improve beneficiary outcomes and experience at the lowest cost. A more detailed discussion of the evidence supporting the episode selection for these models can be found in section III.A.1. of the proposed rule.

The proposed models would also allow CMS to gain additional experience with episode-payment approaches for hospitals with variance in (1) historic care and utilization patterns; (2) patient populations and care patterns; (3) roles within their local markets; (4) volumes of services; (5) levels of access to financial, community, or other resources; and (6) levels of population and health-care-provider density, including local variations in the availability and use of different categories of post-acute care providers. We believe that participation in the proposed EPMs by a large number of hospitals with diverse characteristics would result in a robust data set for evaluating this payment approach and would stimulate the rapid development of new evidence-based knowledge. Testing the proposed EPMs in this manner would also allow us to learn more about patterns of inefficient utilization of health care services and how to incentivize quality improvement for beneficiaries receiving services in AMI, CABG, and SHFFT episodes. This knowledge potentially could inform future Medicare payment policies.

We propose the CR incentive payment model to test the effects on quality of care and Medicare expenditures of providing financial incentives to hospitals for beneficiaries hospitalized for treatment of AMI or CABG to encourage care coordination and greater utilization of medically necessary CR and intensive cardiac rehabilitation (ICR) services for 90 days post-hospital discharge where the beneficiary's overall care is paid under either an EPM or the Medicare FFS program. Despite the evidence from multiple studies that CR services improve health outcomes, the literature also indicates that these services are underutilized, estimating that only about 35 percent of AMI patients older than 50 receive this indicated treatment.[4 5 6] Recent analysis confirms a similar pattern of underutilization for Medicare beneficiaries who are eligible for and could benefit from CR.

Considering the evidence demonstrating that CR/ICR services improve long-term patient outcomes, the room for improvement in CR/ICR service utilization for beneficiaries eligible for this benefit, and the need for ongoing, chronic treatment for underlying coronary artery disease (CAD) among beneficiaries that have had an AMI or a CABG, we believe that there is a need for improved long-term care management and care coordination for beneficiaries that have had an AMI or a CABG and that incentivizing the use of CR/ICR services is an important component of meeting this need. We want to reduce barriers to high-value care by testing a financial incentive for hospitals that encourages the management of beneficiaries that have had an AMI or a CABG in ways that may contribute to long-term improvements in quality and reductions in Medicare spending. We seek public comment on the proposals contained in this proposed rule, and also on any alternatives considered.

B. Summary of the Major Provisions

1. Model Overview—EPM Episodes of Care

Under the proposed EPMs, as described further in section III.B.2. of this proposed rule, an AMI, CABG, or SHFFT model episode would begin with an inpatient admission to an anchor hospital assigned to one of the following MS-DRGs upon beneficiary discharge. Acute care hospital services furnished to beneficiaries in AMI, CABG, and SHFFT episodes currently are paid under the Inpatient Prospective Payment System (IPPS) through several Medicare Severity-Diagnosis Related Groups (MS-DRGs): for AMI episodes, AMI MS-DRGs (280-282) and those Percutaneous Coronary Intervention (PCI) MS-DRGs (246-251) representing IPPS admissions for AMI that are treated with PCIs; CABG MS-DRGs (231-236); and SHFFT MS-DRGs (480-482). Episodes would end 90 days after the date of discharge from the anchor hospital, as defined under § 512.2. Defining EPMs' episodes of care in such a manner offers operational simplicity for both providers and CMS. The proposed EPMs' episodes would include the inpatient stays and all related care covered under Medicare Parts A and B within the 90 days after discharge, including hospital care, post-acute care, and physician services.

2. Model Scope

Consistent with the CJR model, we propose that acute care hospitals would be the episode initiators and bear financial risk under the proposed AMI, CABG and SHFFT models. In comparison to other health care facilities, hospitals are more likely to have resources that would allow them to appropriately coordinate and manage care throughout an episode, and hospital staff members already are involved in hospital-discharge planning and post-acute care recommendations for recovery, key dimensions of high-quality and efficient care. We propose to require all hospitals that are paid under the IPPS, have a CMS Certification Number (CCN), and have an address located in selected geographic areas to participate in the EPMs, with limited exceptions. An eligible beneficiary who receives care at such a hospital will automatically be included in the applicable EPM. We propose to select geographic areas through a random sampling methodology.

Under the CR incentive payment model, we propose to provide a CR incentive payment specifically to selected hospitals with financial responsibility for AMI or CABG model episodes (hereinafter EPM-CR participants) because they are already engaged in managing the AMI or CABG model beneficiary's overall care for a period of time following hospital discharge. Similarly, we believe there are opportunities to test the same financial incentives for hospitals where the beneficiary's overall care is paid under the Medicare FFS program. Thus, Start Printed Page 50801we also propose to provide a CR incentive payment specifically to selected hospitals that are not AMI or CABG model participants (hereinafter FFS-CR participants).

Our proposed geographic-area selection process is detailed further in section III.B.4. of this proposed rule.

3. Payment

We propose to test the AMI, CABG, and SHFFT EPMs for 5 performance years. The first performance year will begin July 1, 2017. During these performance years we propose to continue paying hospitals and other providers and suppliers according to the usual Medicare FFS payment systems. However, after the completion of a performance year, the Medicare claims payments for services furnished to the beneficiary during the episode, based on claims data, would be combined to calculate an actual episode payment. The actual episode payment would then be reconciled against an established EPM quality-adjusted target price. The amount of this calculation, if positive, would be paid to the participant. This would be called a reconciliation payment. If negative, we would require repayment from the participant hospital beginning with episodes ending in the second quarter of performance year 2 of the EPMs. EPM participants' quality performance also would be assessed at reconciliation; each participant would receive a composite quality score and a corresponding quality category. EPM participants that achieve a quality category of “acceptable” or higher would be eligible for a reconciliation payment. We also propose to phase in the requirement that participants whose actual episode payments exceed the quality-adjusted target price pay the difference back to Medicare beginning for performance year 2. Under this proposal, Medicare would not require repayment from hospitals for performance year 1 for actual episode payments that exceed their target price in performance year 1, and an applicable discount factor would be used for calculating repayment amounts for performance years 2 and 3, consistent with our final policies for the CJR model. In contrast to the CJR model, due to the clinical characteristics and common patterns of care in AMI episodes, we propose payment adjustments in the cases of certain transfers and readmissions of beneficiaries to inpatient hospitals for these episodes. These payment adjustments are discussed in detail in section III.D.4.b.(1). of this proposed rule. We also propose to limit how much a hospital can gain or lose based on its actual episode payments relative to quality-adjusted target prices. Finally, we propose additional policies to further limit the risk of high payment cases for all participants and for special categories of participants as described in section III.D. of this proposed rule.

In addition to the EPMs, we propose to test a CR incentive payment model to encourage the utilization of CR/ICR services for beneficiaries hospitalized for treatment of AMI or CABG. To determine the CR incentive payment, we propose to count the number of CR/ICR services for the relevant time periods under the Outpatient Prospective Payment System (OPPS) and PFS on the basis of the presence of paid claims of the HCPCS codes that report CR/ICR services and the units of service billed. The initial level of the per-service CR incentive amount would be $25 per CR/ICR service for each of the first 11 CR/ICR services paid for by Medicare during an AMI or CABG model episode or AMI or CABG care period. After 11 CR/ICR services are paid for by Medicare for a beneficiary, the level of the per-service CR incentive amount would increase to $175 per CR/ICR service for each additional CR/ICR service paid for by Medicare during the AMI or CABG model episode or AMI care period or CABG care period. A more detailed discussion of the CR incentive payment is located in section VI.E.1 of this proposed rule. The CR performance years would be the same as the performance years proposed for the EPMs in section III.D.2.a. of this proposed rule. Further details about the payment structure and design of the CR incentive payment model can be found in section VI. of this proposed rule.

4. Similar, Previous, and Concurrent Models

The proposed EPMs are informed by other models and demonstrations currently and previously conducted by CMS, and would explore additional ways to use episode payment to enhance coordination of care and improve the quality of care.

We recently announced practices that will participate in the Oncology Care Model (OCM), an episode payment model for physician practices administering chemotherapy. Under OCM, practices will enter into payment arrangements that include both financial and performance accountability for episodes of care surrounding chemotherapy administration to cancer patients. We will coordinate with other payers to align with OCM in order to facilitate enhanced services and care at participating practices.[7]

CMMI previously tested innovative episode payment approaches in the Medicare Acute Care Episode (ACE) demonstration,[8] and, as described in this proposed rule, currently is testing additional approaches under the BPCI initiative and the CJR model. The ACE demonstration tested a bundled payment approach for cardiac and orthopedic inpatient surgical services and procedures. All Medicare Part A and Part B services pertaining to the inpatient stay were included in the ACE demonstration episodes of care. Evaluations of the ACE demonstration found that while there was not strong quantitative evidence indicating improvements in quality, there was qualitative evidence that hospitals worked to improve processes and outcomes as a result of their participation in the demonstration.

We currently are testing the BPCI initiative, which is composed of four related payment models that link payments for multiple services that a Medicare beneficiary receives during an episode of care into a bundled payment. Under the initiative, entities enter into payment arrangements with CMS that include financial and performance accountability for episodes of care. Episodes of care under the BPCI initiative begin with either: (1) An inpatient hospital stay or (2) post-acute care services following a qualifying inpatient hospital stay. The BPCI initiative is evaluating the effects of episode-based payment approaches on patient experience of care, outcomes, and cost of care for Medicare FFS beneficiaries. Participating organizations chose from 48 clinical episodes, including hip and femur procedures except major joint, acute myocardial infarction, percutaneous coronary intervention, and coronary artery bypass graft surgery. BPCI Model 2 is an episode payment model in which a qualifying acute care hospitalization initiates a 30-, 60-, or 90-day episode of care. The episode includes the inpatient stay in an acute care hospital and all related services covered under Medicare Parts A and B during the episode, including post-acute care services.[9] Our experience testing BPCI Model 2 informed the design of the three Start Printed Page 50802proposed EPMs. Although some interim evaluation results from the BPCI models are available, final evaluation results for the models within the BPCI initiative are not yet available. However, we believe that CMS' experiences with BPCI support the design of the proposed EPMs. Stakeholders both directly and indirectly involved in testing BPCI models have conveyed that they perceive the initiative to be an effective mechanism for advancing better, more accountable care and aligning providers along the care continuum. This message has been reinforced through CMS site visits to participating entities, the Bundled Payments summit in Washington, in-person meetings with Awardees at CMS, and Awardee-led Affinity Group discussions. The BPCI initiative incorporates 48 clinical episodes, including cardiac and orthopedic episodes similar to those proposed for the AMI, CABG, and SHFFT models. These clinical episodes are being tested by over 1200 Medicare providers, including acute care hospitals, physician group practices, skilled nursing facilities, and home health agencies. Cardiac and orthopedic clinical episodes are among the most popular episodes in BPCI, indicating that BPCI awardees participating in BPCI believe they can reduce cost and improve quality for beneficiaries in these episodes of care.

Our design and implementation of the CJR model, which is an episode payment model for LEJR episodes, also informed the design of the proposed AMI, CABG, and SHFFT EPMs. After releasing a proposed rule in July 2015 and receiving nearly 400 comments from the public, in November 2015 we released final regulations implementing the CJR model. Approximately 800 acute care hospitals (approximately 23 percent of all IPPS hospitals) now participate in the CJR model. The first CJR performance year began on April 1, 2016. The CJR model will continue for 5 performance years, ending on December 31, 2020. The proposed AMI, CABG, and SHFFT models build upon our experience designing and implementing the CJR model, including feedback from providers and other public stakeholders during the CJR model's rulemaking and implementation processes.

Further information of why specific elements of the models and initiatives were incorporated into the EPMs' designs is discussed later in this proposed rule.

5. Overlap With Ongoing CMS Efforts

We propose to exclude from participation in the AMI, CABG, and SHFFT models certain acute care hospitals participating in BPCI Models 2 and 4 for the hip and femur procedures except major joint or for all three of the BPCI cardiac episodes (AMI, PCI, and CABG). We propose to exclude beneficiaries in the proposed EPMs' episodes from being included in certain Innovation Center ACO models, the Next Generation ACO Model and Comprehensive ESRD Care. Other CMS programs, such as the Medicare Shared Savings Program and other accountable care organization (ACO) or total cost of care initiatives will remain eligible for EPM episode initiation. We propose to account for overlap, that is, where EPM beneficiaries also are included in other models and programs to ensure the financial policies of the models are maintained and results and spending reductions are attributed to one model or program. More detail on our proposed policies for accounting for provider- and beneficiary-level overlap is discussed in section III.D.6. of this proposed rule.

The amendments made by the Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) (Pub. L. 114-10, April 16, 2015) created two paths for eligible clinicians to link quality to payments: The Merit-Based Incentive Payment System (MIPS) and Advanced Alternative Payment Models (APMs). These two paths create a flexible payment system called the Quality Payment Program as proposed by CMS in the Quality Payment Program proposed rule (81 FR 28161 through 28586). The MIPS streamlines and improves on three current programs—the Physician Quality Reporting System (PQRS), the Physician Value-based Payment Modifier (VM), and the Medicare Electronic Health Record (EHR) Incentive Program—and continues the focus on quality and value in one cohesive program. Through participation in Advanced APMs, eligible clinicians can become Qualifying APM Participants (QPs) for a year beginning with CY 2019 and receive an APM Incentive Payment (or, in later years, a more favorable payment update under the PFS) for the year.

So that the EPMs may be able to meet the criteria to be Advanced APMs based on the requirements proposed in the Quality Payment Program proposed rule, we propose to require EPM participants to use Certified Electronic Health Record Technology (CEHRT) (as defined in section 1848(o)(4) of the Act) in Track 1 of each EPM. We propose that EPM participants in these tracks must use certified health information technology (IT) functions, in accordance with the definition of CEHRT under our regulation at 42 CFR 414.1305, to document and communicate clinical care with patients and other health care professionals as described in the Quality Payment Program proposed rule (81 FR 28161 and 28299). We also make similar proposals with respect to CJR.

We propose to implement two different tracks within the EPMs whereby EPM participants that meet proposed requirements for use of CEHRT and financial risk would be in Track 1 (an Advanced APM track) and EPM participants that do not meet these requirements would be in Track 2 (a non-Advanced APM track). The different tracks would not change how EPM participants operate within the EPM itself, beyond the requirements associated with selecting to meet CEHRT use requirements. The only distinction between the two tracks is that only Track 1 EPMs could be considered an Advanced APM for purposes of the Quality Payment Program based on the proposed criteria in the Quality Payment Program proposed rule. We make similar proposals with respect to CJR. We would consider modifying requirements proposed in this rule as necessary to reconcile them with policies adopted in the Quality Payment Program final rule. A more detailed discussion of the proposals for how EPMs and CJR could qualify as Advanced APMs, and how eligible clinicians participating in the EPMs and CJR would be identified and affected, can be found in sections III.A.2 and V.O. of this proposed rule.

6. Quality Measures and Reporting Requirements

Similar to the quality measures selected for the CJR model, we propose to use established measures used in other CMS quality-reporting programs for the proposed EPMs' episodes. We propose to use these measures to test EPMs' success in achieving its goals under section 1115A of the Act and to monitor for beneficiary safety. For the SHFFT model, we propose applying the same quality measures selected for the CJR model.

The following proposed quality measures for SHFFT episodes are:

  • THA/TKA Complications: Hospital-Level Risk-Standardized Complication Rate (RSCR) Following Elective Primary Total Hip Arthroplasty (THA) and/or Total Knee Arthroplasty (TKA) (National Quality Forum [NQF] #1550)
  • Hospital Consumer Assessment of Healthcare Providers and Systems (HCAPHS) Survey (NQF #0166)
  • Successful Voluntary Reporting of Patient-Reported Outcomes
Start Printed Page 50803

We propose the following measures for the AMI model:

  • MORT-30-AMI: Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) Following Acute Myocardial Infarction (AMI) Hospitalization (NQF #0230).
  • AMI Excess Days: Excess Days in Acute Care after Hospitalization for Acute Myocardial Infarction (acute care days include emergency department, observation, and inpatient readmission days)
  • HCAPHS Survey (NQF #0166), linear mean roll-up (HLMR) scores like CJR

We propose the following measures for the CABG model:

  • MORT-30-CABG: Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) Following Coronary Artery Bypass Graft Surgery (NQF #2558)
  • HCAPHS Survey (NQF #0166), HLMR scores like CJR

Finally, we are proposing and requesting public feedback on options for including successful implementation testing of the Hybrid AMI measure as a quality measure for the AMI episode. The Hybrid AMI measure will assess a hospital's 30-day risk-standardized acute myocardial infarction mortality rate and will incorporate a combination of claims data and EHR data submitted by hospitals.

Additionally, similar to the CJR model, we propose to adopt a pay-for-performance methodology for EPMs that relies upon a composite quality score to assign respective EPM participants to four quality categories. These quality categories will determine an EPM participant's eligibility for a reconciliation payment should such EPM participant achieve spending below the quality-adjusted target price, as well as the effective discount percentage at reconciliation. Points for quality performance and improvement (as applicable) will be awarded for each episode measure and then summed to develop a composite quality score that will determine the EPM participant's quality category for the episode. Quality performance will make up the majority of available points in the composite quality score, with improvement points available as “bonus” points for the measure. This approach resembles the CJR model methodology.

7. Beneficiary Protections

As with the CJR model, Medicare beneficiaries in the proposed models will retain the right to obtain health services from any individual or organization qualified to participate in the Medicare program. Eligible beneficiaries who receive services from model participants would not have the option to opt out of inclusion in the applicable model. We propose to require participants to supply beneficiaries with written information regarding the design and implications of these models as well as the beneficiaries' rights under Medicare, including their right to use their providers of choice. We would make a robust effort to reach out to beneficiaries and their advocates to help them understand the models. We also propose to use our existing authority, if necessary, to audit participant hospitals if claims analysis indicates an inappropriate change in delivered services. Beneficiary protections are discussed in greater depth in section III.G. of this proposed rule.

8. Financial Arrangements

We propose to use the same general framework finalized in the CJR model to hold participants financially responsible for AMI, CABG and SHFFT model episodes as discussed in section III.I. of this proposed rule. Specifically, only the EPM participants would be directly subject to the requirements of this proposed rule for the proposed EPMs. EPM participants would be responsible for ensuring that other providers and suppliers collaborating with the EPM participants on care redesign for the applicable EPM episodes are in compliance with the applicable EPM's terms and conditions.

We propose adding hospitals to the list of providers and suppliers eligible for gainsharing as EPM collaborators due to the expected participation of multiple hospitals in the episode care for some beneficiaries in AMI and CABG episodes. We further propose adding ACOs to be eligible for gainsharing as EPM collaborators due to the interest of ACOs in gainsharing during the CJR model rulemaking and the ongoing challenges of addressing overlap between episode payment models and ACOs. We also propose provisions that allow for certain gainsharing within ACOs, detailed further in section III.I. of this proposed rule.

In contrast, the CR incentive payment model is specifically tied to increased utilization of CR/ICR services within AMI and CABG model episodes and, therefore, is designed to reward increased referral of AMI and CABG model beneficiaries to CR/ICR programs, as well as supporting beneficiary adherence to the referral and participation in CR/ICR services, rather than the quality and efficiency of EPM episodes themselves. Thus, we do not propose to allow CR incentive payments to be included in sharing arrangements, and the CR incentive payments may be shared with other individual and entities only under circumstances which comply with all existing laws and regulations, including fraud and abuse laws. Financial arrangements are discussed in further detail in section VI.E. of the proposed rule.

9. Data Sharing

Based on our experience with various Medicare programs and models, including the BPCI initiative, the CJR model, the Shared Savings Program, and the Pioneer ACO model, we believe that providing certain beneficiary claims data to model participants will be essential to their success. We propose to share data with participants upon request throughout the performance period of the models to the extent permitted by the Health Insurance Portability and Accountability Act of 1996 (HIPAA) Privacy Rule and other applicable law. We propose to share upon request both raw claims-level data and claims summary data with participants. This approach would allow participants without prior experience analyzing claims to use summary data for analysis of care and spending patterns, while allowing those participants who prefer raw claims-level data the opportunity to analyze claims. We propose to provide participants with up to 3 years of retrospective claims data upon request that will be used to develop their quality-adjusted target price. In accordance with the HIPAA Privacy Rule, we would limit the content of this data to the minimum data necessary for the participant to conduct quality assessment and improvement activities and effectively coordinate care.

10. Program Waivers

Section 1115A of the Act authorizes the Secretary to waive Medicare program requirements as necessary to implement provisions for testing models. Under the CJR model, CMS waived certain program rules regarding the direct supervision requirement for certain post-discharge home visits, telehealth services, and the skilled nursing facility (SNF) 3-day rule. CMS finalized these waivers to offer providers and suppliers more flexibility so that they may increase coordination of care and management of beneficiaries in model episodes. Adopting the CJR waivers for the proposed EPMs required further examination to determine if such adoption would increase financial vulnerability to the Medicare program or would create inappropriate incentives to reduce the quality of Start Printed Page 50804beneficiary care. As discussed in section III.J. of this proposed rule, we propose to do the following:

  • Adopt waivers of the telehealth originating site and geographic site requirement and to allow in-home telehealth visits for all three proposed EPMs, as well as the general waiver to allow post-discharge nursing visits in the home;
  • Provide model-specific limits to the number of post-discharge nursing visits and make model-specific decisions about offering the SNF 3-day stay waiver; and
  • Adopt a waiver for furnishing cardiac and intensive cardiac rehabilitation services to allow a Nurse Practitioner, Clinical Nurse Specialist, or Physician Assistant, in addition to a physician, to perform specific physician functions.

C. Summary of Economic Effects

As shown in our impact analysis, we expect the EPMs to result in savings to Medicare of $170 million over the 5 performance years of the model. We note that a composite quality score will be calculated for each hospital in order to determine eligibility for a reconciliation payment and whether the hospital qualifies for quality incentive payments that will reduce the effective discount percentage experience by the hospital at reconciliation for a given performance year.

More specifically, in performance year 1 of the model, we estimate a Medicare cost of approximately $12 million, as hospitals will not be subject to downside risk in the first year and the first quarter of the second performance year of the model. As we introduce downside risk beginning in the second quarter of performance year 2 of the model, we estimate Medicare savings of approximately $13 million. In performance year 3 of the model, we estimate Medicare savings of $30 million. In performance years 4 and 5 of the model, we will move from target episode pricing that is based on a hospital's experience to target pricing based on regional experience, and we estimate Medicare savings of $61 million and $79 million, respectively.

As a result, we estimate the net savings to Medicare to be $170 million over the 5 performance years of the model. We anticipate there will be a broader focus on care coordination and quality improvement for EPMs among hospitals and other providers and suppliers within the Medicare program that will lead to both increased efficiency in the provision of care and improved quality of the care provided to beneficiaries.

Additionally, the CR incentive model estimates that the impact on the Medicare program may range from up to $27 million of additional spending to $32 million of savings between 2017 and 2024, depending on the change in utilization of CR/ICR services based on the proposed incentive structure.

Finally, the change in the estimated net financial impact to the Medicare program from the CJR model modifications in this proposed rule is $22 million in spending, and the updated assumptions regarding the number of hospitals that will report quality data result in an increase of $14 million dollars in spending. The total estimated net financial impact to the Medicare program from both the modifications in the proposed rule and revised assumptions are $35 million in spending.

We note that under section 1115A(b)(3)(B) of the Act, the Secretary is required to terminate or modify a model unless certain findings can be made with respect to savings and quality after the model has begun. If during the course of testing the model it is determined that termination or modification is necessary, such actions will be undertaken through rulemaking.

II. Background

This proposed rule proposes the implementation of three new EPMs and a CR incentive payment model under the authority of section 1115A of the Act. Under the AMI, CABG, and SHFFT EPMs, acute care hospitals in certain selected geographic areas will be financially accountable for quality performance and spending for applicable episodes of care. We propose to retrospectively apply through a reconciliation process the episode payment methodology; hospitals and other providers and suppliers would continue to submit claims and receive payment via the usual Medicare FFS payment systems throughout the proposed EPMs' performance years. Hospitals participating in the proposed EPMs would receive target prices, which reflect expected spending for care during an episode as well as a discount to reflect savings to Medicare, on a prospective basis, prior to the beginning of a performance year. All related care covered under Medicare Parts A and B and furnished within 90 days after the date of hospital discharge from the anchor hospitalization which initiated the applicable EPM episode would be included in the episode of care. We believe the proposed models will further our goals of improving the efficiency and quality of care for Medicare beneficiaries for these medical conditions and procedures.

III. Provisions of the Proposed Regulations

A. Selection of Episodes, Advanced Alternative Payment Model Considerations, and Future Directions

1. Selection of Episodes for Episode Payment Models in This Rulemaking

a. Overview

CMS has been engaged since 2013 in testing various approaches to episode payment for Medicare FFS beneficiaries for 48 clinical episodes in the BPCI initiative. As of April 1, 2016, the BPCI initiative has 1,522 participants in the risk-bearing phase, comprised of 321 Awardees and 1,201 Episode Initiators. The breakdown of BPCI participants by provider type is as follows: Acute care hospitals (385); skilled nursing facilities (681); physician group practices (283); home health agencies (99); inpatient rehabilitation facilities (9); and long-term care hospitals (1).[10] In BPCI Models 2 and 3, there is participation across all 48 clinical episodes, and in Model 4 there is participation in 19 clinical episodes. The 10 clinical episodes with the most participation are: major joint replacement of the lower extremity; simple pneumonia and respiratory infections; congestive heart failure; chronic obstructive pulmonary disease; bronchitis; asthma; hip and femur procedures except major joint; sepsis; urinary tract infection; acute myocardial infarction (medical management only); medical non-infectious orthopedic; and other respiratory.[11]

In November 2015, CMS released the Final Rule for the Comprehensive Care for Joint Replacement (CJR) model (80 FR 73274 through 73554), the first test of episode payment for Medicare FFS beneficiaries in which providers are required to participate. The CJR model, which began on April 1, 2016, focuses on the episode-of-care for lower-extremity joint replacement (LEJR) procedures. As discussed in the Final Rule (80 FR 73277), LEJR episodes were chosen for the CJR model because they represent one of the most common high-expenditure, high-utilization procedures furnished to Medicare beneficiaries and have significant variation in episode spending. We believe this high volume, coupled with substantial variation in utilization and spending across individual providers and geographic Start Printed Page 50805regions, created a significant opportunity to test whether an episode payment model focused on a defined set of procedures could improve the quality and coordination of care, as well as result in savings to Medicare. Notably, both BPCI and the CJR model are focused on care that is related to an inpatient hospitalization, with CJR and BPCI Model 2 episodes beginning with an inpatient hospitalization (anchor hospitalization) and extending up to 90 days post-hospital discharge.

In this rulemaking, we propose three new EPMs that, like the CJR model, would require provider participation in selected geographic areas. Episodes in the new EPMs would begin with admissions for hospitalizations in IPPS hospitals, and would extend 90 days post-hospital discharge. The episodes included in these three EPMs are AMI, CABG, and SHFFT excluding lower extremity joint replacement. The proposed AMI model includes beneficiaries discharged under AMI MS-DRGs (280-282), representing IPPS admissions for AMI that are treated with medical management. The proposed AMI model also includes beneficiaries discharged under PCI MS-DRGs (246-251) with AMI International Classification of Disease, Tenth Edition, Clinical Modification (ICD-10-CM) diagnosis codes for initial AMI diagnoses in the principal or secondary diagnosis code positions, representing IPPS admissions for AMI that are treated with PCIs. The proposed CABG model includes beneficiaries discharged under CABG MS-DRGs (231-236), representing IPPS admissions for this coronary revascularization procedure irrespective of AMI diagnosis. The proposed SHFFT model includes beneficiaries discharged under hip and femur procedures except major joint replacement MS-DRGs (480-482), representing IPPS admissions for hip-fixation procedures in the setting of hip fractures.

Similar to the selection of LEJR episodes for the CJR model (80 FR 73277), we selected the AMI, CABG, and SHFFT episodes because they represent high-expenditure, high-volume episodes-of-care experienced by Medicare beneficiaries. Based on analysis of historical episodes beginning in CY 2012-2014, the average annual number of historical episodes that began with IPPS hospitalizations and extended 90 days post-hospital discharge, and therefore would have been included in the proposed models, is approximately 168,000 for AMI; 48,000 for CABG; and 109,000 for SHFFT.[12] The total annual Medicare spending for these historical episodes was approximately $4.1 billion, $2.3 billion, and $4.7 billion, respectively.[13] Each of the episodes provides different opportunities in an EPM to improve the coordination and quality of care, as well as efficiency of care during the episode, based on varying current patterns of utilization and Medicare spending.

However, in contrast to LEJR episodes in CJR, which are predominantly elective and during which hospital readmissions are rare and substantial post-acute care provider utilization is common, the proposed AMI, CABG, and SHFFT model episodes have very different current patterns of care. Beneficiaries in these episodes commonly have chronic conditions that contribute to the initiation of the episodes and need both planned and unplanned care throughout the EPM episode following discharge from the initial hospitalization that begins the episode. Both AMI and CABG model episodes primarily include beneficiaries with cardiovascular disease, a chronic condition which likely contributed to the acute events or procedures that initiate the episodes. About half the average AMI model historical episode spending was for the initial hospitalization, with the majority of spending following discharge from the initial hospitalization due to hospital readmissions, while there was relatively less spending on SNF services, Part B professional services, and hospital outpatient services. In CABG model historical episodes, about three-quarters of episode spending was for the initial hospitalization, with the remaining episode spending relatively evenly divided between Part B professional services and hospital readmissions, and a lesser percentage on SNF services. Similar to AMI episodes, post-acute care provider use was relatively uncommon in CABG model historical episodes, while hospital readmissions during CABG model historical episodes were relatively common. SHFFT model historical episodes also were accompanied by substantial spending for hospital readmissions, and post-acute care provider use in these episodes also was high.[14] The number of affected beneficiaries and potential impact of the models on quality and Medicare spending present an important opportunity to further the Administration's goal of shifting health care payments to support the quality of care over the quantity of services by promoting better coordination among health care providers and suppliers and greater efficiency in the care of beneficiaries in these models, while reducing Medicare expenditures.[15] Pay-for-performance episode payment models such as the three EPMs proposed in this rulemaking financially incentivize improved quality of care and reduced cost by aligning the financial incentives of all providers and suppliers caring for model beneficiaries with these goals. This alignment leads to a heightened focus on care coordination and management throughout the episode that prioritizes the provision of those items and services which improve beneficiary outcomes and experience at the lowest cost.

We selected all of the proposed EPM episodes based on their clinical homogeneity, site-of-service, and MS-DRG assignment considerations. We anticipate these proposed new EPMs, like the CJR model, would benefit Medicare beneficiaries by improving the coordination and transition of care among various care settings to facilitate beneficiaries' return to their communities as their recoveries progress, improving the coordination of items and services paid through Medicare FFS, encouraging more provider investment in infrastructure and redesigned care processes for higher quality and more efficient service delivery, and incentivizing higher value care across the inpatient and post-acute care spectrum spanning the episode-of-care (80 FR 73276). However, improving value in the EPMs through these means requires a cohort of beneficiaries with similar clinical features such that coordination and care redesign efforts can be targeted. Therefore, we propose EPM episodes built on common pathologic and treatment processes; that is, beneficiaries included in both the AMI and CABG models have cardiovascular pathologies that drive their clinical courses during the Start Printed Page 50806episodes, and SHFFT model beneficiaries all share similar diagnoses of hip fracture and treatment with hip fixation that drive their clinical courses during their respective episodes.

b. SHFFT Model

The SHFFT model was selected to complement the CJR model. The SHFFT model is being tested in the same hospitals participating in the CJR model as discussed in section III.B.4 of this proposed rule, so that all surgical treatment options for Medicare beneficiaries with hip fracture (hip arthroplasty and fixation) would be included in episode payment models. Hip fracture is a serious and sometimes catastrophic event for Medicare beneficiaries. In 2010, 258,000 people aged 65 and older were admitted to the hospital for hip fracture, with an estimated $20 billion in lifetime cost for all hip fractures in the United States in a single year.[16] In 2013, fracture of the neck of the femur (the most common location for hip fracture) was the eighth most common principal discharge diagnosis for hospitalized Medicare FFS beneficiaries, constituting 2.7 percent of discharges.[17] Mortality associated with hip fracture is 5-10 percent after 1 month and approximately 33 percent at 1 year.[18] Hip arthroplasty and hip fixation, or “hip pinning,” represent the two broad surgical options for treating hip fractures.[19] The CJR model episodes begin with admission to acute care hospitals for LEJR procedures assigned to MS-DRG 469 (Major joint replacement or reattachment of lower extremity with major complications or comorbidities) or MS-DRG 470 (Major joint replacement or reattachment of lower extremity without major complications or comorbidities) upon beneficiary discharge and paid under the IPPS, including total and partial hip replacement in the setting of hip fracture (80 FR 73280). Therefore, the SHFFT model, which would additionally test an episode payment for hip fixation, provides an opportunity to complete the transition to episode payment for the surgical treatment and recovery of the significant clinical condition of hip fracture.

c. AMI and CABG Models

The AMI and CABG models, which we propose to be tested at a single set of hospitals as discussed in section III.B.5 of this proposed rule, were selected to include all beneficiaries who have an AMI treated medically or with revascularization with PCI, as well as all beneficiaries who undergo CABG (whether performed during the care of an AMI or performed electively for stable ischemic heart disease or other indication). Both cardiac models represent clinical conditions that result in a significant burden of morbidity and expenditures in the Medicare population. CABG typically is the preferred revascularization modality for patients with ST elevation AMI where the coronary anatomy is not amenable to PCI or there is a mechanical complication (for example, ventricular septal defect, rupture of the free wall of the ventricle, or papillary-muscle rupture with severe mitral regurgitation); for patients with CAD other than ST elevation AMI where there is left main coronary artery disease or multi-vessel disease with complex lesions; and for patients with clinically significant CAD in at least one vessel and refractory symptoms despite medical therapy and PCI.[20] Despite the greater acute morbidity related to major cardiothoracic surgery, CABG is associated with lower longer-term rates of major adverse cardiac and cerebrovascular events in comparison to PCI for certain groups of patients.[21] Moreover, a recent study found that in a group of patients with ischemic cardiomyopathy, the rates of death from any cause, death from cardiovascular causes, and death from any cause or hospitalization for cardiovascular causes were significantly lower over 10 years among patients who underwent CABG in addition to receiving medical therapy than among those who received medical therapy alone.[22] While about 30 percent of CABGs are performed during the care of AMIs, we propose to include these particular AMI beneficiaries generally in the same episode as CABG for other indications, rather than in the AMI episode, since we anticipate hospitals will seek to improve the quality and efficiency of care for that surgical intervention, regardless of indication.[23]

We propose AMI as the episode for an EPM because we recognize it as a significant clinical condition for which evidence-based clinical guidelines are available for the most common AMI scenarios that begin with a beneficiary's presentation for urgent care, most commonly to a hospital emergency department. The hospital phase involves medical management for all patients, as well as potential revascularization, most commonly with PCI. Secondary prevention and plans for long-term management begin early during the hospitalization and extend following hospital discharge and are addressed in clinical guidelines.[24 25] The AMI model is the first Innovation Center episode payment model that includes substantially different clinical care pathways (medical management and PCI) for a single clinical condition in one episode in a model and, as such, represents an important next step in testing episode payment models for clinical conditions which involve a variety of different approaches to treatment and management.

The American Heart Association estimates that every 42 seconds, someone in the United States has a myocardial infarction.[26] AMI remains Start Printed Page 50807one of the most common hospital diagnoses among Medicare FFS beneficiaries, and almost 20 percent of beneficiaries discharged for AMI are readmitted within 30 days of hospital discharge.[27 28] In 2013, AMI was the sixth most common principal discharge diagnosis for hospitalized Medicare FFS beneficiaries, constituting 2.9 percent of discharges.[29] Of the approximately 395,000 Medicare FFS beneficiaries with short-term acute care hospital discharges (excluding Maryland) for AMI in FY 2014, 60 percent were discharged under MS-DRGs proposed to be included in the AMI model, specifically 33 percent under AMI MS-DRGs and 25 percent under PCI MS-DRGs.[30] An additional 3 percent of beneficiaries were in MS-DRGs assigned for death from AMI in the hospital. Although 5 percent of beneficiaries with hospital discharges for AMI were discharged under CABG MS-DRGs, we note that because both PCI and fibrinolysis can restore blood flow in an acutely occluded coronary artery more quickly than CABG, these interventions are currently preferred to CABG in most cases of AMI. Furthermore, over recent years cardiovascular clinical practice patterns have generally shifted away from surgical treatment of coronary artery occlusion toward percutaneous, catheter-based interventions.[31] The remaining 34 percent of beneficiaries with AMI diagnoses were distributed across a heterogeneous group of over 300 other MS-DRGs, such as septicemia, respiratory system diagnosis with ventilator support, and major cardiovascular procedures. For this latter group of beneficiaries, the AMI diagnosis appeared in a secondary position on the hospital claim in more than 90 percent of the cases, therefore most likely representing circumstances where the beneficiary hospitalized for another clinical condition experienced an AMI during the hospital stay. By focusing the AMI model on AMIs treated medically or with revascularization with PCI, we propose to test a condition-specific EPM that is discretely defined and includes a significant majority of beneficiaries with AMI in the AMI model. In CYs 2012-2014, the average Medicare spending for an AMI episode that extends 90 days post-hospital discharge was approximately $24,200.[32] From the AMI model, we expect to better understand the impact such an EPM can have on efficiency and quality of care for beneficiaries across the entire spectrum of AMI care, including diagnosis, treatment, and recovery, as well as short-term secondary prevention.

Beneficiaries in the proposed AMI and CABG models would all have CAD. In 2010 in the U.S, the prevalence of CAD in the population 65 years and older was about 20 percent.[33] Patients with CAD also often experience other conditions with significant health-related implications, including diabetes. To improve care for patients with CAD, most approaches in the private and public sectors focus on improving the efficiency and quality of care around procedures such as PCI and CABG. The BPCI models are an example of such an approach. As discussed previously in this section, our proposal for the AMI model extends beyond a procedure-based EPM to include beneficiaries hospitalized for medical management or PCI for AMI in a single EPM, and we propose to test the CABG model, which also would include beneficiaries with AMI, at the same participant hospitals. We believe that hospitalization for AMI, whether accompanied solely by medical management or including revascularization during the initial hospitalization or in a planned CABG readmission, is a sentinel event indicating the need for an increased focus on condition-specific management, as well as on care coordination and active management to prevent future acute events, both during the AMI and CABG model episodes and beyond. We also believe that improving the quality and efficiency of CAD care over a long period of time is important given the chronic nature of this condition that has serious implications for beneficiary health.

The AMI and CABG models provide an opportunity for us to incentivize CAD-specific care management and care coordination for AMI and CABG model beneficiaries that lay the groundwork for longer-term improvements in quality and efficiency of care for beneficiaries with CAD. We note that the quality measures proposed for use in the pay-for-performance methodologies of the AMI and CABG models do not currently include longer-term outcomes or patient experience outside of the AMI or CABG model episode itself, as discussed in sections III.E.2.b. and c. of this proposed rule, although we are interested in comments about potential future measures that could incorporate longer-term outcomes. Moreover, as discussed in section VI. of this proposed rule, we also propose to test a cardiac rehabilitation (CR)/intensive cardiac rehabilitation (ICR) incentive payment, hereinafter CR incentive payment, in AMI and CABG model participants located in some of the MSAs selected for AMI and CABG model participation, as well as in hospitals located in some of the MSAs that are not selected for AMI or CABG model participation. We would evaluate the effects of the CR incentive payment in the context of an episode payment model and Medicare FFS on utilization of CR/ICR, as well as short-term (within the period of time extending 90 days following hospital discharge from an AMI or CABG hospitalization) and longer-term outcomes. We believe this test may result in valuable findings about effective strategies to increase utilization of CR/ICR services that have a strong evidence-base for their effectiveness but a long history of underutilization.

2. Advanced Alternative Payment Model Considerations

a. Overview for the EPMs

The MACRA created two paths for eligible clinicians to link quality to payments: The MIPS and Advanced APMs. These two paths create a flexible payment system called the Quality Payment Program as proposed by CMS in the Quality Payment Program proposed rule (81 FR 28161 through 28586).

As proposed in the Quality Payment Program proposed rule, an APM must meet three criteria to be considered an Advanced APM (81 FR 28298). First, the APM must provide for payment for covered professional services based on quality measures comparable to measures described under the Start Printed Page 50808performance category described in section 1848(q)(2)(B)(i) of the Act, which is the MIPS quality performance category. Under the Quality Payment Program proposed rule, we proposed that the quality measures on which the Advanced APM bases payment for covered professional services (as that term is defined in section 1848(k)(3)(A) of the Act) must include at least one of the following types of measures, provided that they have an evidence-based focus and are reliable and valid (81 FR 28302):

  • Any of the quality measures included on the proposed annual list of MIPS quality measures.
  • Quality measures that are endorsed by a consensus-based entity.
  • Quality measures developed under section 1848(s) of the Act.
  • Quality measures submitted in response to the MIPS Call for Quality Measures under section 1848(q)(2)(D)(ii) of the Act.
  • Any other quality measures that CMS determines to have an evidence-based focus and be reliable and valid.

As we discussed in the Quality Payment Program proposed rule, because the statute identifies outcome measures as a priority measure type and we wanted to encourage the use of outcome measures for quality performance assessment in APMs, we further proposed in that rule that, in addition to the general quality measure requirements, an Advanced APM must include at least one outcome measure if an appropriate measure is available on the MIPS list of measures for that specific QP Performance Period, determined at the time when the APM is first established (81 FR 28302 through 28303).

Second, the APM must either require that participating APM Entities bear risk for monetary losses of a more than nominal amount under the APM or be a Medical Home Model expanded under section 1115A(c) of the Act. Except for Medical Home Models, we proposed in the Quality Payment Program proposed rule that, for an APM to meet the nominal amount standard, the specific level of marginal risk must be at least 30 percent of losses in excess of expected expenditures; a minimum loss rate, to the extent applicable, must be no greater than 4 percent of expected expenditures; and total potential risk must be at least 4 percent of expected expenditures (81 FR 28306).

Third, the APM must require participants to use CEHRT (as defined in section 1848(o)(4) of the Act), as specified in section 1833(z)(3)(D)(i)(I) of the Act, to document and communicate clinical care with patients and other health care professionals. Specifically, where the APM participants are hospitals, the APM must require each hospital to use CEHRT (81 FR 28298 through 28299).

In this proposed rule, we propose to adopt two different tracks for the EPMs—Track 1 in which EPMs and EPM participants would meet the criteria for Advanced APMs as proposed in the Quality Payment Program proposed rule, and Track 2 in which the EPMs and EPM participants would not meet those proposed criteria. For the proposed AMI, CABG, and SHFFT models, we propose pay-for-performance methodologies that use quality measures that we believe would meet the proposed Advanced APM quality measure requirements in the Quality Payment Program proposed rule. As discussed in sections III.E.2. and 3. of this proposed rule, all but one of the AMI, CABG, and SHFFT model measures used in the EPM pay-for-performance methodologies are NQF-endorsed and have an evidence-based focus and are reliable and valid. Therefore, we believe they would meet the proposed Advanced APM general quality measure requirements. The Excess Days in Acute Care after Hospitalization for AMI (AMI Excess Days) measure, which is proposed for the AMI model, is not currently NQF-endorsed, but we believe it meets the measure requirements by having an evidence-based focus and being reliable and valid because this measure has been proposed and adopted through rulemaking for use in the Hospital Inpatient Quality Reporting (HIQR) Program.

Each of the proposed EPM pay-for-performance methodologies includes one outcome measure that is NQF-endorsed, has an evidence-based focus, and is reliable and valid. The EPM quality measures are discussed in detail in section III.E. of this proposed rule, where we assign the quality measures to quality domains. For the AMI model, we propose to use the Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) Following Acute Myocardial Infarction (NQF #0230) (MORT-30-AMI) outcome measure. For the CABG model, we propose to use the Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) Following Coronary Artery Bypass Graft (CABG) Surgery (NQF# 2558) (MORT-30-CABG) outcome measure. Finally, for the SHFFT model, we propose to use the Hospital-level RSCR following elective primary THA and/or TKA (NQF #1550) (Hip/Knee Complications) outcome measure. Thus, based on the proposed use of these three outcomes measures in the EPMs, we believe the proposed AMI, CABG, and SHFFT models would meet the requirement proposed for Advanced APMs in the Quality Payment Program proposed rule for use of an outcome measure that also meets the general quality measure requirements.

In terms of the proposed nominal risk criteria for Advanced APMs, beginning in performance year 2 for episodes ending between April 1, 2018 and December 31, 2018, EPM participants would begin to bear downside risk for excess actual EPM-episode spending above the quality-adjusted target price as discussed in section III.D.2.c. of this proposed rule. The marginal risk for excess actual EPM-episode spending above the quality-adjusted target price would be 100 percent over the range of spending up to the stop-loss limit, which would exceed 30 percent marginal risk, and there would be no minimum loss rate. As a result, we believe the EPMs would meet the marginal risk and minimum loss rate elements of the nominal risk criteria for Advanced APMs proposed in the Quality Payment Program proposed rule. Total potential risk for most EPM participants would be 5 percent of expected expenditures beginning in the second quarter of performance year 2, and increasing in subsequent performance years as discussed in section III.D.7.b. of this proposed rule. Therefore, we believe the total potential risk applicable to most EPM participants, with the lowest total potential risk being 5 percent for EPM episodes ending on or after April 1, 2018 in performance year 2, would meet the total potential risk element of the nominal risk amount standard for Advanced APMs proposed in the Quality Payment Program proposed rule because it is greater than the value of at least 4 percent of expected expenditures.

We note that we propose that EPM participants that are rural hospitals, sole community hospitals (SCHs), Medicare Dependent Hospitals (MDHs) and Rural Referral Centers (RRCs) would have a stop-loss limit of 3 percent beginning in the second quarter of performance year 2 as discussed in section III.D.7.c. of this proposed rule. Because 3 percent is less than the proposed threshold of at least 4 percent of expected expenditures for total potential risk proposed for Advanced APMs in the Quality Payment Program proposed rule, those rural hospitals, SCHs, MDHs, and RRCs that are EPM participants subject to special protections would be in Track 2 EPMs that would not meet the proposed nominal risk standard for Advanced Start Printed Page 50809APMs for performance year 2. We recognize that this proposal might initially limit the ability of rural hospitals, SCHs, MDHs, and RRCs to be in Track 1 EPMs that are Advanced APMs. We believe this potential limitation on rural hospitals, SCHs, MDHs, and RRCs is appropriate for the following reasons: (1) Greater risk protections for these hospitals proposed for the EPMs beginning in the second quarter of performance year 2 and subsequent performance years compared to other EPM participants are necessary, regardless of their implications regarding Advanced APMs based on the nominal risk standard proposed in the Quality Payment Program proposed rule, because these hospitals have unique challenges that do not exist for most other hospitals, such as being the only source of health care services for beneficiaries or certain beneficiaries living in rural areas or being located in areas with fewer providers, including fewer physicians and post-acute care facilities; and (2) under the risk arrangements proposed for the EPMs, these hospitals would not bear an amount of risk in performance year 2 that we determined to be more than nominal in the Quality Payment Program proposed rule. However, we seek comment on whether we should allow EPM participants that are rural hospitals, SCHs, MDHs, or RRCs to elect a higher stop-loss limit for the part of performance year 2 where downside risk applies in order to permit these hospitals to be in Track 1 EPMs for that part of performance year 2. We note that by performance year 3, the stop-loss limit for these hospitals with special protections under the EPMs would increase to 5 percent under our proposal, so these hospitals could be in Track 1 EPMs based on the nominal risk standard proposed in the Quality Payment Program proposed rule.

As addressed in the Quality Payment Program proposed rule, it is necessary for an APM to require the use of CEHRT in order to meet the criteria to be considered to be an Advanced APM. Therefore, according to the requirements proposed in the Quality Payment Program proposed rule, so that the EPMs may meet the proposed criteria to be Advanced APMs, we propose to require EPM participants to use CEHRT (as defined in section 1848(o)(4) of the Act) to participate in Track 1 of the EPMs. We propose that Track 1 EPM participants must use certified health IT functions, in accordance with the definition of CEHRT under our regulation at 42 CFR 414.1305, to document and communicate clinical care with patients and other health care professionals as proposed in the Quality Payment Program proposed rule (81 FR 28299). We believe this proposal would allow Track 1 EPMs to be able to meet the proposed criteria to be Advanced APMs.

Without the collection of identifying information on eligible clinicians (physicians, nonphysician practitioners, physical and occupational therapists, and qualified speech-language pathologists) who would be considered Affiliated Practitioners as proposed in the Quality Payment program proposed rule under the EPMs, CMS would not be able to consider participation in the EPMs in making determinations as to whom could be considered a QP (81 FR 28320). As detailed in the Quality Payment Proposed rule, these determinations are based on whether the eligible clinician meets the QP threshold under either the Medicare Option starting in payment year 2019 or the All-Payer Combination Option, which is available starting in payment year 2021 (81 FR 28165). Thus, we make proposals in the following sections to specifically address these issues that might otherwise preclude the EPMs from being considered Advanced APMs, or prevent us from operationalizing them as Advanced APMs. Based on the proposals for Advanced APM criteria in the Quality Payment Program proposed rule, we seek to align the design of the proposed EPMs with the proposed Advanced APM criteria and enable CMS to have the necessary information on eligible clinicians to make the requisite QP determinations.

b. EPM Participant Tracks

To be considered an Advanced APM, the APM must require participants to use CEHRT (as defined in section 1848(o)(4) of the Act), as specified in section 1833(z)(3)(D)(i)(I) of the Act. We propose that all EPM participants must choose whether to meet the CEHRT use requirement. EPM participants that do not choose to meet and attest to the CEHRT use requirement would be in Track 2 of the EPMs. EPM participants selecting to meet the CEHRT use requirement would be in Track 1 of the EPMs and would be required to attest in a form and manner specified by CMS to their use of CEHRT that meets the definition in our regulation at § 414.1305 to document and communicate clinical care with patients and other health professionals, consistent with the proposal in the Quality Payment Program proposed rule for the CEHRT requirement for Advanced APMs (81 FR 28299). EPM participants choosing not to meet and attest to the CEHRT use requirement would not be required to submit an attestation.

We believe that the selection by EPM participants to meet and attest to the CEHRT use requirement would create no significant additional administrative burden on EPM participants. Moreover, the choice of whether to meet and attest to the CEHRT use requirement would not otherwise change any EPM participant's requirements or opportunity under the EPM. However, to the extent that eligible clinicians who enter into financial arrangements related to Track 1 EPM participants are considered to furnish services through an Advanced APM, those services could be considered for purposes of determining whether the eligible clinicians are QPs.

The proposals for CEHRT use and attestation for EPM participants are included in § 512.120(a). We seek comment on our proposals for EPM participant CEHRT use requirements.

c. Clinician Financial Arrangements Lists Under the EPMs

In order for CMS to make determinations as to eligible clinicians who could be considered QPs based on services furnished under the EPMs (to the extent the models are determined to be Advanced APMs), we require accurate information about eligible clinicians who enter into financial arrangements under the Track 1 EPMs under which the Affiliated Practitioners support the participants' cost or quality goals as discussed in section III.I. of this proposed rule. We note that eligible clinicians could be EPM collaborators engaged in sharing arrangements with an EPM participant; PGP members who are collaboration agents engaged in distribution arrangements with a PGP that is an EPM collaborator; or PGP members who are downstream collaboration agents engaged in downstream distribution arrangements with a PGP that is also an ACO participant in an ACO that is an EPM collaborator. These terms as they apply to individuals and entities with financial arrangements under the EPMs are discussed in section III.I. of this proposed rule. A list of physicians and nonphysician practitioners in one of these three types of arrangements could be considered an Affiliated Practitioner List of eligible clinicians who are affiliated with and support the Advanced APM Entity in its participation in the Advanced APM as proposed in the Quality Payment Program proposed rule. Therefore, this list could be used to make determinations of who would be Start Printed Page 50810considered for a QP determination based on services furnished under the EPMs (81 FR 28320).

Thus, we propose that each EPM participant that chooses to meet and attest to the CEHRT use requirement must submit to CMS a clinician financial arrangements list in a form and manner specified by CMS on a no more than quarterly basis. The list must include the following information for the period of the EPM performance year specified by CMS:

  • For each EPM collaborator who is a physician, nonphysician practitioner, or provider of outpatient therapy services during the period of the EPM performance year specified by CMS:

++ The name, tax identification number (TIN), and national provider identifier (NPI) of the EPM collaborator.

++ The start date and, if applicable, end date, for the sharing arrangement between the EPM participant and the EPM collaborator.

  • For each collaboration agent who is a physician or nonphysician practitioner of a PGP that is an EPM collaborator during the period of the EPM performance year specified by CMS:

++ The TIN of the PGP that is the EPM collaborator, and the name and NPI of the physician or nonphysician practitioner.

++ The start date and, if applicable, end date, for the distribution arrangement between the EPM collaborator that is a PGP and the physician or nonphysician practitioner who is a PGP member.

  • For each downstream collaboration agent who is a physician or nonphysician practitioner member of a PGP that is also an ACO participant in an ACO that is an EPM collaborator during the period of the EPM performance year specified by CMS:

++ The TIN of the PGP that is the ACO participant, and the name and NPI of the physician or nonphysician practitioner.

++ The start date and, if applicable, end date, for the downstream distribution arrangement between the collaboration agent that is both PGP and an ACO participant and the physician or nonphysician practitioner who is a PGP member.

  • If there are no individuals that meet the requirements to be reported as EPM collaborators, collaboration agents, or downstream collaboration agents, the EPM participant must attest in a form and manner required by CMS that there are no individuals to report on the clinician financial arrangements list.

As discussed in the Quality Payment program proposed rule, those physicians or nonphysician practitioners who are included on the Affiliated Practitioner List as of December 31 of a performance period would be assessed to determine whether they qualify for APM Incentive Payments (81 FR 28320).

While the required submission of this information may create some additional administrative requirements for certain EPM participants, we expect that Track 1 EPM participants could modify their contractual relationships with their EPM collaborators and, correspondingly, require those EPM collaborators to include similar requirements in their contracts with collaboration agents and in the contracts of collaboration agents with downstream collaboration agents.

The proposal for the submission of a clinician financial arrangements list by EPM participants that meet and attest to the CEHRT use requirement for the EPMs is included in § 512.120(b). We seek comments on the proposal for submission of this information. We are especially interested in comments about approaches to information submission, including the periodicity and method of submission to CMS that would minimize the reporting burden on EPM participants while providing CMS with sufficient information about eligible clinicians in order to facilitate QP determinations to the extent EPMs are considered Advanced APMs.

d. Documentation Requirements

For each EPM participant that chooses to meet and attest to CEHRT use, we propose that the EPM participant must maintain documentation of their attestation to CEHRT use and clinician financial arrangements lists submitted to CMS. These documents would be necessary to assess the completeness and accuracy of materials submitted by an EPM participant in the Track 1 EPM and to facilitate monitoring and audits. For the same reason, we further propose that the EPM participant must retain and provide access to the required documentation in accordance with § 512.110.

The proposal for documentation of attestation to CEHRT use and clinician financial arrangements lists submitted to CMS is included in § 512.120(c). We seek comment on this proposal for required documentation.

3. Future Directions for Episode Payment Models

a. Refinements to the BPCI Initiative Models

The BPCI initiative Models 2, 3, and 4 would not currently qualify as Advanced APMs based on the two of the Advanced APM criteria in the Quality Payment Program proposed rule, payment based on quality measures and CEHRT use (81 FR 28298). Specifically, BPCI participants are not currently required to use CEHRT, and although CMS examines the quality of episode care in the BPCI evaluation, BPCI episode payments are not specifically tied to quality performance. Instead, BPCI episode payments are based solely on episode spending performance, although we expect that reductions in spending would generally be linked to improved quality through reductions in hospital readmissions and complications. However, building on the BPCI initiative, the Innovation Center intends to implement a new voluntary bundled payment model for CY 2018 where the model(s) would be designed to meet the criteria to be an Advanced APM.

b. Potential Future Condition-Specific Episode Payment Models

In the context of our proposal for the AMI and CABG models that include beneficiaries with CAD who experience an acute event or a major surgical procedure, we seek comment on model design features for potential future condition-specific episode payment models that could focus on an acute event or procedure or longer-term care management, including other models for beneficiaries with CAD that may differ from the design of the EPMs proposed in this rulemaking. We believe such future models may have the potential to be Advanced APMs that emphasize outpatient care and, like the proposed AMI and CABG models, could incentivize the alignment of physicians and other eligible professionals participating in the Advanced APM through accountability for the costs and quality of care. Such condition-specific episode payment models may provide for a transition from hospital-led EPMs to physician-led accountability for episode quality and costs, especially given the importance of care management over long periods of time for beneficiaries with many chronic conditions.

We request that commenters provide specific information regarding all relevant issues for potential future condition-specific episode payment models, including identifying beneficiaries for the model; including services in the episode definition; beginning and ending episodes; pricing episodes, including risk-adjustment; designating the accountable entity for the quality and cost of the episode, including the role of physician-led Start Printed Page 50811opportunities; sharing of responsibility for quality and spending between primary care providers, specialty physicians, and other health care professionals; incentivizing the engagement of physicians and other providers and suppliers in episode care; measuring quality and including quality performance and improvement in the payment methodology; interfacing with other CMS models and programs responsible for population health and costs, such as ACOs and Primary Care Medical Homes (PCMHs); and other considerations specific to identifying future models as Advanced APMs; and any other issues of importance for the design of such an EPM.

c. Potential Future Event-Based Episode Payment Models for Procedures and Medical Conditions

Given the proposed EPM methodology discussed in section III.C.4.a. of this proposed rule for the three models that would begin the episodes with initial hospitalizations, the proposed AMI, CABG, and SHFFT episodes are similar to the LEJR episodes in the CJR model because they reflect clinical conditions for which care is almost always begun during an inpatient hospitalization, either on an emergency or elective basis. In addition, the clinical conditions represented by these EPM episodes generally result in straightforward assignment to MS-DRGs at discharge that are specific to clinical conditions included in the episodes. This contrasts with procedure-related clinical conditions for which the site-of-service can be inpatient or outpatient (for example, elective PCI for non-AMI beneficiaries) or hospitalization for medical conditions for which the ultimate MS-DRG assigned is less clear at the beginning of an episode (for example, hospitalization for respiratory symptoms which may lead to discharge from heart failure, pneumonia, or other MS-DRGs based on reporting of ICD-CM diagnosis codes on hospital claims).

To address the issues related to the development of future episode payment models for a broader range of clinical conditions, we seek comment on model design features that would be important for episode payment models targeting procedures that may be performed in both the inpatient and outpatient setting, as well as models focused on hospitalization for acute medical conditions which may overlap or interact (for example, sepsis related to pneumonia or acute kidney injury related to congestive heart failure exacerbation). In particular, episode payment models must clearly define the beginning of the episode as well as set an episode price that is appropriate for beneficiaries included in the episode, which has commonly been based on historical spending for such beneficiaries in both existing CMS models and the three proposed EPMs. These parameters pose specific challenges as the variety of clinical conditions targeted for episode payments expands beyond lower extremity orthopedic procedures and acute cardiac conditions, and we expect that such potential future models would need to be designed differently than the CJR model or the EPMs proposed in this rulemaking.

For example, because procedures such as PCI for non-AMI beneficiaries or cardioverter defibrillator implantations can occur in the inpatient or outpatient setting, an episode payment model would need to include beneficiaries receiving such procedures at all sites-of-service so as to not influence decisions on where procedures are performed based on payment-related rather than clinical considerations. Episode payment models that begin with the same procedure performed in the inpatient or outpatient setting would require methodological development beyond the approaches that have been used thus far in CMS's other EPMs that rely upon the MS-DRG for a hospitalization to begin an episode and identify historical episodes for setting episode prices. Such models that involve episode payment for procedures furnished in the inpatient or outpatient setting may allow for significant physician-led opportunities that would allow the models to be identified as Advanced APMs. We seek comment on how these types of procedures could be included in future episode payment models, including identifying the accountable entity, and the role of physician-led opportunities; defining the episode beginning and end; setting episode prices; applying risk-adjustment to account for differences in expected episode spending for a heterogeneous population of beneficiaries; and any other issues of importance for the design of such an episode payment model.

We also seek comment on potential future episode payment models that would include care for medical conditions that result in the serious health event of an inpatient hospitalization, which often represents, regardless of the specific reason for the hospitalization, a common pathway that includes failure of outpatient care management and care coordination for beneficiaries with chronic conditions. While we do include in the proposed AMI model beneficiaries who solely receive medical treatment, we note that beneficiaries with AMI are almost always hospitalized and their MS-DRGs at discharge are generally predictable and consistent based on their AMI diagnoses. This is not the case for a number of medical conditions for which grouping by MS-DRGs is more complicated or less consistent. Many non-procedural hospitalizations of Medicare beneficiaries are ultimately categorized based on the principal ICD-CM diagnosis code reported on a claim, which in turn is mapped to a Major Diagnostic Category (MDC) based on the involved organ system, which then leads to the assignment of any of various specific MS-DRGs based on the medical groups in the MDC. For example, the medical groups for the Respiratory System MDC are pulmonary embolism, infections, neoplasms, chest trauma, pleural effusion, pulmonary edema and respiratory failure, chronic obstructive pulmonary disease, simple pneumonia, RSV pneumonia and whooping cough, interstitial lung disease, pneumothorax, bronchitis and asthma, respiratory symptoms and other respiratory diagnoses.[34] Unlike a beneficiary who undergoes a surgical procedure or who is hospitalized for a specific medical condition such as AMI, the ultimate MS-DRG at discharge assigned to a beneficiary hospitalized for diagnosis and management of respiratory symptoms may not be clear during the hospitalization itself, or even afterward, until the inpatient claim is submitted and paid by Medicare. This makes it challenging for providers to engage in care delivery redesign targeted to a specific patient population identified by MS-DRG. Additionally, it is possible that beneficiaries hospitalized for certain medical conditions also may follow common clinical pathways before and after discharge for which similar care redesign strategies could be developed and used despite those beneficiaries' assignments to different MS-DRGs for their anchor hospitalizations. Thus, we believe that hospitalization for most medical conditions would require special consideration in the development of potential future episode payment models that goes beyond CMS's current approach of relying upon the MS-DRG for the anchor hospitalization to begin an episode and identify historical episodes for setting episode prices. We seek comment on design features needed to address these considerations, including defining the beginning and end of episodes; setting episode prices, Start Printed Page 50812including risk-adjustment, that would support the provision of appropriate and coordinated care for beneficiaries following hospital discharge for a period of time during the episode; and any other issues of importance for the design of such an episode payment model.

d. Health Information Technology Readiness for Potential Future Episode Payment Models

We are particularly interested in issues related to readiness of providers and suppliers that are not hospitals to take on financial responsibility for episode cost and quality in potential future episode payment models. We have some experience in BPCI Models 2 and 3 with non-hospital providers and suppliers, specifically post-acute care providers and physician group practices (PGPs), who assume financial responsibility for the cost of episode care. In BPCI Model 2, PGPs may directly bear financial responsibility for episode cost for up to 48 clinical conditions for the anchor inpatient admission and up to 90 days post-hospital discharge. In BPCI Model 3, PGPs and post-acute care providers, including skilled nursing facilities, home health agencies, inpatient rehabilitation facilities, and long-term care hospitals, may directly bear financial responsibility for episode cost for up to 48 clinical conditions for a duration that extends up to 90 days following initiation of post-acute care following discharge from an inpatient hospitalization.

Under these circumstances, PGPs and post-acute care providers typically need to use health IT to assist them in effectively coordinating the care of BPCI beneficiaries across settings throughout the episodes. The risk-bearing entities participating in BPCI have expressed readiness to take on financial responsibility for episode cost, and they commonly rely upon health IT for assistance in managing the care for BPCI beneficiaries across settings for episodes that extend for a substantial period of time. However, a recent national survey of IT in nursing homes showed common use of IT for administrative activities but less use for clinical care.[35] Anecdotally, stakeholders have told us that accountable non-hospital providers and suppliers, especially those that are not integrated with health systems, may have less well-developed tools for following patients throughout episodes, potentially resulting in greater challenges in reducing the cost and improving the quality of episode care under the BPCI models. Therefore, we understand that limitations in the availability of health IT that can be used in beneficiary management across care settings may pose a significant barrier to the readiness of non-hospital providers and suppliers to assume financial responsibility for episodes in potential future episode payment models.

In the CJR model, acute care hospitals are financially responsible for cost and quality during LEJR episodes-of-care. CJR model participant hospitals may form partnerships with post-acute care providers such as skilled nursing facilities and home health agencies, as well as physicians and PGPs, to share financial risk and collaborate on care redesign strategies, as in BPCI. Although hospitals are the financially responsible entities under the CJR model, we recognize that partnerships with post-acute care providers could be a crucial driver of episode spending and quality, given that many beneficiaries in the CJR model receive post-acute care services after discharge from the hospital. We also recognize that tools such as health IT may be critical for certain care management and quality strategies targeted toward the goal of lower cost and higher quality episode care. Limitations in the availability of health IT may pose a barrier to effective post-acute care provider collaboration and sharing of financial risk in episode payment models even when hospitals are the financially responsible entities under such models, such as the CJR model and the three new EPMs proposed in this rulemaking.

We recognize that there is wide variation in the readiness of other providers and suppliers to bear financial responsibility for episodes, either directly or indirectly through sharing arrangements with the directly responsible entities where those arrangements may include upside and downside risk. For instance, adoption of health IT among providers in the post-acute care market, such as skilled nursing facilities, continues to lag behind hospitals and providers of ambulatory care services. In addition to facing significant resource constraints, post-acute care providers were not included as an eligible provider type under the Medicare and Medicaid Electronic Health Record (EHR) Incentive Programs. The recent extension of Medicaid 90/10 funding offers new opportunities for states to include post-acute care providers in projects focused on infrastructure development, but will not address the cost of health IT adoption among post-acute care providers.[36]

To ensure that post-acute care providers and other types of providers and suppliers can succeed under future episode payment models, either as the directly financially responsible entity or as collaborators with other directly financially responsible entities, we are interested in opportunities to increase provider readiness as part of the design of potential future episode payment models and the potential refinement of current episode payment models. Specifically, we would like to explore: Incentives to encourage post-acute care providers, as well as other providers and suppliers that furnish services to episode payment model beneficiaries, to make necessary investments in health IT infrastructure; payment mechanisms that could leverage savings achieved under episode payment models to contribute to these investments; and any other strategies to enhance the adoption, implementation, and upgrading of certified health IT. We seek comment on these ideas, as well as the following questions:

  • What are key challenges associated with the inclusion of post-acute care providers as the financially responsible entity or as collaborators with other financially responsible entities in episode payment models today?
  • What would be a sufficient financial incentive or bonus to enhance the adoption, implementation, and upgrading of certified health IT in post-acute care settings?
  • How else can episode payment models encourage the use of certified health IT and information sharing among providers and suppliers caring for episode payment model beneficiaries to improve care coordination and patient outcomes?
  • Within the existing CJR model, are there additional opportunities to encourage investment in adoption, implementation, and upgrading of certified health IT among post-acute care providers to support improvements in care coordination and patient outcomes? What CJR model refinements could enable direct investments to support these improvements, particularly among post-acute care providers who are unaffiliated with CJR model participant hospitals but who provide services to CJR model beneficiaries, including post-acute care providers who may enter into financial arrangements with CJR model participant hospitals as CJR collaborators?Start Printed Page 50813

B. Proposed Definition of the Episode Initiator and Selected Geographic Areas

1. Background

The proposed new EPMs will complement the current CJR model and continue efforts to move Medicare towards paying providers based on quality and value. As discussed during rulemaking for the CJR model, CMS is interested in testing and evaluating the impact of an episode payment approach for a broad range of episodes in a variety of other circumstances. In addition to including hospitals that have not chosen to voluntarily participate in earlier models, we also are interested in expanding the range of episodes included beyond elective surgical procedures such that the impact on a broader range of beneficiaries, hospitals, and circumstances may be tested. We also are interested in evaluating the impact on hospitals when an increasing percentage of care to Medicare beneficiaries is paid for through alternative payment models.

As with CJR, we propose in § 512.105(c) that the hospital be the accountable financial entity and that these episode payment models be implemented in all IPPS hospitals in the geographic areas selected, subject to exclusions as specified in §§ 512.230 and 512.240 of the proposed rule. While these are considered new episode payment models and do not reflect an expansion or extension of any previous models, they do intentionally build significantly upon the work of BPCI and, most significantly, the framework established for CJR under 42 CFR part 510 published on November 24, 2015. Given the extensive consideration given to many of these issues during the CJR model planning and rulemaking periods, we believe this is important as we seek to build a model that is scalable across all providers and episode types. We also seek to limit the burden for hospitals and other providers that may be participating across multiple episode types. Therefore, to the extent applicable and appropriate, we have sought consistency with rules established for the CJR model. We seek comment on those areas where alternative options are proposed or should be considered that would not add additional operational burden or complexity.

2. Proposed Definition of Episode Initiator

Under the proposed EPMs, we propose, consistent with our definition under the CJR model that episodes would begin with the admission to an IPPS acute-care hospital that triggers an AMI, CABG or SHFFT episode as specified in section III.C.4.a. of this proposed rule. As with the CJR model, we propose that hospitals would be the only episode initiators in these episode payment models. For purposes of these episodes payment models the term ”hospital” means a hospital as defined in section 1886(d)(1)(B) of the Act. This statutory definition of hospital includes only acute care hospitals paid under the IPPS. Under this proposal, all acute care hospitals in Maryland would be excluded and payments to Maryland hospitals would be excluded in the regional pricing calculations as described in section III.D.4. of this proposed rule. This is the same policy that is being followed with the CJR model. In addition, we also propose to exclude other all-payer state models which may be implemented in the future. We welcome comments on this proposal and whether there are potential approaches for including Maryland acute-care hospitals or, potentially, other hospitals in future all-payer state models in these episode payment models.

As implemented with the CJR model, we propose to designate IPPS hospitals as the episode initiators to ensure that all services covered under FFS Medicare and furnished by EPM participant hospitals in selected geographic areas to beneficiaries who do not meet the exclusion criteria specified in section III.C.4. of this proposed rule are included. In addition, the episodes must not be BPCI episodes that we are proposing to exclude as outlined in this section and in section III.C.4. of this proposed rule. We believe that utilizing the hospital as the episode initiator is a straightforward approach for these models because patients covered under these DRGs and diagnoses require hospital admission for these services, whether provided on an emergent or planned basis. Under these new models covering medical admissions and services that are not necessarily elective, we will be able to expand our testing of a more generalized bundled payment model. Finally, as described in section III.B.4., our proposed geographic area selection approach relies upon our definition of hospitals as the entities that initiate episodes.

3. Financial Responsibility for the Episode of Care

As with the CJR model, we continue to believe it is most appropriate to identify a single type of provider to bear financial responsibility for making repayment, if any, to CMS under the model and propose to make hospitals, as the episode initiators, financially responsible for the episode of care for the following several reasons:

  • Hospitals play a central role in coordinating episode-related care and ensuring smooth transitions for beneficiaries undergoing services related to SHFFT, AMI and CABG episodes. A large portion of a beneficiary's recovery trajectory from an AMI, CABG, or SHFFT begins during the hospital stay.
  • Most hospitals already have some infrastructure related to health information technology, patient and family education, and care management and discharge planning. This includes post-acute care coordination infrastructure and resources such as case managers, which hospitals can build upon to achieve efficiencies under these EPMs.
  • By definition, these episodes always begin with an acute care hospital stay. While often preceded by an emergency room visit and possible transfer from another hospital's emergency room, or followed by post-acute care, these parties are not necessarily always present and would not be appropriate to target as the financially responsible party for this purpose.

EPM episodes may be associated with multiple hospitalizations through transfers. When multiple hospitalizations occur, we propose that the financial responsibility be given to the hospital to which the episode is attributed as described in section III.C.4. We recognize that, particularly where the admission may be preceded by an emergency room visit and subsequent transfer to a tertiary or other regional hospital facility, patients often wish to return home to their local area for post-acute care. Many hospitals have recently heightened their focus on aligning their efforts with those of community providers, both those in the immediate area as well as more outlying areas from which they receive transfers and referrals, to provide an improved continuum of care. In many cases, this is due to the incentives under other CMS models and programs, including ACO initiatives such as the Shared Savings Program, the Hospital Readmissions Reduction Program (HRRP), and the CJR model. By focusing on the hospital as the accountable or financially responsible entity, we hope to continue to encourage this coordination across providers and seek comment on ways we can best encourage these relationships within the scope of these EPMs.

In support of our proposal that hospitals be the episode initiators under these EPMs, we believe that hospitals Start Printed Page 50814are more likely than other providers to have an adequate number of episode cases to justify an investment in episode management for these EPMs. We also believe that hospitals are most likely to have access to resources that would allow them to appropriately manage and coordinate care throughout these episodes. Finally, the hospital staff is already involved in discharge planning and placement recommendations for Medicare beneficiaries, and more efficient post-acute care service delivery provides substantial opportunities for improving quality and reducing costs under EPMs. For those hospitals that are already participating in CJR, we believe the efforts that have been put in place to support patients receiving LEJR will be supportive of the new EPMs proposed under this rule, particularly for SHFFT episodes which we propose to implement in the same geographic areas as the CJR model.

Finally, as noted when planning for the CJR model, although the BPCI initiative includes the possibility of a physician group practice as a type of episode initiating participant, the physician groups electing to participate in BPCI have done so because their practice structure supports care redesign and other infrastructure necessary to bear financial responsibility for episodes. These physician groups are not necessarily representative of the typical group practice. As with the CJR model, the infrastructure necessary to accept financial responsibility for episodes is not present across all physician group practices, and thus we do not believe it would be appropriate to designate physician group practices to bear the financial responsibility for making repayments to CMS under the proposed EPMs. We seek comment on our proposal to establish financial responsibility and accountability under the AMI, CABG, and SHFFT EPMs consistent with our implementation of the CJR model.

Currently, there are SHFFT, AMI, and CABG episodes being tested in BPCI Models 2, 3 or 4. The last remaining BPCI Model 1 hospital will end December 31, 2016 and will, therefore, not overlap with EPM. In addition, under BPCI, there are episodes for PCI, which, if an AMI were also involved, would fall under the AMI model being proposed here. We are proposing that IPPS hospitals located in an area selected for any one of the episode payment models proposed in this rule that also are episode initiators for episodes in the risk-bearing phase of BPCI Models 2 or 4, be excluded from participating in the AMI, CABG, or SHFFT EPMs if the applicable episode otherwise would qualify to be covered under BPCI. This exclusion would be in effect only during the time that the relevant qualifying episodes are included in one of the BPCI models. Likewise, we are proposing that if the EPM participant is not an episode initiator for overlapping episodes under BPCI Models 2 or 4, but these same episodes are initiated during the anchor hospitalization by a physician group practice (PGP) under BPCI Model 2 (where the services are provided at the episode initiating hospital) then the episode also shall be covered under BPCI and be excluded from the EPMs being proposed under this rule. Otherwise qualifying EPM episodes (that is, those that are not part of an overlapping BPCI AMI, CABG, PCI or SHFFT episode) at the participant hospital would be included in these new EPMs. However, because BPCI participation is voluntary and participating providers may select which episodes to participate in, a BPCI participating provider will participate in any of the proposed AMI, CABG, or SHFFT EPMs for any episodes not otherwise preempted under their BPCI participation. For example, a BPCI Model 2 hospital in an AMI episode model geographic area participating in BPCI only for CABGs will be an EPM participant in the AMI model. Similarly, an acute care hospital participating in BPCI for LEJR but not SHFFT episodes would be exempt from participation in the CJR model in a CJR model geographic area but would participate in the SHFFT model for SHFFT episodes. In addition, providers participating in BPCI may also collaborate with an EPM participant for episodes not covered under BPCI. It should be noted that due to differences in how the AMI episode is defined under the AMI model versus BPCI and the inclusion of PCI MS-DRGs under the latter, a patient with the same discharge MS-DRG and diagnoses may qualify for a PCI episode under BPCI and an AMI episode under the AMI model. Our intent is to give precedence to BPCI regardless of which episode a patient qualifies for if the patient would be covered under BPCI.

In section III.D.6. we discuss in more detail how we propose to handle situations when a beneficiary receives services that would qualify for inclusion in more than one CMS payment model during the same or overlapping periods of time. We welcome input on how these overlaps should be handled to best encourage ongoing care coordination while minimizing the impact on other models and limiting confusion and operational burden for providers.

While we propose that the EPM participant be financially responsible for the episode of care under these EPMs, we also believe that effective care redesign requires meaningful collaboration among acute care hospitals, post-acute care providers, physicians, and other providers and suppliers within communities to achieve the highest value care for Medicare beneficiaries. We believe it may be essential for key providers to be aligned and engaged, financially and otherwise, with the EPM participants, with the potential to share financial responsibility with those EPM participants. We note that all relationships between and among providers and suppliers must comply with all relevant laws and regulations, including the fraud and abuse laws and all Medicare payment and coverage requirements unless otherwise specified further in this section and in sections III.I. and III.J. of this proposed rule. Depending on a hospital's current degree of clinical integration, new and different contractual relationships among hospitals and other health care providers may be important, although not necessarily required, for EPM success in a community. We acknowledge that financial incentives for other providers may be important aspects of the model in order for EPM participants to partner with these providers and incentivize certain strategies to improve episode efficiency.

While we acknowledge the important role of conveners in the BPCI model, and AMI, CABG, and SHFFT model participants may wish to enter into relationships with EPM collaborators and other entities in order to manage the episode of care or distribute risk, we propose that the ultimate financial responsibility of the episode remains with the EPM participant. Exceptions to this general rule for beneficiaries covered under certain risk bearing ACO arrangements are outlined in section III.D.6. As with the CJR model, we do not intend to restrict the ability of EPM participants to enter into administrative or risk sharing arrangements related to these EPMs, except to the extent that such arrangements are already restricted or prohibited by existing law. We refer readers to section III.I. of this proposed rule for further discussion of model design elements that may outline financial arrangements between EPM participants and other providers and suppliers.Start Printed Page 50815

4. Proposed Geographic Unit of Selection and Exclusion of Selected Hospitals

In order to determine the geographic unit of selection for these episode payment models, we conducted an analysis similar to that used for the CJR model. For the CJR model, we considered using a stratified random sampling methodology to select: (1) Certain counties based on their Core-Based Statistical Area (CBSA) status; (2) certain zip codes based on their Hospital Referral Regions (HRR) status or (3) certain states. We concluded that selection based on MSAs provided the best balance between choosing smaller geographic units while still capturing the impact of market patterns reflecting the mobility of patients and providers and limiting the potential risk for patient shifting and steerage between MSAs. HRRs are based on where patients receive selected tertiary care services which do not include orthopedic services. Therefore, HRRs may not be representative of where patients receive specialty orthopedic care or more routine orthopedic services such as hip and knee arthroplasty. Selection of states rather than MSAs would have greatly reduced the number of independent geographic areas subject to selection and, therefore, the statistical power of the evaluation. For similar reasons and to maintain consistency with the CJR model, we are, similarly, recommending implementation at the MSA level.

We also similarly considered whether these new models should be limited to hospitals where a high volume of these episodes occur, which would result in a more narrow test on the effects of an episode-based payment, or whether to include all hospitals in particular geographic areas, which would result in testing the effects of an episode-based payment approach more broadly across an accountable care community seeking to coordinate care longitudinally across settings. However, as with the CJR model, there would be more potential for behavioral changes that could include patient shifting and steering between hospitals in a given geographic area that could impact the test. Additionally, this approach would provide less information on testing payments for these episodes across a wide variety of hospitals with different characteristics. Selecting geographic areas and including all IPPS hospitals in those areas not otherwise excluded due to BPCI overlap as previously described and in section III.D.6. of this proposed rule as model participants would help to minimize the risk of participant hospitals shifting higher cost cases out of the EPM.

In determining where to implement these EPMs, we also considered whether implementation of the CJR model in the same geographic area should be a factor. We realize that there is likely to be considerable overlap in the selection criteria between MSAs where the SHFFT EPM might be appropriate and those MSAs where the CJR model is now being implemented. While limiting burden on hospitals is an important consideration, we also believe that the infrastructure being put in place as a result of the CJR model presents significant advantages for implementation of the SHFFT model. For similar reasons, and in order to minimize patient steerage and/or transfer for reasons due solely to the implementation of these new payment models, we believe that it is appropriate to implement the AMI model and CABG model together in the same geographic areas, albeit not necessarily in the same areas as the CJR model.

Therefore, given the authority in section 1115A(a)(5) of the Act, which allows the Secretary to elect to limit testing of a model to certain geographic areas, we propose that the SHFFT model be implemented in those MSAs where the CJR model is being implemented. We also are proposing that the AMI and CABG models be implemented in MSAs selected independently based on the criteria discussed in this proposed rule. This will result in four separate categories of MSAs: (1) MSAs where only the CJR and SHFFT model episodes are being implemented; (2) MSAs where only the CABG model and AMI model episodes are being implemented; (3) MSAs where the CJR as well as the AMI, CABG, and SHFFT models are being implemented; and (4) MSAs where neither CJR nor any of the new episode payment models are being implemented. We believe this will provide an opportunity to test the impact of implementing EPMs across not only a greater diversity of episodes but also as an increasing percentage of hospital discharges. We seek comment on our proposal to implement the SHFFT model in the same geographic region as the CJR model and to implement both the AMI model and the CABG model in the same MSAs, some of which may overlap with MSAs where the CJR and SHFFT models also are being implemented.

5. Overview and Options for Geographic Area Selection for AMI and CABG Episodes

We propose that the AMI and CABG EPMs be implemented together in the same MSAs. These AMI/CABG-participating MSAs may or may not also be LEJR/SHFFT-participating MSAs. The selection of MSAs for AMI/CABG EPMs would occur through a random selection of eligible MSAs.

We propose to require participation in the AMI and CABG models of all hospitals, with limited exceptions as previously discussed in section III.B.4. of this proposed rule, paid under the IPPS that are physically located in a county in an MSA selected through the methodology outlined in section III.B.5.b. in this proposed rule, to test and evaluate the effects of an episode-based payment approach for the proposed EPMs. We propose to determine that a hospital is located in an area selected if the hospital is physically located within the boundary of any of the counties in that MSA as of the date the selection is made.

Although MSAs are revised periodically, with counties added or removed from certain MSAs, we propose to maintain the same cohort of selected hospitals throughout the 5-year performance periods of the EPMs with limited exceptions as described later in this section. Thus, we propose neither to add hospitals to an EPM if after the start of such EPM new counties are added to one of the selected MSAs nor to remove hospitals from an EPM if counties are removed from one of the selected MSAs. We believe that this approach will best maintain the consistency of the participants in the EPMs, which is crucial for our ability to evaluate their respective results. However, we retain the possibility of adding a hospital that is opened or incorporated within one of the selected counties after the selection is made and during the period of performance. (See section III.D. of this proposed rule for discussion of how target prices will be determined for such hospitals.)

The manner in which CMS tracks and identifies hospitals is through the CMS Certification Number (CCN). In keeping with this approach, these EPMs will administer model related activities at the CCN level including the determination of physical location. The physical location associated with the CCN at the time of an EPM's start will be used to determine whether that CCN is located in a selected MSA. For hospitals that share a CCN across various locations, all hospitals under that CCN would be required to participate in the applicable EPM if the physical address associated with the CCN is in the MSA selected, unless otherwise excluded. Similarly, all hospitals under the same CCN, even if some are physically located in the MSA Start Printed Page 50816selected for participation, would not participate in the applicable EPM if the physical address associated with the CCN is not in the MSA.

We considered including hospitals in a given MSA based on whether the hospitals were classified into the MSA for IPPS wage index purposes. However, such a process would be more complicated, and we could not find any compelling reasons favoring such approach. For example, we could assign hospitals to metro divisions of MSAs when those divisions exist. In addition, there is the IPPS process of geographic reclassification by which a hospital's payments can be based on a geographic area other than the one where the hospital is physically located. For the purpose of the EPMs, it is simpler and more straightforward to use a hospital's physical location as the basis of its assignment to a geographic unit. This decision would have no impact on a hospital's payment under the IPPS. We seek comment on our proposal to include a hospital as an EPM participant based on the physical location associated with the CCN of the hospital in one of the counties included in a selected MSA.

a. Exclusion of Certain MSAs

We considered whether certain MSAs should be exempt from the possibility of selection for the AMI/CABG EPMs' implementation. We considered exclusions based on the anticipated number of AMI episodes and CABG episodes in the MSA. We also considered exclusions based on the degree to which such EPMs' episodes would be impacted by overlaps with other payment initiatives, including BPCI and ACOs.

First, we considered the advisability of MSA exclusions based on the number of episodes in a year. We identified qualifying AMI and CABG episodes that initiated between January 1, 2014, and December 31, 2014. AMI and CABG episodes were attributed to an MSA based on the location of the CCN associated with the initiating hospital using the Provider of Service file. Due to the smaller number of relevant AMI and CABG episodes occurring in MSAs, an exclusion rule that required a large number of episodes in each MSA would result in fewer MSAs eligible for selection than was necessary given the desired number of MSAs and the requirement that to have 50 percent or more of MSAs remain in a pool of possible comparison MSAs. From the perspective of evaluating changes to utilization and spending under EPMs, there is no analytic need to eliminate MSAs with small numbers. In fact, including smaller MSAs has the analytic advantage of giving CMS more experience operating EPMs in the smaller-MSA contexts that will help us generalize our EPM-evaluation findings.

We have a strong interest in being able to observe how well EPMs operate in areas with a lower volume of episodes, and, in particular, the consequences of the model for AMI episodes where CABG is not commonly performed or where standard practice is to refer all CABGs outside of the MSA. Given our desire to assess the operation of the AMI EPM in areas with little or no CABG episodes and the desire to have the two cardiac EPMs be administered together in the same MSAs, we propose that the MSA exclusion rules be based on the number of AMI episodes only. This will allow for the inclusion of MSAs with no CABGs.

There is no analytic requirement for a minimum number of cases and there are advantages to including smaller cities. At the same time, we acknowledge that areas with few AMI cases may believe that they will face challenges under the EPMs. Therefore, we propose an exclusion rule that MSAs with fewer than 75 AMI episodes (determined as discussed in section III.C. of this proposed rule) will be removed from the possibility of selection. Cases in hospitals paid under either the CAH methodology or the Maryland All-Payer Model are not included in the count of eligible episodes. We examined a number of different minimum-episode-number cutoffs. The use of the 75 AMIs in a year was a designed to balance limiting the impact of outlier cases on the MSA average episode spending and the desire to retain a non-negligible representation of MSAs in the under 100,000 population and the 100,000 to 200,000 population ranges in our selection pool. The application of Exclusion Rule 1: “less than 75 qualifying AMI episodes in the reference year” resulted in the removal of 49 MSAs from possible selection.

Second, we assessed exclusion rules based on overlap with BPCI. We propose Exclusion Rule 2 such that MSAs are removed from possible selection if there were fewer than 75 non-BPCI AMI episodes in the MSA in the reference year. For the purposes of this exclusion, the number of non-BPCI episodes was estimated by subtracting BPCI cases from the total number of cases used in Exclusion Rule 1. BPCI cases for this purpose are ones during the reference year associated with a hospital or a PGP BPCI Model 2 or 4 episode initiator participating in an AMI, PCI, or CABG episode as of January 1, 2016. Such criterion removed an additional 26 MSAs from potential selection.

Third, we propose to exclude MSAs from possible selection based on whether the number of non-BPCI AMI episodes calculated under Exclusion Rule 2 is less than 50 percent of the total number of AMI episodes calculated under Exclusion Rule 1. We anticipate that some degree of overlap in the BPCI and other EPMs will be mutually helpful. However, we acknowledge that some providers may have concerns that a BPCI Model 2 AMI and PCI participation rate of more than 50 percent may impair the ability of participants in either the EPMs or the BPCI models to succeed in the objectives of the initiative. As a result of this third criterion, 13 additional MSAs were removed from possible selection.

We considered whether there should be an exclusion rule based on the anticipated degree of overlap between the AMI and CABG EPMs and patients who are aligned prospectively to ACOs that are taking two-sided risk, such as ACOs participating in the Next Generation ACO model or Track 3 of the Shared Savings Program. We examined numbers associated with ACOs meeting this status as of May 1, 2016, and this examination did not result in any additional MSAs falling below the 75 AMI episodes threshold. Consequently, we are not proposing any MSA exclusion rule based on the presence of ACOs.

Please refer to Table 1 for the status of each MSA based on these exclusion criteria, available at http://innovation.cms.gov/​initiatives/​epm. After applying these three exclusions, 294 MSAs out of 384 total MSAs are eligible for selection using our proposed selection methodology.

Start Printed Page 50817

Start Printed Page 50818

Start Printed Page 50819

Start Printed Page 50820

Start Printed Page 50821

Start Printed Page 50822

Start Printed Page 50823

Start Printed Page 50824

Start Printed Page 50825

Start Printed Page 50826

Start Printed Page 50827

b. Proposed Selection Approach

We propose the selection of 98 MSAs through the use of simple random selection from the 294 eligible MSAs.

Simple random selection is often considered to be an appropriate default approach to experimental design unless there is a compelling reason to depart from it. One common alternative approach is to perform random selection separately within subgroups. Selection within subgroups can be a useful approach to limiting differences between intervention and control groups to improve statistical power or for facilitating over or under sampling to allow the evaluation to examine effects of the intervention on particular types of MSAs or because those types of MSAs are of particular interest for policy reasons.

In CJR, we used a stratified random assignment approach in which we organized MSAs into strata based on MSA population size and historic LEJR episode payments. Under the CJR model, we believed a stratified approach was appropriate due to wide regional variation in prices, primarily associated with the use of post-acute services. The stratified approach served as a means to oversample in higher-expense MSAs as these areas have both the most need for and the most opportunity under the CJR model.

In assessing whether stratification would be proposed for the EPMs, we assessed a variety of factors described later in this section. Absent stratification, the rate at which a particular type of MSA will appear in the sample will be proportional to how often in appears among eligible MSAs. If a particular type of MSA is relatively common, it is likely to occur often enough that we do not need to deliberately over-sample for it. In the end, our analyses did not provide sufficient evidence that it is necessary to create selection subgroups of MSAs to guide the selection approach. As a result, we are proposing to use simple random selection from the entire pool of eligible MSAs.

(1) Factors Considered but Not Used

We considered a variety of possible MSA characteristics for possible use in classifying sub-groups. Though we did consider many of these variables important, we believe that a simple random selection, where warranted, is preferable.

Some of the factors we considered that we are not proposing to use in the selection methodology include the following:

  • Measures associated with AMI-episode and CABG episode wage-adjusted spending, respectively. In considering how to operationalize such measures, we considered a number of alternatives including average total episode spending payments in an MSA, average episode spending associated with the initial hospital stay(s) and average episode spending occurring in the period after discharge from the initial hospital.
  • Measures associated with variation in practice patterns associated with AMI and CABG episodes. In considering how to operationalize this measure, we considered a number of alternatives including the extent to which both an AMI and a CABG episode are associated with having a transfer hospital stay at the beginning of the episode, and the extent to which CABG hospitalizations occur following a hospital transfer from either within or from outside the same MSA.
  • Measures associated with relative market share of providers with respect to AMI and/or CABG episodes, including the presence or absence of regional referral centers and the number of providers with the capacity to perform CABGs or otherwise treat complex cardiac patients.
  • Health care supply measures of providers in the MSA including acute or post-acute bed counts, and number of relevant physician specialties such as cardiologists and cardiothoracic surgeons.
  • MSA-level demographic measures such as: (1) average income; (2) distributions of population by age, gender or race; (3) percent dually eligible; and (4) percent with specific health conditions or other demographic composition measures.
  • Measures associated with the degree to which a market might be more capable or ready to implement care-redesign activities. Examples of market-level characteristics that might be associated with anticipated ease of implementation include the MSA-level EHR meaningful-use levels, managed-care penetration, ACO penetration, and experience with other bundling efforts.

Though these measures are not proposed to be part of the selection process, we acknowledge that these and other market-level factors may be important to the proper understanding of the evaluation of the impact of EPMs. We intend to consider these and other measures in determining which MSAs are appropriate comparison markets for the evaluation and for possible subgroup analysis or risk-adjustment purposes. The evaluations will include beneficiary-, provider-, and market-level characteristics in how they will examine the performance of these proposed EPMs.

(2) Sample-Size Calculations and the Number of Selected MSAs

Our analyses of the necessary sample size led us to propose the selection of 98 MSAs, out of the 294 MSAs eligible Start Printed Page 50828for selection and 384 total MSAs, to participate in both the AMI and CABG EPMs. In this section, we discuss the assumptions and modeling that went into our proposal to test these EPMs in 98 MSAs. The discussion of the method of selection of these 98 MSAs is addressed in the following section. In coming to the decision to target 98 MSAs, we are proposing an approach that limits the size of the intervention to the greatest degree possible, while still ensuring that we have sufficient statistical power to reliably evaluate the effects of the EPMs. Going below this threshold would jeopardize our ability to be confident in our results and to be able to generalize from the EPMs to the larger national context.

In calculating the necessary size of the AMI and CABG EPMs, a key consideration was to have sufficient power to be able to detect the desired size impact. The larger the anticipated size of the impact, the fewer MSAs we would have to sample in order to observe it. However, a model sized to be able to only detect large impacts runs the risk of not being able to draw conclusions if the size of the change is less than anticipated. The measure of interest used in estimating sample size requirements for the both the AMI and the CABG EPMs was wage-adjusted total episode spending. The data used for the wage-adjusted total episode spending is the 3-year data pull previously described that covers AMI and CABG episodes with admission dates from July 1, 2012, through December 31, 2014. For the purposes of the sample-size calculation, we aim to be able to reliably identify between a 2-percent and 3-percent reduction in wage-adjusted episode spending after 1 year of experience. We chose this range because those numbers represent the anticipated amount of the discount proposed to apply under various conditions of the AMI and CABG EPMs' implementation.

The next consideration in calculating the necessary sample size is the degree of certainty we will need for the statistical tests that will be performed. In selecting the right sample size, there are two types of errors that need to be considered: “false positives” and “false negatives.” A false positive occurs if a statistical test concludes that a model was successful (that is, saved money) when it in fact was not. A false negative occurs if a statistical test fails to find statistically-significant evidence that the model was successful, when it in fact was successful. In considering the minimum sample size needs of the AMI and CABG EPMs, a standard guideline in the statistical literature suggests calibrating statistical tests to generate no more than a 5-percent chance of a false positive and selecting the sample size to ensure no more than a 20-percent chance of a false negative. In contrast, the proposed sample size for this project was based on a 10-percent chance of a false positive and no more than a 30-percent chance of a false negative in order to minimize reduce sample size requirements to the greatest degree possible.

A third consideration in the sample-size calculation was the appropriate unit of selection and whether it is necessary to base the calculation on the number of MSAs, the number of hospitals, or the number of episodes. We are proposing to base the sample size calculation at the MSA level. The proposed EPMs are an example of what is known as a “nested comparative study.” Under a nested comparative study, assignment to an intervention or comparison arms of the study is based on membership in pre-existing, identifiable group where the groups are not formed at random, but rather through some physical, social, geographic, or other connection among their members. Because these groups are not formed at random, individual members of each group are likely to share important commonalities. In the context of the proposed EPMs spending and outcomes for patients cared for within a given MSA are relatively similar to one another due to such factors as the existence of common practice or referral patterns, the underlying health in the population, and the availability of providers in an area.

In statistical terms, these commonalities create a positive correlation (called an intra-class correlation) among hospitals or beneficiaries in the same MSA. Due to that intra-class correlation, the variability of any aggregate statistic—such as the estimated difference in outcomes between the intervention and comparison arms of the study—has two components—(1) variability attributable to variation among hospitals or beneficiaries in a given MSA; and (2) variability attributable to differences between MSAs. An accurate power analysis must account for both components of variability.

In determining the necessary sample size, we take into consideration the degree to which commonalities within MSAs exist and the number of independent beneficiaries and hospitals expected to be included in the EPM within each MSA. As part of this process, we empirically examined the number of beneficiaries, the number of hospitals, and the number of MSAs, as well as the level of correlation in episode payments between each level. Based on this empirical examination, we determined that the correlation was high enough that the degree of variability would be primarily driven by the number of MSAs in the model, indicating that the MSA is the appropriate unit of analysis for the power calculations.

Using the aforementioned assumptions, a power calculation for AMI was run which indicated that at 98 MSAs we would be able to reliably detect a 3-percent reduction in wage-adjusted episode spending after 1 year with a false-positive rate of 10 percent and a false-negative rate of between 20 percent and 40 percent. We are targeting a false-negative rate of 30 percent. The extent to which this rate can be lowered will depend on the ability of evaluation models to substantially reduce variation through risk adjustment and modeling. We believe it is prudent to choose a sample size where the targeted amount is in the middle of this expected band.

We separately assessed the sample-size needs associated with CABG episodes. At 98 MSAs, we anticipate being able to detect a 2.25-percent reduction in wage-adjusted episode expenditures after 1 year with a false-positive rate of 10 percent and a false-negative rate of between 20-40 percent. The effective number of MSAs where the CABG EPM will be tested will be reduced because approximately 6 percent of eligible MSAs had no CABG episodes in the reference year. However, our power calculations do not lead us to believe we need to increase the sample size based on this fact. The number of CABG MSAs can experience this reduction and maintain equivalent levels of power to the AMI episodes.

(3) Method of Selecting MSAs

As previously discussed, we are seeking to choose 98 MSAs from our pool of eligible MSAs through simple random selection. We propose to make the selection in the final rule using SAS Enterprise Guide 7.1 software to run a computer algorithm SAS Enterprise Guide 7.1 and the computer algorithm used to conduct selection represents an industry-standard for generating advanced analytics and provides a rigorous, standardized tool by which to satisfy the requirements of randomized selection. The key SAS commands employed include a “PROC SURVEYSELECT” statement coupled with the “METHOD=SRS” option used to specify simple random sampling as the sample selection method. A random number seed will be generated using the Start Printed Page 50829birthdate of the person executing the program.[37]

We seek comment on our proposal to implement the AMI and CABG models in the selected MSAs, some of which may overlap with MSAs where the CJR and SHFFT models also are being implemented.

C. Episode Definition for the EPMs

1. Background

Episode payment models incentivize improvement in the coordination and quality of care experienced by a Medicare beneficiary, as well as episode efficiency, by bundling payment for services furnished to the beneficiary for specific clinical conditions over a defined period of time. A key model design feature is the definition of the episodes included in the model. The definition of episodes has two significant dimensions—(1) a clinical dimension that describes which clinical conditions and associated services are included in the model; and (2) a time dimension that describes the beginning, middle, and end of the model.

2. Overview of Three Proposed Episode Payment Models

We propose three new EPMs—AMI, CABG, and SHFFT—that each begin with a hospitalization and extend 90 days after hospital discharge. The proposed AMI model generally includes beneficiaries discharged under an AMI MS-DRG (280-282), representing admission to an IPPS hospital for AMI that is treated with medical management, or an IPPS admission for a PCI MS-DRG (246-251) with an International Classification of Diseases (ICD)-Clinical Modification (CM) AMI diagnosis code describing an initial AMI diagnosis in the principal or a secondary diagnosis code position.

The proposed CABG model generally includes beneficiaries discharged under a CABG MS-DRG (231-236), representing an IPPS admission for this coronary revascularization procedure irrespective of AMI diagnosis. The proposed SHFFT model generally includes beneficiaries discharged under hip and femur procedures except major joint MS-DRG (480-482), representing an IPPS admission for a hip fixation procedure in the setting of a hip fracture.

One reason these particular episodes were chosen for the proposed EPMs is that the initiation of treatment for each of the three clinical conditions included in an episode occurs almost exclusively during a hospitalization, which we believe would minimize the possibility of shifting beneficiaries in or out of the EPM based on the site-of-service where treatment is initiated. The majority of evaluation and treatment for AMI is performed in the inpatient hospital setting, commonly beginning when beneficiaries present with symptoms to the emergency department of a hospital. Patients experiencing an AMI are almost uniformly admitted to the hospital for further evaluation and management.[38] Although PCIs can be performed and may be paid by Medicare in the hospital outpatient setting in addition to being performed during a hospitalization, the majority of patients experiencing an AMI who are candidates for procedural revascularization receive PCI procedures during the initial hospitalization for AMI where evaluation also occurs.[39] CABG procedures are furnished exclusively in the inpatient hospital setting. We note that all of the Current Procedural Terminology (CPT) codes that physicians report for CABG are listed on the hospital Outpatient Prospective Payment System (OPPS) inpatient-only list in Addendum E of the 2016 OPPS final rule with comment period that is posted on the CMS Web site.[40] The hip fixation procedures performed in the SHFFT model also are predominantly furnished in the inpatient hospital setting, and we further note that almost all of the CPT codes that describe these procedures also are on the OPPS inpatient-only list.

Hospitals' ability to identify EPM beneficiaries during the hospitalization that begins the episode (hereinafter the anchor hospitalization) also is an important consideration in developing episode payment models that, like the CJR model, rely upon MS-DRG assignment for IPPS claims following their submission in order to identify beneficiaries for model inclusion. This is especially important for medical management of conditions for which the predictability of the ultimate MS-DRG for the hospitalization is less certain than for surgical or procedural MS-DRGs. AMI represents a relative exception among medical conditions as it is associated with specific clinical and laboratory features that enable hospitals to identify beneficiaries with AMI during the anchor hospitalization whom would likely be included in an AMI model episode through their ultimate discharge under an AMI MS-DRG. We note that ICD-CM coding rules allow AMI diagnosis codes in both the primary and secondary position to map to AMI MS-DRGs.[41] In the case of procedural episodes such as CABG, SHFFT, and AMI model episodes for beneficiaries treated with PCI, the MS-DRG for the procedure performed would determine the ultimate MS-DRG assignment for the hospitalization unless additional surgeries higher in the MS-DRG hierarchy also are reported.[42] Therefore, we propose these three EPMs for clinical conditions where MS-DRG assignment is likely to be certain and known during the anchor hospitalization, even though treatment for AMI may involve only medical management. We believe hospitals participating in the proposed EPMs would be able to identify beneficiaries in EPM episodes through their AMI, CABG, and SHFFT episode MS-DRGs during the anchor hospitalization, allowing active coordination of EPM beneficiary care during and after hospitalization.

3. Clinical Dimensions of AMI, CABG, and SHFFT Model Episodes

As we stated in the CJR model Final Rule, we believe that a straightforward approach for hospitals and other providers to identify Medicare beneficiaries in these episode payment models would be important for the care redesign that is required for EPM success, as well as for operationalization of the proposed payment and other EPM policies (80 FR 73299). Therefore, as in the CJR model, we propose that an EPM episode would be initiated by an admission to an acute care hospital for an anchor hospitalization paid under EPM-specific MS-DRGs under the IPPS (80 FR 73300).Start Printed Page 50830

a. Definition of the Clinical Conditions Included in AMI, CABG, and SHFFT Model Episodes

(1) AMI (Medical Management and PCI) Model

We propose the AMI model to incentivize improvements in the coordination and quality of care, as well as episode efficiency, for beneficiaries treated for AMI with either medical management or coronary artery revascularization with PCI. We propose to define beneficiary inclusion in the AMI model by discharge under an AMI MS-DRG (280-282), representing those individuals admitted with AMI who receive medical therapy but no revascularization, and discharge under a PCI MS-DRG (246-251) with an ICD-10-CM diagnosis code of AMI on the IPPS claim for the anchor hospitalization in the principal or secondary diagnosis code position. We note that we would use AMI International Classification of Diseases, 9th revision clinical modification (ICD-9-CM) diagnosis codes to identify historical episodes for setting AMI model-episode benchmark prices in the early performance years of the AMI model. The Uniform Hospital Discharge Data Set (UHDDS) defines the principal diagnosis for hospitalization as “that condition established after study to be chiefly responsible for occasioning the admission of the patient to the hospital for care” and other (secondary) diagnoses as “all conditions that coexist at the time of admission, that develop subsequently, or that affect the treatment received and/or the length of stay. Diagnoses that relate to an earlier episode which have no bearing on the current hospital stay are to be excluded.” [43] We propose to include those beneficiaries discharged under PCI MS-DRGs with an AMI ICD-10-CM diagnosis code in the principal or secondary diagnosis code position to ensure that beneficiaries with an AMI that is not chiefly responsible for occasioning the hospitalization are included in the AMI model because the AMI itself is likely to substantially influence the hospitalization and post-discharge recovery (and be responsible for leading to the PCI) even if an AMI ICD-10-CM diagnosis code is reported in a secondary diagnosis code position. For example, a beneficiary receiving a PCI with an ICD-10-CM diagnosis code of pneumonia in the principal position and an AMI ICD-10-CM diagnosis code in a secondary position would be included in the AMI model, which would be appropriate because the course of the beneficiary's recovery and management during the AMI model episode would be primarily associated with the AMI and PCI. While pneumonia is typically an acute illness that may sometimes result in hospitalization, underlying chronic conditions may increase the likelihood that a beneficiary would be hospitalized for pneumonia, a condition that is more commonly treated on an outpatient basis. AMI in association with a hospitalization for pneumonia would represent a sentinel event for the beneficiary resulting from underlying CAD that signals a need for a heightened focus on medical management of CAD and other beneficiary risk factors for future cardiac events and that may themselves have increased the beneficiary's risk for pneumonia. Thus, care coordination and management in the 90 days post-hospital discharge for these beneficiaries would be focused on managing CAD and the beneficiary's cardiac function after the AMI.

We acknowledge that this proposal to identify beneficiaries included in the AMI model through a combination of MS-DRGs and AMI ICD-CM diagnosis codes represents a modification of the CJR model episode definition methodology. The CJR model defined episodes based on MS-DRGs alone, specifically MS-DRG 469 (Major joint replacement or reattachment of lower extremity with Major Complications or Comorbidities (MCC)) and MS-DRG 470 (Major joint replacement or reattachment of lower extremity without MCC), because the anchor hospitalization for the CJR model was defined by admission for a surgical procedure alone (80 FR 73280). However, the proposed AMI model is defined by admission for a medical condition that includes a range of treatment options, including medical treatment and PCI. Therefore, to identify beneficiaries admitted for AMI and treated with PCI requires ICD-CM diagnosis codes paired with MS-DRGs to identify the subset of PCI MS-DRG cases associated with AMI that would otherwise be excluded from an AMI model based solely on AMI MS-DRGs.

For the purposes of defining historical AMI model episodes, we propose to exclude beneficiaries discharged under PCI MS-DRGs with an AMI ICD-9-CM diagnosis code in the principal or secondary position if there is an intracardiac ICD-9-CM procedure code in any procedure code field. Intracardiac procedure codes do not represent PCI procedures indicated for the treatment of the coronary artery obstruction that results in AMI, but instead represent a group of procedures indicated for treating congenital cardiac malformations, cardiac valve disease, and cardiac arrhythmias. These intracardiac procedures are performed within the heart chambers rather than PCI procedures for AMI that are performed within the coronary blood vessels. To reflect this clinical distinction, the FY 2016 IPPS update removed intracardiac procedures from MS-DRGs 246-251 and assigned them to new MS-DRGs 273 and 274 (80 FR 49367). Therefore, to be consistent with our proposed definition of AMI model episodes that initiate with PCI MS-DRGs 246-251 (not with MS-DRGs 273 and 274) and an AMI ICD-9-CM diagnosis code in the principal or secondary position, we are proposing to define historical AMI model episodes for beneficiaries discharged under PCI MS-DRGS 246-251 as those that do not include the ICD-9-CM procedure codes in Table 2. These codes are also posted on the CMS Web site at https://innovation.cms.gov/​inititatives/​epm.

Table 2—Proposed ICD-9-CM Procedure Codes in any Position on the IPPS Claim for PCI MS-DRGS (246-251) That Do Not Define Historical AMI Model Episodes

ICD-9-CM Procedure codeICD-9-CM Procedure code description
35.52Repair of atrial septal defect with prosthesis, closed technique.
35.96Percutaneous balloon valvuloplasty.
35.97Percutaneous mitral valve repair with implant.
37.26Catheter based invasive electrophysiologic testing.
37.27Cardiac mapping.
37.34Excision or destruction of other lesion or tissue of heart, endovascular approach.
Start Printed Page 50831
37.36Excision, destruction, or exclusion of left atrial appendage.
37.90Insertion of left atrial appendage device.

In FY 2014, there were approximately 395,000 beneficiaries discharged from a short-term acute care hospitalization (excluding Maryland) with an AMI ICD-9-CM diagnosis code in the principal or secondary position on the IPPS claim. Of these beneficiaries, 58 percent were discharged under MS-DRGs that would initiate an AMI model episode, specifically an AMI MS-DRG (33 percent) and PCI MS-DRG (25 percent). Five percent of beneficiaries were discharged from CABG MS-DRGs and 3 percent were discharged from AMI MS-DRGs representing death during the hospitalization. The remaining 34 percent of beneficiaries with an AMI ICD-CM diagnosis code in the principal or secondary position were distributed across over approximately 300 other MS-DRGs, with the septicemia MS-DRGs accounting for 8 percent and the remainder accounting for 3 percent or less of beneficiaries with an AMI ICD-CM diagnosis code on the IPPS claim.[44] We note that the AMI ICD-9-CM diagnosis code was most commonly in a secondary position for discharges from these other MS-DRGs, likely representing beneficiaries hospitalized for another condition who experienced an AMI during that hospitalization. We note that CMS's AMI quality measures used in the Hospital Inpatient Quality Reporting (HIQR) Program are based on all beneficiaries discharged under any MS-DRG who have an AMI ICD-CM diagnosis code only in the principal position, reflecting the measures' focus on the most homogeneous beneficiary population with AMI as the condition responsible for occasioning the hospital admission. This is in contrast with our proposed use of an AMI ICD-10-CM diagnosis code in the principal or a secondary position for the AMI model in order to identify those beneficiaries receiving a PCI whose hospitalization and post-discharge recovery and management would primarily be associated with the PCI and AMI.

The proposed specifications for AMI episodes, including ICD-9-CM AMI diagnosis codes for historical episodes used to set the initial AMI model-episode benchmark prices and ICD-10-CM AMI diagnosis codes for the proposed performance years of the model, are displayed in Table 3. The ICD-9-CM intracardiac procedure codes used to exclude inpatient claims with PCI MS-DRGS 246-251 from anchoring AMI model historical episodes used to set initial AMI model-episode benchmark prices are displayed in Table 3.

Based on Medicare claims data for historical AMI episodes ending in CYs 2012-2014, the annual number of potentially eligible beneficiary discharges for the AMI model nationally was approximately 168,000.[45] This number is less than the approximately 229,000 discharges for beneficiaries with AMI discharged from AMI MS-DRGs 280-282 and PCI MS-DRGs 246-251 that could be expected to be included in the AMI model for several reasons. Discharges do not result in historical episodes when a beneficiary does not meet the beneficiary care inclusion criteria discussed in section III.C.4.a.(1) of this proposed rule; is not discharged alive from PCI MS-DRGS 246-251; is discharged from a transfer hospital during a chained anchor hospitalization; or is discharged from a readmission during an AMI model episode that does not initiate new model episodes.

The proposed list of ICD-9-CM and ICD-10-CM AMI diagnosis codes used to identify beneficiaries discharged under a PCI MS-DRG (MS-DRGs 246-251) in historical episodes and during the performance years of the model that will be included in the AMI model episodes are discussed in section III.C.4.a.(2) of this proposed rule. To make changes to this list as necessary based on annual ICD-10-CM coding changes or to address issues raised by the public throughout the EPM performance years, we propose implementing the following sub-regulatory process, which mirrors the sub-regulatory process as described in the CJR model final rule for updating hip fracture ICD-9-CM and ICD-10-CM diagnosis codes (80 FR 73340) and for updating the exclusions list (80 FR 73305 and 73315). We propose to use this process on an annual, or more frequent, basis to update the AMI ICD-10-CM diagnosis code list and to address issues raised by the public. As part of this process we propose the following standard when revising the list of ICD-10-CM diagnosis codes representing AMI: The ICD-10-CM diagnosis code is sufficiently specific that it represents an AMI. We propose to then post a list of potential AMI ICD-10-CM diagnosis codes to the CMS Web site at https://innovation.cms.gov/​inititatives/​epm to allow for public input on our planned application of these standards, and then adopt the AMI ICD-10-CM diagnosis code list with posting to the CMS Web site of the final AMI ICD-CM diagnosis code list after our consideration of the public input. We would provide sufficient time for public input based on the complexity of potential revisions under consideration, typically at least 30 days, and, while we would not respond to individual comments as would be required in a regulatory process, we could discuss the reasons for our decisions about changes in response to public input with interested stakeholders.

The proposals for identifying the beneficiaries included in the AMI model and the sub-regulatory process for updating the AMI ICD-10-CM diagnosis code list are included in § 512.100(c)(1) and (d), respectively. We seek comment on our proposals to identify beneficiaries included in the AMI model and the sub-regulatory process for updating the AMI ICD-10-CM diagnosis code list. The proposal to exclude inpatient claims with PCI MS-DRGS 246-251 from anchoring AMI model historical episodes used to set initial AMI model-episode benchmark prices when there is an ICD-9-CM intracardiac procedure code on the claim is included in § 512.100(d)(4). We seek comment on our proposal to exclude inpatient claims with PCI MS-DRGS 246-251 from anchoring AMI model historical episodes used to set initial AMI model-episode benchmark prices when there is an ICD-9-CM intracardiac procedure code on the claim.Start Printed Page 50832

(2) CABG Model

We propose the CABG model to incentivize improvements in the coordination and quality of care, as well as episode efficiency, for beneficiaries treated with CABG irrespective of AMI during the CABG hospitalization, thereby including beneficiaries undergoing elective CABG in the CABG model as well as beneficiaries with AMI who have a CABG during their initial AMI treatment. The CABG model is similar to the CJR model in that the anchor hospitalization is defined by admission for a surgical procedure, which is defined by the MS-DRGs for that procedure alone (80 FR 73280). All CABG procedures are performed in the inpatient hospital setting. Thus, we propose to include beneficiaries admitted and discharged from an anchor hospitalization paid under CABG MS-DRGs (231-236) under the IPPS in the CABG model. Based on Medicare claims data for historical CABG episodes beginning in CYs 2012-2014, the annual number of potentially eligible beneficiary discharges for the CABG model nationally was approximately 48,000.[46]

The proposal for identifying beneficiaries included in the CABG model is included in § 512.100(c)(2). We seek comment on our proposal to identify beneficiaries included in the CABG model.

(3) SHFFT (Excludes Lower Extremity Joint Replacement) Model

We propose the SHFFT model to incentivize improvements in the coordination and quality of care, as well as episode efficiency, for beneficiaries treated surgically for hip and femur fractures, other than hip arthroplasty. Together, the CJR and SHFFT models cover all surgical treatment options (that is, hip arthroplasty and fixation) for Medicare beneficiaries with hip fracture.

The SHFFT model is similar to the CJR model in that the anchor hospitalization is defined by admission for a surgical procedure, which is defined by the MS-DRGs for that procedure alone (80 FR 73280). Additionally, most SHFFT procedures are furnished in the inpatient hospital setting, consisting primarily of hip fixation procedures, with or without reduction of the fracture, as well as open and closed surgical approaches. Thus, we propose to include beneficiaries admitted and discharged from an anchor hospitalization paid under SHFFT MS-DRGs (480-482) under the IPPS in the SHFFT model. Based on Medicare claims data for historical SHFFT episodes beginning in CYs 20122014, the annual number of potentially eligible beneficiary discharges for the SHFFT model nationally was approximately 109,000.[47]

The proposal for identifying beneficiaries included in the SHFFT model is included in § 512.100(c)(3). We seek comment on our proposal to identify beneficiaries included in the SHFFT model.

b. Definition of the Related Services Included in EPM Episodes

The general principles for the proposed definition of related services are the same for the AMI, CABG, and SHFFT models, so we address them in a single discussion in this section. Like the CJR model, we are interested in testing inclusive AMI, CABG, and SHFFT model episodes to incentivize comprehensive, coordinated, patient-centered care for the beneficiary throughout the episode (80 FR 73303). Therefore, we propose to exclude Medicare items and services furnished during the EPM episodes only when unrelated to the EPM episode diagnosis and procedures based on clinical rationale that would result in standard exclusions from all of the episodes in a single EPM. Thus, we propose to include all items and services paid under Medicare Part A and Part B unless they fall under an exclusion because they are unrelated to the EPM episodes.

Also like the CJR model, we propose that the items and services ultimately included in the EPM episodes after the exclusions are applied are called related items and services, and that Medicare spending for related items and services be included in the historical data used to set EPM-episode benchmark prices and in the calculation of actual EPM episode payments that would be compared against the quality-adjusted target price to assess the performance of EPM participants (80 FR 73303 and 73315). Additionally, we propose that Medicare spending for unrelated items and services (excluded from the EPMs' episode definitions) would not be included in the historical data used to set EPM-episode benchmark prices or in the calculation of actual EPM episode payments. We propose that related items and services for EPM episodes would include the following items and services paid under Medicare Part A and Part B, after the EPM-specific exclusions are applied:

  • Physicians' services.
  • Inpatient hospital services.
  • Inpatient psychiatric facility (IPF) services.
  • Long-Term Care Hospital (LTCH) services.
  • Inpatient Rehabilitation Facility (IRF) services.
  • Skilled Nursing Facility (SNF) services.
  • Home Health Agency (HHA) services.
  • Hospital outpatient services.
  • Independent outpatient therapy services.
  • Clinical laboratory services.
  • Durable medical equipment.
  • Part B drugs.
  • Hospice.

We note that inpatient hospital services would include services paid through IPPS operating and capital payments. The AMI, CABG, and SHFFT model episodes also could include certain per-member-per-month model payments as discussed in section III.D.6.d. of this proposed rule. These proposed items and services for the EPMs are the same items and services included in CJR model episodes (80 FR 73303 and 73315).

Similar to the CJR model and for the reasons explained in the CJR Final Rule, we propose to exclude drugs that are paid outside of the MS-DRGs included in the EPM episode definitions, specifically hemophilia clotting factors, identified by CPT code, diagnosis code, and revenue center on IPPS claims, from the EPM episodes (80 FR 73303 and 73315). Hemophilia clotting factors, in contrast to other drugs that are administered during a hospitalization and paid through the MS-DRG, are paid separately by Medicare in recognition that clotting factors are costly and essential to appropriate care of certain beneficiaries. Therefore, we believe there are no EPM episode efficiencies to be gained in the variable use of these high cost drugs.

We also propose to exclude IPPS new technology add-on payments for drugs, technologies, and services from these EPM episodes, excluding them from both the actual historical episode data used to set EPM-episode benchmark prices and from actual EPM episode payments that are reconciled to the quality-adjusted target prices like the CJR model (80 FR 73303-73304 and 73315). This would apply to both the anchor hospitalization and any related Start Printed Page 50833readmissions during the EPM episodes. New technology add-on payments are made separately and in addition to the MS-DRG payment under the IPPS for specific new drugs, technologies, and services that substantially improve the diagnosis or treatment of Medicare beneficiaries and would be inadequately paid under the MS-DRG system. We believe it would not be appropriate for the EPM to potentially diminish beneficiaries' access to new technologies or to burden hospitals who choose to use these new drugs, technologies, or services with concern about these payments counting toward EPM participants' actual EPM episode payment. Additionally, new drugs, technologies, or services approved for the add-on payments vary unpredictably over time in their application to specific clinical conditions.

Finally, we propose to exclude OPPS transitional pass-through payments for medical devices as defined in § 419.66 from the EPM episodes because, through the established OPPS review process, we have determined that these technologies have a substantial cost but also lead to substantial clinical improvement for Medicare beneficiaries. This proposal also is consistent with the CJR model final exclusions policy (80 FR 73308 and 73315).

We propose to follow the same general principles in determining other proposed excluded Part A and Part B services from the EPM episodes that we use in the CJR model in order to promote coordinated, high-quality, patient-centered care (80 FR 73304). These include identifying excluded (unrelated) services rather than included (related) services based on clinical review. We would operationalize these principles for the new EPMs, as we do for the CJR model, by excluding unrelated inpatient hospital admissions during the EPM episode by identifying MS-DRGs for exclusion on an EPM-specific basis (80 FR 73304 through 73312 and 73315). We would further exclude unrelated Part B services during the EPM episode based on the diagnosis code on the claim by identifying categories of ICD-CM codes for exclusion (identified by code ranges) on an EPM-specific basis. ICD-9-CM diagnosis code exclusions would apply to historical episodes used to construct EPM-episode benchmark prices, while ICD-10-CM diagnosis code exclusions would apply to EPM episodes during the EPMs' performance years. We propose to identify unrelated Part B services and readmissions based on the BPCI Model 2 Part B exclusions lists that apply to the anchor MS-DRG that initiates the EPM episode, or to the price MS-DRG if it is different than the anchor MS-DRG as described further in section III.D.4.b.(2)(a) of this proposed rule. This proposal is consistent with our use of the BPCI Model 2 LEJR ICD-9-CM, ICD-10-CM, and MS-DRG exclusions lists in the CJR model (80 FR 73304 and 73315).

The BPCI episode-specific exclusions lists were initially developed more than 3 years ago for BPCI through a collaborative effort of CMS staff, including physicians from medical and surgical specialties, coding experts, claims processing experts, and health services researchers. The lists have been shared with thousands of entities and individuals participating in episodes in one or more phases of BPCI, and have undergone refinement in response to stakeholder input about specific diagnoses for exclusion, resulting in only minimal changes over the last 3 years. Thus, the BPCI exclusions lists have been vetted broadly in the health care community; refined based on input from a wide variety of providers, researchers and other stakeholders; and successfully operationalized in the BPCI models. We propose their use in the AMI, CABG, and SHFFT models based on our confidence related to our several years of experience that these definitions are reasonable and workable for AMI, CABG, and SHFFT model episodes, for both providers and CMS, and based on our rulemaking for the CJR model. We note that the BPCI Model 2 exclusions lists for the 48 clinical conditions being tested in the BPCI models include lists that apply to every MS-DRG that could be an anchor MS-DRG (or price MS-DRG, if applicable) for the proposed AMI, CABG, and SHFFT model episodes.

Similar to the CJR model, we propose to include in EPM episodes all Part A services furnished post-hospital discharge during the EPM episode, as these services are typically intended to be comprehensive in nature (80 FR 73304 and 73315). We specifically propose to exclude unrelated hospital readmissions for MS-DRGs that group to the following categories of diagnoses: Oncology, trauma medical admissions, surgery for chronic conditions unrelated to a condition likely to have been affected by care furnished during the EPM episode, and surgery for acute conditions unrelated to a condition resulting from or likely to have been affected by care during the EPM episode. The rationale for these exclusions is the same as the rationale for their exclusion in the CJR model (80 FR 73304).

Specifically with respect to Part B services, similar to the CJR model, we propose to exclude acute disease diagnoses unrelated to a condition resulting from or likely to have been affected by care during the EPM episode, and certain chronic disease diagnoses, as specified by CMS on a diagnosis-by-diagnosis basis, depending on whether the condition was likely to have been affected by care during the EPM episode or whether substantial services were likely to be provided for the chronic condition during the EPM episode (80 FR 73305 and 73315). Thus, we would include all Part B services with principal diagnosis codes on the associated Part B claims that are directly related (clinically and per coding conventions) to EPM episodes, claims for diagnoses that are related to the quality and safety of care furnished during EPM episodes, and claims for services for diagnoses that are related to preexisting chronic conditions such as diabetes, which may be affected by care furnished during EPM episodes.

In general, the anchor MS-DRG that initiates the AMI, CABG, or SHFFT episode would determine the exclusions list that applies to the EPM episode. For example, AMI model episodes may have different exclusions lists applied based on whether the AMI model episode is initiated by admission to the participant hospital that results in discharge from an AMI anchor MS-DRG or a PCI anchor MS-DRG with AMI ICD-10-CM diagnosis code. If a price MS-DRG applies to the AMI model episode that includes a chained anchor hospitalization as described in section III.D.4.b.(2)(a) of this proposed rule, the exclusions list that applies to the price MS-DRG would apply to the AMI model episode. Complete lists of proposed excluded MS-DRGs for readmissions and proposed excluded ICD-CM codes for Part B services furnished during EPM episodes after EPM beneficiary discharge from an anchor or chained anchor hospitalization in the AMI, CABG, and SHFFT models are posted on the CMS Web site at https://innovation.cms.gov/​initiatives/​epm.

Like the CJR model policy, we propose that these exclusion lists would be updated by sub-regulatory guidance on an annual basis, at a minimum, to reflect annual changes to ICD-10-CM coding and annual changes to the MS-DRGs under the IPPS, as well as to address any other issues that are brought to our attention throughout the course of the EPMs' performance period (80 FR 73304 through 73305 and 73315). The standards for this updating process reflect the aforementioned general principles for determining excluded services. That is, we propose to not Start Printed Page 50834exclude any items or services that are directly related to the EPM episode diagnosis or procedure (for example, a subsequent admission for heart failure or repeat revascularization) or the quality or safety of care (for example, sternal wound infection following CABG); or to chronic conditions that may be affected by the EPM diagnosis or procedure and the post-discharge care (for example, diabetes). We propose to exclude items and services for chronic conditions that are generally not affected by the EPM diagnosis or procedure and the post-discharge care (for example, prostate removal for cancer), and for acute clinical conditions not arising from existing EPM episode-related chronic clinical conditions or complications from the EPM episode (for example, appendectomy).

Similar to the CJR model, we propose that the potential revised exclusions, which could include additions to or deletions from the exclusions lists, would be posted to the CMS Web site to allow for public input (80 FR 73305 and 73315). Through the process for public input on potential revised exclusions and then posting of the final revised exclusions, we propose to provide information to the public about when the revisions would take effect and to which episodes they would apply.

The proposal for included services for an EPM is included in § 512.210(a). The proposal for excluded services from the EPM episode is included in § 512.210(b). The proposal for updating the lists of excluded services for EPMs is included in § 512.210(c). We seek comment on our proposals for included and excluded services for the AMI, CABG, and SHFFT models and updating the lists of excluded services.

4. EPM Episodes

a. Beneficiary Care Inclusion Criteria and Beginning of EPM Episodes

(1) General Beneficiary Care Inclusion Criteria

Because of the clinical variability leading up to these EPM episodes and the challenge of identifying unrelated services given the multiple chronic conditions experienced by many EPM beneficiaries, we propose to follow the CJR model precedent and not begin an EPM episode prior to the anchor hospitalization (80 FR 73315 and 73318). We propose that all services that are already included in the IPPS payment based on established Medicare policies (for example, 3-day payment window payment policies) would be included in these EPM episodes, and that the defined population of Medicare beneficiaries whose care would be included in the EPMs would meet all of the following criteria on admission to the anchor or chained anchor hospitalization:

  • Enrolled in Medicare Part A and Part B.
  • Eligible for Medicare not on the basis of end-stage renal disease.
  • Not enrolled in any managed care plan (for example, Medicare Advantage, Health Care Prepayment Plans, cost-based health maintenance organizations).
  • Not covered under a United Mine Workers of America health plan, which provides health care benefits for retired mine workers.
  • Have Medicare as their primary payer.
  • Not aligned to an ACO in the Next Generation ACO model or an ACO in a track of the Comprehensive ESRD Care Initiative incorporating downside risk for financial losses.
  • Not under the care of an attending or operating physician, as designated on the inpatient hospital claim, who is a member of a physician group practice that initiates BPCI Model 2 episodes at the EPM participant for the MS-DRG that would be the anchor MS-DRG under the EPM.
  • Not already in any BPCI model episode.
  • Not already in an AMI, SHFFT, CABG or CJR model episode with an episode definition that does not exclude the MS-DRG that would be the anchor MS-DRG under the applicable EPM.

For a discussion of our proposal to exclude certain ACO-aligned beneficiaries from EPM episodes, we refer to section III.D.6.c.(3) of this proposed rule. For a discussion of our proposals for addressing potential overlap of beneficiaries in episode payment models that are relevant to these last two criteria, we refer to sections III.D.6.c.(1) and (2) of this proposed rule.

The proposal for beneficiary care inclusion policies is included in § 512.230. We seek comment on our proposal of beneficiary care inclusion policies.

(2) Beginning AMI Model Episodes

We propose that, as long as the beneficiary meets the general beneficiary care inclusion criteria, then an AMI model episode would begin with admission of a Medicare beneficiary to an IPPS hospital for the following MS-DRGs, where the specific MS-DRG is called the anchor MS-DRG for the episode:

  • AMI MS-DRGs—

++ 280 (Acute myocardial infarction, discharged alive with MCC);

++ 281 (Acute myocardial infarction, discharged alive with CC); and

++ 282 (Acute myocardial infarction, discharged alive without CC/MCC).

  • PCI MS-DRGs, when the claim includes an AMI ICD-10-CM diagnosis code in the principal or secondary position on the IPPS claim as specified in Table 3—

++ 246 (Percutaneous cardiovascular procedures with drug-eluting stent with MCC or 4+ vessels/stents);

++ 247 (Percutaneous cardiovascular procedures with drug-eluting stent without MCC);

++ 248 (Percutaneous cardiovascular procedures with non-drug-eluting stent with MCC or 4+ vessels/stents);

++ 249 (Percutaneous cardiovascular procedures with non-drug-eluting stent without MCC);

++ 250 (Percutaneous cardiovascular procedures without coronary artery stent with MCC); and

++ 251 (Percutaneous cardiovascular procedures without coronary artery stent without MCC).

Table 3 displays the ICD-9-CM codes that we propose to use to identify historical AMI episodes for beneficiaries discharged from PCI MS-DRGs, as well as the ICD-10-CM diagnosis codes that would be used to identify AMI model episodes for beneficiaries discharged from PCI MS-DRGs throughout the duration of the AMI model. The proposed sub-regulatory process for updating this AMI ICD-10-CM diagnosis code list is described previously in section III.C.3.a.(1) of this proposed rule.

We first identified the ICD-9-CM diagnosis codes for the initial AMI episode-of-care that were historically used to report care for a newly diagnosed AMI patient admitted to the hospital. These codes all have a fifth digit of “1” and were applicable until the patient was discharged from acute medical care, including for any transfers to and from other acute care facilities that occurred. These AMI ICD-9-CM diagnosis codes would be used to identify historical AMI episodes for developing AMI model-episode benchmark prices for anchor PCI MS-DRGs. We propose to cross-walk the ICD-9-CM diagnosis codes for the initial AMI episode-of-care to the ICD-10-CM diagnosis codes that would be reported for similar beneficiaries during the AMI model performance years. The proposed crosswalk in Table 3 is consistent with the crosswalk CMS posted for public comment regarding ICD-9-CM to ICD-10-CM diagnosis Start Printed Page 50835codes used for HIQR Program measures, including AMI quality measures.[48]

Table 3—Proposed ICD-9-CM and ICD-10-CM AMI Diagnosis Codes in the Principal or Secondary Position on the IPPS Claim for PCI MS-DRGS (246-251) That Initiate AMI Model Episodes

ICD-9-CM Diagnosis codeICD-9-CM DescriptionICD-10-CM Diagnosis codeICD-10-CM Description
410.01Acute myocardial infarction of anterolateral wall, initial episode of care121.09ST elevation (STEMI) myocardial infarction involving other coronary artery of anterior wall.
122.0Subsequent ST elevation (STEMI) myocardial infarction of anterior wall.
410.11Acute myocardial infarction of other anterior wall, initial episode of care121.01ST elevation (STEMI) myocardial infarction involving left main coronary artery.
121.02ST elevation (STEMI) myocardial infarction involving left anterior descending coronary artery.
121.09ST elevation (STEMI) myocardial infarction involving other coronary artery of anterior wall.
122.0Subsequent ST elevation (STEMI) myocardial infarction of anterior wall.
410.21Acute myocardial infarction of inferolateral wall, initial episode of care121.10ST elevation (STEMI) myocardial infarction involving other coronary artery of inferior wall.
122.1Subsequent ST elevation (STEMI) myocardial infarction of inferior wall.
410.31Acute myocardial infarction of inferoposterior wall, initial episode of care121.11ST elevation (STEMI) myocardial infarction involving right coronary artery.
122.1Subsequent ST elevation (STEMI) myocardial infarction of inferior wall.
410.41Acute myocardial infarction of other inferior wall, initial episode of care121.19ST elevation (STEMI) myocardial infarction involving other coronary artery of inferior wall.
122.1Subsequent ST elevation (STEMI) myocardial infarction of inferior wall.
410.51Acute myocardial infarction of other lateral wall, initial episode of care121.29ST elevation (STEMI) myocardial infarction involving other sites.
122.8Subsequent ST elevation (STEMI) myocardial infarction of other sites.
410.61True posterior wall infarction, initial episode of care121.29ST elevation (STEMI) myocardial infarction involving other sites.
122.8Subsequent ST elevation (STEMI) myocardial infarction of other sites.
410.71Subendocardial infarction, initial episode of care121.4Non-ST elevation (NSTEMI) myocardial infarction.
122.2Subsequent non-ST elevation (NSTEMI) myocardial infarction.
410.81Acute myocardial infarction of other specified sites, initial episode of care121.21ST elevation (STEMI) myocardial infarction involving left circumflex coronary artery.
121.29ST elevation (STEMI) myocardial infarction involving other sites.
122.8Subsequent ST elevation (STEMI) myocardial infarction of other sites.
410.91Acute myocardial infarction of unspecified site, initial episode of care121.3ST elevation (STEMI) myocardial infarction of unspecified site.
122.9Subsequent ST elevation (STEMI) myocardial infarction of unspecified site.

The proposal for beginning AMI model episodes is included in § 512.240(a)(1). We seek comment on our proposal to begin AMI model episodes.

(3) Beginning CABG Model Episodes

We propose that, as long as a beneficiary meets the general beneficiary care inclusion criteria, a CABG model episode would begin with the admission of a Medicare beneficiary to an IPPS hospital for a CABG that is paid under the following CABG MS-DRGs and the specific MS-DRG is called the anchor MS-DRG for the episode:

  • 231 (Coronary bypass with percutaneous transluminal coronary angioplasty (PTCA) with MCC).
  • 232 (Coronary bypass with PTCA without MCC).
  • 233 (Coronary bypass with cardiac catheterization with MCC).
  • 234 (Coronary bypass with cardiac catheterization without MCC).
  • 235 (Coronary bypass without cardiac catheterization with MCC).
  • 236 (Coronary bypass without cardiac catheterization without MCC).

The proposal for beginning CABG episodes is included in § 512.240(b)(1). We seek comment on our proposal to begin CABG model episodes.Start Printed Page 50836

(4) Beginning SHFFT Episodes

We propose that as long as a beneficiary meets the general inclusion criteria, a SHFFT model episode would begin with the admission of a Medicare beneficiary to an IPPS hospital for surgical treatment of hip or femur fracture (other than joint replacement) that is paid under the following SHFFT MS-DRGs and where the specific MS-DRG is called the anchor MS-DRG for the episode:

  • 480 (Hip and femur procedures except major joint with MCC).
  • 481 (Hip and femur procedures except major joint with complication or comorbidity (CC).
  • 482 (Hip and femur procedures except major joint without CC or MCC).

The proposal for beginning SHFFT model episodes is included in § 512.240(c)(1). We seek comment on our proposal to begin SHFFT model episodes.

(5) Special Policies for Hospital Transfers of Beneficiaries With AMI

The asymmetric distribution of cardiac care across hospitals makes transfer, either from an inpatient admission or from the emergency department (without inpatient admission) of one hospital to another, a common consideration in the treatment course for beneficiaries with an initial diagnosis of AMI. Therefore, transfer for cardiac care is an important consideration for the AMI and CABG models.

The availability of revascularization and intensive cardiac care are particularly important considerations in the transfer of beneficiaries with an AMI. A substantial portion of hospitals do not have revascularization capability (that is, a cardiac catheterization lab for PCI or cardiothoracic surgeons who can perform CABG) or cardiovascular intensive care units (CVICU) and, therefore, must transfer beneficiaries to provide access to these services. In the PCI and CABG examples, the discharge from the transfer hospital that accepted the beneficiary would result in discharge under the MS-DRGs for PCI (246-251) or CABG (231-236). For the CVICU example, the transfer hospital's discharge MS-DRG would be AMI (280-282). There is evidence of the asymmetric distribution of cardiac care in the 2014 IPPS and critical access hospital claims data: while 4,332 hospitals submitted at least one claim for an AMI MS-DRG, only 1,755 (41 percent) and 1,156 (27 percent) of these hospitals filed at least one claim for PCI or CABG MS-DRGs, respectively.[49]

The potential transfer scenarios are best illustrated by the care pathways experienced by beneficiaries with AMI. These beneficiaries typically present to a hospital's emergency department where the evaluation identifies the AMI diagnosis and determines the initial indicated treatments. Depending on the beneficiary's clinical needs and the hospital's treatment capacity, the beneficiary could be—

  • Admitted to the initial treating hospital, with no transfer to another hospital during the initial hospitalization for AMI. We refer to this scenario as no transfer;
  • Admitted to the initial treating hospital and later transferred to a transfer hospital. We refer to this scenario as inpatient-to-inpatient transfer and the transfer hospital as an i-i transfer hospital; or
  • Transferred from the initial treating hospital to a transfer hospital without admission to the initial treating hospital. We refer to this scenario as outpatient-to-inpatient transfer and the transfer hospital as an o-i transfer hospital.

Our proposals and alternatives considered for these scenarios are described in detail in this section. In our proposals for AMI or CABG model episodes for initial AMI care, our overarching policy is that every AMI or CABG model episode would begin at the first AMI or CABG model participant to which the beneficiary is admitted for an AMI MS-DRG, PCI MS-DRG with an AMI ICD-CM diagnosis code, or CABG MS-DRG. The AMI or CABG model participant where the episode begins would then be financially responsible for the AMI or CABG model episode unless the episode is canceled.

Based on our analysis of Medicare claims data, about 75 percent of historical AMI episodes and CABG episodes for beneficiaries with AMI begin through the emergency department of the hospital where the anchor hospitalization for the AMI or CABG model episode would occur. In another 18 percent of historical AMI episodes and CABG episodes for beneficiaries with AMI, the anchor hospitalization occurs at a transfer hospital following an emergency department visit at another hospital without admission to that hospital for an MS-DRG that would initiate an AMI or CABG model episode.[50]

In each of these scenarios, policies to determine which episode type applies, the beginning of the episode, and the specific hospital with financial responsibility for the episode must be determined (for example, AMI or CABG, if CABG is provided as an initial treatment in an outpatient-to-inpatient or inpatient-to-inpatient scenario). In this section, we discuss each of the scenarios in detail and provide a summary of the scenarios in Table 4.

In the no transfer scenario, the episode would begin upon admission to an AMI or CABG model participant under circumstances that meet the criteria discussed in sections III.C.4.a.(1) and (2) or (3) of this proposed rule, and the AMI or CABG model episode that applies would be determined by the specific MS-DRG for the anchor hospitalization. Financial responsibility for the episode would be attributed to the sole treating hospital involved in the initial AMI care. Under this proposal, the treating hospital's quality measure performance would determine the effective discount factor to be applied to the AMI or CABG model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule.

The inpatient-to-inpatient transfer scenario has several potential outcomes. If the beneficiary initially presents for AMI care to a hospital that is not an AMI model participant and is admitted and then transferred to an i-i transfer hospital that is an AMI or CABG model participant, the episode would first initiate at the i-i transfer hospital and, therefore, the i-i transfer hospital would be financially responsible for the AMI or CABG model episode. The i-i transfer hospital's quality measure performance would determine the effective discount factor to be applied to the AMI or CABG model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule.

Conversely, if a beneficiary initially presents for AMI care to an AMI model participant and is admitted and then transferred to an i-i transfer hospital (hereinafter a chained anchor hospitalization) and the i-i transfer hospital is not an AMI or CABG model participant, the episode would initiate at the initial treating hospital and would only be canceled for beneficiaries discharged from the i-i transfer hospital Start Printed Page 50837under MS-DRGs that are not anchor MS-DRGs for AMI or CABG model episodes is discussed in section III.C.4.b. of this proposed rule. The initial treating hospital's quality measure performance would determine the effective discount factor to be applied to the AMI or CABG model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule. We also refer to section III.D.4.b.(2)(a) of this proposed rule for further discussion of price MS-DRGs that may differ from the anchor MS-DRG in AMI model episodes that include a chained anchor hospitalization, in order to provide pricing adjustments for episodes where the initial treating hospital is responsible for the AMI model episode.

Inpatient-to-inpatient transfers between AMI and CABG model participant hospitals are further considered in this section and specifically include beneficiaries experiencing an AMI who are transferred for revascularization (that is, PCI or CABG) or a higher level of medical AMI care. We note that of all beneficiaries experiencing an AMI in historical episodes, about half received no revascularization (PCI or CABG) during the anchor hospitalization or the 90-day post-hospital discharge period, about 40 percent received a PCI, and less than 10 percent had CABG surgery.[51] Moreover, three-quarters of CABG procedures and over 90 percent of PCIs for beneficiaries experiencing an AMI occurred at the hospital that first admitted the beneficiary for an inpatient hospitalization.[52]

However, given the asymmetric distribution of cardiac care capacity there will be beneficiaries who initiate an AMI model episode by admission to an initial treating hospital but then require transfer to an i-i transfer hospital for additional treatment during the AMI model episode, resulting in a chained anchor hospitalization. For historical AMI episodes ending in CY 2014, only about 12 percent of beneficiaries who would have initiated an AMI model episode through admission and assignment to an AMI MS-DRG at the initial treating hospital were transferred to an i-i transfer hospital, with 30 percent and 20 percent receiving PCI or CABG, respectively, at the i-i transfer hospital. Another 20 percent were discharged from the i-i transfer hospital in the chained anchor hospitalization under an AMI MS-DRG. The remaining 30 percent of beneficiaries were discharged from the i-i transfer hospital in the chained anchor hospitalization under other MS-DRGs that would not have initiated AMI or CABG model episodes, including cardiac valve surgery, septicemia, and renal failure. From the perspective of hospital capacity and transfer patterns, most hospitals transferred less than 10 percent of beneficiaries initiating a historical AMI episode under an AMI MS-DRG at the first admitting hospital, and only a handful of hospitals transferred the majority of their patients in this scenario.[53] This small number of hospitals that transferred the majority of their patients includes a range of urban and rural hospitals with 50 to 250 beds.

The need to transfer a beneficiary in an AMI model episode during the anchor hospitalization for appropriate care that results in a chained anchor hospitalization where the hospitals are both AMI or CABG model participants raises considerations about whether attribution of the AMI model episode should be to the first treating hospital that admitted the beneficiary or the i-i transfer hospital, as well as considerations about the specific model (AMI or CABG) for attribution of the episode in some circumstances. For example, if the first treating hospital initiates an AMI model episode by admitting a beneficiary and then transfers the beneficiary to another hospital where the beneficiary is treated and ultimately discharged from acute care, ending the chained anchor hospitalization under a CABG MS-DRG, then we need to determine whether the beneficiary would be included in the AMI or CABG model, which hospital assumes financial responsibility for the beneficiary's episode, and under what circumstances, if any, would the AMI model episode be canceled if a transfer occurs.

In considering the model episode that includes the beneficiary's care and accountability for the beneficiary in inpatient-to-inpatient transfer scenarios between AMI and CABG model participant hospitals that result in a chained anchor hospitalization for AMI, several factors are relevant, including the timing of final discharge disposition of the beneficiary, including to post-acute care; the location of the post-acute care; the identity and location of the physician who is most responsible for managing the beneficiary's care after discharge; and consistency across other CMS transfer policies. We note that while 64 percent of CABG beneficiaries in historical episodes received post-acute care services following discharge from the anchor hospitalization (most commonly home health services—43 percent received home health services only and 13 percent a combination of home health and SNF services), only 36 percent of historical AMI beneficiaries received post-acute services.[54] Of further relevance for beneficiaries with an AMI diagnosis is that significant follow up care is usually performed by cardiologists who manage the patient's underlying cardiovascular disease, rather than the interventional cardiologist or cardiothoracic surgeon that perform the revascularization procedure. PCI procedures, billed by interventional cardiologists, have a 0-day global period, reflecting that follow up care is not typically furnished by interventional cardiologists. We further note that patients in commercial programs that require travel to regional centers of excellence for CABG generally only stay in the remote location away from the patient's home for a week or so post-hospital discharge. We expect that beneficiaries hospitalized for treatment of AMI, even if they are transferred to a revascularization hospital resulting in a chained anchor hospitalization, would receive most follow up care in their local communities, a view that was supported by many commenters on the CJR model proposed rule who asserted that many patients requiring post-acute care prefer to return to their home communities for that care following hospital discharge (80 FR 23457). Finally, consistency across other CMS program policies when a beneficiary with an AMI experiences an inpatient-to-inpatient transfer is relevant to developing policies for the proposed AMI and CABG models. Specifically, we note that the Hospital-Level, Risk-Standardized Payment Associated with a 30-Day Episode of Care for AMI (NQF #2431) measure used in the hospital value-based purchasing (HVBP) Program attributes payments for transferred beneficiaries to the hospital that Start Printed Page 50838admitted the patient for the initial AMI hospitalization.[55]

Based on these considerations, we propose that once an AMI model episode is initiated at an AMI model participant hospital through an inpatient hospitalization, the AMI model episode would continue under the financial responsibility of that participant hospital, regardless of whether the beneficiary is transferred to another AMI or CABG model participant hospital for further medical management of AMI, or for a PCI or CABG during a chained anchor hospitalization. Under this proposal, the initial treating hospital's quality measure performance would determine the effective discount factor to be applied to the AMI model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule. Our proposal to cancel AMI model episodes for beneficiaries discharged from the i-i transfer hospital under MS-DRGs that are not anchor MS-DRGs for AMI or CABG model episodes is discussed in section III.C.4.b. of this proposed rule. We also refer to section III.D.4.b.(2)(a) of this proposed rule for further discussion of price MS-DRGs that may differ from the anchor MS-DRG in AMI model episodes that include a chained anchor hospitalization, in order to provide pricing adjustments for episodes where the initial treating hospital is responsible for the AMI model episode.

We note that we do not propose to cancel the AMI model episode even if the transfer and admission to the i-i transfer hospital would otherwise initiate a CABG model episode at the i-i transfer hospital. We believe that once the AMI model episode has been initiated, all related care during the episode (including hospital care for transfers and related readmissions for CABG) should be fully attributed to the AMI model episode in the manner described in this section for the episode and that the first hospital that initiated the AMI model episode should be financially responsible for the AMI episode. Therefore, we do not propose to cancel the AMI model episode if a CABG is performed during a chained anchor hospitalization, nor do we propose that a beneficiary could simultaneously be in an AMI and CABG model episode for overlapping periods of time due to the different MS-DRGs that apply during the chained anchor hospitalization. Instead, we would make an AMI model episode pricing adjustment for these circumstances by paying the AMI model participant based on a price MS-DRG that is different from the anchor MS-DRG to reflect Medicare payment for the CABG as discussed in section III.D.4.b.(2)(a) of this proposed rule.

We considered several alternatives to our proposal for AMI model episode attribution for inpatient-to-inpatient transfer scenario where both hospitals are AMI or CABG model participants. First, we considered canceling the AMI model episode initiated at the initial treating hospital when a transfer occurs, and basing any AMI or CABG model episode initiation on the MS-DRG for the final i-i transfer hospital admission in the chained anchor hospitalization as long as that latter hospital is an AMI or CABG model participant. This would place financial responsibility for the episode on the i-i transfer hospital if the beneficiary goes on to be discharged from acute care at that hospital. Attributing episodes under this alternative policy would assign beneficiaries to the final i-i transfer hospital for the AMI or CABG model episode based on the model episode definitions in sections III.C.4.a.(2) and (3) of this proposed rule. That is, if the beneficiary is discharged from the final admission in the chained anchor hospitalization under an AMI MS-DRG or a PCI MS-DRG, then the AMI model episode initiated at the initial treating hospital would be canceled and the i-i transfer hospital accepting the beneficiary on referral would initiate an AMI model episode. Similarly, if the beneficiary is discharged from the final admission in the chained anchor hospitalization under a CABG MS-DRG, then the AMI model episode initiated at the first hospital would be canceled and the i-i transfer hospital accepting the beneficiary on referral would initiate a CABG model episode. Under this alternative, the i-i transfer hospital's quality measure performance would determine the effective discount factor to be applied to the AMI or CABG model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule. However, we do not propose this alternative because we believe that post-acute care and care management following hospital discharge are more likely to be effectively provided near the beneficiary's home community, rather than near the i-i transfer hospital accepting the beneficiary upon referral.

Second, we considered proposing an episode hierarchy such that, during a chained anchor hospitalization, the most resource-intensive MS-DRG during the whole chained anchor hospitalization would determine the model episode and the financially responsible hospital for the episode. For example, if we establish CABG, PCI, and AMI MS-DRGs in descending order of inpatient hospital resource-intensity, we would initiate a model episode based on the most resource-intensive MS-DRG during the chained anchor hospitalization and attribute the model episode to the hospital discharging the beneficiary under that MS-DRG. Under this scenario, either the initial treating or i-i transfer hospital's quality measure performance would determine the effective discount factor to be applied to the AMI or CABG model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule, depending on the specific hospital discharging the beneficiary under the most resource-intensive MS-DRG during the chained anchor hospitalization. However, we do not propose this alternative because we believe, like the first alternative we considered, this could frequently lead to episode responsibility being attributed to the i-i transfer hospital when the local hospital first caring for the beneficiary with AMI may be better positioned to coordinate care in the beneficiary's home community.

Thus, our proposal would place responsibility for care during the 90-day post-hospital discharge period in the AMI model episode on the AMI model participant hospital to which the beneficiary initially presented for AMI care and was admitted, rather than on the i-i transfer hospital to which the beneficiary was transferred after initiating the AMI model episode. Given the broad episode definition of AMI model episodes, we believe that the post-discharge care required following hospitalization that includes CABG, PCI, or medical management is best coordinated and managed by the hospital that originally admitted the beneficiary for the AMI. Such post-discharge care could include follow up for adherence to cardiac rehabilitation referral and management of the beneficiary's underlying CAD and comorbidities. Even in the case of the more common surgical complications of CABG, such as wound infection, the beneficiary commonly would be admitted to the local hospital for treatment.

We further propose that, as discussed in section III.I.3 of this proposed rule, hospitals may be collaborators in the AMI, CABG, and SHFFT models in order to increase the financial alignment of hospitals and other EPM collaborators with EPM participants that are Start Printed Page 50839financially responsible for EPM episodes. Therefore, we expect that community hospital participants in the AMI model would be able to enter into collaboration agreements with i-i transfer hospitals accepting AMI model beneficiaries on referral to allow sharing of episode reconciliation payments or repayment responsibility with the i-i transfer hospitals if those hospitals play a significant role in care redesign of AMI or CABG care pathways or management of beneficiaries throughout AMI or CABG model episodes, including during the 90 days post-hospital discharge. We expect that community hospitals would need to coordinate closely with i-i transfer hospitals accepting AMI model beneficiaries on referral as the beneficiaries in AMI model episodes are discharged from those hospitals, in order to improve the quality and efficiency of AMI model episodes. This coordination could potentially be enhanced if i-i transfer hospitals are AMI model collaborators with financial incentives that are aligned with those of the AMI model participants through sharing arrangements.

The proposal for AMI model episode attribution in circumstances that involve inpatient-to-inpatient transfers of beneficiaries with AMI is included in § 512.240(a)(2). We seek comment on our proposal for AMI model episode attribution in circumstances that involve inpatient-to-inpatient transfers of beneficiaries with AMI, including comment on the alternatives considered.

In the outpatient-to-inpatient transfer scenario where a beneficiary with AMI is transferred from the emergency department of the initial treating hospital without admission to that hospital as an inpatient to an o-i transfer hospital for admission, we propose that the AMI or CABG model episode would begin at the o-i transfer hospital based on the MS-DRG (and AMI ICD-CM diagnosis code if a PCI MS-DRG applies) that is assigned to that anchor hospitalization. That is, if a beneficiary receives initial AMI care in a hospital emergency department without admission and is transferred to an AMI or CABG model participant (the o-i transfer hospital) for admission, then the AMI or CABG model episode would begin in the first hospital involved in the beneficiary's AMI or CABG care that admits the beneficiary as an inpatient, specifically the o-i transfer hospital. Therefore, the o-i transfer hospital would be financially responsible for the AMI or CABG model episode. This proposed attribution is in accordance with the proposed AMI and CABG model rules, as discussed in sections III.C.4.a.(2) and (3) of this proposed rule, that initiate an AMI model episode with a hospitalization that results in discharge from an AMI MS-DRG or PCI MS-DRG with an AMI ICD-CM diagnosis code in the principal or secondary position from an AMI model participant or a CABG model episode with a hospitalization that results in discharge from a CABG MS-DRG. Under this proposal, the o-i transfer hospital's quality measure performance would determine the effective discount factor to be applied to the AMI or CABG model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule. Under this proposal, regardless of whether the initial treating hospital is an AMI or CABG model participant, an AMI or CABG model episode would only be initiated at the o-i transfer hospital if that hospital is an AMI or CABG model participant.

We considered an overarching alternative policy that would begin every AMI or CABG model episode at the first AMI or CABG model participant at which either:

  • The beneficiary presented to the emergency department for initial AMI care before being transferred to an o-i transfer hospital; or
  • The beneficiary was admitted for an AMI MS-DRG, PCI MS-DRG with an AMI ICD-CM diagnosis code, or a CABG MS-DRG.

The AMI or CABG model participant where the episode begins would then be financially responsible for the AMI or CABG model episode unless the episode is canceled. Under this alternative, there would no changes to our proposals for attributing episodes with no transfers or inpatient-to-inpatient transfers.

However, under this alternative, if the beneficiary presented for initial AMI care to the emergency department of an AMI or CABG model participant, the AMI or CABG model episode would begin at this initial treating hospital when a beneficiary is transferred from the emergency department for his or her first inpatient hospitalization which occurs at an o-i transfer hospital. This would place financial responsibility for the AMI or CABG model episode on the initial treating hospital despite the fact that the beneficiary was transferred from that hospital without being admitted, and the initial treating hospital's quality measure performance would determine the effective discount factor to be applied to the AMI or CABG model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule.

Identifying the emergency department visit at the initial treating hospital would require using Field (Form Locator) 15—Point of Origin for Admission or Visit code on the CMS 1450 IPPS claim from the o-i transfer hospital to identify transfer from another hospital and linking that claim to the hospital outpatient claims from the initial treating hospital for the emergency department visit and other hospital outpatient services that occurred within a certain period of time prior to the o-i transfer hospital admission and that are related to the AMI care. The episode would be assigned to the AMI model even if the beneficiary received a CABG at the o-i transfer hospital, and we would assign financial responsibility for the AMI model episode to the initial treating hospital. Under this alternative, the initial treating hospital's quality measure performance would determine the effective discount factor to be applied to the AMI model benchmark episode price for the episode at reconciliation as described in section III.D.4.b.(10) of this proposed rule. We would also need to identify other types of related services to include in the episode that would begin prior to the o-i transfer hospital admission, such as physicians' services for care in the emergency department.

This alternative would have the benefit of consistently including all care in each AMI or CABG model episode that occurs following presentation of a beneficiary with AMI to the emergency department of an AMI or CABG model participant in the AMI or CABG model episode, regardless of whether an AMI or CABG model episode involves no transfer, o-i transfer, or i-i transfer. However, because this alternative would begin the AMI model episode prior to the initial hospital admission, we would need to establish additional policies for identifying the beneficiaries who initiate these episodes and define the timeframe and services that would be included in the AMI or CABG model episode prior to admission to the o-i transfer hospital.

We do not propose this alternative because we believe the policies necessary to begin the AMI or CABG model episode at the first treating hospital when an inpatient hospitalization does not occur would be complex, challenging to operationalize, and require assumptions about the relationship of care to the AMI based solely on administrative claims data that are insufficient to ensure we can accurately identify related care. We believe it remains problematic to define the services to be included in AMI or CABG model episodes if those services precede an inpatient hospitalization that Start Printed Page 50840would otherwise initiate the AMI or CABG model episode. For example, we would need to define the timeframe for beginning an AMI or CABG model episode with an emergency department visit for AMI that results in a transfer to the o-i transfer hospital, as well as the Part A and Part B services to be included in the AMI or CABG model episode that would result. As we discuss in section III.C.4.a.(1) of this proposed rule, we do not propose to begin any EPM episode prior to the anchor hospitalization because of the clinical variability leading up to all EPM episodes and the challenge of identifying unrelated services prior to the inpatient hospitalization. Thus, we do not propose to make an exception for transfers from the emergency department of the initial treating AMI or CABG model participant hospital when the beneficiary with AMI is not admitted to that hospital.

We seek comment on the proposal for AMI and CABG model episode initiation and attribution for the outpatient-to-inpatient transfer scenario, as well as the alternative considered that would begin an episode upon presentation of a beneficiary for initial AMI care to the emergency department of an AMI or CABG model participant when the care results in an outpatient-to-inpatient transfer.

Table 4 provides a summary of our proposals for episode initiation and attribution at the beginning of AMI care for no transfer, inpatient-to-inpatient transfer, and outpatient-to-inpatient transfer scenarios, including a description of how these relate to the participation in the AMI or CABG models of hospitals providing initial AMI care.

Table 4—Proposed Initiation and Attribution of AMI and CABG Model Episodes That Involve No Transfer, or Outpatient-to-Inpatient or Inpatient-to-Inpatient Transfers at the Beginning of AMI Care

ScenarioEpisode initiation and attribution
No transfer (participant): Beneficiary admitted to an initial treating hospital that is a participant in the AMI or CABG model for an AMI MS-DRG, PCI MS-DRG with AMI ICD-CM diagnosis code, or CABG MS-DRGInitiate AMI or CABG model episode based on anchor hospitalization MS-DRG. Attribute episode to the initial treating hospital.
No transfer (nonparticipant): Beneficiary admitted to an initial treating hospital that is not a participant in the AMI or CABG model for an AMI MS-DRG, PCI MS-DRG with AMI ICD-CM diagnosis code, or CABG MS-DRGNo AMI or CABG model episode is initiated.
Inpatient-to-inpatient transfer (nonparticipant to participant): Beneficiary admitted to an initial treating hospital that is not an AMI or CABG model participant and later transferred to an i-i transfer hospital that is an AMI or CABG model participant for an AMI MS-DRG, PCI MS-DRG with AMI ICD-CM diagnosis code, or CABG MS-DRGInitiate AMI or CABG model episode based on the MS-DRG at i-i transfer hospital. Attribute episode to the i-i transfer hospital.
Inpatient-to-inpatient transfer (participant to participant or participant to nonparticipant): Beneficiary admitted to an initial treating hospital that is an AMI or CABG model participant for an AMI MS-DRG, PCI MS-DRG with AMI ICD-CM diagnosis code, or CABG MS-DRG and later transferred to an i-i transfer hospital for an AMI, PCI, or CABG MS-DRG, regardless of whether the i-i transfer hospital is an AMI or CABG model participantInitiate AMI or CABG model episode based on anchor hospitalization MS-DRG at initial treating hospital. If the chained anchor hospitalization results in a final AMI, PCI, or CABG MS-DRG, calculate episode benchmark price based on the AMI, PCI or CABG MS-DRG with the highest IPPS weight. If the final MS-DRG is not an AMI, PCI, or CABG MS-DRG, cancel the episode. Attribute episode to the initial treating hospital.
Outpatient-to-inpatient transfer (nonparticipant to participant or participant to participant): Beneficiary transferred without admission from the initial treating hospital, regardless of whether the initial treating hospital is an AMI or CABG model participant, to a o-i transfer hospital that is an AMI or CABG model participant and is discharged from the o-i transfer hospital for an AMI MS-DRG, PCI MS-DRG with AMI ICD-CM diagnosis code, or CABG MS-DRGInitiate AMI or CABG model episode based on anchor hospitalization MS-DRG at o-i transfer hospital. Attribute episode to the o-i transfer hospital.
Outpatient-to-inpatient transfer (participant to nonparticipant): Beneficiary transferred without admission from the initial treating hospital that is an AMI or CABG participant to an o-i transfer hospital that is not an AMI or CABG model participantNo AMI or CABG model episode is initiated.

b. Middle of EPM Episodes

Similar to the CJR model, we propose that once an EPM episode begins, it would continue until the end of the episode as described in the following section, unless certain circumstances arise during the episode (80 FR 73318). When an EPM episode is canceled, we propose that the services furnished to beneficiaries prior to and following the EPM episode cancellation would continue to be paid by Medicare as usual but there would be no actual EPM episode spending calculation that would be reconciled against the EPM quality-adjusted target price.

Specifically, we propose that the following circumstances occurring during an EPM episode would cancel the EPM episode:

  • The beneficiary ceases to meet any of the general beneficiary inclusion criteria described in section III.C.4.a.(1) of this proposed rule, except the three criteria regarding inclusion in other episode payment model episodes.
  • The beneficiary dies during the anchor hospitalization.
  • The beneficiary initiates any BPCI model episode.

For purposes of cancellation of EPM episodes for beneficiary overlap with other episode payment models, we propose that if a beneficiary in an EPM episode would initiate any BPCI model episode, the EPM episode would be canceled. We refer to section III.D.6.c.(1) of this proposed rule for further discussion of our proposals addressing potential overlap of beneficiaries in the proposed EPMs with BPCI. We also refer to section III.D.6.c.(3) of this proposed rule for discussion of our proposal to cancel EPM episodes for beneficiaries who become aligned with specified ACOs during EPM episodes.

Our proposal to only cancel the EPM episode if a beneficiary dies during the anchor hospitalization differs from the final CJR model policy that cancels an Start Printed Page 50841episode if a beneficiary dies any time during the episode (80 FR 73318). As discussed in the CJR model Final Rule for LEJR episode, we believe that it also would be appropriate to cancel an episode in the AMI, CABG, and SHFFT models when a beneficiary dies during the anchor hospitalization as there would be limited incentives for efficiency that could be expected during the anchor hospitalization itself (80 FR 73318). We agreed with commenters on the CJR model proposed rule that we should cancel CJR model episodes for death any time during those episodes, because beneficiary deaths following LEJR would be uncommon and expected to vary unpredictably, leading to extremely high or low episode spending that was not typical for a LEJR episode. A recent analysis that pooled results from 32 studies showed the incidence of mortality during the first 30 and 90 days following hip replacement to be 0.30 percent and 0.65 percent, respectively, confirming our expectation of low mortality rates during LEJR episodes.[56] In contrast, the 30-day national CABG and AMI mortality rates as displayed on Hospital Compare are significantly higher at approximately 3 percent and 14 percent respectively.[57] Several CMS programs use 30-day mortality measures for CABG and AMI as measures of hospital quality, and these measures are proposed for use in the pay-for-performance methodology for the CABG and AMI models as discussed in section III.E.3.f. of this proposed rule. Similarly, a 2009 study shows a 30-day hip fracture mortality rate for Medicare beneficiaries of approximately 5 percent, significantly higher than the mortality rate following LEJR procedures.[58] Thus, we would expect that deaths during SHFFT model episodes would be more common than in CJR model episodes. Because beneficiaries in AMI, CABG, and SHFFT model episodes are at significant risk of death during these episodes that extends 90 days post-hospital discharge, we consider mortality to be a harmful beneficiary outcome that should be targeted for improvement through care redesign incentivized by the EPMs for these clinical conditions. Therefore, we do not believe it would be appropriate to exclude beneficiaries from AMI, CABG, or SHFFT model episodes who die any time during the episode like we do in the CJR model. Instead, we propose to maintain beneficiary episodes in the EPMs even if death occurs during the episodes, meaning we would calculate actual EPM episode spending when beneficiaries die following discharge from the anchor hospitalization but within the 90-day post-hospital discharge episode duration and reconcile it against the quality-adjusted target price. We believe this proposal would encourage EPM participants to actively manage EPM beneficiaries to reduce their risk of death, especially as death is often preceded by expensive care for emergencies and complications. Because of the higher mortality rates for all of the proposed EPM episodes than for LEJR episodes in the CJR model, we do not consider mortality following hospital discharge to be atypical and, therefore, we propose to cancel EPM episodes only for death during the anchor hospitalization.

We further propose that the following circumstances also would cancel an AMI model episode in the circumstances of a chained anchor hospitalization when the beneficiary is discharged from acute care under an MS-DRG from the final transfer hospital in the chained anchor hospitalization that could not, itself, initiate an AMI or CABG model episode, regardless of whether the final transfer hospital is an AMI or CABG model participant (that is, the episode would be canceled if the final transfer hospital MS-DRG is any MS-DRG other than an AMI MS-DRG, PCI MS-DRG, or CABG MS-DRG);

While we would begin an AMI model episode with the first hospitalization in the chained anchor hospitalization that would initiate an episode as discussed in section III.C.4.a.(5) of this proposed rule, we understand that a variety of types of care at i-i transfer hospitals could occur following the discharge from the hospital that began the AMI model episode during the chained anchor hospitalization, most commonly further medical management of AMI and revascularization that could be appropriately included in the AMI model episode. We further note that less than 0.2 percent of beneficiaries in historical AMI episodes have more than one inpatient-to-inpatient transfer during the chained anchor hospitalization.[59] However, in some cases transfer to another hospital during an AMI episode could result in a final i-i transfer hospital MS-DRG for care that would not itself have initiated an AMI (or CABG) model episode if all inpatient hospital care were furnished at a single hospital. For example, a beneficiary in an AMI model episode could be transferred to another hospital where the beneficiary undergoes cardiac valve surgery or treatment for renal failure or stroke. In some of these cases, further treatment at the i-i transfer hospital could be due to potentially avoidable complications resulting from insufficient care management during the AMI model episode that is initiated at the first hospital. In other cases the care at the i-i transfer hospital could be unavoidable and clinically appropriate, resulting from the beneficiary's evolving AMI or other associated chronic conditions and the specific capabilities of the hospital that initiated the AMI model episode. Therefore, we believe it would be most appropriate to cancel AMI model episodes under the circumstances when a beneficiary in an AMI model episode is discharged from acute care under an MS-DRG from the final i-i transfer hospital in the chained anchor hospitalization that is not an AMI, PCI, or CABG MS-DRG that could initiate an AMI or CABG model episode (that is, the episode would be canceled if the final transfer hospitalization MS-DRG is any MS-DRG other than an AMI, PCI, or CABG MS-DRG). We note that we would not require an AMI ICD-10-CM diagnosis code on all claims in a chained anchor hospitalization for a beneficiary in an AMI model episode in order to provide to an adjusted payment at the price MS-DRG for the AMI model episode as discussed in section III.D.4.b.(2)(a) of this proposed rule. We also would not cancel the AMI model episode if an AMI ICD-10-CM diagnosis code is not on the claim for the final transfer hospitalization, as long as the discharge is under an AMI, PCI, or CABG MS-DRG. Because the beneficiary would be in an AMI model episode during a chained anchor hospitalization, we would treat the beneficiary who is transferred to an i-i transfer hospital according to all policies that apply to the diagnosis of AMI in the CABG and AMI models, regardless of whether an AMI ICD-10-CM diagnosis code was on the PCI or CABG MS-DRG claim from the final i-i transfer hospital. Overall, this proposal would treat the hospital that initiated the AMI model episode and then transferred the beneficiary most similarly to a hospital that furnished all of the beneficiary's inpatient care itself, Start Printed Page 50842with respect to whether or not the beneficiary's care is ultimately included as an episode in the AMI model.

Finally, we do not propose to cancel an AMI episode altogether for a CABG readmission during the 90-day post-hospital discharge period or cancel the AMI model episode and initiate a CABG model episode because planned CABG readmission following an anchor hospitalization that initiates an AMI model episode may be an appropriate clinical pathway for certain beneficiaries. Instead, we propose to provide an adjusted AMI model-episode benchmark price that includes a CABG readmission in such circumstances so as not to financially penalize participant hospitals for relatively uncommon, costly, clinically appropriate care patterns for beneficiaries in AMI model episodes. We refer to section III.D.4.b.(2)(c) of this proposed rule for discussion of the adjusted AMI model-episode benchmark price that would apply in the case of CABG readmission during an AMI model episode.

The proposals for cancellation of EPM episodes are included in § 512.240(a)(3), (b)(2), and (c)(2). We seek comment on our proposals for cancellation of EPM episodes.

c. End of EPM Episodes

(1) AMI and CABG Models

We propose a 90-day post-hospital discharge episode duration for AMI model episodes. AMI in general, whether managed medically or with revascularization, has a lengthy recovery period, during which the beneficiary has a higher than average risk of additional cardiac events and other complications, as well as higher utilization of diagnostic testing and related cardiac procedures. AMI frequently serves as a sentinel event that marks the need for a heightened focus on medical management of coronary artery disease and other beneficiary risk factors for future cardiac events, cardiac rehabilitation over multiple months, and beneficiary education and engagement. Given the broad episode definition for AMI model episodes that includes beneficiaries receiving both medical and PCI management for an acute event, we do not believe that an episode longer than 90 days would be feasible due to the higher risk of including unrelated services in the episode beyond several months after hospital discharge. However, we believe that 90-day post-hospital discharge episodes would provide substantial incentives for aggressive medical management, cardiac rehabilitation, and beneficiary education and engagement, whereas a shorter episode duration would have less effect. We acknowledge that ongoing disease management for beneficiaries with cardiovascular disease must extend long after the conclusion of the proposed AMI model episodes. Nevertheless, we believe the proposed 90-day post-hospital discharge episode duration remains appropriate for an episode payment model focused around a hospitalization. We expect that the medical management and care coordination during AMI model episodes would continue to be provided as beneficiaries transition out of AMI model episodes, potentially into a primary care medical home or other model or program with accountability for population health, such as an ACO.

We further note based on analysis of historical episodes that about 10 percent of beneficiaries hospitalized with AMI who received a CABG received the CABG between 2 and 90 days post-discharge from the anchor hospitalization (these beneficiaries would be in AMI model episodes), while the remaining 90 percent of CABGs for beneficiaries hospitalized with AMI were provided during the initial hospitalization (these beneficiaries would in CABG model episodes). In contrast, fewer than 3 percent of those AMI model beneficiaries who received an inpatient or outpatient PCI during an AMI model episode received the PCI between 2 and 90 days post-discharge from the anchor hospitalization, while more than 97 percent received the PCI during the anchor hospitalization.[60] We refer to section III.D.4.b.(2)(c) of this proposed rule for further discussion of pricing adjustments and alternatives considered for setting EPM-episode benchmark prices for AMI model episodes where PCI or CABG occurs during the AMI episode but post-discharge from the anchor or chained anchor hospitalization.

Finally, for similar reasons, we believe CABG model episodes should extend 90 days post-hospital discharge. About one-third of CABG procedures are performed in the context of a hospital admission for AMI, leading to the same considerations discussed previously in this section around the appropriate episode duration for beneficiaries with AMI. The remaining CABG model beneficiaries are likely to have significant ischemic heart disease, making the occurrence of CABG itself a sentinel event, like AMI, that marks the need for a heightened focus on medical management of CAD and other beneficiary risk factors for future cardiac events, cardiac rehabilitation over multiple months, and beneficiary education and engagement. Moreover, CABG procedures have 90-day global periods under the Physician Fee Schedule, consistent with the lengthy period of recovery associated with major chest surgery. Thus, a 90-day post-hospital discharge episode duration is consistent with the recovery period from CABG surgery. We acknowledge that ongoing disease management for beneficiaries with cardiovascular disease must extend long after the conclusion of the proposed CABG model episodes. Nevertheless, we believe the proposed 90-day post-hospital discharge episode duration remains appropriate for an episode payment model focused around a hospitalization. We expect that the medical management and care coordination during CABG model episodes would continue to be provided as beneficiaries transition out of CABG model episodes, potentially into a primary care medical home or other model or program with accountability for population health, such as an ACO.

As in the CJR model, we propose that the day of discharge from the anchor hospitalization counts as day 1 of the post-hospital discharge period (80 FR 73324). However, in the case of an AMI model episode that includes a chained anchor hospitalization, we would count the day of discharge from the final hospitalization in the chained anchor hospitalization as day 1 of the post-hospital discharge period. Since the post-hospital discharge period is intended to extend 90 days for recovery following hospital discharge, we believe it is appropriate under these circumstances to begin the 90-day count when the beneficiary is ultimately discharged from acute care for the first time during the AMI model episode. However, the hospital that initiated the AMI model episode in the chained anchor hospitalization would continue to be responsible in the AMI model for the episode discussed previously in section III.C.4.a.(5) of this proposed rule.

The proposals for the end of AMI and CABG model episodes are included in §§ 512.240(a)(1) and (b)(1), respectively. We seek comment on our proposals to end AMI and CABG model episodes.

(2) SHFFT Model

We believe that SHFFT model beneficiaries are similar to CJR model beneficiaries who undergo hip replacement for fracture. We believe Start Printed Page 50843that the same episode duration as the CJR model of 90 days is appropriate for SHFFT model episodes in order to include the full time for recovery of function for these beneficiaries, which extends beyond 60 days based on patterns of post-acute care provider use (80 FR 73319 through 73324). Therefore, we propose a 90-day post-hospital discharge duration for SHFFT model episodes.

The proposal for the end of SHFFT model episodes are included in § 512.240(c)(1). We seek comment on our proposal to end SHFFT model episodes.

III. Provisions of the Proposed Regulations

D. Methodology for Setting EPM Episode Prices and Paying EPM Participants in the AMI, CABG, and SHFFT Models

1. Background

a. Overview

We propose that the AMI, CABG, and SHFFT models would provide incentives for EPM participants to work with other health care providers and suppliers to improve the quality and efficiency of care for Medicare beneficiaries by paying EPM participants or holding them responsible for repaying Medicare based on EPM participants' performance with respect to the quality and spending for AMI, CABG, and SHFFT episodes in a manner similar to the CJR model. Given the general similarity between the design of the CJR model and these EPMs, there is precedent for adopting the general payment and pricing parameters used under the CJR model, with modification to appropriately pay for EPM episodes that include the different clinical conditions treated in AMI, CABG, and SHFFT model episodes. The following sections describe our proposals for the:

  • Performance year, retrospective episode payments, and two-sided risk EPMs.
  • Adjustments to actual EPM-episode payments and to historical episode payments used to set episode prices.
  • EPM episode price-setting methodologies.
  • Process for reconciliation.
  • Adjustments for overlaps with other Innovation Center models and CMS programs.
  • Limits or adjustments to EPM participants' financial responsibility.

b. Key Terms for EPM Episode Pricing and Payment

For purposes of ease of understanding of the technical discussion that follows around EPM episode pricing and payment, we are providing the following definitions of terms that are used in sections that precede their technical definition and cross-references to other sections of this proposed rule for more detailed discussion of the policies associated with these terms.

  • Anchor hospitalization—hospitalization that initiates an EPM episode and has no subsequent inpatient-to-inpatient transfer chained anchor hospitalization.
  • Chained anchor hospitalization—an anchor hospitalization that initiates an AMI model episode and has at least one subsequent inpatient-to-inpatient transfer.
  • Anchor MS-DRG—MS-DRG assigned to the first hospitalization discharge, which initiates an EPM episode.
  • Price MS-DRG—for EPM episodes without a chained anchor hospitalization, the price MS-DRG is the anchor MS-DRG. For AMI model episodes with a chained anchor hospitalization, the price MS-DRG is the MS-DRG assigned to the AMI model episode according to the hierarchy described in III.D.4.b.(2)(i).
  • Episode benchmark price—dollar amount assigned to EPM episodes based on historical EPM-episode data (3 years of historical Medicare payment data grouped into EPM episodes according to the EPM episode definitions as discussed in sections III.C.3. and III.C.4. of this proposed rule) prior to the application of the effective discount factor, as described throughout sections III.D.4.b through e. of this proposed rule.
  • CABG readmission AMI model episode benchmark price—episode benchmark price assigned to certain AMI model episodes with price MS-DRG 280-282 or 246-251 and with a readmission for MS-DRG 231-236, as described in sections III.D.4.b.(2)(c) and III.D.4.e. of this proposed rule.
  • Quality-adjusted target price—dollar amount assigned to EPM episodes as the result of reducing the episode benchmark price by the EPM participant's effective discount factor based on the EPM participant's quality performance, as described in sections III.D.4.b.(10) and III.E.3.f. of this proposed rule.
  • Excess EPM-episode spending—dollar amount corresponding to the amount by which actual EPM-episode payments for all EPM episodes attributed to an EPM participant exceed the quality-adjusted target prices for the same EPM episodes, as discussed in section III.D.2.c. of this proposed rule.

2. Performance Years, Retrospective Episode Payments, and Two-Sided Risk EPMs

a. Performance Period

Consistent with the methodology for the CJR model, we propose 5 performance years (PYs) for the EPMs, which would include EPM episodes for the periods displayed in the following Table 5:

Table 5—Performance Years for EPMS

Performance year (PY)Calendar yearEPM episodes included in performance year
12017EPM episodes that start on or after July 1, 2017 and end on or before December 31, 2017.
22018EPM episodes that end between January 1, 2018 and December 31, 2018, inclusive.
32019EPM episodes that end between January 1, 2019 and December 31, 2019, inclusive.
42020EPM episodes that end between January 1, 2020 and December 31, 2020, inclusive.
52021EPM episodes that end between January 1, 2021 and December 31, 2021, inclusive.

As displayed in Table 5, some EPM episodes that would begin in a given calendar year may be captured in the following performance year due to some EPM episodes ending after December 31st of a given calendar year. For example, EPM episodes beginning in December 2017 and ending in March 2018 would be part of performance year 2. We believe that the proposed period of time for the EPMs, which generally aligns with the performance period for other Innovation Center models, for example, the CJR and Pioneer ACO models, should be sufficient to test and gather the data needed to evaluate the EPMs (80 FR 73325). In contrast, we would be concerned whether an EPM with fewer than 5 performance years would be sufficient for these purposes.Start Printed Page 50844

We also recognize that our proposal would allow only 6 months of EPM episodes for PY1 as compared to 9 months for the CJR model. We considered extending the first PY, for example, to 18 months. As discussed further in section III.D.2.c. of this proposed rule, however, we are instead proposing to delay the requirement for participants to begin accepting downside risk until the second quarter of PY2. As such, EPM participants would have a comparable transition period to that of CJR participants with respect to when they must accept downside risk while still allowing us to make timely reconciliation payments to EPM participants as well as to most effectively align EPM reconciliation with the reconciliation processes for other models and programs with which the EPMs overlap (for example, the Shared Savings Program, Pioneer ACO model, Comprehensive Primary Care Initiative, and Oncology Care Model). We believe that it is important to synchronize the timing of reconciliation for EPMs with other efforts that need this information when making their financial calculations. We seek comment on this proposal.

b. Retrospective Payment Methodology

Consistent with the CJR model, we propose to apply a retrospective payment methodology to the proposed EPMs (80 FR 73329). Under this proposal, all providers and suppliers caring for Medicare beneficiaries in EPM episodes would continue to bill and be paid as usual under the applicable Medicare payment systems. After the completion of an EPM performance year, Medicare claims for services furnished to EPM beneficiaries would be grouped into EPM episodes and aggregated, and EPM participants' actual EPM episode-payments compared to quality-adjusted target prices (which account for the level of EPM episode quality), as described in section III.D.5.a. of this proposed rule. Based on an EPM participant's performance (taking into account quality and spending), we would determine if Medicare would make a payment to the participant (reconciliation payment), or if the participant owes money to Medicare (resulting in Medicare repayment).

We considered an alternative option of paying for EPM episodes prospectively by paying one lump sum amount to the EPM participant for the expected spending for the EPM episode which extends 90 days post-hospital-discharge. However, as was the case when we established regulations for the CJR model, we continue to believe that such an option would be challenging to implement at this time given the payment infrastructure changes for both EPM participants and Medicare that would need to be developed to pay and manage prospective episode payments under these EPMs (80 FR 73329). Moreover, we continue to believe that a retrospective payment approach can accomplish the objective of testing episode payments in a broad group of hospitals, including financial incentives to streamline care delivery around that episode, without requiring core billing and payment changes by providers and suppliers, which would create substantial administrative burden.

We seek comment on this proposal.

c. Two-Sided Risk EPMs

As we did for the CJR model, we propose to establish two-sided risk for hospitals participating in the EPMs. Under this proposal, for each of performance years 1 through 5, we would make EPM-episode reconciliation payments to EPM participants that achieve reduced actual EPM payments relative to their quality-adjusted target prices (80 FR 73229-7333). Likewise, beginning with episodes ending in the second quarter of performance year 2 and extending through each of performance years 3 through 5, we would hold EPM participants responsible for repaying Medicare when their actual EPM-episode payments exceed their quality-adjusted target prices. As such, our proposal differs from CJR in that we are proposing a modestly shorter period in which EPM participants would accept downside risk in order to allow them a comparable transition period to that of CJR participants in which to do so. Accordingly, we will refer to the two portions of performance year 2 as—

  • Performance Year 2 (NDR) or PY2 (NDR) for the first quarter, that is January 1, 2018 to March 31, 2018, in which EPM participants assume no downside risk and therefore would have no Medicare repayment responsibility; and
  • Performance Year 2 (DR) or PY2 (DR) for the second, third and fourth quarters, that is April 1, 2018 to December 31, 2018, in which EPM participants assume downside risk and would have Medicare repayment responsibility. We believe that our proposal to establish two-sided risk would provide appropriate incentives for EPM participants to improve their care quality and efficiency under the EPMs. We also continue to believe, as we indicated in the CJR Final Rule, that we would diminish these incentives if we instead proposed to establish one-sided risk, in which an EPM participant could qualify for a reconciliation payment but not be held responsible for Medicare repayments (80 FR 73329). In recognition that EPM participants may need to make infrastructure, care coordination and delivery, and financial preparations for the EPMs, which can take several months or longer to implement, we do believe that it is reasonable to delay EPM participant responsibility for repaying excess EPM-episode spending in performance year 1 to more strongly align EPM-participant incentives with care quality. Thus, similar to what we did for the CJR model, we are proposing to phase-in this repayment responsibility beginning in the second quarter of EPM performance year 2 as displayed in Table 6.

We refer to section III.E.3.f. of this proposed rule for additional information on the effective discount factors used to calculate quality-adjusted target prices, as well as the quality categories that determine an EPM participant's effective discount factor that would be applied to the EPM benchmark episode price at reconciliation to calculate the repayment amount during the phase-in period in EPM performance year 2 (quarters 2 through 4) and performance year 3. Table 6 also presents the phase-in of the proposed stop-loss limits and discount percentages, which are discussed in detail in section III.D.7.b. and III.D.4.b.(10) of this proposed rule.

We seek comment on this proposal.

Table 6—Stop-Loss Thresholds and Discount Percentage Ranges for Medicare Repayments by PY

PY1PY2 (NDR)PY2 (DR) %PY3 %PY4 %PY5 %
Stop-loss thresholdn/a as no downside risk in PY1 and PY2 (DR)5102020
Start Printed Page 50845
Discount percentage (range) for Repayment, Depending on Quality Category0.5-2.00.5-2.01.5-3.01.5-3.0
* Stop-loss thresholds for certain hospitals, including rural and sole-community hospitals are 3% for PY2 (DR) and 5% for PY3-PY5.

3. Adjustments to Actual EPM-Episode Payments and to Historical Episode Payments Used to Set Episode Prices

a. Overview

We propose to calculate actual EPM-episode payments and historical episode payments (3 years of historical Medicare payment data grouped into EPM episodes according to the EPM episode definitions as discussed in sections III.C.3. and III.C.4. of this proposed rule) to calculate EPM quality-adjusted target prices for each performance year of the EPMs as we did for the CJR model—that is, for each non-cancelled EPM episode, we would calculate these amounts based on Medicare payments for Parts A and B claims for services included in the EPM episode definition. As was the case for the CJR model, we also propose to include certain payment adjustments in the EPMs for: (1) Special payment provisions under existing Medicare payment systems; (2) payments for services that straddle episodes; and (3) high payment episodes (80 FR 73330 through 73336). We also propose to additionally include an adjustment for reconciliation payments and Medicare repayments when updating EPM participant episode benchmark and quality-adjusted target prices (80 FR 73330 through 73331). We refer to section III.D.6. of this proposed rule for discussion of adjustments for overlaps with other Innovation Center models and CMS programs.

b. Special Payment Provisions

Many of the existing Medicare payment systems have special payment provisions that have been created by regulation or statute to improve quality and efficiency in service delivery. IPPS hospitals are subject to incentives under the HRRP, the HVBP Program, the Hospital-Acquired Condition (HAC) Reduction Program, and the HIQR Program and Outpatient Quality Reporting (OQR) Program. IPPS hospitals and CAHs are subject to the Medicare Electronic Health Record (EHR) Incentive Program. Additionally, the majority of IPPS hospitals receive additional payments for Medicare Disproportionate Share Hospital (DSH) and Uncompensated Care, and IPPS teaching hospitals can receive additional payments for Indirect Medical Education (IME). IPPS hospitals that meet certain requirements related to low volume Medicare discharges and distance from another hospital receive a low volume add-on payment. Also, some IPPS hospitals qualify to be sole community hospitals (SCHs) or Medicare Dependent Hospitals (MDHs), and they may receive enhanced payments based on cost-based hospital-specific rates for services; whether a SCH or MDH receives enhanced payments may vary year to year, in accordance with § 419.43(g) and § 412.108(g), respectively.

Medicare payments to providers of post-acute care services, including IRFs, SNFs, IPFs, HHAs, LTCHs, and hospice facilities, are conditioned, in part, on whether the provider satisfactorily reports certain specified data to CMS: Inpatient Rehabilitation Facility Quality Reporting Program (IRF QRP); Skilled Nursing Facility Quality Reporting Program (SNF QRP); Inpatient Psychiatric Facility Quality Reporting Program (IPF QRP); Home Health Quality Reporting Program (HH QRP); Long-Term Care Hospital Quality Reporting Program (LTCH QRP); and Hospice Quality Reporting Program. Additionally, IRFs located in rural areas receive rural add-on payments, IRFs serving higher proportions of low-income beneficiaries receive increased payments according to their low-income percentage (LIP), and IRFs with teaching programs receive increased payments to reflect their teaching status. SNFs receive higher payments for treating beneficiaries with human immunodeficiency virus (HIV). HHAs located in rural areas also receive rural add-on payments.

Ambulatory Surgical Centers (ASCs) have their own Quality Reporting Program (ASC QRP). Physicians also have a set of special payment provisions based on quality and reporting: Medicare EHR Incentive Program for Eligible Professionals; Physician Quality Reporting System (PQRS); and Physician Value-based Modifier Program.

Consistent with how we determine payments under the CJR model, we propose to adjust both the actual and historical EPM-episode payments used to set EPM-episode benchmark and quality-adjusted target prices by excluding these special payments from EPM-episode calculations using the CMS Price Standardization methodology (80 FR 73333). We believe that in applying this methodology to exclude these payments from our calculations, we would best maintain appropriate incentives for both the proposed EPMs and the existing incentive programs. Also, not excluding add-on payments based on the characteristics of providers caring for EPM beneficiaries, such as more indigent patients, having low Medicare hospital volume, being located in a rural area, supporting greater levels of physician training, and having a greater proportion of beneficiaries with HIV, from actual EPM-episode payments could inappropriately result in certain EPM participants that receive more add-on payments having worse episode payment performance compared to quality-adjusted target prices than what their performance would otherwise have been. Additionally, not excluding enhanced payments for MDHs and SCHs could result in higher or lower quality-adjusted target prices just because EPM participants received their enhanced payments in 1 historical year but not the other, regardless of actual utilization. We also believe that excluding special payments would ensure an EPM participant's actual episode payment performance is not artificially improved or worsened because of payment reduction penalties or incentives or enhanced or add-on payments, the effects of which we are not intending to test under the proposed models. In addition to the various incentives, enhanced payments, and add-on payments, sequestration came into effect for Medicare payments for discharges on or after April 1, 2013, per the Budget Control Act of 2011 and delayed by the American Taxpayer Relief Act of 2012. Sequestration applies a 2-percent Start Printed Page 50846reduction to Medicare payment for most Medicare FFS services.

For more information on the CMS Price (Payment) Standardization Detailed Methodology, we refer to the QualityNet Web site at http://www.qualitynet.org/​dcs/​ContentServer?​c=​Page&​pagename=​QnetPublic%2FPage%2FQnetTier4&​cid=​1228772057350 and to 80 FR 73331.

Accordingly, we propose to exclude these special payments from EPM-episode calculations using the CMS Price Standardization methodology at § 512.300(e)(2). We seek comment on our proposal to exclude special payments using the CMS Price Standardization methodology.

c. Services That Straddle Episodes

A service that straddles an EPM episode is one that begins before the start of or continues beyond the end of an EPM episode that extends 90 days post-hospital discharge. Under the CJR model, we prorate payments so that they include only the portion of the payment that is included in the CJR model episode, using separate approaches to prorate payments under each payment system, for example, IPPS, non-IPPS and other inpatient services, and home health services (80 FR 73333 through 73335). We propose to apply the CJR model methodologies for prorating payments when calculating actual EPM-episode payments and when calculating historical EPM-episode payments used to set EPM-episode benchmark and quality-adjusted target prices. We believe these methodologies would most accurately account for spending within EPM episodes under the proposed EPMs.

The proposed methodologies for prorating payments are included in § 512.300(f). We seek comment on our proposed methodologies for prorating payments.

d. High-Payment EPM Episodes

For the CJR model, we defined a high-payment episode as an episode with payments 2 standard deviations or more above the mean calculated at the regional level (80 FR 73336 through 73337). As with the CJR model, we propose applying a high-payment episode ceiling when calculating actual EPM-episode payments and when calculating historical EPM-episode payments used to set EPM-episode benchmark and quality-adjusted target prices. We propose to apply the ceiling according to the following groupings that align with our proposed EPM price-setting methodology.

First, for SHFFT model episodes, we propose to calculate and apply the ceiling separately for each SHFFT price MS-DRG at the regional level.

Second, for AMI model episodes with price MS-DRGs 280-282 or 246-251 without readmission for CABG MS-DRGs, we propose to calculate and apply the ceiling separately for each price MS-DRG at the regional level.

Third, for CABG model episodes, we propose to apply ceilings separately to the payments that occurred during the anchor hospitalization of the CABG model episode and to the payments that occurred after the anchor hospitalization. For the anchor hospitalization portion of CABG model episodes, we propose to calculate and apply the ceiling separately by each price MS-DRG in 231-236 at the regional level. For the post-anchor hospitalization portion we propose to calculate and apply the ceiling separately for the following groupings at the regional level:

  • With AMI ICD-CM diagnosis code on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235).
  • With AMI ICD-CM diagnosis code on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236).
  • Without AMI ICD-CM diagnosis code on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235).
  • Without AMI ICD-CM diagnosis code on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236).

Fourth, for AMI model episodes with price MS-DRG 231-236, we propose to apply ceilings separately to the payments that occurred during the chained anchor hospitalization and to the payments that occurred after the chained anchor hospitalization. For the anchor hospitalization portion of the episode, we propose to apply the regional level ceiling calculated for the anchor hospitalization portion of a CABG model episode for the corresponding price MS-DRG, as described previously. For the post-anchor hospitalization portion of the episode, we propose to apply the regional level ceiling calculated for the post-anchor hospitalization portion of a CABG model episode for the corresponding price MS-DRG with AMI diagnosis.

Fifth, for AMI model episodes with price MS-DRG 280-282 or 246-251 and with readmission for CABG MS-DRGs, we propose to apply the ceiling separately to the payments during the CABG readmission and all other payments during the episode. For payments during the CABG readmission portion of the AMI model episode we propose to apply the regional level ceiling calculated for the anchor hospitalization portion of a CABG model episode for the corresponding CABG readmission MS-DRG, as described previously. For all other payments during the AMI model episode, we propose to apply the regional level ceiling calculated for AMI model episodes with price MS-DRG 280-282 or 246-251 and without readmission for CABG MS-DRGs corresponding to the AMI price MS-DRG.

We believe that this ceiling would protect EPM participants from variable repayment risk for especially-high payment EPM episodes where the clinical scenarios for these cases each year may differ significantly and unpredictably.

The proposal for capping high payment EPM episodes is included in § 512.300(e)(1). We seek comment on our proposal to cap high payment EPM episodes.

e. Treatment of Reconciliation Payments and Medicare Repayments When Calculating Historical EPM-Episode Payments To Update EPM-Episode Benchmark and Quality-Adjusted Target Prices

For the CJR model, we exclude CJR model reconciliation payments and Medicare repayments from the expenditure data used to update historical claims when calculating CJR model target prices, although we received comments on the proposed rule encouraging us to include these payments. For example, commenters supported their inclusion because CJR-participating hospitals otherwise would be providing care coordination services that would not be paid directly or accounted for under applicable Medicare FFS payments systems and thus might be funded through reconciliation payments. Further, by excluding reconciliation payments from our calculations, commenters suggested that we may underestimate their actual resource costs when updating target prices for the care necessary during episodes. The CJR Final Rule discussed our view that including reconciliation payments would have the effect of Medicare paying CJR model participant hospitals their target prices, regardless of whether such participant was below, above, or met their episode target price. We also noted that we had not discussed any alternatives in the CJR model proposed rule, and that we might Start Printed Page 50847consider including these payments in updating historical claims through future rulemaking (80 FR 73332).

After further consideration, we are proposing to include both reconciliation payments and Medicare repayments when calculating historical EPM-episode payments to update EPM-episode benchmark and quality-adjusted target prices. We concur with the views expressed by commenters on the CJR model proposed rule that including these payments would more fully recognize the total resource costs of care under an EPM than would their exclusion. As indicated in section V.5 of this proposed rule, we are also proposing to modify our policy for the CJR model to also include reconciliation payments and Medicare repayments when updating target prices under that model We also considered an option where we would include only reconciliation payments when updating but not Medicare repayments; however, we believe this option would not achieve our intention of more fully capturing the costs of care under the EPM. We would further note that the inclusion of both reconciliation payments and Medicare repayments could have differential effects on an EPM participant's benchmark and quality-adjusted target prices based on whether or not it received a reconciliation payment or made a Medicare repayment. For example, all else equal, including an EPM reconciliation payment when updating an EPM participant's EPM-episode benchmark and quality-adjusted target prices would modestly increase the quality-adjusted target prices in performance years 3 through 5 in comparison to not including the reconciliation payment. Conversely, all else equal, including a Medicare repayment when updating an EPM participant's EPM-episode benchmark and quality-adjusted target prices would reduce the next performance year's quality-adjusted target price in comparison to not including the Medicare repayment.

Following analogous logic, we also propose to include BPCI Net Payment Reconciliation Amounts in our calculations when updating EPM-episode benchmark and quality-adjusted target prices. We would note, however, that the effects of these proposals would largely be confined to PY3 of the EPMs and diminish as EPM-participant historical EPM-episode updates are eventually determined based on regional payments in subsequent years of the EPMs. This is because the net sum of EPM reconciliation payments, Medicare repayments, and BPCI Net Payment Reconciliation Amounts would represent a small portion of the total historical EPM-episode payments captured in regional pricing.

When updating EPM-episode benchmark and quality adjusted target prices for CABG model episodes, we propose to apportion EPM reconciliation payments and BPCI Net Reconciliation Payment Amounts proportionally to the anchor hospitalization and post-anchor hospitalization portions of CABG model historical episodes. We also propose to calculate the proportions based on regional average historical episode payments that occurred during the anchor hospitalization portion of CABG model episodes and regional average historical episode payments that occurred during the post-anchor anchor hospitalization portion of CABG model episodes that were initiated during the 3 historical years. This aligns with the general proposal to calculate the CABG model-episode benchmark price as the sum of the corresponding CABG anchor hospitalization benchmark price and the corresponding CABG post-anchor hospitalization benchmark price, as discussed in III.D.4.b.(2)(ii) and III.D.4.d. of this proposed rule.

The proposal to include both reconciliation payments and Medicare repayments when calculating historical EPM-episode payments to update EPM-episode benchmark and quality-adjusted target prices is included in § 512.300(c)(8). We seek comment on our proposal to include both reconciliation payments and Medicare repayments when calculating historical EPM-episode payments to update EPM-episode benchmark and quality-adjusted target prices.

4. EPM-Episode Price-Setting Methodologies

a. Overview

Whether an EPM participant receives a reconciliation payment or is made responsible to repay Medicare under the proposed EPM is based on the EPM participant's actual EPM-episode payments relative to quality-adjusted target prices, as well as the EPM participant's eligibility for reconciliation payment based on acceptable, good, or excellent quality performance. While our proposals for relating EPM participant quality performance to EPM payments are further discussed in section III.E.3.f of this proposed rule, the remainder of this section will discuss the proposed approach to establishing EPM-episode benchmark and quality-adjusted target prices.

For the purposes of price-setting, any references in this proposed rule to AMI ICD-CM diagnosis codes means those ICD-9-CM and ICD-10-CM diagnosis codes for historical EPM episodes or ICD-10-CM diagnosis codes for EPM episodes during the EPM performance years that can be found in the specific EPM episode definitions parameters spreadsheet. Also, for the purposes of price-setting, any references in this proposed rule to intracardiac ICD-CM procedure codes means those ICD-9-CM procedure codes for historical EPM episodes that can be found in the specific EPM episode definitions parameters spreadsheet. The EPM episode definitions parameters spreadsheets are posted on the CMS Web site at https://innovation.cms.gov/​inititatives/​epm.

We propose to establish EPM-episode benchmark and quality-adjusted target prices for each EPM participant based on the following MS-DRGs and diagnoses included in the AMI, CABG, and SHFFT models as discussed in sections III.C.3 and III.C.4. of this proposed rule:

(1) AMI Model

  • AMI MS-DRGs—

++ 280 (Acute myocardial infarction, discharged alive with MCC);

++ 281 (Acute myocardial infarction, discharged alive with CC);

++ 282 (Acute myocardial infarction, discharged alive without CC/MCC); and

  • PCI MS-DRGs, when the claim includes an AMI ICD-CM diagnosis code in the principal or secondary position on the inpatient claim and when the claim does not include an intracardiac ICD-CM procedure code in any position on the inpatient claim—

++ 246 (Perc cardiovasc proc with drug-eluting stent with MCC or 4+ vessels/stents);

++ 247 (Perc cardiovasc proc with drug-eluting stent without MCC);

++ 248 (Perc cardiovasc proc with non-drug-eluting stent with MCC or 4+ vessels/stents);

++ 249 (Perc cardiovasc proc with non-drug-eluting stent without MCC);

++ 250 (Perc cardiovasc proc without coronary artery stent with MCC); and

++ 251 (Perc cardiovasc proc without coronary artery stent without MCC).

(2) CABG Model DRGs—

  • 231 (Coronary bypass with PTCA with MCC);
  • 232 (Coronary bypass with PTCA without MCC);
  • 233 (Coronary bypass with cardiac cath with MCC);
  • 234 (Coronary bypass with cardiac cath without MCC);Start Printed Page 50848
  • 235 (Coronary bypass without cardiac cath with MCC); and
  • 236 (Coronary bypass without cardiac cath without MCC).

(3) SHFFT Model DRGs—

  • 480 (Hip and femur procedures except major joint with MCC);
  • 481 (Hip and femur procedures except major joint with CC); and
  • 482 (Hip and femur procedures except major joint without CC or MCC).

We propose to generally apply the CJR model methodology to set EPM-episode benchmark and quality-adjusted target prices, with the addition of some adjustments based on the specific clinical conditions and care patterns for EPM episodes included in the AMI, CABG, and SHFFT models (80 FR 73337 through 73338). The proposed price-setting methodology incorporates the following features:

  • Set different EPM benchmark and quality-adjusted target prices for EPM episodes based on the assigned price MS-DRG in one of the included MS-DRGs to account for patient and clinical variations that impact EPM participants' costs of providing care. Inpatient claims with PCI MS-DRGs 246-251 that contain an intracardiac ICD-CM procedure code in any position would not anchor an historical episode, nor be considered when assigning a price MS-DRG. This is because beginning in FY 2016, inpatient claims containing an intracardiac ICD-10-CM procedure code in any position no longer map to MS-DRGs 246-251.
  • Adjust EPM benchmark and quality-adjusted target prices for certain EPM episodes involving chained anchor hospitalizations, specific readmissions, or the presence of an AMI ICD-CM diagnosis code for CABG MS-DRGs.
  • Use 3 years of historical Medicare FFS payment data grouped into EPM episodes according to the EPM episode definitions in sections III.C.3 and III.C.4. of this proposed, termed historical EPM episodes and historical EPM-episode payments. The specific set of 3 historical years would be updated every other performance year.
  • Apply Medicare payment system (for example, IPPS, OPPS, IRF PPS, SNF, MPFS.) updates to the historical EPM-episode data to ensure we incentivize EPM participants based on historical utilization and practice patterns, not Medicare payment system rate changes that are beyond such participants' control. Because different Medicare payment system updates become effective at two different times of the year, we would calculate one set of EPM-benchmark and quality-adjusted target prices for EPM episodes initiated between January 1 and September 30 and another set for EPM episodes initiated between October 1 and December 31.
  • Blend together EPM-participant hospital-specific and regional historical EPM-episode payments, transitioning from primarily hospital-specific to completely regional pricing over the course of the 5 performance years, to incentivize both historically-efficient and less-efficient EPM participants to furnish high quality, efficient care in all years of the EPM Regions would be defined as each of the nine U.S. Census divisions.
  • Normalize for hospital-specific wage-adjustment variations in Medicare payment systems when combining hospital-specific and regional historical EPM episodes.
  • Pool together EPM episodes by groups of price MS-DRGs to allow a greater volume of historical cases and allow us to set more stable prices.
  • Apply an effective discount factor on EPM-episode benchmark prices to serve as Medicare's portion of reduced expenditures from the EPM episode, with any remaining portion of reduced Medicare spending below the quality-adjusted target price potentially available as reconciliation payments to the EPM participant where the anchor hospitalization occurred.
  • Further discussion on each of the proposed features and sequential steps to calculate EPM-episode benchmark and quality-adjusted target prices can be found in sections III.D.4.b through e. of this proposed rule, which immediately follow.

We also propose to calculate and communicate EPM-episode benchmark and quality-adjusted target prices to EPM participants prior to the performance period in which the prices apply (that is, prior to January 1, 2018, for prices covering EPM episodes that start between January 1, 2018, and September 30, 2018; prior to October 1, 2018, for prices covering EPM episodes that start between October 1, 2018, and December 31, 2018). We believe that prospectively communicating EPM-episode benchmark and quality-adjusted target prices to EPM participants would help them make infrastructure, care coordination and delivery, and financial refinements they may deem appropriate to prepare for the new episode target prices under the model.

The proposal to prospectively communicate quality-adjusted target prices are included in § 512.300(c)(9). We seek comment on our proposal to prospectively communicate these prices.

b. EPM-Episode Benchmark and Quality-Adjusted Target Price Features

(1) Risk-Stratifying EPM-Episode Benchmark Prices Based on MS-DRG and Diagnosis

To account for some of the clinical and resource variations that would be expected to occur under the EPMs, we propose generally to apply the episode pricing methodology that was applied to the CJR model to develop the EPM-episode benchmark prices, hereinafter called the standard EPM-episode benchmark price. In addition, for each EPM participant, we propose to risk-stratify and establish special EPM-episode benchmark prices for episodes in different pricing scenarios as described in this section, as well as sections III.D.4.c. through e. of this proposed rule. For purposes of this proposed rule, risk-stratification means the methodology for developing the EPM-episode benchmark price that accounts for clinical and resource variation in historical EPM episodes so that the quality-adjusted target price (calculated from the EPM-episode benchmark price) can be compared to actual EPM episode payments for EPM beneficiaries with similar care needs to those in historical EPM episodes.

For the SHFFT model, we propose to set the price MS-DRG equal to the anchor MS-DRG. We propose to calculate standard SHFFT model-episode benchmark prices based on price MS-DRGs following the general payment methodology that was applied to the CJR model with risk stratification according to the anchor MS-DRG (80 FR 73337 through 73358).

Similarly, for AMI model episodes without chained anchor hospitalizations and without readmissions for CABG MS-DRGs, we propose to set the price MS-DRG equal to the anchor MS-DRG. We propose to calculate standard AMI model-episode benchmark prices based on price MS-DRGs following the general payment methodology that was applied to the CJR model with risk stratification according to the anchor MS-DRG (80 FR 73337 through 73358). We propose to apply the CJR model payment methodology separately to AMI model episodes with anchor AMI MS-DRGs 280-282 and anchor PCI MS-DRGs 246-251 with a corresponding AMI ICD-CM diagnosis code on the inpatient claim for the anchor hospitalization and without an intracardiac ICD-CM procedure code in any position on the inpatient claim for the anchor hospitalization.

For episodes in the AMI model with chained anchor hospitalizations and no readmissions for CABG MS-DRGs, we Start Printed Page 50849propose to set the price MS-DRG based on the hierarchy described in section III.D.4.b.(2)(a) and to calculate AMI model-episode benchmark prices based on price MS-DRGs as described in sections III.D.4.b.(2)(a) and III.D.4.c. of this proposed rule.

For AMI model episodes without chained anchor hospitalizations and with readmissions for CABG MS-DRGs, we propose to set the price MS-DRG as the anchor MS-DRG and to calculate CABG readmission AMI model-episode benchmark prices as described in sections III.D.4.b.(2)(b), III.D.4.b.(2)(c), and III.D.4.e of this proposed rule.

For AMI model episodes with chained anchor hospitalizations that do not include CABG MS-DRGs and with readmissions for CABG MS-DRGs, we propose to set the price MS-DRG based on the hierarchy described in section III.D.4.b.(2)(a) and to calculate CABG readmission AMI model-episode benchmark prices as described in sections III.D.4.b.(2)(b), III.D.4.b.(2)(c), and III.D.4.e. of this proposed rule.

For CABG model episodes, we propose to set the price MS-DRG as the anchor MS-DRG and to calculate CABG model-episode benchmark prices as the sum of the CABG anchor hospitalization portion price and the CABG post-anchor hospitalization portion price, which would be calculated by applying the general payment methodology that was applied to the CJR model separately to the expenditures that occurred during the anchor hospitalization of the CABG model episode and to the expenditures that occurred after the anchor hospitalization as discussed in sections III.D.4.b.(2)(b) and III.D.4.d. of this proposed rule (80 FR 73337 through 73358).

Finally, we propose that after assigning an EPM-episode benchmark price to each EPM episode, the EPM-episode quality-adjusted target price would be the EPM-episode benchmark price reduced by the effective discount factor for the corresponding EPM that corresponds to the EPM participant's quality category, as discussed in sections III.D.4.b.(10) and III.E.3.f. of this proposed rule.

(2) Adjustments To Account for EPM-Episode Price Variation

We also have considered further adjustments to account for clinical and resource variation that could affect EPM participants' costs for EPM episodes. As was the case for the CJR model, we continue to believe that no standard risk adjustment approach that is widely-accepted throughout the nation exists for the proposed EPM episodes (80 FR 73338 through 73339). Thus, we are not proposing to make risk adjustments based on beneficiary-specific demographic characteristics or clinical indicators. Likewise, we continue to believe that CMS Hierarchical Condition Categories (HCC) used to adjust for risk in the Medicare Advantage program would not be appropriate for risk-adjusting EPM episodes as such categories are used to predict total Medicare expenditures in an upcoming year for MA plans and may not be appropriate for use in predicting expenditures over a shorter period of time, such as the EPM episodes. Further, the validity of HCC scores for predicting Medicare expenditures for shorter episodes-of-care or specifically for the AMI, CABG, and SHFFT model episodes that we are proposing has not been determined. Thus, we do not propose to risk-adjust EPM-episode benchmark or quality-adjusted target prices using HCC scores for the currently proposed EPMs. We refer to the CJR Final Rule for additional discussion of our assessment of risk-adjustment options for the CJR model, which informs our views on their appropriateness for the proposed EPMs (80 FR 73338 through 73340).

However, we believe there are circumstances that could account for spending variation in EPM episodes where certain pricing adjustments could be appropriate. We have identified several scenarios where increased EPM-episode efficiencies would be limited for certain groups of EPM beneficiaries and a standard EPM-episode benchmark price based on the anchor MS-DRG would, therefore, not account for circumstances where clinically-appropriate care could consistently result in higher EPM-episode payments. For example, as discussed in section III.C.4.a.(5) of this proposed rule, variation could arise from the asymmetric distribution of cardiac care across hospitals, which makes transfers, either from a hospitalization or from the emergency department (without inpatient admission) of one hospital to another, a common consideration in the treatment course for beneficiaries with an initial diagnosis of AMI, resulting in a chained anchor hospitalization for inpatient-to-inpatient transfers. Alternately, we recognize that certain episodes involving hospital readmissions for clinically-appropriate planned follow-up care may have higher episode spending than episodes with a single hospitalization or with chained anchor hospitalizations involving transfers that do not have any readmissions. Further, a beneficiary who has a CABG in the context of hospitalization for an AMI may have different spending in the 90 days post-hospital-discharge due to different health needs than a beneficiary who has an elective CABG. Accordingly, we propose specific policies and payment adjustments in recognition of the systematic, consistent variation in EPM-episode spending that could result from such circumstances.

(a) Adjustments for Certain AMI Model Episodes With Chained Anchor Hospitalizations

In section III.C.4.a.(5) of this proposed rule, we proposed that once an AMI model episode is initiated at an AMI model participant, the AMI model episode continues under the responsibility of that specific participant, regardless of whether the beneficiary is transferred to another hospital for further medical management of AMI or revascularization through PCI or CABG during a chained anchor hospitalization. Given there could be significant differences between the discharge MS-DRG from the hospital that initiates the AMI episode and the hospital to which a beneficiary is transferred, as well as the Medicare payment associated with these different MS-DRGs and the post-discharge spending for these beneficiaries, we believe it would be appropriate to adjust the AMI model-episode benchmark prices for certain AMI model episodes involving a chained anchor hospitalization.

More specifically, we believe that it would be appropriate to make an adjustment when a final hospital discharge MS-DRG in the chained anchor hospitalization is an anchor MS-DRG under either the AMI or CABG model. Thus, for episodes involving a chained anchor hospitalization with a final discharge diagnosis of any of AMI MS-DRG 280-282, PCI MS-DRG 246-251 without an intracardiac ICD-CM procedure code in any position on the inpatient claim, or CABG MS-DRG 231-236, we propose to set a chain-adjusted AMI model-episode benchmark price or “price MS-DRG” based on the AMI, PCI, or CABG MS-DRG in the chained anchor admission with the highest IPPS weight. If a CABG MS-DRG occurs in a chained anchor hospitalization that was initiated with an AMI MS-DRG or PCI MS-DRG without an intracardiac ICD-CM procedure code in any position on the corresponding inpatient claim, we propose that the AMI model episode would begin with and be attributed to the first hospital, and we propose to set the price MS-DRG to the CABG MS-DRG in the chained anchor Start Printed Page 50850hospitalization with the highest IPPS weight.

If the price MS-DRG is an AMI or PCI MS-DRG, we propose to set the episode benchmark price as the standard AMI model-episode benchmark price for the price MS-DRG, subject to a possible adjustment for readmission for CABG MS-DRGs, as described in section III.D.4.b.(2)(c) of this proposed rule. If the price MS-DRG is a CABG MS-DRG, we propose to set the AMI model-episode benchmark price as the CABG model-episode benchmark price for the corresponding CABG MS-DRG, with no further adjustment in the event of a readmission for CABG MS-DRGs.

Table 7 displays the weights for CABG, PCI, and AMI MS-DRGs established in the FY 2016 IPPS final rule, which are subject to change each FY through the annual IPPS rulemaking (80 FR 49325 through 49886).

Table 7—FY 2016 IPPS Weights for MS-DRGS 231-236, 246-251, and 280-282

MS-DRGMS-DRG titleWeights
231CORONARY BYPASS W PTCA W MCC7.8056
232CORONARY BYPASS W PTCA W/O MCC5.7779
233CORONARY BYPASS W CARDIAC CATH W MCC7.3581
234CORONARY BYPASS W CARDIAC CATH W/O MCC4.9076
235CORONARY BYPASS W/O CARDIAC CATH W MCC5.8103
236CORONARY BYPASS W/O CARDIAC CATH W/O MCC3.8013
246PERC CARDIOVASC PROC W DRUG-ELUTING STENT W MCC OR 4+ VESSELS/STENTS3.2494
247PERC CARDIOVASC PROC W DRUG-ELUTING STENT W/O MCC2.1307
248PERC CARDIOVASC PROC W NON-DRUG-ELUTING STENT W MCC OR 4+ VES/STENTS3.0696
249PERC CARDIOVASC PROC W NON-DRUG-ELUTING STENT W/O MCC1.9140
250PERC CARDIOVASC PROC W/O CORONARY ARTERY STENT W MCC2.6975
251PERC CARDIOVASC PROC W/O CORONARY ARTERY STENT W/O MCC1.6863
280ACUTE MYOCARDIAL INFARCTION, DISCHARGED ALIVE W MCC1.6971
281ACUTE MYOCARDIAL INFARCTION, DISCHARGED ALIVE W CC1.0232
282ACUTE MYOCARDIAL INFARCTION, DISCHARGED ALIVE W/O CC/MCC0.7557

We believe that this proposal could minimize potential disincentives to AMI model participants from transferring patients when different or higher levels of care are needed. This is because the AMI model-episode benchmark prices we set would be more representative of the AMI spending based on the totality of care furnished during the chained anchor hospitalization and post-discharge period within the AMI model episode and for which the AMI model participants would be held accountable. We also believe that our proposal could encourage AMI model participants that frequently transfer patients after admission to improve their efficiency and the quality of care by transferring beneficiaries needing higher levels of care prior to hospital admission and managing those beneficiaries admitted to reduce the need for later transfers.

As an alternative, we also considered an approach where we would set the target price taking into consideration IPPS payments for both the MS-DRG assigned to the first admission in the chained anchor hospitalization and the MS-DRG assigned to the final admission in the chained anchor hospitalization. We could apply this approach to all AMI model participant hospitals or to only a subset of hospitals based on special situations that could lead to more common transfer scenarios that are unavoidable, such as small bed-size, rural location, interventional or cardiac surgery capacity, or other characteristic of the hospitals. All AMI model episodes involving chained anchor hospitalizations would include at least two IPPS payments for the chained anchor hospitalization, compared to one IPPS payment for most AMI episodes with only an anchor hospitalization that does not result in an inpatient-to-inpatient transfer. The alternative approach would likely result in a higher AMI-model episode benchmark price than under our proposal for AMI model episodes including a chained anchor hospitalization. Therefore, we believe this alternative approach could have the effect of further reducing potential disincentives to hospitals from transferring patients when different or a higher level of care is needed; however, we are not convinced this approach would ultimately improve care quality and efficiency under the AMI model.

First, we are concerned that this alternative approach could serve as an incentive for hospitals to admit and then transfer patients when doing so might not be medically necessary, which would neither enhance care quality nor efficiency. A recent study showed that non-procedure hospitals, defined as hospitals that lack onsite cardiac catheterization and coronary revascularization facilities, vary substantially in their use of the transfer process for Medicare beneficiaries admitted with AMI.[61] Beneficiaries transferred from hospitals that had a high transfer rate experienced greater use of invasive cardiac procedures after admission to the transfer hospital than beneficiaries transferred from hospitals with a low transfer rate. However, higher transfer rates were not associated with a significantly lower risk-standardized mortality rate at 30 days, and at one year, there was only a 1.1 percent mortality rate difference between hospitals with higher and lower transfer rates. As such, we believe this alternative approach could be appropriate for only a subset of AMI model participant hospitals based on specific hospital characteristics that could lead to a higher frequency of unavoidable transfers for AMI model beneficiaries rather than appropriate for hospitals overall. In addition, if we were to adopt this alternative approach, we believe it would also be necessary to incorporate methods for monitoring changes in the frequency of AMI model participant hospital patient transfers over the model's performance years, as well as assessing the appropriateness of those transfers. For example, to address changes in transfer frequency, we might compare how often an AMI model participant hospital transferred a beneficiary following an inpatient admission within each performance year relative to the frequency of transfers during its initial 3-year historical period. To address Start Printed Page 50851appropriateness of transfers, we might consider reviewing and comparing a sample of a hospital's transfers within a performance year as compared to the historical period. Furthermore, we might also propose future changes to this approach where changes in the frequency or appropriateness of transfers were identified.

Second, in contrast to our proposal, we believe that this alternative approach would not have the benefit of encouraging AMI model participant hospitals to make an early decision and transfer patients prior to rather than following inpatient admission when doing so prior to admission would be appropriate for the beneficiary's clinical circumstances and the hospital's capabilities. While we recognize that in some cases, an AMI model beneficiary admitted to the initial treating hospital may need to be transferred to a referral hospital that can provide a different or higher level of care, we believe it is important that the AMI model's payment methodology support the goal of rapid decision-making by the AMI model participant hospital about the AMI model beneficiary's care pathway based on clinical guidelines that often incorporate a time dimension in the guidelines for care.

Thus, on balance, we believe our proposed methodology would best establish appropriate incentives to improve care quality and efficiency under the AMI model by encouraging timely decisions about admission to the initial treating hospital and incentivizing only those transfers that are necessary to meet AMI model beneficiary's health care during the course of their hospitalization. Our proposal would adjust the AMI model-episode benchmark price that applies to the episode when a chained anchor hospitalization occurs and results in more costly care at the transfer hospital than would be expected based on the anchor MS-DRG at the initial treating hospital who would be accountable for the episode under the AMI model, thus accounting for the care at the referral hospital.

In contrast, some chained anchor hospitalizations could begin an episode based on an MS-DRG that anchors an episode in the model such as an AMI MS-DRGs that subsequently also includes an MS-DRG that does not anchor an episode under the model (for example, heart failure, renal failure, or cardiac valve replacement). Some of these non-anchor MS-DRGs could be related to the AMI episode but are unavoidable, for example, cardiac valve surgery, while others could potentially reflect complications resulting from inadequate care management during the episode (for example, heart or renal failure).

As discussed in section III.C.4.b. of this proposed rule, we propose to cancel an AMI model episode when the final MS-DRG in a chained anchor hospitalization is from an MS-DRG that would not an anchor MS-DRG under the AMI or CABG model. We believe that, in tandem, these proposals would allow for appropriate pricing of AMI model episodes that continue and include chained anchor hospitalizations.

The proposals to establish pricing for AMI model episodes involving chained anchor hospitalizations are included in § 512.300(c)(7)(i). We seek comment on our proposals for pricing AMI episodes involving chained anchor hospitalizations and the alternative proposals we considered. We also seek comment on the alternative considered that would account for both the MS-DRGs at the first and last hospitals caring for the AMI model beneficiary during the chained anchor hospitalization in setting the AMI-model episode benchmark price for episodes involving a chained anchor hospitalization. In particular, under such an alternative, we seek comment on the clinical circumstances in which inpatient-to-inpatient transfers are unavoidable and whether or not there are hospital characteristics that would lead us to expect higher frequencies of unavoidable inpatient-to-inpatient transfers for AMI model beneficiaries than hospitals overall. We also seek comment on how we could discourage unintended consequences under this alternative, such as less timely decisions about the most appropriate hospital to treat the beneficiary and increased beneficiary transfers that are unnecessary or inappropriate for improved quality of AMI model episode care.

(b) Adjustments for CABG Model Episodes

Among Medicare beneficiaries historically discharged under a CABG MS-DRG, average episode spending was substantially higher for those beneficiaries who also had AMI ICD-CM diagnosis codes on their inpatient claims ($57,000) than those who did not ($44,000).[62] About 30 percent of CABG beneficiaries had AMI ICD-CM diagnosis codes on their claims, while about 70 percent did not, and this percentage of CABG beneficiaries with AMI varied substantially across IPPS hospitals furnishing CABG procedures.[63] While average spending, in total, was substantially higher for CABG beneficiaries with AMI than without AMI, average spending during the anchor hospitalization was not substantially higher. Rather, much of this variation in CABG model episode spending occurred after discharge from the anchor hospitalization and correlated both with the presence of AMI and whether the CABG beneficiary was discharged from the anchor hospitalization in a CABG MS-DRG with major complication or comorbidity (MS-DRGs 231, 233, or 235) as opposed to a CABG MS-DRG without major complication or comorbidity (MS-DRGs 232, 234, or 236). Specifically, we found that average CABG episode spending after discharge from the anchor hospitalization was—

  • $9,000 for non-AMI CABG beneficiaries discharged from MS-DRGs 232, 234, or 236;
  • $11,000 for CABG beneficiaries with AMI discharged from MS-DRGs 232, 234, or 236;
  • $16,000 for non-AMI CABG beneficiaries discharged from MS-DRGs 231, 233, or 235; and
  • $20,000 for CABG beneficiaries with AMI discharged from MS-DRGs 231, 233, or 235.[64]

Thus, for CABG model episodes, we propose to set CABG model-episode benchmark prices by first splitting historical CABG model-episode expenditures into expenditures that occurred during anchor hospitalizations and expenditures that occurred after discharge from the anchor hospitalizations.

We propose to calculate the CABG anchor hospitalization benchmark price by following the general payment methodology that was applied to the CJR model, with expenditures limited to those that occurred during the anchor hospitalization and risk stratification according to the price CABG MS-DRG (80 FR 73337 through 73358).

We also propose to calculate the CABG post-anchor hospitalization benchmark price by following the general payment methodology that was applied to the CJR model, with Start Printed Page 50852expenditures limited to those that occurred after the anchor hospitalization and risk-stratification according to the presence of an AMI ICD-CM diagnosis code on the anchor inpatient claim and whether the price MS-DRG is a CABG MS-DRG with major complication or comorbidity (231, 233, or 235) or a CABG MS-DRG without major complication or comorbidity (232, 234, or 236) (80 FR 73337 through 73358).

We propose that the CABG model-episode benchmark price for an episode would be the sum of the corresponding CABG anchor hospitalization benchmark price and the corresponding CABG post-anchor hospitalization benchmark price, as discussed in this section and in III.D.4.d.

The proposals to establish pricing for CABG model episodes are included in § 512.300(c)(7)(ii). We seek comment on our proposals to establish pricing for CABG model episodes.

(c) Adjustments for Certain AMI Model Episodes With CABG Readmissions

In section III.C.4.b of this proposed rule, we discuss AMI model episodes where a beneficiary is discharged from an AMI model participant under an AMI MS-DRG and is later readmitted for a CABG. In that section, we did not propose to cancel the AMI model episode altogether for a CABG readmission during the 90-day post-hospital discharge period or cancel the AMI model episode and initiate a CABG model episode because planned CABG readmission following an anchor hospitalization that initiates an AMI episode may be an appropriate clinical pathway for certain beneficiaries. For example, we noted that historically approximately 10 percent of those AMI beneficiaries who received CABGs during AMI episodes would receive the CABGs between 2 and 90 days post-discharge from the anchor hospitalization, and most of those readmissions did not occur through hospital emergency departments. Even though CABG readmissions are not excluded from AMI model episodes (because they are clinically-related to the AMI model episode), we propose to provide an adjusted AMI model-episode benchmark price in such circumstances so as not to financially penalize AMI model participants for relatively uncommon, costly, clinically-appropriate care patterns for AMI model beneficiaries. Accordingly, we are proposing to establish an adjusted CABG-readmission AMI model-benchmark episode price for AMI model episodes with a price MS-DRG of 280-282 or 246-251 that have readmission for a CABG MS-DRG 231-236.

Specifically, if a CABG readmission occurs during an AMI model episode with a price MS-DRG of 280-282 or 246-251, we propose to calculate a CABG-readmission AMI model-episode benchmark price equal to the sum of the standard AMI model-episode benchmark price corresponding to the price MS-DRG (AMI MS-DRGs 280-282 or PCI MS-DRGs 246-251) and the CABG anchor hospitalization benchmark price corresponding to the MS-DRG of the CABG readmission. Because the adjustment would be based on the anchor hospitalization benchmark price, which does not include costs associated with the post-discharge period for CABG, this adjustment approach would avoid “double counting” post-discharge costs. Because adjusting for spending that occurred during a CABG readmission accounts for most of the spending variation between AMI model episodes with a CABG readmission and AMI model episodes without a CABG readmission, we propose no additional adjustment to the price for AMI model episodes with a CABG readmission.

In the event of any other readmission other than CABG during an AMI model episode that is not excluded from the AMI model episode definition, we would apply the usual rules of EPM-episode pricing that would include the spending for the related readmission in the actual AMI model-episode spending, without other adjustments. Fewer than 3 percent of those AMI model beneficiaries who receive inpatient or outpatient PCIs during AMI episodes receive the PCIs between 2 and 90 days post-discharge from the anchor or chained anchor hospitalizations, and we do not propose to make a pricing adjustment for PCIs that occur later in the AMI model episodes after discharge from the anchor or chained anchor hospitalizations. Since a PCI for an AMI typically is provided during the anchor or chained anchor hospitalization and most PCIs later in an episode occur in the context of a beneficiary presenting through the emergency department, we believe that the beneficiary likely has experienced a complication of care resulting in a PCI that may potentially be avoided through care management during the AMI model episode. Given that our intention is to offer appropriate incentives for care quality and efficiency by holding AMI model participants accountable for readmissions that could be related to the quality of care provided prior to the readmission, we believe that an adjustment other than for a CABG readmission would not be appropriate.

The proposal for adjusting episodes involving CABG readmissions is included in § 512.300(c)(7)(iii). We seek comment on our proposal for adjusting episodes involving CABG readmissions.

(d) Potential Future Approaches to Setting Target Prices for AMI and Hip Fracture Episodes

As previously described, our proposed approach for pricing AMI and CABG model episodes for beneficiaries with AMI sets different episode target prices depending upon whether the beneficiary is managed medically, undergoes PCI, or undergoes CABG during the acute phase of the episode, as well as whether the episode involves a chained anchor hospitalization or CABG readmission. Similarly, the target price set for beneficiaries experiencing hip fracture would depend on whether the patient undergoes hip fixation (and therefore initiates a SHFFT model episode) or hip arthroplasty (and therefore initiates a CJR model episode). We believe that this is a prudent approach that both recognizes the resource costs of services provided while encouraging care redesign during the portions of these episodes that we believe present the greatest opportunities to improve the quality and efficiency of the care delivered.

However, we note that the general principle guiding our payment reform efforts is that the payment system should hold providers accountable for the overall quality and cost of the care their beneficiaries receive rather than setting their payment based on the specific services delivered or settings in which they are delivered. We believe that this approach gives providers maximum flexibility to redesign care in ways that both produce the best outcomes for patients and controls the growth in spending for these services.

For this reason, we are interested in exploring future approaches to episode payment that would set an inclusive target price for episodes for beneficiaries with AMI that does not depend on whether the beneficiary is managed medically or receives PCI or CABG during the acute portion of the episode and, similarly, future approaches that would set prices for episodes for beneficiaries with hip fracture that do not depend on whether the beneficiary undergoes hip fixation or hip arthroplasty. While we believe that the choice of treatment during the acute phase of these episodes may be determined predominantly by clinical factors such that financial factors may play a smaller role in shaping episode Start Printed Page 50853care redesign than they do following hospital discharge, we nevertheless believe it would be valuable to consider testing an inclusive episode payment model. Providers may be able to redesign and implement care pathways that we might not have otherwise anticipated, especially as the evidence-base for AMI and hip fracture treatment continues to grow and evolve.

We seek comment on this type of approach to setting an inclusive episode target price and on any episode payment model design features that would be needed to make such an approach successful. In particular, we seek comment on potential approaches to risk-adjustment aimed at ensuring that providers are appropriately paid for caring for high-complexity episode beneficiaries in the context of this alternative approach. We would seek to ensure that all providers caring for these episode beneficiaries, including those providers for which we propose additional protections and those that serve a high percentage of potentially vulnerable populations of medically and socially complex patients as discussed in section III.D.7.c. of this proposed rule, would not bear undue financial risk and to mitigate any incentives to avoid caring for high-complexity patients. In addition, we seek comment on whether and how our methodology linking quality performance to payment under the proposed EPMs and the CJR model might need to be modified in the context of this alternative approach that would set an inclusive episode target price, in order to appropriately incentivize the delivery of high-quality care and discourage stinting on appropriate care.

(e) Summary of Pricing Methodologies for AMI, CABG, and SHFFT Model Episode Scenarios

Tables 8 through 10 summarize the standard pricing methodologies and the adjustments that would occur that are proposed in sections III.D.4.b.(1) and (2) of this proposed rule for AMI, CABG, and SHFFT model episodes.

Table 8—AMI Model Pricing Scenarios

AMI pricing scenarioPrice
AMI Scenarios without Chained Anchor Hospitalization
Single hospital AMI MS-DRG or PCI MS-DRG (with AMI diagnosis)Episode benchmark price is standard episode benchmark price based on anchor MS-DRG (which is the price MS-DRG).
AMI Scenarios with Chained Anchor Hospitalizations
A chained anchor hospitalization where the discharge from the first hospital is an AMI MS-DRG or PCI MS-DRG (with AMI diagnosis) that results in a final discharge from an AMI, PCI, or CABG MS-DRG (transfer PCI and CABG MS-DRGs not required to have AMI ICD-CM diagnosis code)Episode benchmark price is the standard episode benchmark price or the CABG model episode benchmark price corresponding to price MS-DRG, assigned as the AMI, PCI, or CABG MS-DRG with highest IPPS weight.
If the price MS-DRG is a CABG MS-DRG, the CABG model episode benchmark price is the sum of the CABG anchor hospitalization price for the MS-DRG and the CABG post-anchor hospitalization price based on with AMI ICD-CM diagnosis code and whether the CABG MS-DRG is w/MCC or not.
AMI Scenarios with Readmissions
An AMI MS-DRG or PCI MS-DRG (with AMI diagnosis) anchored episode without a chained anchor hospitalization ongoing with CABG readmissionEpisode benchmark price is the sum of the standard episode benchmark price corresponding to the price MS-DRG and the CABG anchor hospitalization benchmark price corresponding to the CABG readmission MS-DRG.
AMI MS-DRG or PCI MS-DRG (with AMI diagnosis) anchored AMI episode with chained anchor hospitalization (not containing a CABG MS-DRG) ongoing with CABG readmissionEpisode benchmark price is the sum of the standard episode benchmark price for the price MS-DRG assigned to the chained anchor hospitalization and the CABG anchor hospitalization benchmark price corresponding to the CABG readmission MS-DRG.

Table 9—CABG Model Pricing Scenarios

CABG pricing scenarioPrice
Single hospital CABG MS-DRG with AMI diagnosisEpisode benchmark price is the sum of the CABG anchor hospitalization benchmark price for the MS-DRG and the CABG post-anchor hospitalization benchmark price based on the presence of an AMI ICD-CM diagnosis code and whether the anchor MS-DRG is w/MCC or w/o MCC.
Single hospital CABG MS-DRG without AMI diagnosisEpisode benchmark price is the sum of the CABG anchor hospitalization benchmark price for the MS-DRG and the CABG post-anchor hospitalization benchmark price based on no AMI ICD-CM diagnosis code and whether the anchor MS-DRG is w/MCC or w/o MCC.
Start Printed Page 50854

Table 10—SHFFT Model Pricing Scenarios

SHFFT Pricing scenarioPrice
SHFFT MS-DRGEpisode benchmark price is standard episode benchmark price based on anchor MS-DRG (which is the price MS-DRG).

(3) Three Years of Historical Data

As was the case for the CJR model, we propose to use 3 years of historical EPM episodes for calculating EPM participants' EPM-episode benchmark prices, with each set of historical episodes updated every other year (80 FR 73340 through 73341). Under our proposal, each of the first 2 years of historical data would be trended to the most recent of the 3 years, based on national trend factors for each combination of price MS-DRGs and payments would be updated for each payment system (for example, IPPS, PFS, etc.) based on annual changes in input costs (see sections III.D.4.b(4) and III.D.4.b(5) of this proposed rule that immediately follow). Under our proposal, we would establish historical EPM-episode payments based on episodes that started between—

  • January 1, 2013 and December 31, 2015 for performance years 1 and 2;
  • January 1, 2015 and December 31, 2017 for performance years 3 and 4; and
  • January 1, 2017 and December 31, 2019 for performance year 5.

We believe that 3 years of historical EPM-episode data should provide sufficient historical episode volume to reliably calculate EPM-episode benchmark prices, and that updating these data every other year would allow us to make the most current claims data available in a way that incorporates the effects of regular Medicare payment system updates and changes in utilization without creating uncertainty in pricing for EPM participants. We would further note that the effects of updating EPM-participant hospital-specific data on an EPM-episode's benchmark prices would diminish over time as the contribution of regional pricing on EPM benchmark prices will increase from one-third for performance years 1 and 2 to two-thirds in performance year 3, and 100 percent in performance years 4 and 5.

The proposal for 3 years of historical data updated every other year under the proposed EPMs is included in § 512.300(c)(1).

We seek comment on our proposal for 3 years of historical data updated every other year.

(4) Trending Historical Data to the Most Recent Year

We recognize that some payment variation could exist in the 3 years of historical EPM-episode data due to annual Medicare payment system updates (for example, IPPS, OPPS, IRF PPS, SNF PPS) and national changes in utilization patterns. Thus, EPM episodes in the third year of the 3 historical years might have higher average payments than those from the earlier 2 years, in part due to Medicare payment rate increases over the course of the 3-year period. Also, EPM-episode payments could change over time due to national trends reflecting changes in industry-wide practice patterns. For example, readmissions for all patients, including those in CABG model episodes, may decrease nationally due to improved industry-wide surgical protocols that reduce the chance of infections. We do not intend for the incentives under the EPMs to be affected by Medicare payment system rate changes that are beyond EPM participants' control or to provide reconciliation payments to (or require repayments from) EPM participants for achieving lower (or higher) Medicare expenditures solely because they followed national changes in practice patterns. Instead, we aim to incentivize EPM participants to improve care quality and efficiency based on their hospital-specific inpatient and post-discharge care practices under the EPMs.

To mitigate the effects of Medicare payment system updates and changes in national utilization practice patterns on the 3 years of historical episode data, we propose to apply a national trend factor to each of the years of historical EPM-episode payments as we do with the CJR model (80 FR 73341 through 73342). Specifically, we propose to inflate the 2 oldest years of historical EPM-episode payments for EPM episodes to the most recent year of the 3 historical years using changes in the national EPM-episode payments for each different type of EPM episode. That is, we propose to apply separate national trend factors for the following pricing scenarios:

  • SHFFT model episodes, separately by each price MS-DRG in 480-482.
  • AMI model episodes without CABG readmissions, separately by each price MS-DRG in 280-282 and 246-251; and
  • The anchor hospitalization portion of CABG model episodes, separately by each price MS-DRG in 231-236.
  • The post-anchor hospitalization portion of CABG model episodes, separately for:

++ With AMI ICD-CM diagnosis code on the anchor inpatient claim and CABG price MS-DRG with major complication or comorbidity (231, 233, or 235);

++ With AMI ICD-CM diagnosis code on the anchor inpatient claim and CABG price MS-DRG without major complication or comorbidity (232, 234, or 236);

++ Without AMI ICD-CM diagnosis code on the anchor inpatient claim and CABG price MS-DRG with major complication or comorbidity (231, 233, or 235); and

++ Without AMI ICD-CM diagnosis code on the anchor inpatient claim and CABG price MS-DRG without major complication or comorbidity (232, 234, or 236).

For example, when using Calendar Year (CY) 2013 through 2015 historical EPM-episode data to establish EPM-episode benchmark prices for performance years 1 and 2, we would calculate an aggregate national average SHFFT model episode payment in historical episodes with price MS-DRG 480 for each of the 3 historical years. To trend historical payments to the most recent year in an historical window, we would create a ratio based on national average historical EPM-episode payment for that episode type in a previous year and for the most recent year. Thus, in this example, we would create a ratio of national average SHFFT model historical episode payment with price MS-DRG 480 in CY 2015 as compared to that national average SHFFT model historical episode payment in CY 2013 in order to trend the CY 2013 historical SHFFT model episode payments to CY 2015. Similarly, we would determine the ratio of the national average SHFFT model historical episode payment for CY 2015 to national average SHFFT model historical episode payment in CY 2014 to trend 2014 SHFFT model episode payments to CY 2015. This process would be repeated for each pricing scenario previously listed.Start Printed Page 50855

We believe this method for trending data would capture updates in Medicare payment systems as well as national utilization pattern changes that might have occurred within that 3-year period. Moreover, as with the CJR model, we believe that adjusting for national rather than regional trends in utilization would be most appropriate as any Medicare payment system updates and significant changes in utilization practice patterns would not be region-specific but rather be reflected nationally.

The proposal for trending historical data is included in § 512.300(c)(11). We seek comment on our proposal for trending historical data.

(5) Update Historical EPM-Episode Payments To Account for Ongoing Payment System Updates

As previously mentioned, we propose to prospectively update the historical EPM-episode payments to account for ongoing updates to Medicare payment systems (for example, IPPS, OPPS, IRF PPS, SNF, PFS, etc.) in order to ensure we incentivize EPM participants based on historical utilization and practice patterns, not Medicare payment system rate changes that are beyond hospitals' control. Under our proposal, we would apply the same methodology developed for the CJR model to incorporate Medicare payment updates (80 FR 73342 through 73446).

Because Medicare payment systems rates are not updated at the same time during the year—for example, rates under the IPPS, IRF PPS, and SNF payment systems are updated effective October 1, while the hospital OPPS and MPFS rates are updated annually effective January 1—we propose to generally update historical EPM-episode payments and calculate EPM-episode benchmark prices separately for EPM episodes initiated between January 1 and September 30 versus October 1 and December 31 of each performance year, and at other intervals if determined necessary. The EPM-episode benchmark price in effect as of the day the EPM episode is initiated would be the EPM-episode benchmark price for the whole EPM episode. Note that for performance year 5, the second set of EPM-episode benchmark prices would be for EPM episodes that start and end between and including October 1 and December 31 because the fifth performance period of the SHFFT, CABG, and AMI models would end on December 31, 2021. Also, an EPM episode benchmark price for a given EPM performance year could be applied to EPM episodes included in another performance year. For example, an EPM episode initiated in November 2017, and ending in February 2018 would have an EPM-episode benchmark price based on the second set of 2017 EPM-episode benchmark prices (for EPM episodes initiated between October 1, 2017, and December 31, 2017), and it would be captured in the CY 2018 EPM performance year (performance year 2) because it ended between January 1, 2018, and December 31, 2018. We refer to section III.D.2.a. of this proposed rule for further discussion on the definition of EPM performance years.

We propose to update historical EPM-episode payments by applying separate Medicare payment system update factors each January 1 and October 1 to each of the following six components of each EPM participant's historical EPM-episode payments:

  • Inpatient acute.
  • Physician.
  • IRF.
  • SNF.
  • HHA.
  • Other services.

A different set of update factors would be calculated for January 1 through September 30 versus October 1 through December 31 EPM episodes each EPM performance year. The six update factors for each of the previously stated components would be EPM-participant hospital-specific and would be weighted by the percent of the Medicare payment for which each of the six components accounts in the EPM participant's historical EPM episodes. The weighted update factors would be applied to historical EPM-participant hospital-specific average payments to incorporate ongoing Medicare payment system updates. A weighted update factor would be calculated by multiplying the component-specific update factor by the percent of the EPM participant's historical EPM-episode payments the component represents, and summing together the results. Each of an EPM participant's six update factors would be based on how inputs have changed in the various Medicare payment systems for the specific EPM participant.

As an example, we will assume for purposes of this example that 50 percent of an EPM participant's historical EPM-episode payments were for inpatient acute care services, 15 percent were for physician services, 35 percent were for SNF services, and 0.0 percent were for the remaining services. We will also assume for purposes of this example that the update factors for inpatient acute care services, physician services, and SNF services are 1.02, 1.03, and 1.01, respectively. The weighted update factor in this example would be the following: (0.5 * 1.02) + (0.15 * 1.03) + (0.35 * 1.01) = 1.018. The EPM participant in this example would have its historical average EPM-episode payments multiplied by 1.018 to incorporate ongoing payment system updates. The specific order of steps, and how this step fits in with others, is discussed further in sections III.D.4.c through d. of this proposed rule. Also, as discussed further in sections III.D.4.c. through d. the update factors would vary by price MS-DRG. For example, in CABG model episodes, the update factors would be calculated separately for the anchor hospitalization portion of episodes and the post-anchor hospitalization portion of episodes, as described in section III.D.4.d.

Region-specific update factors for each of the previously stated components and weighted update factors would also be calculated in the same manner as the EPM-participant hospital-specific update factors. Instead of using historical EPM episodes attributed to a specific hospital, region-specific update factors would be based on all historical EPM episodes initiated at any IPPS hospital within the region with historical EPM episodes, regardless of whether or not the MSAs in which the hospitals are located were selected for inclusion in the models. We refer to the CJR Final Rule for further discussion of our specific methodology and considerations for adopting this methodology for updating historical EPM-episode payments for ongoing payment system updates (80 FR 73342 through 73446).

The proposal for updating episode payments for ongoing annual Medicare payment updates is included in § 512.300(c)(10). We seek comment on our proposal for updating episodes payments for ongoing annual Medicare payment updates.

(6) Blend Hospital-Specific and Regional Historical Data

We propose to calculate EPM-episode benchmark prices using a blend of EPM-participant hospital-specific and regional historical average EPM-episode payments, including historical EPM-episode payments for all IPPS hospitals that are in the same U.S. Census division, which is discussed further in section III.D.4.b.(7) of this proposed rule. Specifically, we propose to blend two-thirds of the EPM-participant hospital-specific historical EPM-episode payments and one-third of the regional historical EPM-episode payments to set an EPM participant's EPM-episode benchmark prices for the first 2 performance years of the proposed EPMs (CYs 2017 and 2018). For performance year 3 of the EPMs (CY Start Printed Page 508562019), we propose to adjust the proportion of the EPM-participant hospital-specific and regional historical EPM-episode payments used to calculate the EPM-episode benchmark prices from two-thirds EPM-participant hospital-specific and one-third regional to one- third EPM-participant hospital-specific and two-thirds regional. Finally, we propose to use only regional historical EPM-episode payments for performance years 4 and 5 of the EPMs (CYs 2020 and 2021) to set an EPM participant's EPM episode-benchmark prices, rather than a blend between the hospital-specific and regional historical EPM episode payments.

Consistent with our methodology for the CJR model, we propose two exceptions. First, we propose to use only regional historical EPM-episode payments to calculate EPM episode- benchmark prices for EPM participants with low historic EPM-episode volume (80 FR 73544). For SHFFT model episodes, this exception applies to SHFFT model participants with fewer than 50 historical SHFFT model episodes in total across the 3 historical years. For AMI model episodes anchored by MS-DRGs 280-282, this exception applies to AMI model participants with fewer than 75 of these particular AMI model historical episodes in total across the 3 historical years. For AMI model episodes anchored by PCI MS-DRGs 246-251, this exception applies to AMI model participants with fewer than 125 of this particular AMI model historical episodes in total across the 3 historical years. For CABG model episodes, this exception applies to CABG model participants with fewer than 50 historical CABG model episodes in total across the 3 historical years. The proposed thresholds for low historic volume in this proposed rule are higher than the CJR model threshold for low historical LEJR episode volume of 20 episodes in total across the 3 historical years. The higher thresholds are based on the volume thresholds from the BPCI Model 2 Risk Track B for 90-day episodes, which increase when the ratio of within-hospital episode spending variation to between-hospital episode spending variation increases. That is, as EPM episode payment variation increases within a hospital relative to EPM-episode payment variation between hospitals, it is necessary to have more EPM episodes at that hospital to estimate a stable EPM- episode benchmark price using data from only that hospital. We propose to set higher thresholds for the SHFFT, AMI, and CABG models based on internal analysis from BPCI episode data that shows higher within-hospital episode spending variation relative to between-hospital episode spending variation for episodes anchored by the EPM MS-DRGs, compared to episodes anchored by MS-DRGs 469 and 470 included in the CJR model.[65]

Second, in the case of an EPM participant that has undergone a merger, consolidation, spin-off, or other reorganization that results in a new hospital entity without 3 full years of historical claims data, we propose that EPM participant hospital-specific historical EPM-episode payments would be determined using the historical EPM episode payments attributed to their predecessor(s), as in the CJR model (80 FR 73544).

The aforementioned proposals align with our method for blending EPM participant hospital-specific and regional data under the CJR model. We refer to the CJR model Final Rule for further discussion on alternatives to and reasons for adopting this methodology for the CJR model, which informs our proposal with respect to the proposed EPMs (80 FR 73346-73349).

The proposal for blending payments when establishing participants' benchmark and quality-adjusted targets and certain exceptions is included in § 512.300(c)(2), (3), and (4). We note that the specific order of steps, and how this step fits in with others, is discussed further in section III.D.4.c. of this proposed rule. We seek comment on our proposal for blending payments when establishing participants' benchmark and quality-adjusted targets as well as the proposed exceptions.

(7) Define Regions as U.S. Census Divisions

As we do for the CJR model, for all 5 performance years, we proposed to define “region” as one of the nine U.S. Census divisions [66] in Figure 1 (80 FR 73349 through 73350).

Start Printed Page 50857

We believe U.S. Census divisions provide the most appropriate balance between very large areas with highly disparate utilization patterns and very small areas that would be subject to price distortions due to low volume or hospital-specific utilization patterns. We clarify that we would ascribe the same regional component of EPM-episode benchmark prices for EPM participants in MSAs that span U.S. Census divisions. That is, selected MSAs that span U.S. Census divisions would be attributed to one U.S. Census division for purposes of calculating the regional component of an EPM-episode benchmark price. Specifically, we will attribute an MSA to the U.S. Census division in which the majority of people in the MSA reside.

The proposal to define a region as one of the nine U.S. Census divisions is included in § 512.300(c)(2). We seek comment on our proposal to define region in this manner.

(8) Normalize for Provider-Specific Wage Adjustment Variations

Some variation in historical EPM-episode payments across hospitals in a region may be due to wage adjustment differences in Medicare payments. In setting Medicare payment rates, Medicare typically adjusts facilities' costs attributable to wages and wage-related costs (as estimated by the Secretary from time to time) by a factor (established by the Secretary) that reflects the relative wage level in the geographic are of the facility or practitioner (or the beneficiary's residence, in the case of home health and hospice services) compared to a national average wage level. Such adjustments are essential for setting accurate payments, as wage levels vary significantly across geographic areas of the country. However, having the wage level for one hospital influence the regional-component of another hospital's EPM episode-benchmark price with a different level would introduce unintended pricing distortion not based on utilization pattern differences.

To preserve how wage levels affect provider payment amounts, while minimizing the distortions introduced when calculating the regional-component of blended EPM-episode benchmark prices, we propose to normalize for wage indices at the claim level for both historical EPM-episode payments and actual EPM-episode payments. As discussed in section III.D.3.b. of proposed rule, we propose to utilize the CMS Price (Payment) Standardization Detailed Methodology to calculate EPM-episode benchmark and quality-adjusted target prices and actual EPM-episode spending. This methodology removes wage level differences in calculating standardized payment amounts.

We believe it is important to reintroduce wage index variations near the end of the EPM-episode price-setting methodology and when calculating actual EPM-episode payments during an EPM performance year, to account for the differences in cost for care redesign across different geographic areas of the country. For example, hiring additional hospital staff to aid in patient follow-up during the post-discharge period of an AMI model episode would be significantly more costly in San Francisco than in rural Idaho. If we do not reintroduce wage index variations into EPM-episode benchmark price and actual EPM-episode payment calculations, we would calculate reconciliation and repayment amounts that would not capture labor cost variation throughout the country, and EPM participants in certain regions may see less opportunity and financial incentive to invest in care redesign. Thus, when setting EPM-episode benchmark prices and calculating actual EPM-episode payments, we propose to reintroduce the hospital-specific wage variations by multiplying EPM-episode payments by the wage normalization factor when calculating the EPM-episode benchmark prices and actual EPM -episode payments for each EPM participant, as described in section III.D.4.c. of the proposed rule.

We propose to use the following algorithm to create a wage normalization factor: 0.7 * IPPS wage index + 0.3. The 0.7 approximates the Start Printed Page 50858labor share in IPPS, IRF PPS, SNF, and HHA Medicare payments. The specific order of steps, and how this step fits in with others, is discussed further in section III.D.4.c. of the proposed rule. We refer to the CJR model Final Rule for more detailed information on our normalization process adopted for the CJR model (80 FR 73350 through 73352).

The proposal to normalize for provider-specific wage adjustment variations is included in § 512.300(c)(12). We seek comment on our proposal to normalize for these variations.

(9) Combining Episodes To Set Stable Benchmark and Quality-Adjusted Target Prices

For the purposes of having sufficient episode volume to set stable EPM-episode benchmark and quality-adjusted target prices, we propose generally to follow the process from the CJR model to calculate severity factors, EPM-participant hospital-specific weights, and region-specific weights that allow us to surmount issues of low volume for EPM episodes with particular characteristics by aggregating EPM episodes and portions of EPM episodes across dimensions that include anchor MS-DRGs, the presence of AMI ICD-CM diagnosis code on the anchor inpatient claim, and the presence of a major complication or comorbidity for anchor CABG MS-DRGs (80 FR 73352 through 73353). Where the CJR Final Rule refers to anchor factors, for the purposes of this proposed rule we refer to severity factors to avoid confusion when performing calculations pertaining to expenditures that occurred during the anchor hospitalization and after the anchor hospitalization in CABG model episodes.

For SHFFT model episodes, we propose to combine episodes with price MS-DRGs 480-482 to use a greater historical episode volume to set more stable SHFFT episode benchmark and quality-adjusted target prices. To do so, we propose to calculate severity factors for episodes with price MS-DRGs 480 and 481 equal to—

The national average would be based on SHFFT model episodes attributed to any IPPS hospital. The resulting severity factors would be the same for all SHFFT model participants. For each SHFFT model participant, a hospital weight would be calculated using the following formula, where SHFFT model episode counts are SHFFT-model-participant hospital-specific and based on the SHFFT model episodes in the 3 historical years used in SHFFT model episode benchmark and quality-adjusted target price calculations:

A SHFFT model participant's hospital-specific average episode payment would be calculated by multiplying such participant's hospital weight by its combined historical average episode payment (sum of historical episode payments for historical episodes with price MS-DRGs 480-482 divided by the number of historical episodes with price MS-DRGs 480-482). The calculation of the hospital weights and the hospital-specific pooled historical average episode payments would be comparable to how case-mix indices are used to generate case-mix adjusted Medicare payments. The hospital weight essentially would count each episode with price MS-DRGs 480 and 481 as more than one episode (assuming episodes with price MS- DRGs 480 and 481 have higher average payments than episodes with price MS-DRG 482) so that the pooled historical average episode payment, and subsequently the SHFFT model episode benchmark and quality-adjusted target prices, are not skewed by the SHFFT model participant's relative breakdown of historical episodes with price MS-DRGs 480 and 481 versus historical episodes with price MS-DRG 482.

We would calculate region-specific weights and region-specific pooled historical average payments following the same steps proposed for hospital-specific weights and hospital-specific pooled average payments. Instead of grouping episodes by the attributed hospital as is proposed for hospital-specific calculations, region-specific calculations would group together SHFFT model episodes that were attributed to any IPPS hospital located within the region. The hospital-specific and region-specific pooled historical average payments would be blended together as discussed in section III.D.4.b.(6) of the proposed rule. The specific order of steps, and how this step fits in with others, is discussed further in section III.D.4.c. of the proposed rule.

Afterwards, the blended pooled calculations would be ”unpooled” by setting the episode benchmark price for episodes with price MS-DRG 482 to the resulting calculation, and by multiplying the resulting calculation by the severity factors to produce the episode benchmark prices for episodes with price MS-DRGs 480 and 481. Applying the discount factor as discussed in III.D.4.b.(10) and III.D.4.c. would result in the SHFFT model quality-adjusted target prices for episodes with price MS-DRGs 480-482.

For episodes in the AMI model with price MS-DRGs in 280-282 or 246-251 Start Printed Page 50859and without readmissions for CABG MS-DRGs, we propose to follow an analogous procedure to the SHFFT model with the following modifications. First we propose to group episodes with price MS-DRGs 280-282 separately from episodes with price MS-DRGs 246-251 for the calculations. Second, we propose to calculate severity factors for episodes with price MS-DRGs 280-282 as—

Third, we propose to calculate hospital-specific weights and region-specific weights for episodes with price MS-DRGs 280-282 as—

Fourth, we propose to calculate severity factors for episodes with price MS-DRG 246-251 as—

Fifth, we propose to calculate hospital-specific weights and region-specific weights for episodes with price MS-DRG 246-251 as—

Start Printed Page 50860

After blending historical and regional pooled episode payments for episodes with price MS-DRGs 280-282, the blended pooled calculations would be ”unpooled” by setting the episode benchmark price for price MS-DRG 282 to the resulting calculation, and by multiplying the resulting calculation by the severity factors to produce the episode benchmark prices for price MS-DRGs 280 and 281.

After blending historical and regional pooled episode payments for episodes with price MS-DRGs 246-251, the blended pooled calculations would be ”unpooled” by setting the episode benchmark price for price MS-DRG to the resulting calculation, and by multiplying the resulting calculation by the severity factors to produce the episode benchmark prices for price MS-DRGs 246-251.

Applying the discount factor as discussed in III.D.4.b.(10) and III.D.4.c would result in the quality-adjusted target prices for price MS-DRGs 280-282 and 246-251.

For episodes in the CABG model with price MS-DRGs in 231-236, we propose to calculate severity factors, hospital-specific weights, and region-specific weights separately for the anchor hospitalization portion of CABG model episodes and the post-anchor hospitalization portion of CABG model episodes.

For the anchor hospitalization portion of CABG model episodes, we propose to follow an analogous procedure to the SHFFT model with the anchor hospitalization portion of CABG model episodes grouped by the price MS-DRG. Specifically, we propose to calculate anchor hospitalization severity factors for price MS-DRGs 231-235 as—

We also propose to calculate hospital-specific weights and region-specific weights for the anchor hospitalization portion of CABG model episodes as—

After blending historical and regional pooled anchor hospitalization payments for the CABG model episodes, the blended pooled calculations would be ”unpooled” by setting the price MS-DRG 236 anchor hospitalization benchmark price to the resulting calculation, and by multiplying the resulting calculation by the severity factors to produce the anchor hospitalization benchmark prices for price MS-DRGs 231-235.

For the post-anchor hospitalization portion of CABG model episodes, we propose to follow an analogous procedure to the SHFFT model with the post-anchor hospitalization portion of Start Printed Page 50861CABG model episodes grouped in the following manner—

  • With AMI diagnosis on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235)
  • With AMI diagnosis on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236)
  • Without AMI diagnosis on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235)
  • Without AMI diagnosis on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236)

Specifically, we propose to calculate post-anchor hospitalization severity factors as—

We also propose to calculate hospital-specific weights and region-specific weights for the post-anchor hospitalization portion of CABG model episodes as—

After blending historical and regional pooled post-anchor hospitalization payments for the CABG model episodes, the blended pooled calculations would be ”unpooled” by setting the without AMI ICD-CM diagnosis code on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236) post-anchor hospitalization benchmark price to the resulting calculation, and by multiplying the resulting calculation by the severity factors to produce the post-anchor hospitalization benchmark prices for:

  • With AMI diagnosis on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235)
  • With AMI diagnosis on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236)
  • Without AMI diagnosis on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235)

We propose to calculate episode benchmark prices for CABG model episodes by summing combinations of CABG anchor hospitalization benchmark prices and CABG post-anchor hospitalization benchmark prices. Applying the discount factor as discussed in III.D.4.b.(10) and III.D.4.d of this proposed rule would result in the quality-adjusted target prices for CABG model episodes.

For episodes in the AMI model with CABG readmissions, we propose to perform no additional blending of hospital-specific and regional-specific episode payments. We propose to calculate the AMI model episode benchmark and quality-adjusted target prices for such episodes as described in section III.D.4.e. of this proposed rule.

The proposals to combine episodes to set stable benchmark and quality-adjusted target prices are included in § 512.300(c)(13). We seek comment on our proposals for combining episodes for these purposes.

(10) Effective Discount Factors

As discussed in section III.D.2.c. of this proposed rule, we propose to make EPM participants partly or fully accountable for EPM-episode payments in relationship to the EPM quality-adjusted target price. As part of this, in setting an episode quality-adjusted target price for an EPM participant, we propose to apply an effective discount factor to an EPM participant's hospital-Start Printed Page 50862specific and regional blended historical EPM-episode payments for a performance period. We expect EPM participants to have a significant opportunity to improve the quality and efficiency of care furnished during episodes in comparison with historical practice, because the EPMs would facilitate the alignment of financial incentives among providers caring for EPM beneficiaries. Our proposed effective discount factors are intended to serve as Medicare's portion of reduced expenditures from an EPM episode with any EPM-episode expenditures below the quality-adjusted target price potentially available as reconciliation payments to the EPM participant where the anchor hospitalization occurred.

For the EPMs, we propose to establish a 3 percent effective discount factor to calculate the quality-adjusted target prices for EPM participants in the below acceptable and acceptable quality categories, as discussed in section III.E.3.f. of this proposed rule and similar to the CJR model (80 FR 73355). The effective discount factor to calculate the quality-adjusted target price for EPM participants in the good and excellent quality categories would be 2 percent and 1.5 percent, respectively.

Because of the proposed phase-in of repayment responsibility as discussed in section III.D.2.c. of this proposed rule, with no responsibility in either performance year 1 or performance year 2 (NDR) and only partial repayment responsibility in performance year 2 (DR) and all of performance year 3, an EPM participant with actual EPM-episode payments that exceed the quality-adjusted target prices multiplied by the EPM participant's number of EPM episodes to which each quality-adjusted target price would apply in performance year 2 (DR) and performance year 3 would owe Medicare less that would otherwise result from this calculation. As discussed in section III.E.3.f of this rule, an “applicable discount factor” applies to repayment amounts in performance years 2 and 3 while this repayment responsibility is being phased-in. We refer to section III.E.1. and specifically Tables 20 through 28 in this proposed rule for further illustration of the discount percentages that would apply for reconciliation payment and Medicare repayment over the 5 EPM performance years. We believe this methodology offers EPM participants an opportunity to create savings for themselves and Medicare, while also maintaining or improving quality of care for EPM model beneficiaries.

The proposal to establish discount factors that would apply to the quality categories is included in § 512.300(d). We seek comment on our proposal to establish discount factors that apply to the quality categories.

c. Approach To Combine Pricing Features for all SHFFT Model Episodes and AMI Model Episodes Without CABG Readmissions

The following presents our proposed methodology for combining the pricing features presented in section III.D.4.b. of this proposed rule with respect to SHFFT model episodes and AMI model episodes without a CABG readmission.

  • Step 1—Calculate historical EPM-episode payments for episodes that were initiated during the 3-historical-years of each applicable EPM (that is, individually for each of the SHFFT and AMI models) (section III.D.4.b.(3) of this proposed rule) for all IPPS hospitals for all Medicare Part A and B services included in the EPM episodes. Limit the potential AMI model episodes to those episodes with price MS-DRGs in 280-282 or 246-251 and without readmissions for CABG MS-DRGs. We note that specific PBPM payments may be excluded from historical EPM-episode payment calculations as discussed in section III.D.6.d. of this proposed rule.
  • Step 2—Remove the effects of special payment provisions (section III.D.3.b. of this proposed rule) and normalize for wage index differences (section III.D.4.b.(8) of this proposed rule) by standardizing Medicare FFS payments at the claim-level.
  • Step 3—Prorate Medicare payments for included episode services that span a period of care that extends beyond the episode (section III.D.3.c. of this proposed rule.).
  • Step 4—Trend forward the 2 oldest historical years of data to the most recent year of historical data (section III.D.4.b.(4) of this proposed rule). Separate national trend factors would be applied for each combination of price MS-DRGs.
  • Step 5—Cap high episode payment episodes with a region- and price-MS-DRG-specific high payment ceiling (section III.D.3.d. of this proposed rule), using the episode output from the previous step.
  • Step 6—Group episodes based on price MS-DRGs (SHFFT MS-DRGs 480-482; AMI MS-DRGs 280-282; PCI MS-DRGs 246-251). Within each group of episodes, calculate severity factors and EPM-participant hospital-specific weights (section III.D.4.b.(9) of this proposed rule) using the episode output from the previous step to pool together episodes in each group of price MS-DRGs, resulting in EPM-participant hospital-specific pooled historical average episode payments for each group of price MS-DRGs. Similarly, calculate region-specific weights to calculate region-specific pooled historical average episode payments for each group of price MS-DRGs.
  • Step 7—Calculate EPM-participant hospital-specific and region-specific weighted update factors (section III.D.4.b.(5) of this proposed rule). Multiply each EPM-participant hospital-specific and region-specific pooled historical average episode payment by its corresponding EPM-participant hospital-specific and region-specific weighted update factors to calculate EPM-participant hospital-specific and region-specific updated, pooled, historical average episode payments.
  • Step 8—Blend together each EPM-participant hospital-specific updated, pooled, historical average episode payment with the corresponding region-specific updated, pooled, historical average episode payment according to the proportions for the EPM performance year (III.D.4.b.(6) of this proposed rule). EPM participants that do not have the minimum episode volume across the historical 3 years would use 0.0 percent and 100 percent as the proportions for hospital and region, respectively.
  • Step 9—Multiply the outputs of step (8) by the wage normalization factor described in section III.D.4.b.(8) of this proposed rule to reintroduce geographic variation. For purposes of this proposed rule, we will define the three outputs of this step as the standard episode benchmark price for—

++ SHFFT model episodes with price MS-DRG 482

++ AMI model episodes with price MS-DRG 282 without readmission for CABG, and

++ AMI model episodes with price MS-DRG 251 without readmission for CABG.

  • Step 10—Multiply the output of step (9) by the appropriate severity factors (step (6) of this calculation process and detailed in section III.D.4.b.(9) of this proposed rule) to calculate the standard episode benchmark prices for—

++ SHFFT model episodes with price MS-DRGs 480-481

++ AMI model episodes with price MS-DRGs 280-281 without readmission for CABG

++ AMI model episodes with price MS-DRGs 246-250 without readmission for CABG

Start Printed Page 50863
  • Step 11—Multiply the outputs of step (9) and (10) by 1 minus the applicable effective discount factor based on the EPM participant's quality category as described in sections III.D.4.b.(10) and III.E.3.f. of this proposed rule. For purposes of this proposed rule, we will define the outputs of this step as the episode quality-adjusted target prices for:

++ SHFFT model episodes with price MS-DRGs 480-482

++ AMI model episodes with price MS-DRGs 280-282 without readmission for CABG, and

++ AMI model episodes with price MS-DRGs 246-251 without readmission for CABG

d. Approach To Combine Pricing Features for CABG Model Episodes

The following presents our proposed methodology for combining the pricing features presented in section III.D.4.b of this proposed rule with respect to CABG model episodes.

(1) Anchor Hospitalization Portion of CABG Model Episodes

  • Step 1—Calculate historical episode payments that occurred during the anchor hospitalization of CABG model episodes that were initiated during the 3 historical years (section III.D.4.b.(2) of this proposed rule) for all IPPS hospitals for all Medicare Part A and B services included in the episodes. We note that specific PBPM payments may be excluded from historical episode payment calculations as discussed in section III.D.6. of this proposed rule.
  • Step 2—Apply steps III.D.4.c.(2) through (4) to the results of step (1) with trend factors calculated based on the anchor hospitalization portion of CABG model episodes with price MS-DRGs 231-236.
  • Step 3—Group the anchor hospitalization portion of episodes based on price MS-DRGs 231-236 and apply steps III.D.4.c.(6) through (10) to the anchor hospitalization portion of the CABG model episodes with severity factors, hospital-specific weighted update factors, and region-specific weighted update factors calculated to apply based only on the anchor hospitalization portion of CABG model episodes with price MS-DRGs 231-236. For purposes of this proposed rule, we will define the output of this step as CABG anchor hospitalization benchmark prices for CABG model episodes with price MS-DRGs 231-236.

(2) Approach To Combine Pricing Features for Post-Anchor Hospitalization Portion of CABG Model Episodes

  • Step 1—Calculate historical episode payments that occurred after the anchor hospitalization for CABG model episodes that were initiated during the 3 historical years (section III.D.4.b.(2) of this proposed rule) for all IPPS hospitals for all Medicare Parts A and B services included in the episodes. We note that specific PBPM payments may be excluded from historical episode payment calculations as discussed in section III.D.6. of this proposed rule.
  • Step 2—Apply steps III.D.4.c.(2) through (4) to the results of step (1) with trend factors calculated based on the post-anchor hospitalization portion of CABG model episodes with price MS-DRGs 231-236, as described in section III.D.4.b.(4) of this proposed rule.
  • Step 3—Group the post-anchor hospitalization portion of episodes based on—

++ With AMI diagnosis on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235)

++ With AMI diagnosis on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236)

++ Without AMI diagnosis on the anchor inpatient claim and price MS-DRG with major complication or comorbidity (231, 233, or 235)

++ Without AMI diagnosis on the anchor inpatient claim and price MS-DRG without major complication or comorbidity (232, 234, or 236).

Then apply steps III.D.4.c.(6)-(10) to the post-anchor hospitalization portion of the CABG model episodes with severity factors, hospital-specific weights, and region-specific weights calculated to apply based on the groups previously described in this step. For purposes of this proposed rule, we will define the output of this step as CABG post-anchor hospitalization benchmark prices for CABG model episodes corresponding to the groups described in this step.

(3) Combine CABG Anchor Hospitalization Benchmark Price and CABG Post-Anchor Hospitalization Benchmark Price

  • Step 1—Sum the CABG anchor hospitalization benchmark price corresponding to each price CABG MS-DRG and the CABG post-anchor hospitalization price corresponding to each of the post-anchor hospitalization groupings described in III.D.4.d.(2). For purposes of this proposed rule, we will define the outputs of those calculations to be CABG model episode benchmark prices for—

++ CABG model episodes with price MS-DRG 231 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 232 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 233 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 234 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 235 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 236 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 231 and without AMI diagnosis

++ CABG model episodes with price MS-DRG 232 and without AMI diagnosis

++ CABG model episodes with price MS-DRG 233 and without AMI diagnosis

++ CABG model episodes with price MS-DRG 234 and without AMI diagnosis

++ CABG model episodes with price MS-DRG 235 and without AMI diagnosis, and

++ CABG model episodes with price MS-DRG 236 and without AMI diagnosis

The CABG episode benchmark prices for each price CABG MS-DRG with AMI diagnosis would also apply as AMI model episode benchmark prices for AMI model episodes with price MS-DRGs 231-236.

  • Step 2—Multiply the results of step 1 by the appropriate effective discount factor that reflects the EPM participant's quality category as described in sections III.D.4.b.(10) and III.E.3.f. of this proposed rule. For purposes of this proposed rule, we will define the outputs of this step to be CABG model episode quality-adjusted target prices for—

++ CABG model episodes with price MS-DRG 231 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 232 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 233 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 234 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 235 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 236 and with AMI diagnosis

++ CABG model episodes with price MS-DRG 231 and without AMI diagnosis

++ CABG model episodes with price MS-DRG 232 and without AMI diagnosisStart Printed Page 50864

++ CABG model episodes with price MS-DRG 233 and without AMI diagnosis

++ CABG model episodes with price MS-DRG 234 and without AMI diagnosis

++ CABG model episodes with price MS-DRG 235 and without AMI diagnosis, and

++ CABG model episodes with price MS-DRG 236 and without AMI diagnosis

The episode quality-adjusted target prices for each anchor CABG MS-DRG with AMI diagnosis would also apply as AMI model episode quality-adjusted target prices for AMI model episodes with price MS-DRGs 231-236. The effective discount factor applied to calculate the AMI model episode quality-adjusted target prices for AMI model episodes with price MS-DRGs 231-236 could differ from the effective discount factor applied to calculate CABG model episode quality-adjusted target prices for CABG model episodes if the participant had different levels of quality performance on the AMI and CABG model composite quality scores that determine the discount factor for the quality-adjusted target prices.

e. Approach To Combine Pricing Features for AMI Model Episodes With CABG Readmissions

The following presents our proposed methodology for combining the pricing features presented in section III.D.4.b of this proposed rule with respect to AMI model episodes with a CABG readmission.

In general, the AMI model episode benchmark price for AMI model episodes with CABG readmission is the sum of the applicable standard AMI model episode benchmark price for an AMI episode without readmission corresponding to the AMI price MS-DRG and the applicable CABG anchor hospitalization benchmark price for a CABG model episode corresponding to the CABG readmission MS-DRG in the AMI model.

  • Step 1—For each combination of AMI price MS-DRG and CABG readmission MS-DRG, sum the corresponding AMI model episode benchmark price and CABG anchor hospitalization benchmark price. This results in 54 possible CABG readmission AMI model episode benchmark prices, corresponding to—

++ Price MS-DRG 280; Readmission MS-DRG 231

++ Price MS-DRG 280; Readmission MS-DRG 232

++ Price MS-DRG 280; Readmission MS-DRG 233

++ Price MS-DRG 280; Readmission MS-DRG 234

++ Price MS-DRG 280; Readmission MS-DRG 235

++ Price MS-DRG 280; Readmission MS-DRG 236

++ Price MS-DRG 281; Readmission MS-DRG 231

++ Price MS-DRG 281; Readmission MS-DRG 232

++ Price MS-DRG 281; Readmission MS-DRG 233

++ Price MS-DRG 281; Readmission MS-DRG 234

++ Price MS-DRG 281; Readmission MS-DRG 235

++ Price MS-DRG 281; Readmission MS-DRG 236

++ Price MS-DRG 282; Readmission MS-DRG 231

++ Price MS-DRG 282; Readmission MS-DRG 232

++ Price MS-DRG 282; Readmission MS-DRG 233

++ Price MS-DRG 282; Readmission MS-DRG 234

++ Price MS-DRG 282; Readmission MS-DRG 235

++ Price MS-DRG 282; Readmission MS-DRG 236

++ Price MS-DRG 246; Readmission MS-DRG 231

++ Price MS-DRG 246; Readmission MS-DRG 232

++ Price MS-DRG 246; Readmission MS-DRG 233

++ Price MS-DRG 246; Readmission MS-DRG 234

++ Price MS-DRG 246; Readmission MS-DRG 235

++ Price MS-DRG 246; Readmission MS-DRG 236

++ Price MS-DRG 247; Readmission MS-DRG 231

++ Price MS-DRG 247; Readmission MS-DRG 232

++ Price MS-DRG 247; Readmission MS-DRG 233

++ Price MS-DRG 247; Readmission MS-DRG 234

++ Price MS-DRG 247; Readmission MS-DRG 235

++ Price MS-DRG 247; Readmission MS-DRG 236

++ Price MS-DRG 248; Readmission MS-DRG 231

++ Price MS-DRG 248; Readmission MS-DRG 232

++ Price MS-DRG 248; Readmission MS-DRG 233

++ Price MS-DRG 248; Readmission MS-DRG 234

++ Price MS-DRG 248; Readmission MS-DRG 235

++ Price MS-DRG 248; Readmission MS-DRG 236

++ Price MS-DRG 249; Readmission MS-DRG 231

++ Price MS-DRG 249; Readmission MS-DRG 232

++ Price MS-DRG 249; Readmission MS-DRG 233

++ Price MS-DRG 249; Readmission MS-DRG 234

++ Price MS-DRG 249; Readmission MS-DRG 235

++ Price MS-DRG 249; Readmission MS-DRG 236

++ Price MS-DRG 250; Readmission MS-DRG 231

++ Price MS-DRG 250; Readmission MS-DRG 232

++ Price MS-DRG 250; Readmission MS-DRG 233

++ Price MS-DRG 250; Readmission MS-DRG 234

++ Price MS-DRG 250; Readmission MS-DRG 235

++ Price MS-DRG 250; Readmission MS-DRG 236

++ Price MS-DRG 251; Readmission MS-DRG 231

++ Price MS-DRG 251; Readmission MS-DRG 232

++ Price MS-DRG 251; Readmission MS-DRG 233

++ Price MS-DRG 251; Readmission MS-DRG 234

++ Price MS-DRG 251; Readmission MS-DRG 235, and

++ Price MS-DRG 251; Readmission MS-DRG 236

  • Step 2—Multiply the results of step 1 by the effective discount factor that reflects the EPM participant's quality category, as described in sections III.D.4.b.(10) and III.E.3.f. of this proposed rule. For purposes of this proposed rule, we will define the outputs of this step to be AMI model episode quality-adjusted target prices for the same combinations of AMI price MS-DRG and readmission MS-DRG in step (1).

5. Process for Reconciliation

a. Net Payment Reconciliation Amount (NPRA)

Consistent with the CJR model, we propose to conduct reconciliation for each EPM by calculating an EPM-specific NPRA for each EPM participant (80 FR 73381 through 73383). After the completion of an EPM performance year, we propose to retrospectively calculate an EPM participant's actual EPM-episode payments based on the EPM episode definitions as discussed in sections III.C.3. and III.C.4. of this proposed rule and the payment policies applicable to calculating actual EPM-episode payments as discussed in the subsections of section III.D.3 of this proposed rule.

We propose to compare each EPM participant's actual EPM episode payments to its quality-adjusted target prices. We propose, as discussed in Start Printed Page 50865section III.D.4. of this proposed rule, that an EPM participant would have multiple quality-adjusted target prices for EPM episodes ending in a given performance year, based on the anchor MS-DRG for the EPM episode, whether the EPM episode included a chained anchor hospitalization; whether the EPM episode included readmission for CABG MS-DRGs; whether the EPM episode included an AMI ICD-CM diagnosis code on the anchor inpatient claim; the performance year when the EPM episode was initiated; when the EPM episode was initiated within a given performance year (January 1 through September 30 of the performance year, October 1 through December 31 of the performance year, October 1 through December 31 of the prior performance year); and the potential effective discount factors. The difference between each EPM episode's actual EPM episode payment and the relevant quality-adjusted target price under the EPM (calculated as quality-adjusted target price subtracted by actual EPM episode payment) would be aggregated for all EPM episodes in each EPM for an EPM participant within the performance year, representing the NPRA. For performance year 2, we would perform two separate aggregations in order to create two NPRAs—one reflecting episodes that ended during performance year 2 (NDR), and a second for episodes that ended during performance year 2 (DR).

We propose to not include any reconciliation payments or repayments to Medicare under the EPMs for a given performance year when calculating actual episode spending and, therefore the NPRA for a subsequent performance year. We want to incentivize providers to provide high-quality and efficient care in all years of the EPMs. If reconciliation payments for a performance year were counted as Medicare expenditures in a subsequent performance year, an EPM participant would experience higher Medicare expenditures in the subsequent performance year as a consequence of providing high-quality and efficient care in the prior performance year, negating some of the incentive to perform well in the prior year. Therefore, we propose to not have the NPRA for a given performance year be impacted by EPM repayments or reconciliation payments made in a prior performance year. For example, if an EPM participant receives a $10,000 reconciliation payment in the second quarter of 2018 for achieving episode spending below the quality-adjusted target price for performance year 1, that $10,000 reconciliation payment amount would not be included in the performance year 2 calculations of actual EPM-episode payments.

The NPRA would be subject to the stop-loss and stop-gain limits described in section III.D.7.b. of this proposed rule.

b. Payment Reconciliation

We propose to retrospectively reconcile an EPM participant's actual EPM-episode payments against the quality-adjusted target prices 2 months after the end of the performance year. Specifically, we would capture claims submitted by March 1st following the end of the performance year and carry out the NPRA calculation as described previously to make an EPM reconciliation payment or hold hospitals responsible for repayment, as applicable, in quarter 2 of that calendar year.

We also propose that during the following performance year's reconciliation process, we would calculate the prior performance year's actual EPM episode payments a second time to account for final claims run-out and any canceled EPM episodes, due to overlap with other models or other reasons as specified in section III.C.4.b of this proposed rule. This calculation, termed the subsequent reconciliation, would occur approximately 14 months after the end of the prior performance year. As discussed later in this section, the amount from this calculation, if different from zero, would be applied to the NPRA for the subsequent performance year, as well as the post-episode spending and ACO overlap calculation in order to determine the amount of the payment Medicare would make to the EPM participant or such participant's repayment amount. We note that the subsequent reconciliation calculation would be combined with the previous calculation of NPRA for a performance year to ensure the stop-loss and stop-gain limits discussed in section III.D.7.b. of this proposed rule are not exceeded for a given performance year.

For the performance year 1 reconciliation process, we would calculate an EPM participant's NPRA as previously described, and if positive, such participant would receive the amount as a reconciliation payment from Medicare, subject to the stop-gain limit for performance year 1. If negative, the EPM participant would not be responsible for repayment to Medicare, consistent with our proposal to phase in financial responsibility beginning in the second quarter of performance year 2.

For the performance year 2 reconciliation process, we would calculate two separate NPRAs for an EPM participant—one for episodes that ended during performance year 2 (NDR) and a second for episodes that ended during performance year 2 (DR). While these NPRAs would be separately determined for each of these two periods, whether an EPM participant receives a Medicare reconciliation payment or makes a Medicare repayment in performance year 2 would be determined based on the sum of these two separately determined NPRAs. The NPRA for both performance year 2 (NDR) and performance year 2 (DR) would be subject to the same stop-gain limit of 5 percent, but because EPM participants would only have repayment responsibility for negative NPRA in performance year 2 (DR), the stop-loss limit of 5 percent would only apply to performance year 2 (DR). Thus, if an EPM participant's NPRA for the first quarter of performance year 2 is positive, that amount would be counted toward a reconciliation payment from Medicare, subject to the stop-gain limit for performance year 2. If negative, the EPM participant would not be responsible for repayment to Medicare of the amount determined for performance year 2 (NDR). If an EPM participant's NPRA is positive for episodes ending during performance year 2 (DR), that amount would be counted toward a reconciliation payment from Medicare, subject to the stop-gain limit for performance year 2. If negative, the EPM participant would be responsible for repayment to Medicare of the amount determined for episodes ending during performance year 2 (DR), subject to the stop loss limits for performance year 2 (DR).

During the subsequent reconciliation process for performance year 2, we would also calculate the prior performance year's actual EPM episode payments a second time separately for episodes that ended during performance year 2 (NDR) and for episodes that ended during performance year 2 (DR).

Also, starting with the EPM reconciliation process for performance year 2, in order to determine the reconciliation or repayment amount, the amount from the subsequent reconciliation calculation would be combined with the NPRA for that subsequent year. The result of the post-episode spending calculation for performance year 1, as proposed in section III.D.7.e., and the dollar amount of the EPM discount percentage that was paid out as shared savings to an ACO during the prior year as specified in section III.D.6.b. of this proposed rule, would also be added to the NPRA and subsequent reconciliation calculation in Start Printed Page 50866order to create the reconciliation payment or repayment amount. If the amount is positive, and if the EPM participant is in the acceptable or better quality category for the EPM (discussed further in section III.E.3.f of this proposed rule), the EPM participant would receive the amount as a reconciliation payment from Medicare. If the amount is negative, Medicare would hold the EPM participant responsible for repaying the absolute value of the repayment amount following the rules and processes for all other Medicare debts. For example, when we conduct reconciliation for performance year 2 in early 2019, we would calculate the performance year 2 NPRA and the subsequent reconciliation calculation, post-episode spending, and ACO overlap calculation for performance year 1. These amounts would be added together to create the reconciliation payment or repayment amount.

Note that given our proposal to not hold EPM participants financially responsible for repayment for the first performance year, during the reconciliation process for performance year 2, the subsequent reconciliation calculation amount (for performance year 1) would be compared against the performance year 1 NPRA to ensure that the sum of the NPRA calculated for performance year 1 and the subsequent reconciliation calculation for year 1 is not less than zero. Likewise given our proposal to not hold EPM participants financially responsible for repayment for episodes ending during performance year 2 (NDR), during the reconciliation process for performance year 3, the subsequent reconciliation calculation amount for performance year 2 (NDR) would be compared against the performance year 2 (NDR) NPRA to ensure that the sum of the NPRA calculated for performance year 2 (NDR) and the subsequent reconciliation calculation for performance year 2 (NDR) is not less than zero.

For performance year 2 (DR) and performance years 3 through 5, though, we propose that Medicare would hold the participant responsible for repaying part or all of the absolute value of the repayment amount, as proposed in section III.D.2.c. of this proposed rule, following the rules and processes for all other Medicare debts. Table 11 illustrates a simplified example of how the subsequent reconciliation calculation may affect the following year's reconciliation payment. Note that this example assumes the EPM participant is not responsible for post-episode spending or ACO overlap for performance year 1.

Table 11—Sample Reconciliation Results

Performance year 1 (2017) NPRAPerformance year 1 subsequent reconciliation calculationDifference between PY1 subsequent reconciliation calculation and NPRAPerformance year 2 (2018) NPRA *Reconciliation payment made to EPM participant in quarter 2 2019
Hospital A$50,000$40,000($10,000)$25,000$15,000
* Note the calculation of NPRA for performance year 2 represents the combined amounts of the NPRA for performance year 2 (NDR) and performance year 2 (DR).

The second column represents the NPRA calculated for performance year 1, meaning that EPM participant Hospital A's aggregated episode payment was $50,000 below the sum of quality-adjusted target prices for all of Hospital A's EPM episodes. The third column represents the subsequent reconciliation calculation, indicating that when calculating actual EPM-episode payments during performance year 1 a second time, we determined that Hospital A's aggregated EPM-episode payment was $40,000 below the sum of quality-adjusted target prices for all of Hospital A's EPM episodes, due to claims run out, accounting for model overlap, or other reasons. The fourth column represents the difference between the subsequent reconciliation calculation and the raw NPRA calculated for performance year 1. This difference is then combined with the amount in the fifth column to create the reconciliation payment amount for PY2, which is reflected in the sixth column.

This reconciliation process would account for overlap between the CJR model and other CMS models and programs as discussed in section III.D.6.b of this proposed rule, and would also involve updating performance year EPM-episode claims data. We also note that in cases where an EPM participant has appealed one or more of its EPM quality measure results through the HIQR Program appeal process (which is not part of the proposed EPM appeals process), where such HIQR Program appeal findings would result in a different effective discount factor for the EPM participant to calculate the quality-adjusted target prices from EPM-episode benchmark prices, the subsequent reconciliation calculation would account for these changes as well.

For example, for performance year 1 for these EPMs in 2017, we would capture claims submitted by March 1, 2018, and reconcile payments for EPM participants approximately 6 months after the end of the performance year 1 in quarter 2 of calendar year 2018. We would carry out the subsequent reconciliation calculation in the following year in quarter 2 of calendar 2019, simultaneously with the reconciliation process for the second performance year, 2018. Table 12 displays the reconciliation timeframes for the EPMs.

Table 12—Proposed Timeframe for Reconciliation for EPMs

EPM performance yearEPM performance periodReconciliation claims submitted byNPRA calculationSecond reconciliation, ACO overlap, and post-episode spending calculationsCalculation amounts included in reconciliation payment and repayment amounts
Year 1 *Episodes beginning on or after July 1, 2016 and ending through December 31, 2017March 1, 2018Q2 2018March 1, 2019Q2 2019
Start Printed Page 50867
Year 2Episodes ending January 1, 2018 through December 31, 2018March 1, 2019Q2 2019March 1, 2020Q2 2020
Year 3Episodes ending January 1, 2019 through December 31, 2019March 1, 2020Q2 2020March 2, 2021Q2 2021
Year 4Episodes ending January 1, 2020 through December 31, 2020March 2, 2021Q2 2021March 1, 2021Q2 2021
Year 5Episodes ending January 1, 2021 through December 31, 2021March 1, 2022Q2 2022March 1, 2023Q2 2023
* Note that the reconciliation for Year 1 would not include repayment responsibility from EPM participants.

We propose this approach in order to balance our goals of providing reconciliation payments in a reasonable timeframe, while being able to account for overlap and all Medicare claims attributable to EPM episodes. We believe that beginning to pull claims 2 months after the end of the performance year would provide sufficient claims run-out to conduct the reconciliation in a timely manner, given that our performance year includes EPM episodes ending, not beginning, by December 31st. We note that in accordance with the regulations at § 424.44 and the Medicare Claims Processing Manual (Pub. L. 100-04), Chapter 1, Section 70, Medicare claims can be submitted no later than 1 calendar year from the date-of-service. We recognize that by pulling claims 2 months after the end of the performance year to conduct reconciliation, we would not have complete claims run-out. However, we believe that the 2 months of claims run-out would be an accurate reflection of EPM-episode payments and consistent with the claims run-out timeframes used for reconciliation in other payment models, such as BPCI Models 2 and 3 and the CJR model. The alternative would be to wait to reconcile until we have full claims run out 12 months after the end of the performance year, but we are concerned that this approach would significantly delay earned reconciliation payments under the EPMs.

However, we propose to conduct a subsequent reconciliation calculation 14 months after the end of a performance year to account for canceled episodes, post-episode spending, overlap with other CMS models and programs, and any remaining claims available at that time. The proposals for the annual reconciliation and subsequent reconciliation calculation are included in § 512.305 and § 512.307. We seek comment on these proposals for an annual reconciliation and subsequent calculation.

c. Reconciliation Report

For EPM participants to receive timely and meaningful feedback on their performance under the models as well as better understand the basis of their reconciliation payment or Medicare repayment for a given performance year, if any, we propose to annually issue to EPM participants a reconciliation report, similar to the CJR Reconciliation Report we make available to CJR participants (80 FR 73408). We propose that these reports would contain the following information:

  • Information on the EPM participant's composite quality score described in section III.E.3.a. through III.E.3.e of this proposed rule.
  • The total actual episode payments for the EPM participant.
  • The NPRA.
  • Whether the EPM participant is eligible for a reconciliation payment or must make a repayment to Medicare.
  • The NPRA and subsequent reconciliation calculation amount for the previous performance year, as applicable.
  • The post-episode spending amount and ACO overlap calculation for the previous performance year, as applicable.
  • The reconciliation payment or repayment amount.

For performance year 2, we propose that the reconciliation report would include information separately for the performance year 2 (NDR) and performance year 2 (DR) portions of that year.

As discussed in section III.D.8 of this proposed rule, EPM participants would review their reconciliation report and would be required to provide written notice of any error, in a calculation error form that must be submitted in a form and manner specified by CMS. Unless the EPM participant provides such notice, the reconciliation report would be deemed final within 45 calendar days after it is issued, and CMS would proceed with payment or repayment. The proposal to issue a reconciliation report is included in § 512.305(f). We seek comments on our proposal to issue a reconciliation report to EPM participants and what other information, if any, would be helpful to include in this report.

6. Adjustments for Overlaps With Other Innovation Center Models and CMS Programs

a. Overview

Three issues may arise in overlap situations that must be addressed under EPM. First, we acknowledge that there may be circumstances where a hospital in a geographic area selected for the AMI, CABG or SHFFT model is also participating in BPCI for the same episode. We refer to this as “provider overlap.” Second, there may be situations where a Medicare beneficiary receives care that could potentially be counted under more than one episode or total cost of care payment model. We refer to this as “beneficiary overlap.” Finally, EPM reconciliation payments and Medicare repayments made under Parts A and B and attributable to a specific beneficiary's episode may be at risk of not being accounted for by other models and programs when determining the beneficiary's cost of care under Medicare. Therefore, a payment reconciliation policy is necessary in order to credit the entity that is closest to that care for the episode of care in terms of time, location, and care management responsibility.

We establish our proposal for provider overlap at § 512.100(b) and § 512.230(g). We establish our proposal for beneficiary overlap at § 512.230(f), § 512.230(h), and § 512.230(i). We establish our proposal for payment reconciliation at § 512.210 and Start Printed Page 50868§ 512.305. We seek comment on our proposals to account for overlap between EPMs and other CMS models or programs.

b. Provider Overlap

(1) BPCI Participant Hospitals in Geographic Areas Selected for EPMs

Provider overlap exists when a hospital in a geographic area selected for the AMI, CABG or SHFFT model is also an episode initiator in BPCI for an episode anchored by that EPM's DRG. BPCI is an episode payment model testing AMI, CABG, SHFFT, and 45 other episodes in acute care, post-acute care, or both acute care and post-acute care settings.

Similar to CJR, we propose that in the geographic areas where the AMI, CABG and SHFFT models will be implemented, an acute care hospital participating in BPCI Model 2 or 4 will participate in an EPM for episodes anchored by EPM MS-DRGs that are not covered under the hospital's current BPCI agreement. If a BPCI hospital in an EPM-selected area withdraws from BPCI episodes anchored by EPM MS-DRGs, the BPCI hospital will participate in the EPMs for which it was previously excluded. This proposal promotes accountable care by ensuring beneficiary coverage by BPCI or an EPM in selected areas.

We establish the proposal for hospitals in geographic areas selected for EPMs that are also participating in BPCI episodes anchored by EPM DRGs at § 512.100(b). We seek comment on this proposal.

(2) BPCI Physician Group Practice (PGP) Episode Initiators in Hospitals Participating in EPMs

It is possible that a physician in a BPCI PGP may treat a Medicare beneficiary in a hospital participating in one or more EPM. We propose that if a beneficiary is admitted to an EPM participant for an episode anchored by EPM MS-DRGs covered under the PGP's BPCI agreement and the attending or operating physician on the admission's inpatient claim is a member of the BPCI PGP, the BPCI episode will take precedence over the EPM episode for which the hospital would otherwise be the accountable entity. In other words, if, for any portion of the EPM episode, a beneficiary would also be in a BPCI PGP episode, we will cancel or never initiate the EPM episode. For example—

  • A beneficiary is admitted for a CABG to an EPM participant in the CABG model. The attending or operating physician on the inpatient claim for the admission is in a BPCI Model 2 PGP participating in CABG. The episode is initiated under BPCI; an EPM episode does not initiate.
  • A beneficiary is admitted for an AMI to an EPM participant in the AMI model. The beneficiary receives a PCI while hospitalized. The attending or operating physician on the inpatient claim for the admission is in a BPCI Model 2 PGP participating in PCI episodes but not medical AMI episodes. A PCI episode is initiated under BPCI; an EPM episode does not initiate.
  • A beneficiary is admitted for an AMI to an EPM participant in the AMI model. A PCI was not part of the beneficiary's treatment. The attending or operating physician on the inpatient claim for the admission is in a BPCI Model 2 PGP participating in PCI episodes only. The episode is initiated under the AMI model. A PCI episode under BPCI Model 2 would not initiate unless a PCI were performed on the beneficiary, and
  • A beneficiary is admitted for an AMI to an EPM participant in the AMI model. A CABG was not part of the beneficiary's treatment. The attending or operating physician on the inpatient claim is in a BPCI Model 2 PGP participating in CABG episodes only. The episode is initiated under the AMI model. A CABG episode under BPCI Model 2 would not be initiated unless a CABG was performed on the beneficiary while hospitalized.

We establish the proposal for BPCI PGP episode initiators in hospitals participating in EPMs at § 512.230(g). We seek comment on this proposal.

(c) Beneficiary Overlap

(1) Beneficiary Overlap With BPCI

We also need to account for instances where a different model's episode could initiate during an ongoing EPM episode. We propose that any BPCI Model 2, 3 or 4 episode, regardless of its anchor DRG exclusion status from an EPM episode definition, takes precedence over an AMI, CABG or SHFFT episode such that it would cancel or prevent the initiation of an AMI, CABG or SHFFT episode. For example—

  • If a beneficiary is in an ongoing AMI model episode and is treated for SHFFT by a hospital, PGP physician, or post-acute care provider participating in a BPCI SHFFT episode, the initial AMI model episode will be canceled. The second entity will initiate a new episode under BPCI subject to the payment rules under that model, and
  • If a beneficiary is in an ongoing BPCI AMI episode and is readmitted for SHFFT to an EPM participant in the SHFFT model, the BPCI episode would continue and the SHFFT model episode would not initiate.

Participants in BPCI have an expectation that eligible episodes will be part of the BPCI model test, whereas based on our proposal EPM participants would be aware that episodes may be canceled when there is overlap with BPCI episodes. We aim to preserve the integrity of ongoing model tests without introducing major modifications that could make evaluation of existing models more challenging. Given the current scheduled end date for the BPCI, we are proposing to give precedence to episodes covered under BPCI Models 2, 3 and 4 initiated on or before September 30, 2018.

We acknowledge this BPCI-EPM overlap policy differs from the CJR beneficiary overlap policy, where a beneficiary may be in a CJR LEJR episode and a non-LEJR BPCI episode concurrently. However, in CJR this overlap is rare. Within the 90-day post-hospital discharge period, included readmissions occur for less than 1 percent of LEJR beneficiaries. In contrast, included readmissions occur for approximately 25 percent of AMI and CABG beneficiaries. The high incidence of included readmissions for AMI, CABG and SHFFT episodes necessitates a different policy to avoid double-paying savings and double-counting losses, as well as not initiating new episodes when the readmission is a complication of care during the first episode that could be managed.

We considered alternative options for dealing with situations in which a beneficiary in an EPM episode could also be in a BPCI episode, including allowing the first episode initiated to take precedence regardless of the model under which it occurred. This would encourage more accountable care, particularly with AMI, CABG, and SHFFT episodes that are more likely to involve readmissions for complications than generally occur with LEJR. However, preventing BPCI episodes from being initiated, particularly those initiated by post-acute care providers which, by definition, occur after an anchor hospitalization, could substantially reduce the number of such episodes and our ability to fully test BPCI. Moreover, operational challenges due to different timelines for payment reconciliation are of concern.

We establish the proposal for beneficiary overlap with BPCI at § 512.230(h). We seek comment on this proposal.Start Printed Page 50869

(2) Beneficiary Overlap With the CJR Model and Other EPMs

As discussed in section III.C.4. of this proposed rule, if a beneficiary is in a SHFFT, AMI or CABG model or CJR episode and has a hospital readmission that is not excluded from the ongoing episode definition and would otherwise initiate an episode in a different EPM or the CJR model, that hospital readmission will not initiate another episode or cancel the ongoing episode. If a beneficiary is in a SHFFT, AMI or CABG model episode or CJR episode and has a hospital readmission that is excluded from the ongoing episode definition and could initiate an episode in a different EPM or the CJR model, the subsequent EPM or CJR episode will initiate, the ongoing episode would continue, and both episodes will occur concurrently. For example—

  • The CJR model episode definition does not exclude the MS-DRGs that would initiate a SHFFT model episode. If a beneficiary is in the CJR model and receives SHFFT at an EPM participant in the SHFFT model during the ongoing CJR episode, the CJR episode will continue and the SHFFT model episode will not initiate;
  • SHFFT model episode definition does not exclude the MS-DRGs that would initiate a CJR LEJR episode. If a beneficiary is in the SHFFT model and receives an LEJR at a CJR hospital during the ongoing SHFFT episode, the SHFFT episode will continue and the CJR episode will not initiate;
  • The SHFFT model episode definition does not exclude the MS-DRGs that would initiate an AMI model episode. If a beneficiary is in the SHFFT model and is readmitted for an AMI to an EPM participant in the AMI model during the ongoing SHFFT model episode, the SHFFT model episode will continue and the AMI model episode will not initiate;
  • The AMI model episode definition does not exclude the MS-DRGs that would initiate a CABG model episode. If a beneficiary is in the AMI model and is readmitted for a CABG to the same or another EPM participant in the CABG model during the ongoing AMI model episode, the AMI model episode will continue and the CABG model episode will not initiate.

We believe that an overlap policy that gives precedence to the ongoing episode over subsequent episodes initiated during the post-hospital discharge period, except where the second admission is explicitly excluded, aligns with our stated goal of encouraging more accountable care. Moreover, this policy would establish an operationally straightforward policy for future EPMs.

We establish the proposal for beneficiary overlap with the CJR model and other EPMs at § 512.230(i). We seek comment on this proposal.

(3) Beneficiary Overlap With Shared Savings Models and Programs

We expect many beneficiaries in an AMI, CABG or SHFFT model episode will also be aligned or attributed to a Shared Savings Program participant or a participant in an ACO model initiated by the CMS Innovation Center. For the purposes of this discussion, the term ACO will be used generically to refer to either Shared Savings Program or Innovation Center ACO models. Shared savings payments to ACOs and shared savings losses repaid by ACOs to CMS have the potential to overlap with EPM reconciliation payments. As with CJR, we propose to attribute savings achieved during an EPM episode to the EPM participant, and include EPM reconciliation payments for ACO-aligned beneficiaries as ACO expenditures. In order to address comments received during rulemaking for CJR, we propose to test an alternative strategy to address ACO overlap. Specifically, we propose to exclude beneficiaries from EPMs who are aligned to ACOs in the Next Generation ACO model and End Stage Renal Disease (ESRD) Seamless Care Organizations (ESCOs) in the Comprehensive ESRD Care initiative in tracks with downside risk for financial losses. We do not propose to exclude beneficiaries aligned to Shared Savings Program ACOs in Tracks 1, 2, or 3 at this time. However, we seek comment on excluding beneficiaries from EPMs that are prospectively assigned to SSP Track 3 as well as to other financial risk tracks. The Shared Savings Program is a national program. We do not believe that testing a new approach to addressing overlap, which could potentially disrupt ACO investments, operations, and care redesign activities, would be appropriate at this time prior to a test with a smaller population. We plan to monitor and learn from the test of excluding beneficiaries prospectively assigned to an ACO from risk tracks and consider these results and comments in future rule-making.

Several strong considerations drive us to otherwise follow CJR precedent for addressing ACO overlap. First, CMS continues to avoid double payment of savings and double recoupment of losses, which is an important principle of successful payment reform. Second, in implementing the EPMs, there would be no additional operational effort due to consistency in ACO overlap policies across models. In this respect, we anticipate little to no difficulty in replicating prior policy as new episode payment models are introduced. Third, this would have no negative financial impact on EPM participants, an important consideration for future EPMs. The payment reconciliation for EPM participants is described in section III.D.5. of this proposed rule.

Therefore, we propose to follow the policy set forth in the CJR Final Rule for accounting for overlap between EPMs and the Shared Savings Program and ACO models other than the Next Generation ACO model and CEC listed previously.

Additionally, for programmatic consistency among ACO models and programs, given that our ACO models generally are tested for the purpose of informing future potential changes to the Shared Savings Program, we believe that the ACO model overlap adjustment policy should be aligned with the Shared Savings Program policy. Thus, we propose that under EPMs, we would make an adjustment to the reconciliation amount to account for any of the applicable discount for an episode resulting in Medicare savings that is paid back through shared savings under the Shared Savings Program or any other ACO model, but only when an EPM hospital also participates in the ACO and the beneficiary in the EPM episode is also aligned to that ACO. This adjustment would be necessary to ensure that the applicable discount under the EPM is not reduced because a portion of that discount is paid out in shared savings to the ACO and thus, indirectly, back to the hospital.

However, we propose not to make an adjustment under EPMs when a beneficiary receives an AMI, SHFFT, or CABG at a hospital participating in the corresponding EPM and is aligned to an ACO in which the hospital is not participating. While this proposal would leave overlap unaccounted for in such situations, we do not believe it would be appropriate to hold responsible for repayment the hospital that managed the beneficiary during the episode through an EPM adjustment, given that the participant may have engaged in care redesign and reduced spending during the EPM episode. The participant may be unaware that the beneficiary is also aligned to an ACO. However, we recognize that as proposed this policy would allow an unrelated ACO full credit for the Medicare savings achieved during the episode. The evaluation of each of the EPMs, as discussed in section IV. of this proposed rule, would examine overlap in such Start Printed Page 50870situations and the potential effect on Medicare savings.

We note that our proposed policy as outlined in this proposed rule would entail CMS reclaiming from the EPM participant any discount percentage paid out as shared savings for the Shared Savings Program or ACO models only when the hospital is an ACO participant and the beneficiary is aligned with that ACO, while other total cost of care models such as the Comprehensive Primary Care Plus initiative (CPC+) would adjust for the discount percentage in their calculations. We believe that other ACO models in testing that share operating principles with the Shared Savings Program should follow the same policies as the EPM Shared Savings Program adjustment for certain overlapping ACO beneficiaries. As the landscape of CMS models and programs changes, we may revisit this policy through future rulemaking.

However, there are circumstances when an alternative option may be appropriate to consider. Therefore, we are also considering an EPM-ACO overlap policy that would exclude from EPMs beneficiaries who are aligned to ACOs in the Next Generation ACO model and ESCOs in the Comprehensive ESRD Care Initiative in tracks with downside risk for financial losses. Some ACOs have successfully managed acute care and post-acute care expenditures below regional or national mean costs, and expressed that the current CJR and BPCI ACO overlap policies deprives them of a key source of savings. We are aware of situations in certain markets that seem to reduce opportunities for ACOs to achieve savings given historic experience that indicates these particular ACOs are able to manage the care within episodes as successfully as EPM participants. Attributing savings to participants in episode payment models, such as CJR participants and EPM participants under this proposed rule, creates a problem where the ACO is accountable for coordinating a beneficiary's care over a performance year but is not able to benefit from savings achieved from episodes completed during the performance year. Data shows that post-acute care spending is among the most significant sources of savings for ACOs currently, and where they focus significant investments.[68 69]

Certain considerations weigh against exclusion of all ACO-aligned beneficiaries from participation in EPM episodes. Such a blanket exclusion would remove a large proportion of Medicare FFS beneficiaries from the EPMs, many of whom would inevitably receive care at EPM participants. This would dilute the power of the EPM test and generalizability of EPM findings. Additionally, differences between ACO beneficiary alignment algorithms do not support a blanket exclusion. It is more operationally feasible to identify and exclude beneficiaries who are prospectively aligned to ACOs. In retrospective alignment models, beneficiaries may be aligned to an ACO at the end of the performance year, before the performance year, or preliminarily aligned to one ACO before the performance year and subsequently aligned to a different ACO after all qualifying services are considered. In retrospective alignment, there will be significant numbers of beneficiaries aligned at final reconciliation to a given ACO who were not identified as preliminarily aligned to that ACO prior to the performance year. That is, they were identified either as unaligned to any ACO or aligned to a different ACO. In prospective alignment models and tracks, the list of aligned beneficiaries is available prior to the start of the performance year and a beneficiary's alignment does not change on the basis of his or her utilization in the performance year (subject to various exclusions made on a quarterly basis, such as a beneficiary's election into a Medicare Advantage plan).

Because ACOs in two-sided risk arrangements have stronger incentives than those in one-sided risk arrangements to reduce total cost of care, especially given the possibility of paying CMS shared losses, we believe that ACOs in two-sided risk arrangements may be best positioned to assume the risk associated with EPM episodes, while ACOs in one-sided risk arrangements may be less well-positioned to do so. ACOs in one-sided risk arrangements, such as those in the Shared Savings Program Track 1, do not bear the risk of owing losses to CMS. In contrast, ACOs in two-sided risk arrangements, such as the Next Generation ACO model, are held to as much as 80 percent to 100 percent of first dollar losses. Thus, we believe that pursuing a blanket exclusion from EPMs of aligned beneficiaries from all ACOs would inappropriately disadvantage EPM participants that carry significant financial risk under EPM.

This proposed ACO overlap policy would grant ACOs in models and tracks with the highest levels of downside risk for financial losses—the Next Generation ACO model and tracks of the Comprehensive ESRD Care Initiative with downside risk for financial losses—paramount financial opportunity in exchange for accepting total cost of care responsibility for their beneficiaries. EPM participants may still realize opportunities to save by partnering with ACOs, but outside of the EPM arrangement. Specifically, we refer to section IIII.I. of this proposed rule which describes opportunities for gainsharing allowed under these models.

This policy tests the effects of such an ACO-aligned beneficiary exclusion policy within a broader test of the effectiveness of EPMs. We can learn its impact on EPM participants and ACOs that have beneficiaries excluded from EPMs, as well as ACOs that do not have beneficiaries excluded from EPMs. This will improve our understanding about the appropriate entity to hold accountable for the costs within the episode. For this reason we are recommending this test be limited to the AMI, CABG, and SHFF, and CJR models, and ACO models being conducted under CMS' Innovation Center, and are not proposing to implement the policy more broadly to other ACOs, such as those in the Shared Savings Program. In proposing the exclusion of beneficiaries in only a limited number of ACO initiatives we attempt to balance the desire to build a new payment reform initiative while mitigating the potential challenges to existing shared savings models and programs. We seek comment on this proposal as well as input on extending the proposal to CJR and other ACOs accepting two-sided risk, such as those ACOs in the Shared Savings Program Track 3.

We have investigated CMS data related to the services under consideration in the AMI, CABG and SHFFT models. A small fraction of total beneficiaries aligned to ACOs qualifying for this exclusion in fact have relevant anchor hospitalizations that would initiate an EPM in a given calendar year. For instance, from 2013 through 2015, about 2.4 percent of beneficiaries aligned to Pioneer ACO model participants had an anchor hospitalization that would have Start Printed Page 50871initiated an AMI, CABG or SHFFT model.

We have considered several additional options to account for EPM-ACO beneficiary overlap prior to proposing the strategy outlined previously. We considered whether to split the risk, including at an equal sharing rate, at the time of financial reconciliation between EPM participants and ACOs when episodes included overlapping beneficiaries. This has the advantage of mitigating the supposed “carve out” of ACO expenditures, but requires CMS to arbitrarily declare a level of risk sharing. We are also concerned about the operational feasibility of such calculations, given that reconciliation would have to occur in tandem, resulting in long delays in payments or recoupments for both EPM participants and ACOs. We also considered whether to attribute to ACOs the more favorable of either the episode-specific target price or the actual expenditures incurred by the beneficiary during the episode time period. However, this policy would result in significant losses to the Medicare Trust Fund, as the double payment of savings/losses would be a certainty.

We establish the proposal to exclude from the EPMs beneficiaries who are aligned to an ACO in the Next Generation ACO Model or Comprehensive ESRD Care Initiative at § 512.230(f). We establish the proposal to attribute savings achieved during an EPM episode to the EPM participant, and include EPM reconciliation payments for other ACO-aligned beneficiaries as ACO expenditures at § 512.305 and § 512.307. We seek comment on our proposals to account for beneficiary overlap with shared savings models and programs.

d. Payment Reconciliation of Overlap With Non-ACO CMS Models and Programs

In general, Per-Beneficiary Per-Month (PBPM) payments are for new or enhanced provider or supplier services that share the goal of improving quality of care overall and reducing Medicare expenditures for services that could be avoided through improved care coordination. Some of these PBPM payments may be made for services furnished to a beneficiary that is in another Innovation Center model at the that same time that the beneficiary is in an EPM, but the clinical relationship between the services paid by the PBPM payments and the EPM will vary. For purposes of this proposed rule, we consider clinically related those services paid by PBPM payments that are for the purpose of care coordination and care management of any beneficiary diagnosis or hospital admission not excluded from an EPM's episode definition, as discussed in section III.C. of this proposed rule.

As with CJR, we propose to include PBPM payments for new and enhanced services in EPM reconciliation calculations if we determine, on a model by model basis, that the services paid by PBPM payments are (1) not excluded from an EPM model's episode definition; (2) rendered during the episode; and (3) paid for from the Medicare Part A or Part B Trust Funds. That is, we would include the clinically related services paid by a PBPM payment if the services would not otherwise be excluded based on the principal diagnosis code on the claim, as discussed in section III.C. of this proposed rule. The PBPM payments for clinically related services would not be excluded from the EPMs' historical episodes used to calculate target prices when the PBPM payments are made from the Part A or Part B Trust Fund, and they would not be excluded from calculation of actual episode expenditures during an EPM's performance period. PBPM model payments that we determine are clinically unrelated would be excluded, regardless of the funding mechanism or diagnosis codes on claims for those payments. We note that in the case of PBPM model payments, principal diagnosis codes on a Part B claim (which are used to identify exclusions from EPMs, as discussed in section III.C. of this proposed rule) would not be the only mechanism for exclusion of a service from an EPM. All such PBPM model payments we determine are clinically unrelated would be excluded as discussed in this proposed rule. Finally, all services paid by PBPM payments funded through the Innovation Center's appropriation under section 1115A of the Act would be excluded from the EPMs, without a specific determination of their clinical relationship to an EPM. We believe including such PBPM payments funded under the Innovation Center's appropriation and not included on claims would be operationally burdensome and could significantly delay any reconciliation payments and repayments for the EPMs. In addition, because these services are not paid for from the Medicare Parts A or B Trust Funds, we are not confident that they would be covered by Medicare under existing law. Therefore, we believe the services paid by these PBPM payments are most appropriately excluded from the EPMs. Our proposal for the treatment of services paid by PBPM payments in the EPMs would pertain to all existing models with PBPM payments, as well as future models and programs that incorporate PBPM payments. We believe that this proposal is fully consistent with our goal of including all related Part A and Part B services in the EPMs, as discussed in section III.C. of this proposed rule.

As with CJR, the OCM and MCCM services and conditions are excluded from the AMI, CABG, and SHFFT episode definitions and thus their payments are excluded from EPM reconciliation (listed on the CMS Web page at https://innovation.cms.gov/​Files/​x/​cjr-pbpmexclusions.xlsx). While the OCM will pay for new or enhanced services through PBPM payments funded by the Medicare Part B Trust Fund, we do not believe these services are clinically related to the EPMs. The OCM incorporates episode-based payment initiated by chemotherapy treatment, a service generally reported with ICD-9-CM and ICD-10-CM codes that will be excluded from the AMI, CABG, and SHFFT episode definition in section III.C. of this proposed rule. We believe the care coordination and management services paid by OCM PBPM payments would be focused on chemotherapy services and their complications, so the services would be clinically unrelated to AMI, CABG and SHFFT model episodes. Therefore, we propose that services paid by PBPM payments under the OCM be excluded from the AMI, CABG and SHFFT models. Similarly, we propose to exclude services paid by PBPM payments under the MCCM. The MCCM focuses on providing care coordination and palliative care services for beneficiaries with certain conditions certified as terminally ill with a life expectancy of 6 months or less that have not elected the Medicare hospice benefit. The MCCM seeks to test whether providing palliative care services, without beneficiaries having to forgo curative care, incentivizes beneficiaries to elect hospice sooner. This is aimed at addressing the large percentage of hospice beneficiaries who elect the hospice benefit too late to fully benefit from the range of services that hospice has to offer at end of life. Since the purpose of the MCCM is to test whether providing palliative care services to beneficiaries who are otherwise eligible to elect the Medicare hospice benefit without requiring the beneficiary to forgo curative care results in beneficiaries electing the hospice benefit sooner, we will not include such Start Printed Page 50872payments in the AMI, CABG and SHFFT models' episode spending calculations. In addition, unlike the regular hospice benefits, which are furnished to beneficiaries in lieu of curative care and which therefore can be coordinated during an AMI, CABG or SHFFT model episode, the services furnished under the MCCM will be in addition to curative services. We note that we are including such curative services in the EPM episode, as they are consistent with our episode definition described in III.C. of this proposed rule, but not the services represented by the PBPM, which are provided in addition to curative services. Beneficiaries electing the hospice benefit could have lower episode spending because they have forgone curative care, however beneficiaries included in the MCCM may have higher episode spending because they are receiving both curative care and the services represented by the PBPM. We do not want to create incentives that deter providers from enrolling beneficiaries in the MCCM.

We acknowledge there may be new models that could incorporate a PBPM payment for new or enhanced services. We would plan to make our determination about whether services paid by a new model PBPM payment that is funded under the Medicare Trust Funds are clinically related to EPM episodes through the same sub regulatory approach that we are proposing to use to update the episode definitions (excluded MS-DRGs and ICD-CM diagnosis codes). We would assess each model's PBPM payment to determine if it would be primarily used for care coordination or care management services for excluded clinical conditions in the EPMs based on the standards we propose to use to update EPM episode definitions that are discussed in section III.C. of this proposed rule.

If we determine that a PBPM payment would primarily be used to pay for services to manage an excluded clinical condition, we would exclude the PBPM payment from the EPM on the basis that it pays for unrelated services. If we determine that the PBPM payment could primarily be used for services to manage an included clinical condition, we would include the PBPM payment in the EPM if the diagnosis code on the claim for the PBPM payment was not excluded from the episode, following our usual process for determining excluded claims for Part B services in accordance with the EPM episode definitions discussed in section III.C. of this proposed rule. We would post our proposed determination about whether the PBPM payment would be included in the episode to the CMS Web site to allow for public input on our planned application of these standards, and then adopt changes to the overlap list with posting to the CMS Web site of the final updated list after our consideration of the public input.

The payment reconciliation is described in section III.D.5. of this proposed rule. As with CJR, it is important that other models and programs in which providers are accountable for the total cost of care be able to account for the full Medicare payment, including EPM-related reconciliation payments and repayments as described in section III.D.5. of this proposed rule, for beneficiaries who are also in EPM episodes.

We establish the proposal for accounting for non-ACO services and payments in the EPM reconciliation process at § 512.210. We seek comment on this proposal.

7. Limits or Adjustments to EPM Participants' Financial Responsibility

a. Overview

We recognize that hospitals that would be designated for participation in the proposed EPMs currently vary with respect to their readiness to function under an EPM with regard to their organizational and systems capacity and structure, as well as their beneficiary population served. Some EPM participants may be more quickly able to demonstrate high quality performance and savings than others, even though we proposed that the EPM-episode benchmark prices be based predominantly on the hospital's own historical EPM-episode utilization in the early years of the EPMs. We also note that providers may be incentivized to excessively reduce or shift utilization outside of an EPM's episode by the proposed payment policies of the EPMs. In order to mitigate any excessive repayment responsibility for EPM participants or reduction or shifting of care outside an EPM episode, especially beginning in performance year 2 of the EPMs when we propose to begin to phase in responsibility for repaying Medicare for excess EPM-episode payments, we propose several specific policies as follows.

b. Limit on Actual EPM-Episode Payment Contribution to Repayment Amounts and Reconciliation Payments

(1) Limit on Actual EPM-Episode Payment Contribution to Repayment Amounts

As discussed in section III.D.3.d. of this proposed rule regarding our proposed pricing adjustment for high payment EPM episodes, EPM participants would not bear financial responsibility for actual EPM-episode payments greater than a ceiling set at 2 standard deviations above the mean regional EPM-episode payment. Nevertheless, EPM participants would begin to bear repayment responsibility beginning performance year 2 (DR) for those EPM episodes where actual EPM-episode payments are greater than the EPM quality-adjusted target prices up to the level of the regional EPM-episode ceiling. When aggregated across all EPM episodes in a model, the total money owed to Medicare by an EPM participant for actual EPM-episode payments above the applicable EPM quality-adjusted target price could be substantial if a hospital's EPM episodes generally had high payments. As an extreme example, if a hospital had all of its EPM episodes paid at 2 standard deviations above the mean regional EPM-episode payment, the EPM participant would need to repay Medicare a large amount of money, especially if the number of EPM episodes was large.

To limit a hospital's overall repayment responsibility for actual EPM-episode payments under the EPMs, (hereafter called a “stop-loss limit”), we propose to establish the same stop-loss limits that were adopted for the CJR model (80 FR 73401); except, that they would apply beginning in the second rather than first quarter of performance year 2. Specifically, we propose a 5 percent stop-loss limit in performance year 2 (DR), a 10 percent stop-loss limit in performance year 3, and a 20 percent stop-loss limit for performance years 4 and 5 for each EPM. That is, beginning in the second quarter of performance year 2 as we phase in repayment responsibility, the EPM participant would owe Medicare under each proposed EPM no more than 5 percent of the sum of the EPM quality-adjusted target prices for all of the EPM participant's EPM episodes during performance year 2 (DR). This responsibility gradually phases up to 20 percent by performance year 4.

For performance year 2, the comparison against the stop loss limit would only apply for NPRA attributable to episodes ending in performance year 2 (DR). When we calculate the NPRA for performance year 2 as described in section III.D.5. of this proposed rule, we would ensure the NPRA attributable to episodes ending during performance year 2 (NDR) is not less than zero and that NPRA attributable to episodes ending during performance year 2 (DR) does not exceed the stop-loss limit of 5 Start Printed Page 50873percent of the sum of quality-adjusted target prices for episodes that ended during performance year 2 (DR).

Similarly, when we conduct the subsequent reconciliation calculation to reassess actual EPM-episode payments for performance year 2 (which will occur concurrently with the reconciliation for performance year 3), we would combine the performance year 2 (NDR) NPRA and the result of the subsequent reconciliation calculation for performance year 2 (NDR) to ensure the result is not less than zero. Also, we would combine the performance year 2 (DR) NPRA and the result of the subsequent reconciliation calculation for performance year 2 (DR) to ensure the stop-loss limit is not exceeded.

For performance years 3 through 5, it would not be necessary to split the performance years to ensure that the stop-loss limit is not exceeded as a single stop-loss limit would apply in each year. For example, when we calculate the NPRA for performance year 3, as described in section III.D.5. of this proposed rule, we would ensure the NPRA does not exceed the stop-loss limit of 10 percent of the sum of quality-adjusted target prices. Similarly when we conduct the subsequent reconciliation calculation to reassess actual EPM-episode payments for performance year 3 (which will occur concurrently with the reconciliation for performance year 4), we would combine the performance year 3 NPRA and the result of the subsequent reconciliation calculation for performance year 3 to ensure the stop-loss limit is not exceeded.

Note that, as described in sections III.D.5.b. and III.D.7.e., the result of the post-episode spending calculation and ACO overlap calculation that would occur concurrently with the subsequent reconciliation calculation for a given performance year would not be subject to the stop-loss limit. The result of these calculations will be added to the NPRA and subsequent reconciliation calculation to create the repayment amount or reconciliation payment. We believe that these limits both offer EPM participants reasonable protections while maintaining incentives to improve care quality and efficiency. We would note that in addition to the CJR model, we apply a similar ultimate 20 percent stop-loss limit to payments under the BPCI initiative.

The proposal to limit hospitals' overall payment responsibility under the models is included in § 512.305(c)(2)(iii)(A). We seek comment on our proposal to limit hospitals' overall payment responsibility.

(2) Limitation on Reconciliation Payments

We believe limits on reconciliation payments made under the proposed EPMs would also be appropriate for several reasons. Under our proposal, in performance year 1, EPM participants have no repayment responsibility for excess EPM episode spending above the EPM quality-adjusted target price. CMS bears full financial responsibility for Medicare actual EPM-episode payments for an EPM episode that exceeds the EPM quality-adjusted target price, and we believe our responsibility should have judicious limits. Therefore, we believe it would be reasonable to cap an EPM participant's reconciliation payment due to actual EPM-episode payments for a given performance year as a percentage of EPM-episode payment on the basis of responsible stewardship of CMS resources. In addition, we note that beginning in performance year 1, EPM participants would be eligible for reconciliation payments due to the NPRA if actual EPM-episode payments are less than the quality-adjusted target prices. This proposal for reconciliation payments due to the NPRA provides a financial incentive to EPM participants from the beginning of the model to manage and coordinate care throughout the EPM episode with a focus on ensuring that EPM beneficiaries receive the lowest intensity, medically appropriate care throughout the EPM episode that results in high quality outcomes. Therefore, we also believe it would be reasonable to cap an EPM participant's reconciliation payment resulting from actual EPM-episode payments based on concerns about potential excessive reductions in utilization under the proposed EPMs that could lead to beneficiary harm.

In determining what would constitute an appropriate reconciliation payment limit due to actual episode spending (hereafter called a “stop-gain limit”), we believe it should provide significant opportunity for EPM participants to receive reconciliation payments for greater episode efficiency that includes achievement of quality care and actual EPM-episode payment reductions below the quality-adjusted target price, while avoiding the creation of significant incentives to sharply reduce utilization that could be harmful to EPM beneficiaries. We also believe that establishing parallel stop-gain and stop-loss limits is important to provide proportionately similar protections to CMS and EPM participants for their financial responsibilities under the EPMs as well as to protect the health of beneficiaries. Accordingly, we propose to establish symmetrical stop-gain limits. Specifically, we propose a 5 percent stop-gain limit in performance years 1 and 2, a 10 percent stop-gain limit in performance year 3, and a 20 percent stop-gain limit for performance years 4 and 5 for each EPM. That is, in performance year 1 as we phase in the stop-gain limits, the reconciliation payment that the EPM participant would be eligible to receive under each proposed EPM would be no more than 5 percent of the sum of the EPM quality-adjusted target prices for all of the EPM participant's EPM episodes during the performance year. This limit gradually phases up to 20 percent by performance year 4. As indicated in the CJR Final Rule, we want to ensure that any savings achieved by EPM participants in the early years of the EPM are not due to random variation, and that changes undertaken to improve efficiency include achievement in care quality and not sharp decreases in utilization that could be harmful to beneficiaries (80 FR 73402).

We clarify that, as with the stop-loss limit as discussed in this section, we propose that we would determine whether an EPM participant has met the stop-gain limit by assessing the NPRA and subsequent reconciliation for a given performance year, if any. We believe this approach aligns with our goal to place limits on the amount a participant may earn as a reconciliation payment due to reduced actual EPM-episode payments.

We would also note that we plan to monitor beneficiary access and utilization of services and the potential contribution of the stop-gain limit to any inappropriate reduction in EPM- episode services. We refer to section III.G. of this proposed rule for our proposals on monitoring and addressing hospital performance under the proposed EPMs.

The proposal to establish a cap on an EPM participant's reconciliation payment due to actual EPM-episode payments for a given performance year as a percentage of EPM-episode payment is included in § 512.305(c)(2)(iii)(B). We seek comment on this proposed cap.

c. Additional Protections for Certain EPM Participants

(1) Proposed Policies for Certain EPM Participants to Further Limit Repayment Responsibility

While the aforementioned proposals generally provide additional safeguards to ensure that EPM participants would have limited repayment responsibility due to the raw NPRA, we are proposing Start Printed Page 50874additional protections for certain groups of EPM participants that may have a lower risk tolerance and less infrastructure and support to achieve efficiencies for high-payment EPM episodes. Specifically, we are proposing additional protections for rural hospitals, SCHs, Medicare Dependent Hospitals, and Rural Referral Centers (RRCs). We note that these categories of hospitals often have special payment protections or additional payment benefits under Medicare because we recognize the importance of preserving Medicare beneficiaries' access to care from these hospitals.

For the purpose of these models, we propose to define a Rural Hospital as an IPPS hospital that is either located in a rural area in accordance with § 412.64(b) or in a rural census tract within an MSA defined at § 412.103(a)(1) or has reclassified to rural in accordance with § 412.103.

We propose to define a Sole Community Hospital as it is defined in § 412.92. That is, hospitals paid under the IPPS can qualify for SCH status if they meet one of the following criteria:

  • Located at least 35 miles from other like hospitals.
  • Located in a rural area, located between 25 and 35 miles from other like hospitals, and no more than 25 percent of residents or Medicare beneficiaries who become hospital inpatients in the hospital's service area are admitted to other like hospitals located within a 35-mile radius of the hospital or the hospital has fewer than 50 beds and would meet the 25 percent criterion if not for the fact that some beneficiaries or residents were forced to seek specialized care outside of the service area due to the unavailability of necessary specialty services at the hospital.
  • Hospital is rural and located between 15 and 25 miles from other like hospitals but because of local topography or periods of prolonged severe weather conditions, the other like hospitals are inaccessible for at least 30 days in each of 2 out of 3 years.
  • Hospital is rural and the travel time between the hospital and the nearest like hospital is at least 45 minutes.

We propose to define a Medicare Dependent Hospital (MDH) as it is defined in § 412.108. That is, an MDH is a hospital that meets the following criteria:

  • Located in a rural area.
  • Has 100 beds or less.
  • Is not a SCH.
  • Sixty percent of the hospital's inpatient days or discharges were attributable to individuals entitled to Medicare Part A benefits during specified time periods as provided in § 412.108.

We propose to define a Rural Referral Center as it is defined in § 412.96. Specifically, RRCs are defined as IPPS hospitals with at least 275 beds that meet the following criteria:

  • Fifty percent of the hospital's Medicare patients are referred from other hospitals or from physicians who are not on the staff of the hospital.
  • At least 60 percent of the hospital's Medicare patients live more than 25 miles from the hospital.
  • At least 60 percent of all services the hospital furnishes to Medicare patients are furnished to patients who live more than 25 miles from the hospital.

If a hospital does not meet these criteria, a hospital can also qualify for RRC status if a hospital meets the following criteria:

  • For specified period of time, the hospital has a case-mix that equals at least the lower of the median case mix index (CMI) value for all urban hospitals nationally; or the median CMI value for urban hospitals located in its region, excluding those hospitals receiving indirect medical education payments.
  • Its number of discharges is at least—

++ 5,000 (or 3,000 for an osteopathic hospital); or

++ The median number of discharges for urban hospitals in the census region in which it is located, set by the CMS through IPPS rulemaking.

  • Additionally, a hospital must meet one of the following criteria:

++ More than 50 percent of its active medical staff are specialists who meet the conditions specified at § 412.96(c)(3).

++ At least 60 percent of all discharges are for inpatients who reside more than 25 miles from the hospital.

++ At least 40 percent of all inpatients treated are referred from other hospitals or from physicians who are not on the hospital's staff.

Additional information on these hospitals can be found in the CJR Final Rule at 80 FR 73403 through 73405.

In the CJR Final Rule, we established the same stop-gain limits for these hospitals as for hospitals in general (that is, 5 percent in performance years 1 and 2, 10 percent in performance year 3, and 20 percent in performance years 4 and 5); however, we limited losses for rural hospitals, SCHs, Medicare Dependent Hospitals and RRCs to 3 percent in performance year 2, and 5 percent in performance years 3 through 5 (80 FR 73406). In that Final Rule, we noted that these hospitals can face unique challenges that do not exist for most other hospitals. For example, these hospitals may be the only source of healthcare services for beneficiaries or certain beneficiaries living in rural areas, and may be in areas with fewer providers including fewer physicians and post-acute care facilities. Further, these hospitals may have more limited options in coordinating care and reducing spending while maintaining quality of care. We continue to believe that urban hospitals may not have similar concerns as they are often in areas with many other providers and have a greater opportunity to develop efficiencies under the EPMs. Given these circumstances, for the CJR model we determined that we should have a more protective stop-loss limit policy for these hospitals. Given the similarity between the CJR model and the proposed EPMs, we have similar concerns, which we believe should be addressed by establishing greater protections for these hospitals when they are EPM participants. Accordingly, we are proposing the same stop-loss thresholds for these hospitals participating in the proposed EPMs as were adopted for the CJR model except that the thresholds would begin in performance year 2 (DR)—specifically, 3 percent in performance year 2 (DR), and 5 percent for performance years 3 through 5 for each EPM.

The proposal to establish separate financial loss limits for certain hospitals that could be less able to tolerate risk is included in § 512.305(c)(2)(iii)(C). We seek comment on our proposed limit on financial loss for these hospitals.

(2) Considerations for Hospitals Serving a High Percentage of Potentially Vulnerable Populations

In addition to the aforementioned hospitals, we recognize that other EPM participants, for which we do not propose additional protections, could also face factors affecting their ability to achieve savings under the proposed EPMs, and that these factors could be unrelated to their practice patterns but instead could reflect the EPM participants' responsibilities for a relatively high percentage of potentially vulnerable populations with higher than average historical spending and/or less opportunities for efficiencies. For example, this could include hospitals that serve a relatively high percentage of beneficiaries that are dually eligible for both Medicare and Medicaid or whose total Medicare payments include a relatively high proportion of disproportionate share hospital payments under 1886(d) (5) (F) of the Act. Some of these hospitals are located in rural areas and would thus likely be Start Printed Page 50875classified as a type of hospital for which we propose additional protections. However, most hospitals that serve a relatively high percentage of beneficiaries that are dually eligible for both Medicare and Medicaid or whose total Medicare payments include a relatively high proportion of disproportionate share hospital payments are located in urban areas, and very few are classified as a rural hospital, RRC, MDH, or SCH that would be subject to the additional protections we propose. For the first 2 performance years of the EPMs, where quality-adjusted target prices are set predominantly based on EPM-participant hospital-specific data, factors affecting these hospitals may be of less concern than in the final 3 performance years of the EPMs where pricing is either predominantly or totally based on regional data.

The potential challenges posed by these kinds of factors is highlighted in Section 2(d) of the Improving Medicare Post-Acute Care Transformation “IMPACT” Act of 2014 (Pub. L. 113-183). Specifically, Section 2(d) requires the Secretary to conduct a study that examines the effect of individuals' socioeconomic status, including their Medicaid eligibility, on quality measures and resource use and other measures for individuals under the Medicare program, in recognition that less healthy individuals may require more intensive interventions. The Secretary is required to submit a report on the results of this study within 2 years of enactment of the IMPACT Act. The IMPACT Act also requires the Secretary to conduct a second study that examines the impact of various risk factors, as well as race, health literacy, limited English proficiency (LEP), and Medicare beneficiary activation, on quality measures and resource use and other measures under the Medicare program in order to recognize that less healthy individuals may require more intensive interventions. The Secretary must submit a report on the results of this study within 5 years of enactment of the IMPACT Act.

If these studies find a relationship between the factors examined in the studies and quality measures and resource use and other measures, then the Secretary shall provide recommendations for, among other things, how CMS should account for such factors in quality measures, resource use measures, and other measures under Medicare; and in determining payment adjustments based on such measures in other applicable provisions related to the program. Likewise, taking into account these studies and their recommendations as well as other relevant information, the Secretary is required to routinely, as determined appropriate and based on an individual's health status and other factors, assess appropriate adjustments to quality measures, resource use measures, and other measures under the Medicare program; and assess and implement appropriate adjustments to Medicare payments based on these measures. The Assistant Secretary for Planning and Evaluation is responsible for these studies and a report on the results of the first one is forthcoming. Upon issuance of these studies' reports, we plan to consider their results as we implement the proposed EPMs. We also plan to monitor the influence of beneficiary characteristics such as socioeconomic status on EPM participants' performance during our implementation and evaluation of the EPMs. Given that the performance of EPM participants would be compared largely against their own historical episode cost performance data for the first 2 years of the models, we do not anticipate that the aforementioned factors should materially affect participants' ability to achieve savings. However, as we increasingly begin to rely more on regional cost performance data to determine episode benchmarks and quality-adjusted target prices in performance year 3, these factors could become more germane. Thus, in the event we identify the need for adjustments, we could consider proposing additional policies through subsequent rulemaking. Additionally, we plan to use information collected as part of our efforts to monitor beneficiary access to care and quality of care as discussed in sections III.G.4. and III.G.5. of this proposed rule to inform if potential adjustments would be needed in future years of the model.

Protections for EPM participants are discussed in section III.D.7.b.(1) of this proposed rule. We seek comment about all issues specific to hospitals serving a high percentage of potentially vulnerable populations and their opportunities to advance the goals of the EPMs. In particular, we seek comment, including data analysis, about approaches to identifying these hospitals; their opportunities to achieve high quality episode performance; specific considerations about their opportunities to achieve efficient care for the clinical conditions included in the AMI, CABG, and SHFFT models; potential approaches to risk adjustment as elaborated upon in section III.D.4.b.(2)(d) of this proposed rule; potential approaches to additional protections that could be considered for the future modeled after our proposals in section III.D.7.b.(1) of this proposed rule for certain other EPM participants or other alternatives; and evaluation methodologies to ensure that we include appropriate comparison groups and monitor and evaluate the most relevant outcomes.

d. Application of Stop-Gain and Stop-Loss Limits

Because hospitals could be participating in the proposed AMI, CABG, and SHFFT models concurrently with the CJR model, an additional consideration concerns the level at which the stop-loss and stop-gain thresholds would be applied, for example, at the hospital level, as is currently the case for the CJR model, or at some other level, for example, at the model level. Our intention is to establish appropriate incentives and protections for hospitals under the proposed EPMs and the CJR model without creating unnecessary administrative complexity. This issue becomes especially relevant to the proposed EPMs and CJR model given that the CJR model and proposed EPMs would be operating at different points within their performance periods. That is, episodes under the proposed EPMs would always lag 1 performance year behind those in the CJR model. Thus, SHFFT model participants that would begin the first SHFFT model performance year in 2017 would already be participating in their second performance year under the CJR model. Consequently, in this example, a stop-loss limit could apply to the performance year 2 episodes under the CJR model but not to the performance year 1 SHFFT model episodes under the SHFFT model as SHFFT model participants would not have repayment responsibility in SHFFT model performance year 1 under our proposal. In contrast, for this example, the stop-gain limits would be the same for both the SHFFT and CJR model since the limit for both performance year 1 and 2 would be 5 percent.

Continuing with this example for a later performance year (performance year 4 for the CJR model and performance year 3 for the SHFFT model), any stop-loss limits that applied would be different. That is, the stop-loss limits for the CJR model episodes in performance year 4 would be 20 percent in contrast to the 10 percent stop-loss limit that would apply to the SHFFT model episodes in performance year 3. The proposed stop-gain limits would likewise diverge in this example as they Start Printed Page 50876are proposed to be symmetrical with the stop-loss limits.

Given these differences, we considered two options for setting stop-gain and stop-loss limits for hospitals participating in more than one of the AMI, CABG, SHFFT, and CJR models. Under the first option, we would determine stop-loss and stop-gain limits, in total, at the participant level based on weighted thresholds. Specifically, CMS would calculate a single weighted stop-loss/gain threshold based on the total spending under each model. Thus, using the aforementioned example where CJR model episodes would be in performance year 4 of their model and SHFFT model episodes would be in performance year 3, assuming 50 percent of total spending under the CJR and SHFFT models is for CJR model episodes and the remaining 50 percent is for SHFFT model episodes, the weighted stop-loss limit for the two models at the hospital level would be 15 percent: (0.50 × 0.20 for CJR model episodes) + (0.5 × 0.10 for SHFFT model episodes) = 0.15. Although this option would allow the application of a single stop-loss threshold to a hospital's total repayment under the models, we are concerned that computing a single limit such as this could either dilute or magnify the intended protections of the stop-loss limit under each model. As such, a hospital that would have been protected from repayment exceeding 10 percent of its SHFFT model quality-adjusted target prices multiplied by the number of SHFFT model episodes for performance year 3 would only be protected for costs above the higher 15 percent level. Conversely, a hospital that would have been protected only for repayment above 20 percent of its CJR model quality-adjusted target prices multiple by the number of CJR model episodes for performance year 3 would be protected against repayment above the lower 15 percent threshold.

Alternatively, we considered establishing stop-loss and stop-gain thresholds at the model level; that is, separately for each of the AMI, CABG, and SHFFT models, in addition to the limits that already exist for the CJR model. Under this option, we would separately apply the CJR-applicable stop-loss and stop-gain limits to CJR model episodes, the AMI-applicable limits to AMI model episodes, and so forth. Thus, considering the aforementioned example, the stop-loss limit for CJR model episodes in performance year 4 would be 20 percent for the hospital's CJR model episodes, while the stop-loss limit for SHFFT model episodes for performance year 3 would be 10 percent. While we might choose to aggregate these amounts to conduct a single financial transaction with a hospital participating in more than one model, we believe this option that would apply stop-loss and stop-gain limits at the model level for hospitals participating in more than one model is superior to first option in that it better maintains appropriate incentives and protections under each of the models.

The proposal to establish stop-gain and stop-loss limits at the model level is included in § 512.305(c)(2)(iii)(D). We seek comment on our proposal to establish stop-gain and stop-loss limits at the model level.

e. EPM Participant Responsibility for Increased Post-Episode Payments

We note that while episodes under the proposed EPMs would extend 90 days post-discharge from the anchor or chained anchor hospitalization, some EPM participants may have an incentive to withhold or delay medically-necessary care until after an EPM episode ends to reduce its actual EPM-episode payments. This inappropriate shifting could include both those services that are related to the episode (for which the hospital would bear financial responsibility as such services would be included in the actual EPM-episode payment calculation) and those that are unrelated (which would not be included in the actual EPM-episode payment calculation), because an EPM participant engaged in shifting of medically-necessary services outside the EPM episode for potential financial reward may be unlikely to clearly distinguish whether the services were related to the EPM episode or not in the hospital's decisions.

We believe that this inappropriate shifting would not be typical, especially given the relatively long EPM episode duration. However, in order to identify and address inappropriate shifting of care, we propose to calculate for each EPM performance year the total Medicare Parts A and B expenditures in the 30-day period following completion of each EPM episode for all services covered under Medicare Parts A and B, regardless of whether the services are included in the proposed EPM episode definition (sections III.C.3. and III.C.4 of this proposed rule). This proposal is consistent with our processes for BPCI Model 2 and the CJR model (80 FR 73407 through 73408).

We propose that the post-episode spending calculation for a performance year would occur at the same time we perform the subsequent reconciliation calculation for that same year. We believe this timeframe will allow sufficient time for claims run out in order to set a reliable regional threshold for determining the post-episode spending. For example, we would conduct reconciliation for performance year 1 in the spring of 2018. The post-episode spending calculation for performance year 1 would occur during the next reconciliation process (spring 2019), when we conduct the subsequent reconciliation calculation for performance year 1 and account for overlap with other models and programs.

Our proposed calculation would include prorated payments for services that extend beyond the EPM episode as discussed in section III.D.3.c. of this proposed rule. Specifically, we would identify whether the average 30-day post-episode spending for an EPM participant in any given EPM performance year is greater than 3 standard deviations above the regional average 30-day post-episode spending, based on the 30-day post-episode spending for episodes attributed to all regional hospitals participating in the EPM in the same region as the EPM participant. We propose that if the EPM participant's average post-episode spending exceeds this threshold, the EPM participant would repay Medicare for the amount that exceeds such threshold. We note that an EPM participant's responsibility for post-episode spending would not be subject to the stop-loss and stop-gain limits proposed in section III.D.7.b. of this proposed rule. Although we believe cases in which an EPM participant would be responsible for repayment of post-episode spending that exceed the threshold would be rare, our intention is to identify and hold EPM participants responsible for situations in which those participants have significantly increased spending on services in the 30 days following the end of an EPM episode in order to inappropriately shift services out of EPM episodes. We do not believe such behavior should be subject to stop-loss limits. This policy is consistent with our proposal for the CJR model in section V.D.1. of this proposed rule.

Based on our experience with BPCI, we have not found that this proposal, including our proposal to include all Medicare Parts A and B expenditures to measure 30-day post-episode spending, would inappropriately penalize EPM participants. To that end, however, we believe our proposed threshold of 3 standard deviations above the regional average is a high threshold, and we only propose that an EPM participant would repay Medicare for the amount that Start Printed Page 50877exceeds such threshold. We further note that those EPM participants that are eligible for reconciliation payments in an EPM performance year and also have average 30-day post-episode spending that is higher than 3 standard deviations above the regional average 30-day post-episode spending would have their reconciliation payments reduced by the amount by which spending exceeds 3 standard deviations.

The proposals to determine if a participant's post-episode spending 30 days after the end of an episode exceeds 3 standard deviations of average spending in their region for that period, and require those participants exceeding that threshold to repay Medicare for the amounts in excess of 3 standard deviations are included in § 512.307(c). We seek comment on our proposals to determine if a participant exceeds this threshold and to repay amounts in excess of the threshold.

8. Appeals Process

a. Overview

Consistent with the BPCI initiative and CJR model, we propose to institute appeals processes for the EPMs that would allow EPM participants to appeal matters related to payment, CR incentive payments, reconciliation amounts, repayment amounts, determinations associated with quality measures affecting payment, as well as non-payment related issues, such as enforcement matters. These matters are discussed throughout section III.D. and III.F. respectively.

We seek comment on the proposal to institute appeals processes, in the following discussion, for the EPMs.

b. Notice of Calculation Error (First Level Appeal)

We propose the following calculation error process for EPM participants to contest matters related to payment or reconciliation, of which the following is a non-exhaustive list: The calculation of the EPM participant's reconciliation amount or repayment amount as reflected in the reconciliation report; the calculation of the EPM participant's CR incentive payment as reflected in the CR incentive payment report; the calculation of NPRA; the calculation of the percentiles of quality measure performance to determine eligibility to receive a reconciliation payment; and the successful reporting of the voluntary PRO THA/TKA data to adjust the reconciliation payment. EPM participants would review their reconciliation report and CR incentive payment report and be required to provide written notice of any error, in a calculation error form that must be submitted in a form and manner specified by CMS. Unless the EPM participant provides such notice, the reconciliation report and CR incentive report would be deemed final within 45 calendar days after it is issued, and CMS would proceed with payment or repayment. If CMS receives a timely notice of an error in the calculation, CMS would respond in writing within 30 calendar days to either confirm or refute the calculation error, although CMS would reserve the right to an extension upon written notice to the participant. We propose that if an EPM participant does not submit timely notice of a calculation error, that is notice within 45 calendar days of the issuance of the reconciliation report and CR incentive payment report the EPM participant would be precluded from later contesting any of the following matters contained in the reconciliation report or CR incentive payment report for that performance year; any matter involving the calculation of the EPM participant's reconciliation amount or repayment amount as reflected in the reconciliation report; any matter involving the calculation of the EPM participant's CR incentive payment as reflected in the CR incentive payment report; any matter involving the calculation of NPRA; the calculation of the percentiles of quality measure performance to determine eligibility to receive a reconciliation payment; and the successful reporting of the voluntary PRO THA/TKA data to adjust the reconciliation payment. Given that EPM participants bear the financial risk in the EPM model, only EPM participants may use the dispute resolution process described in this section.

In summary, we propose the following requirements in § 512.310 (a) for notice of calculation error:

  • Subject to the limitations on review in subpart D of this part, if an EPM participant wishes to dispute the calculation that involves a matter related to payment, a CR incentive payment, reconciliation amounts, repayment amounts, or determinations associated with quality measures affecting payment, the EPM participant is required to provide timely written notice of the error, in a form and manner specified by CMS.
  • Unless the EPM participant provides such notice, CMS deems final the reconciliation report and CR incentive payment report 45 calendar days after the reconciliation report or CR incentive payment report is issued and proceeds with the payment or repayment processes as applicable.
  • If CMS receives a notice of a calculation error within 45 calendar days of the issuance of the reconciliation report or CR incentive payment report, CMS responds in writing within 30 calendar days to either confirm that there was an error in the calculation or verify that the calculation is correct, although CMS reserves the right to an extension upon written notice to the EPM participant.
  • Only EPM participants may use the notice of calculation error process described in this part.

We seek comment on the proposed notice of calculation error requirements.

c. Dispute Resolution Process (Second Level of Appeal)

We propose the following dispute resolution process. First, we propose that only an EPM participant may utilize the dispute resolution process. Second, in order to access the dispute resolution process a participant must have timely submitted a calculation error form, as previously discussed, for any matters related to payment. We propose these matters would include any amount or calculation indicated on a reconciliation report or CR incentive payment report, including calculations not specifically reflected on a reconciliation report or CR incentive payment report but which generated figures or amounts reflected on a reconciliation report or a CR incentive payment report. The following is a non-exhaustive list of the matters we propose would need to be first adjudicated by the calculation error process as previously detailed: Calculations of reconciliation or repayment amounts; calculation of CR incentive payment amounts; calculations of NPRA; and any calculations or percentile distribution involving quality measures that we propose could affect reconciliation or repayment amounts. If an EPM participant wants to engage in the dispute resolution process with regard to one of these matters, we propose it would first need to submit a calculation error form. Where the EPM participant does not timely submit a calculation error form, we propose the dispute resolution process would not be available to the EPM participant with regard to those matters for the reconciliation report or CR incentive payment report for that performance year.

If the EPM participant did timely submit a calculation error form and the EPM participant is dissatisfied with CMS's response to the EPM participant's notice of calculation error, the EPM participant would be permitted to Start Printed Page 50878request reconsideration review by a CMS reconsideration official. The reconsideration review request would be submitted in a form and manner and to an individual or office specified by CMS. The reconsideration review request would provide a detailed explanation of the basis for the dispute and include supporting documentation for the EPM participant's assertion that CMS or its representatives did not accurately calculate the NPRA, the CR incentive payment, or post-episode spending amount in accordance with EPM rules. The following is a non-exhaustive list of representative payment matters:

  • Calculations of NPRA, calculations of the CR incentive payment, post-episode spending amount, target prices or any items listed on a reconciliation report or CR incentive payment report.
  • The application of quality measures to a reconciliation payment, including the calculation of the percentiles thresholds of quality measure performance to determine eligibility to receive reconciliation payments, or the successful reporting of the voluntary PRO THA/TKA data to adjust the reconciliation payment.
  • Any contestation based on the grounds that CMS or its representative made an error in calculating or recording such amounts.

Where the matter is unrelated to payment, such as termination from the model, the EPM participant need not submit a calculation error form. We propose to require the EPM participant to timely submit a request for reconsideration review, in a form and manner to be determined by CMS. Where such request is timely received, we propose CMS would process the request as discussed later in this section.

We propose that the reconsideration review would be an on-the-record review (a review of briefs and evidence only). The CMS reconsideration official would make reasonable efforts to notify the EPM participant in writing within 15 calendar days of receiving the EPM participant's reconsideration review request of the date and time of the review, the issues in dispute, the review procedures, and the procedures (including format and deadlines) for submission of evidence (the “Scheduling Notice”). The CMS reconsideration official would make reasonable efforts to schedule the review to occur no later than 30 days after the date of the Scheduling Notice. The provisions at § 425.804(b), (c), and (e) (as in effect on the publication date of this proposed rule) would apply to reviews conducted pursuant to the reconsideration review process for EPM. The CMS reconsideration official would make reasonable efforts to issue a written determination within 30 days of the review. The determination would be final and binding.

We solicit comment on our proposals related to appeals rights under this model. The two-step appeal process for payment matters—(1) calculation error form, and (2) reconsideration review—is used broadly in other CMS models. We seek comment on whether we should develop an alternative appeal process. We are also interested in whether there should be appeal rights for reductions or eliminations of NPRA as a result of enforcement actions, as discussed in section III.F. of this proposed rule, and if so, whether the process for such appeals should differ from the processes proposed here.

In summary, we propose the following requirements in § 512.310(b) for the reconsideration process:

  • If the EPM participant is dissatisfied with CMS's response to the notice of a calculation error, the EPM participant may request a reconsideration review in a form and manner as specified by CMS.
  • The reconsideration request must provide a detailed explanation of the basis for the dispute and include supporting documentation for the EPM participant's assertion that CMS or its representatives did not accurately calculate the NPRA, the reconciliation payment, the CR incentive payment or the repayment amount in accordance with subpart d of this part.
  • If CMS does not receive a request for reconsideration from the EPM participant within 10 calendar days of the issue date of CMS's response to the EPM participant's notice of calculation error, then CMS's response to the calculation error is deemed final and CMS proceeds with reconciliation payment or repayment processes, as applicable, as described in § 512.305.
  • The CMS reconsideration official notifies the EPM participant in writing within 15 calendar days of receiving the EPM participant's review request of the following:

++ The date, time, and location of the review.

++ The issues in dispute.

++ The review procedures.

++ The procedures (including format and deadlines) for submission of evidence.

  • The CMS reconsideration official takes all reasonable efforts to schedule the review to occur no later than 30 days after the date of receipt of notification.
  • The provisions at § 425.804(b), (c), and (e) of this chapter are applicable to reviews conducted in accordance with the reconsideration review process for the EPM.
  • The CMS reconsideration official issues a written determination within 30 days of the review. The determination is final and binding.

Only EPM participants may utilize the dispute resolution process described in this subpart. We seek comment on the proposed reconsideration process for the EPMs.

d. Exception to the Notice of Calculation Error Process and Notice of Termination

Similar to the CJR model and BPCI initiative, if the EPM participant contests a matter that does not involve an issue contained in, or a calculation which contributes to, an EPM reconciliation report or a CR incentive report, a notice of calculation error is not required. Consistent with III.D.8(c) in this proposed rule, in instances where a notice of calculation error is not required, for example an EPM participant's termination from the EPM, we propose the EPM participant provide a written notice to CMS requesting review within 10 calendar days of the notice. CMS has 30 days to respond to the EPM participant's request for review. If the EPM participant fails to notify CMS, the decision is deemed final.

In summary, we propose the following requirements in § 512.310(c) for an exception to the notice of calculation error process.

  • If the EPM participant contests a matter that does not involve an issue contained in, or a calculation which contributes to, a reconciliation report or CR incentive payment report, a notice of calculation error is not required. In these instances, if CMS does not receive a request for reconsideration from the EPM participant within 10 calendar days of the notice of the initial determination, the initial determination is deemed final and CMS proceeds with the action indicated in the initial determination.

In summary, we propose the following requirements in § 512.310(d) for notice of termination:

  • If an EPM participant receives notification that it has been terminated from the EPM and wishes to appeal such termination, it must provide a written notice to CMS requesting review of the termination within 10 calendar days of the notice. CMS has 30 days to respond to the EPM participant's request for review. If the participant fails to notify CMS, the termination is deemed final.Start Printed Page 50879

We seek comment on the proposed exception to the notice of calculation error process and notice of termination.

e. Limitations on Review

In summary, we propose the following requirements in § 512.310(e) for limitations on review:

  • In accordance with section 1115A(d)(2) of the Act, there is no administrative or judicial review under sections 1869 or 1878 of the Act or otherwise for the following:

++ The selection of models for testing or expansion under section 1115A of the Act.

++ The selection of organizations, sites, or participants to test those models selected.

++ The elements, parameters, scope, and duration of such models for testing or dissemination.

++ Determinations regarding budget neutrality under section 1115A(b)(3) of Act.

++ The termination or modification of the design and implementation of a model under section 1115A(b)(3)(B) of Act.

++ Decisions to expand the duration and scope of a model under section 1115A(c) of the Act, including the determination that a model is not expected to meet criteria described in paragraph (e)(1) or (2) of this section.

We seek comment on the proposed limitations on review.

III. Provisions of the Proposed Regulations

E. EPM Quality Measures, Public Display, and Use of Quality Measures in the EPM Payment Methodology

1. Background

As discussed in the CJR model final rule, Medicare payment policy has moved away from FFS payments unlinked to quality and towards payments that are linked to quality of care (80 FR 73358). Through the Medicare Modernization Act and the Affordable Care Act, we have implemented specific IPPS programs like the HIQR Program (section 1886(b)(3)(B) of the Act), the HVBP Program (subsection (o) of section 1886), the Hospital Acquired Condition Reduction Program (HACRP) (subsection (q) of section 1886), and the Hospital Readmissions Reduction Program (HRRP) (subsection (p) of section 1886), where quality of care is linked to payment. We have also implemented the Shared Savings Program, an ACO program that links shared savings payment to quality performance. The CJR model similarly incorporates pay-for-performance through the potential for financial reward to participants based on the hospital's level of quality performance, while also including an incentive for quality improvement if the hospital's current level of quality is relatively low (80 FR 73374).

We propose pay-for-performance methodologies similar to the CJR model for the proposed EPMs. Specifically, we propose to financially reward higher quality in an EPM episode by reducing the effective discount factor used to calculate EPM quality-adjusted target prices at reconciliation. We would establish the effective discount factor based on the EPM participant's overall quality performance and improvement on the EPM's quality measures as reflected in the EPM participant's EPM composite quality score. We would calculate the EPM participant's composite quality score for each EPM performance year at the time of reconciliation. The EPM composite quality score would also determine whether an EPM participant is eligible for a reconciliation payment if savings are achieved beyond the EPM quality-adjusted target price by setting a minimum EPM composite quality score for reconciliation payment eligibility.

We note that we continue to believe that EPMs should include pay-for-performance methodologies that incentivize improvements in patient outcomes while simultaneously lowering health care spending (80 FR 73465). We believe that improved quality of care, specifically achieved through coordination and communication among providers in conjunction with patients and their caregivers, can favorably influence performance on patient outcomes. Like the CJR model, we also believe that the proposed three new EPMs would provide the opportunity for EPM participants to improve the quality of care based on timely reported patient experience, including communications with doctors and nurses, and responsiveness of hospital staff (80 FR 73465). Finally, we strive to align as many measures as possible in CMS's proposed new EPMs with those in ongoing models and programs. Our goal is to focus provider improvement efforts and minimize burden on EPM participants in needing to become familiar with and report new measures, while still allowing us to appropriately capture meaningful quality data and use it in the EPMs' pay-for-performance methodologies.

More specifically, similar to our final decision for the CJR model, we are not proposing to use any readmissions measures that could apply to clinical conditions in these EPMs but that are already in place or have been finalized for the HRRP, specifically the Hospital 30-day all-cause risk-standardized readmission rate (RSRR) following AMI hospitalization (NQF #0505) and the Hospital 30-day all-cause, unplanned, RSRR following CABG surgery (NQF #2515), due to the incentives, already in place by the HRRP, for hospitals to lower excess readmission rates (80 FR 73479). While we consider these readmissions measure rates to be important metrics for providing information about AMI and CABG hospital performance in the HRRP and HIQR Program for payment and public reporting, respectively, other proposed measures for the AMI and CABG models support the intent of these models to reduce actual payments in an EPM episode while ensuring that quality of care for AMI and CABG model beneficiaries is improved.

Furthermore, while we recognize the lack of complete alignment between EPM beneficiaries and the proposed cohorts for the EPM quality measures, we believe the proposed measures provide meaningful information about EPM participant quality performance and improvement that are relevant to EPM beneficiaries. For the AMI and CABG models in particular, beneficiaries included in the proposed episode-specific measures would significantly overlap with beneficiaries in AMI and CABG model episodes. We note that for purposes of the EPMs where we need to identify episodes that are included in the EPMs, we use the terms anchor and chained anchor hospitalization to identify hospitalizations that initiate EPM episodes for beneficiaries whose care is included in the EPMs. In describing the quality measures in detail in section III.E.4. of this proposed rule, we use the term index hospitalization to identify hospitalizations of beneficiaries whose outcomes are included in the measures. Thus, anchor hospitalizations and index hospitalizations would have varying degrees of overlap depending on the specific quality measure.

Moreover, we note that hospitals are the unit of analysis for the EPMs and that the proposed measures are hospital-centric measures, both because these are currently available measures that are aligned with those in other CMS programs and because one of the major goals of the EPMs is to encourage collaboration among different types of providers in order to achieve better care and reduced expenditures, while holding acute care hospitals financially responsible. For further discussion of our proposal that hospitals be Start Printed Page 50880accountable for EPM episodes, we refer to section III.B.3. of this proposed rule.

We recognize that there are also some gaps in the current proposed measures relative to other settings in which patients receive care post-hospital discharge during EPM episodes, as well as around important complications of care for clinical conditions included in the three models. However, we believe that these hospital-level measures reasonably assess how well EPM participants provide care for EPM beneficiaries since the measures, depending on the EPM, assess—(1) important patient outcomes, including mortality as well as complications and days of acute care following discharge from the index hospitalization which can be costly; and (2) patients' perspectives on their hospital experience, which include patient feedback on communication with doctors, communication with nurses, responsiveness of hospital staff, communication about medicines, discharge information, cleanliness of the hospital environment, quietness of the hospital environment, and transition to post-hospital care. As we gain more experience with the EPMs, as well as the CJR model currently in testing, and future EPMs, we plan to work to create a more robust set of episode quality measures for these and future models. We will continue to assess the evolving inventory of measures and will continue to refine quality measures for potential future rulemaking based on public comments, changes to the EPMs' payment methodologies, recommendations from EPM participants and their collaborators, and new CMS episode measure development activities as we learn more about the impact of EPMs on quality improvement and episode efficiency. We refer to section III.E.4.e. of this proposed rule for a discussion of potential future EPM episode measures.

2. Selection of Proposed Quality Measures for the EPMs

a. Overview of Quality Measure Selection

The outcome and patient experience measures proposed for the EPMs were selected in order to: (1) Promote alignment with the financial and quality goals of the EPMs; (2) leverage hospitals' familiarity with the measures due to their use in other CMS hospital quality programs, including programs that tie payment to performance such as the HVBP Program; (3) streamline EPM measures for EPM participants testing more than one EPM; and (4) ensure consistency with CMS's priorities to reduce AMI and CABG mortality and complications while improving patient experience, as well as with CMS's priorities to reduce major LEJR surgery complications while improving patient experience for SHFFT model beneficiaries, like those in the CJR model.

b. AMI Model Quality Measures

In order to encourage care collaboration among multiple providers of AMI model beneficiaries, we propose three required measures and one measure that relies on voluntary data submission, in order to determine AMI model participant episode quality performance and improvement that would be linked to the AMI model payment methodology as discussed in section III.E.3.f.(2) of this proposed rule. We propose the following measures for the AMI model:

  • Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) Following Acute Myocardial Infarction (NQF #0230) (MORT-30-AMI).
  • Excess Days in Acute Care after Hospitalization for AMI (AMI Excess Days).
  • HCAHPS Survey (NQF #0166).
  • Voluntary Hybrid Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate Following Acute Myocardial Infarction (AMI) Hospitalization (NQF #2473) (Hybrid AMI Mortality) data submission.

We refer to sections III.E.4.a. and d. of this proposed rule for a detailed discussion of our proposals regarding these measures for the AMI model, including their importance as measures of the quality-of-care for beneficiaries treated for AMI. The proposals for the AMI model measures are included in § 512.411, and the proposals for reporting the measures are included in § 512.400. We seek comment on our proposals for AMI model quality measures.

c. CABG Model Quality Measures

In order to encourage care collaboration among multiple providers of CABG model beneficiaries, we propose two required measures, in order to determine CABG model participant episode quality performance and improvement that would be linked to the CABG model payment methodology as discussed in section III.E.3.f.(3) of this proposed rule. We propose the following measures for the CABG model:

  • Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) Following Coronary Artery Bypass Graft (CABG) Surgery (NQF #2558) (MORT-30-CABG).
  • HCAHPS Survey (NQF #0166).

We refer to sections III.E.4.b. and d. of this proposed rule for a detailed discussion of our proposals regarding these measures for the CABG model, including their importance as measures of the quality-of-care for beneficiaries treated with CABG.

The proposals for the CABG model measures are included in § 512.412., and the proposals for reporting the measures are included in § 512.400. We seek comment on our proposals for CABG model quality measures.

d. SHFFT Model Quality Measures

In order to encourage care collaboration among multiple providers of SHFFT model beneficiaries, we propose two required measures and one measure that relies on voluntary data submission, in order to determine SHFFT model participant episode quality performance and improvement that would be linked to the SHFFT model payment methodology as discussed in section III.E.3.f.(4) of this proposed rule. While we recognize that none of the proposed measures specifically target the care of SHFFT model beneficiaries, these measures are the same as those used for the CJR model because SHFFT model episodes will be tested along with the LEJR episodes in the CJR model (80 FR 73501 and 73507) at mostly the same hospitals. In addition, as discussed further in section III.E.3.e.(3) of this proposed rule, we propose to calculate a hospital-level composite quality score that would apply to episode payment for both the CJR and SHFFT models, consistent with our proposal of the same measures for the two models. We believe that due to the inclusion of beneficiaries with hip fracture in both the CJR and SHFFT models and our desire to streamline EPM participant measure reporting, as well as the focus of both models on major lower extremity orthopedic surgery, the same set of quality measures can be used for both models to incentivize quality improvement in lower extremity orthopedic surgery care and episode efficiency. We are also considering future measure development focused specifically on hip and femur fracture patients. We expect that many of the physicians and other providers collaborating with participant hospitals in the SHFFT and CJR models will be the same, such that certain care pathways and episode efficiencies may be coordinated for SHFFT and CJR model beneficiaries regardless of the model, potentially resulting in quality improvement for beneficiaries in both models. We propose the following measures for the SHFFT model:Start Printed Page 50881

  • Hospital-level RSCR following elective primary THA and/or TKA (NQF #1550) (Hip/Knee Complications).
  • HCAHPS Survey (NQF #0166).
  • Total Hip Arthroplasty (THA)/Total Knee Arthroplasty (TKA) voluntary patient-reported outcome (PRO) and limited risk variable data submission (Patient-reported outcomes and limited risk variable data following elective primary THA/TKA).

We considered an alternative approach to the required quality measures for the SHFFT model given that the proposed measures do not specifically target the SHFFT model beneficiaries. This alternative approach would not account for any hip-specific measures (such as, Hospital-level RSCR following elective primary THA and/or TKA (NQF #1550) (Hip/Knee Complications)) and would instead only measure patient experience through the HCAHPS Survey (NQF #0166). Although there may be some rationale for excluding measures that do not specifically target SHFFT model beneficiaries, we do not propose this approach to SHFFT model quality measures because we believe that it is critical to include a measure of both clinical and patient experience outcomes in the setting of lower extremity orthopedic surgery episodes. Additionally, we believe that using quality measures for SHFFT model episodes that do not align with those in the CJR model could generate confusion at CJR model participant hospitals where we propose that the SHFFT model be tested as discussed in section III.B.4. of this proposed rule.

We refer to sections III.E.4.c. and d. of this proposed rule for a detailed discussion of our proposals regarding these measures for the SHFFT model, including their importance as measures of the quality-of-care for beneficiaries undergoing major lower extremity joint replacement surgery.

The proposals for the SHFFT model measures are included in § 512.413, and the proposals for reporting the measures are included in § 512.400. We seek comment on our proposals for SHFFT model quality measures.

3. Proposed Use of Quality Measures in the EPM Payment Methodologies

a. Overview of EPM Composite Quality Score Methodology

We believe that the proposed EPMs provide another mechanism for hospitals to improve quality of care, while also achieving cost efficiency. Incentivizing high-value care through episode payments for AMI, CABG, and hip fracture care is a primary objective of these proposed EPMs. Therefore, incorporating quality performance into the episode payment structure is an essential component of the proposed EPMs, just as it is for the CJR model (80 FR 73370). For the reasons stated previously, we believe it is important for the AMI, CABG, and SHFFT models to link the financial reward opportunity with performance and improvement in the quality of care for Medicare beneficiaries treated for AMI, CABG, and hip fracture.

As discussed in section III.D.4.a. of this proposed rule, which outlines the pricing methodologies for EPM episodes, for each EPM participant we propose to set an EPM-episode benchmark price for each EPM episode. We would apply the EPM participant's effective discount factor based on the participant's quality performance and improvement for the EPM performance year to the EPM-episode benchmark episode price to calculate the quality-adjusted target price for each EPM episode. We refer to section III.E.3.f. of this proposed rule for further discussion of the relationship between an EPM participant's quality performance and improvement and the effective discount factor. Each EPM episode includes an anchor hospitalization for either AMI (AMI MS-DRG or PCI MS-DRG with AMI ICD-10-CM diagnosis code in the principal or secondary diagnosis code position), CABG (CABG MS-DRG), or SHFFT (SHFFT MS-DRG) and a 90-day period after discharge from the anchor or chained anchor hospitalization. As discussed in section III.C.4.a.(5) of this proposed rule, a chained anchor hospitalization is an anchor hospitalization that initiates an AMI model episode and has at least one subsequent inpatient-to-inpatient transfer. An EPM quality-adjusted target price would represent expected spending on all related Part A and Part B items and services furnished during EPM episodes based on historical EPM episodes, and would incorporate the EPM participant's effective discount factor for the EPM performance year. Participants that achieve actual EPM-episode payments below the quality-adjusted target price for a given performance year may be eligible for a reconciliation payment from CMS, subject to the proposed stop-gain limit policy as discussed in section III.D.7.b. of this proposed rule. Participants that achieve actual EPM-episode payments that exceed the quality-adjusted target price for a given performance year may be required to repay Medicare a portion or all of the excess EPM-episode spending.

We propose an EPM composite quality score methodology for linking quality and payment in the EPMs that is similar to that methodology finalized for the CJR model (80 FR 73363 to 73381). Similar to the CJR model, the EPM-specific composite quality score methodology would allow both performance and improvement on each EPM's required quality measure to be meaningfully valued in the EPMs' pay-for-performance methodology, incentivizing and rewarding cost savings in relation to the quality of episode care provided by the EPM participant (80 FR 73374 and 73370). Specifically, the EPM composite quality score is made up of the composite performance score (which includes both patient experience and outcome measures, including points for voluntarily reported measures) and an improvement score.

We believe the actual level of quality performance achieved should be most highly valued in the EPM composite quality score to reward those EPM participants furnishing high quality care to EPM beneficiaries, with a smaller contribution to the EPM composite quality score made by improvement points if measure result improvement is achieved. We acknowledge that substantial improvement on a quality measure result is not the sole indicator that an EPM episode-of-care is high quality; yet, the improvement spurred by the hospital's participation in the EPM deserves to be valued as the EPM participant's performance is moving in a direction that is good for the health of beneficiaries. Like the CJR model, the EPMs involve a wide range of participants that must participate if they are located in the selected MSAs, and the participants would be starting from many different current levels of quality performance. We note that the Shared Savings Program utilizes a similar scoring and weighting methodology, which is described in detail in the CY 2011 Shared Savings Program Final Rule (see § 425.502). The HVBP Program and the HACRP also utilize a similar scoring methodology, which applies weights to various measures and assigns an overall score to a hospital (79 FR 50049 and 50102). Despite the small number of quality measures proposed for the EPMs, the measures represent both clinical outcomes and patient experience, and each carries substantial value in the EPM composite quality score.

Although performance and improvement on each measure would be valued in the EPM composite quality score methodology, it is the EPM participant's overall quality Start Printed Page 50882performance under the EPM that would be considered in the pay-for-performance approach, rather than performance on each quality measure individually determining the financial opportunity under the EPM. The EPM composite score methodology also provides a framework for incorporating additional measures of meaningful outcomes for EPM episodes in the future. Finally, while we believe that high performance on all of the quality measures represents goals of clinical care that should be achievable by all EPM participants that heighten their focus on these measures, we appreciate that many participants have room for significant improvement in their current measure performance. The EPM composite score methodology would provide the potential for financial reward for more EPM participants that reach overall acceptable or better quality performance, thus incentivizing their continued efforts to improve the quality and efficiency of EPM episodes.

We seek comment on our proposal to use an EPM-specific composite quality score in the pay-for-performance methodologies of the AMI, CABG, and SHFFT models.

b. Determining Quality Measure Performance

Similar to our reasoning in the CJR model, we believe that relative measure performance for the EPM measures would be the most appropriate way to incorporate quality performance into the EPMs because we do not have sufficient information about participant performance to set and use an absolute performance result on each measure (80 FR 73371). Moreover, we believe that participants nationally are currently working to improve their performance on the quality measures proposed for the EPMs on an ongoing basis as these are included in other CMS programs such as the HIQR and HVBP Programs. Therefore, while we expect that EPM participants would have a heightened focus on performance on these measures as a result of the financial incentives resulting from the EPM payment methodology, we are not yet certain what performance outcomes can be achieved under best practices.

Thus, at the time of reconciliation for an EPM performance year, we propose to assign each EPM participant's measure point estimate from the most recent year as discussed in section III.E.5. of this proposed rule to a performance percentile based on the national distribution of measure results for subsection (d) hospitals that are eligible for payment under the IPPS reporting the measure that meet the minimum patient case or survey count. This proposal applies to the MORT-30-AMI (NQF #0230) and AMI Excess Days measure results for the AMI model; the MORT-30-CABG (NQF #2558) measure result for the CABG model; the Hip/Knee Complications (NQF #1550) measure result for the SHFFT model; and the HCAHPS Survey (NQF #0166) measure result for all of the EPMs.

The measure-specific parameters that would apply to developing the national distributions are displayed in Table 13.

Table 13—Requirements for Use of Subsection (d) Hospitals That Are Eligible for Payment Under the IPPS Measure Results in Developing National Distribution of Required Measures for EPMS

MeasureRequirements for use in national distribution
MORT-30-AMI (NQF #0230)At least 25 patient cases in the 3-year measure performance period.
AMI Excess DaysAt least 25 patient cases in the 3-year measure performance period.
MORT-30-CABG (NQF #2558)At least 25 patient cases in the 3-year measure performance period.
Hip/Knee Complications (NQF #1550)At least 25 patient cases in the 3-year measure performance period.
HCAHPS Survey (#0166)At least 100 completed surveys in the 4-quarter reporting period.

We would assign any low volume EPM participant without a reportable value for the measure, new hospitals that are identified as EPM participants, or EPM participants where CMS has suppressed the measure value due to an error in the data used to calculate the measure to the 50th performance percentile of the measure result, so as not to disadvantage an EPM participant based on its low volume or lack of applicable cases because that participant may in actuality provide high quality care. We believe that relative measures of quality performance are most appropriate for the EPMs as participants continue to make progress nationally on improving patient outcomes and experience. Proposed measure-specific assignment of points in the EPMs' composite quality scores based on relative quality measure performance are discussed in sections III.E.3.e.(1), (2), and (3) of this proposed rule.

We seek comment on our proposed overall approach to determining quality measure performance based on assigning the EPM participant's measure point estimate to a measure performance percentile based on the national distribution of measure results from subsection (d) hospitals eligible for payment under the IPPS.

c. Determining Quality Measure Improvement

Consistent with our reasoning for the CJR model, we believe it would be important in the EPMs to directly reward EPM participants for quality improvement, similar to the pay-for-performance policies under other programs such as the HVBP Program and the Shared Savings Program, in order to provide a significant incentive for quality improvement for EPM participants at all current levels of quality performance (70 FR 73379). For the CJR model, we adopted a refinement to the composite quality score methodology that would supplement the composite quality score's valuing of quality performance in the pay-for-performance methodology of the CJR model (80 FR 73379). As in the CJR model, we believe the heightened focus on EPM episode cost and quality performance by participants in the EPMs may lead to substantial year-over-year quality measure improvement over the EPM performance years. Nevertheless, we believe that the actual level of quality performance achieved in the EPMs should be most highly valued in the EPM composite quality score to reward those participants furnishing high-quality care to EPM beneficiaries, with a small contribution to the composite quality score made by improvement points if measure result improvement is achieved. Thus, we propose adding into the EPM-specific composite quality score up to 10 percent of the maximum value for each EPM quality measure to which improvement could apply (excluding the voluntary data submission measures) for those EPM participants that demonstrate substantial improvement from the prior year's measure performance on that measure (80 FR 73379 through 73380). The maximum EPM composite quality score would be capped at 20 points Start Printed Page 50883under this proposal. Proposed measure-specific assignment of points for improvement in the EPMs' composite quality scores are discussed in sections III.E.3.e.(1), (2), and (3).

For the AMI and CABG models, we propose to define measure improvement differently than in the CJR model, using an approach that is more similar to the methodologies of other CMS programs such as the HVBP Program. The CJR model defined measure improvement for model participants relative to a national performance distribution (80 FR 73380). In contrast, we propose to define measure improvement as any improvement in an AMI or CABG model participant's own measure point estimate from the previous year, regardless of the participant's measure point estimate starting and ending values, if the AMI or CABG model participant falls into the top 10 percent of participants based on the national distribution of measure improvement over the 2 years for subsection (d) hospitals that are eligible for payment under the IPPS reporting the measure that meet the minimum patient case or survey count. We propose this approach because it represents the greatest confidence that we are capturing meaningful improvement on a measure by an AMI or CABG model participant in comparison with performance changes of other hospitals yet, unlike the CJR and proposed SHFFT model methodologies, is founded on an AMI or CABG model participant's own measure performance change from year-to-year. We believe that moving toward incorporating a model participant's own measure performance improvement in the pay-for-performance methodologies for EPMs strengthens the incentives in the models for quality improvement, especially for EPM participants at the lower end of current measure performance.

For the SHFFT model, we propose to modify the definition of improvement used in the CJR model in two ways (80 FR 73379 through 73380). First, we propose to define measure improvement as improving 2 deciles or more in comparison to the national distribution of measure results from the prior year, based on a comparison of relative quality measure performance over the most recent 2 years of available quality measure result data. This is the same methodology as finalized for the CJR model, except that it reduces the threshold for improvement from 3 deciles to 2 deciles in order to reward a broader range of improvement. Second, we propose to award up to 10 percent of the maximum measure performance score on the outcome and patient experience measures described in III.E.3.e.(3) of this proposed rule, with a cap of the SHFFT model composite quality score at 20 points. This alters the CJR model methodology, which calculates the measure performance score, voluntary reporting points, and measure improvement score separately for a total potential maximum score of 22. Taken together, these two changes bring calculation of the SHFFT model composite quality score into greater alignment with existing CMS programs, such as the HVBP Program, by expanding the number of SHFFT model participants eligible for quality improvement points but reducing the number of participants who receive both the highest quality performance score on a measure and points for measure improvement simultaneously.

In section V.E. of this proposed rule, we propose changes to the CJR model composite quality score calculation consistent with the SHFFT model methodology described here, allowing use of the same definition of quality improvement for the SHFFT and CJR models, because these models would be tested in mostly the same hospitals. We believe this approach would provide SHFFT model participants at all current levels of quality performance, including those historically lagging, with significant incentives to achieve improvement quality of care under the SHFFT model. Using a common approach to measuring quality improvement for the SHFFT and CJR models would provide a single participant-level composite quality score that can be applied at reconciliation for each model to determine the payment policies that would apply to the participant for the CJR and SHFFT model episodes, taking into consideration the different model performance years.

The proposals to determine quality measure improvement for the AMI, CABG, and SHFFT models are included in § 512.315(b)(3), (c)(3), and (d)(3), respectively. We seek comment on our proposals to determine quality measure improvement for the AMI, CABG, and SHFFT models.

d. Determining Successful Submission of Voluntary Data for AMI and SHFFT Models (1) Hybrid AMI Mortality (NQF #2473) Voluntary Data

Similar to the CJR model, we propose that AMI model participants that successfully submit the Hybrid AMI Mortality (NQF #2473) measure voluntary data would be eligible for points in the AMI model composite quality score (80 FR 73375, 73381). Encouraging collection and submission of the Hybrid AMI Mortality (NQF #2473) measure voluntary data through the AMI model would increase hospital familiarity with submitting hybrid quality measures based on claims data and data submitted from electronic health records; further develop an outcome measure that provides meaningful information on outcomes for AMI hospitalizations that are commonly experienced by Medicare beneficiaries; provide another quality measure that may be incorporated into the AMI model pay-for-performance methodology in future years, pending successful implementation testing of the measure; and inform the quality strategy of future payment models.

The proposed requirements for determining successful submission of Hybrid AMI Mortality (NQF #2473) measure voluntary data are included in § 512.411(b)(2) and discussed in detail in section III.E.4.a.(3)(vii) of this proposed rule. We seek comment on our proposals for determining successful submission of voluntary data for each AMI model performance year.

(2) Patient-Reported Outcomes and Limited Risk Variable Voluntary Data Following Elective Primary THA/TKA

Like the CJR model, we propose that SHFFT model participants that successfully submit Patient-reported outcomes and limited risk variable voluntary data following elective primary THA/TKA be eligible for points in the SHFFT model composite quality score (80 FR 73375, 73381). We note that SHFFT model participants that are also participating in the CJR model would not need to submit data twice to satisfy the successful submission requirements of both models. If those hospitals successfully submit voluntary data for the CJR model they would be credited with successful submission under the SHFFT model.

The proposed requirements for determining successful submission of Patient-reported outcomes and limited risk variable voluntary data following elective primary THA/TKA are included in § 512.13(b)(2) and discussed in detail in section III.E.4.c.(2)(viii) of this proposed rule. We seek comment on our proposals for determining successful submission of voluntary data for each SHFFT model performance year.

e. Calculation of the EPM-Specific Composite Quality Score

(1) AMI Model Composite Quality Score

We propose to assign each participant an AMI model composite quality score, calculated as the sum of the individual quality measure performance scores Start Printed Page 50884(including successful submission of Hybrid AMI Mortality (NQF #2473) measure voluntary data if applicable) and improvement scores. The quality measure performance scores would be set to reflect the intended weights for each of the quality measures and the successful submission of the Hybrid AMI Mortality (NQF #2473) voluntary data in the AMI model composite quality score. Each quality measure performance would be assigned a weight in the AMI model composite quality score, and possible scores for the measures would be set to reflect those weights. We would weight AMI model participant performance on each of the three required measures and successful submission of Hybrid AMI Mortality (NQF #2473) voluntary data according to the measure weights displayed in Table 14.

Table 14—Measures and Associated Performance Weights in AMI Model Composite Quality Score

Quality measureWeight in composite quality scoreQuality domain/weight
MORT-30-AMI (NQF #0230)50%Outcome/80%.
AMI Excess Days20%
Hybrid AMI Mortality (NQF #2473) Voluntary Data10%
HCAHPS Survey (NQF #0166)20%Patient Experience/20%.

We would assign the lowest weight of 10 percent to the submission of Hybrid AMI Mortality (NQF #2473) measure voluntary data because these data represent an AMI model participant's meaningful participation in advancing the quality measurement of AMI outcomes in keeping with our goal to move toward the use of electronic health records (EHRs) for measures, and in response to stakeholder feedback to include clinical data in outcome measures. Given the importance of AMI mortality as an extremely serious AMI outcome, we propose to assign the highest individual measure weight of 50 percent to the MORT-30-AMI (NQF #0230) measure. We propose to assign another 20 percent of the weight to the AMI Excess Days measure that is also included in the outcome quality domain. The remaining 20 percent of the AMI model composite quality score weight would be assigned to the HCAHPS Survey (NQF #0166) measure because we believe that incorporating this quality measure, which reflects performance regarding patients' perspectives on care, including communication, care transitions, and discharge information, is a meaningful patient experience measure of AMI model episode quality. This proposal of weights for the outcome and patient experience quality domains for the AMI model composite quality score is similar to the proposal of weights for the CABG model composite quality score described later in this section. We would assign the highest overall weight to the outcome quality domain (consisting of two measures and voluntary data submission) because the measures in this quality domain are specific to meaningful outcomes for AMI model beneficiaries. We do not propose to assign the HCAHPS survey (NQF #0166) measure the highest weight of the quality and patient experience domains, as the measure is not specific to AMI model episodes, but rather to all clinical conditions treated by AMI model participants. Unlike the CJR model where the quality measure weights in the CJR model composite quality score relatively evenly balance the outcome and patient experience quality domains, we would assign the highest weight in the AMI model to the outcome quality domain (consisting of two measures and voluntary data submission) because the measures in this quality domain are specific to meaningful, serious outcomes for AMI model beneficiaries, especially mortality which is not an outcome measure used in the CJR model composite quality score (80 FR 73375).

Under such an approach, we would first score individually each AMI model participant on the MORT-30-AMI (NQF #0230) measure; AMI Excess Days measure; and HCAHPS Survey (NQF #0166) measure based on the AMI model participant's performance percentile as compared to the national distribution of subsection (d) hospitals that are eligible for payment under the IPPS measure performance, assigning scores according to the point values displayed in Table 15. These individual measure scores have been set to reflect the measure weights included in Table 14 so they can ultimately be summed without adjustment in calculating the AMI model composite quality score. We note that in a chained anchor hospitalization where we propose in section III.C.4.a.(5) of this proposed rule that once an AMI model episode is initiated at a participant hospital, the AMI model episode would continue under the responsibility of that participant hospital, the transfer hospital's quality measure performance would not be included in assessing the AMI model participant's measure performance for the AMI model composite quality score. However, because the MORT-30-AMI (NQF #0230) measure attributes deaths to the initial hospital that admitted the beneficiary as an inpatient for AMI treatment in a transfer scenario, AMI model beneficiaries who die following treatment at a transfer hospital would be included in the AMI model participant's measure result and, therefore, their care represented in this quality measure.

Table 15—Individual Measure Performance Scoring for Three Required AMI Quality Measures

Performance percentileMORT-30-AMI (points)AMI excess days (points)HCAHPS survey (points)
≥90th10.004.004.00
≥80th and <90th9.253.703.70
≥70th and <80th8.503.403.40
≥60th and <70th7.753.103.10
≥50th and <60th7.002.802.80
≥40th and <50th6.252.502.50
Start Printed Page 50885
≥30th and <40th5.502.202.20
<30th0.000.000.00

Given the current national distribution of subsection (d) hospitals eligible for payment under the IPPS performance on these measures, we believe that small point increments related to higher measure performance deciles would be the most appropriate way to assign more points to reflect meaningfully higher quality performance on the measures. The absolute differences for each decile among the three measures reflect the intended weight of the measure in the AMI model composite quality score. These three measures are well-established measures in use under CMS hospital programs, so we do not believe that scores below the 30th percentile reflect quality performance such that they should be assigned any individual quality measure score points under the AMI model.

Additionally, we would assign a measure quality score of 2 points for AMI model participants that successfully submit Hybrid AMI Mortality (NQF #2473) measure voluntary data and 0 points for participants that do not successfully submit these data. Because we would not use the actual Hybrid AMI Mortality (NQF #2473) measure result as an outcome measure in assessing AMI episode quality performance under the AMI model, we propose this straightforward binary approach to scoring the submission of Hybrid AMI Mortality (NQF #2473) measure voluntary data for hybrid outcome measure testing.

CMS may, in future regulations, require hospitals to report additional data elements from EHRs and propose additional hybrid measures in this and other models and programs, such as the HIQR Program. If, in future regulations, hospitals are required to report these same five data elements (age; heart rate; systolic blood pressure; troponin, creatinine) and six linking variables (CMS Certification Number (CCN), Medicare Health Insurance Claim (HIC) Number, date of birth, sex, admission date, and discharge date) that are included in the Hybrid AMI Mortality (NQF #2473) measure to support measurement through another CMS program, such as the HIQR Program, CMS may propose changes to the AMI model measures and the methodology for assigning the AMI model composite quality score.

Finally, we would award improvement scores on a measure-by-measure basis to those AMI model participants that demonstrate improvement on the measure; improvement points would be awarded for up to 10 percent of the maximum measure performance points available, with the total AMI model composite quality score capped at 20. Thus, improvement scores would be up to 1.0 points for the MORT-30-AMI (NQF #0230) measure; up to 0.4 points for the AMI Excess Days measure; and up to 0.4 points for the HCAHPS Survey (NQF #0166) measure.

We would sum the performance and improvement scores on the three quality measures and the score on successful submission of Hybrid AMI Mortality (NQF #2473) measure voluntary data to calculate an AMI composite quality score for each AMI model participant.

The proposal for the methodology to calculate the AMI model composite quality score is included in § 512.315(b)(1)-(4). We seek comment on our proposed methodology to calculate the AMI model composite quality score.

(2) CABG Model Composite Quality Score

We propose to assign each participant a CABG model composite quality score, calculated as the sum of the individual quality measure performance and improvement scores. The quality measure performance scores would be set to reflect the intended weights for each of the quality measures. Each quality measure performance would be assigned a weight in the CABG model composite quality score and possible scores for the measures would be set to reflect those weights. We would weight CABG model participant performance on each of the two required measures according to the measure weights displayed in Table 16.

TABLE 16—Measures and Associated Performance Weights in CABG Model Composite Quality Score

Quality measureWeight in composite quality scoreQuality domain/weight
MORT-30-CABG (NQF #2558)75%Outcome/75%.
HCAHPS Survey (NQF #0166)25%Patient Experience/25%.

We propose to assign 75 percent of the weight in the CABG model composite quality score to the outcome quality domain, assigning all weight to the MORT-30-CABG (NQF #2558) measure, and the remaining 25 percent of the CABG model composite quality score weight to the HCAHPS Survey (NQF #0166) measure representing the patient experience quality domain. This proposal of weights for the outcome and patient experience quality domains for the CABG model composite quality score is similar to the proposal of weights for the AMI model composite quality score described previously in this section. CABG mortality is an extremely serious outcome and, like our proposal for the Mort-30-AMI (NQF #230) measure in the AMI model composite quality score, we propose that the MORT-30-CABG (NQF #2558) measure would have the highest individual measure weight in the CABG model composite quality score. We would assign 25 percent of the weight to the HCAHPS survey measure (NQF #0166) because we believe that incorporating this quality measure, which reflects performance regarding patients' perspectives on care, including communication, care transitions, and discharge information, is a meaningful Start Printed Page 50886patient experience measure of CABG model episode quality. We would assign the highest overall weight to the outcome quality domain (consisting of one measure) because it is specific to meaningful outcomes for CABG surgery for CABG model beneficiaries. We do not propose to assign the HCAHPS survey (NQF #0166) measure the highest weight of the quality and patient experience quality domains, as the measure is not specific to CABG model episodes, but rather to all clinical conditions treated by CABG model participants. Unlike the CJR model where the measure weights in the CJR model composite quality score relatively evenly balance the outcome and patient experience quality domains, CABG mortality representing the outcome quality domain is a serious outcome specific to CABG model beneficiaries such that we believe it deserves a high weight in the proposed CABG model composite quality score (80 FR 73375).

Under such an approach, we would first score individually each CABG model participant on the MORT-30-CABG (NQF #2558) measure; and HCAHPS Survey (NQF #0166) measure based on the participant's performance percentile as compared to the national distribution of subsection (d) hospitals that are eligible for payment under the IPPS measure performance, assigning scores according to the point values displayed in Table 17. These individual measure scores have been set to reflect the measure weights included in Table 16 so they can ultimately be summed without adjustment in calculating the CABG model composite quality score.

Table 17—Individual Scoring for Two Required CABG Quality Measures

Performance percentileMORT-30-CABG (points)HCAHPS survey (points)
≥90th15.005.00
≥80th and <90th13.884.63
≥70th and <80th12.754.25
≥60th and <70th11.633.88
≥50th and <60th10.503.50
≥40th and <50th9.383.13
≥30th and <40th8.252.75
<30th0.000.00

Given the current national distribution of subsection (d) hospitals that are eligible for payment under the IPPS performance on these measures, we believe that small point increments related to higher measure performance deciles would be the most appropriate way to assign more points to reflect meaningfully higher quality performance on the measures. The absolute differences for each decile among the two measures reflect the intended weight of the measure in the CABG model composite quality score. These two measures are well-established measures in use under CMS hospital programs, so we do not believe that scores below the 30th percentile reflect quality performance such that they should be assigned any individual quality measure score points under the CABG model.

Finally, we would award improvement scores on a measure-by-measure basis to those CABG model participants that demonstrate improvement on the measure; improvement points would be awarded for up to 10 percent of the maximum measure performance points available, with the total CABG model composite quality score capped at 20. Thus, improvement scores would be up to 1.5 points for the MORT-30-CABG (NQF #2558) measure; and up to 0.5 points for the HCAHPS Survey (NQF #0166) measure.

We would sum the performance and improvement scores on the two quality measures to calculate a CABG model composite quality score for each CABG model participant.

The proposal for the methodology to calculate the CABG model composite quality score is included in § 512.315(c)(1) through (4). We seek comment on our proposed methodology to calculate the CABG model composite quality score.

(3) SHFFT Model Composite Quality Score

We propose to adopt the same calculation of the SHFFT model composite quality score as the CJR model, including the proposed changes to the CJR model composite quality score methodology described in section V.E. of this proposed rule. For those participants in both SHFFT and CJR models, the SHFFT model composite quality score calculated each year would be the same as the CJR model composite quality score (80 73370 through 73381). We propose to assign each SHFFT model participant a SHFFT model composite quality score, capped at 20 points and calculated as the sum of the individual quality measure and improvement scores as well as successful submission of THA/TKA voluntary PRO and limited risk variable data if applicable. The quality measure performance scores would be set to reflect the intended weights for each of the quality measures. Each quality measure performance would be assigned a weight in the SHFFT model composite quality score and possible scores for the measures would be set to reflect those weights. We would weight SHFFT model participant performance on each of the two required measures and successful submission of THA/TKA voluntary PRO and limited risk variable data according to the measure weights displayed in Table 30.

Table 18—Measures and Associated Performance Weights in SHFFT Model Composite Quality Score

Quality measureWeight in composite quality scoreQuality domain/weight
Hip/Knee Complications (NQF #155050%Outcome/50%.
THA/TKA voluntary PRO and limited risk variable submission10%Patient Experience/50%.
HCAHPS Survey (NQF #0166)40%
Start Printed Page 50887

Consistent with the CJR model, we propose to assign 50 percent of the weight in the SHFFT model composite quality score to the outcome quality domain, assigning 50 percent of the weight to the Hip/Knee Complications (NQF #1550) measure. We propose to assign 50 percent of the weight to the patient experience quality domain, specifically 10 percent of the weight in that quality domain to the THA/TKA voluntary PRO and limited risk variable submission. We would assign 40 percent of the weight to the HCAHPS survey measure (NQF #0166) representing the patient experience (80 FR 73375). We would assign 40 percent to the HCAHPS survey measure (NQF #0166) because we believe that incorporating this quality measure, which reflects performance regarding patients' perspectives on care, including communication, care transitions, and discharge information, is a highly meaningful outcome measure of SHFFT episode quality under the SHFFT model, and because doing so ensures that there is a consistent methodology for linking quality performance and improvement to payment for SHFFT model participants that are also participating in the CJR model. As in the CJR model, we believe this weighting appropriately balances patient experience with meaningful health outcomes for beneficiaries (80 FR 73375).

Under such an approach, we would first score individually each SHFFT model participant on the Hip/Knee Complications (NQF #1550) measure; and HCAHPS Survey (NQF #0166) measure based on the participant's performance percentile as compared to the national distribution of subsection (d) hospitals that are eligible for payment under the IPPS measure performance, assigning scores according to the point values displayed in Table 19. These individual measure scores have been set to reflect the measure weights included in Table D6 so they can ultimately be summed without adjustment in calculating the SHFFT model composite quality score. We note that the point score for each decile for the two measures for the SHFFT model is the same as that used for other CJR model.

Table 19—Individual Scoring for Two Required SHFFT Quality Measures

Performance percentileHip/knee complications (points)HCAHPS survey quality score (points)
≥90th10.008.00
≥80th and <90th9.257.40
≥70th and <80th8.506.80
≥60th and <70th7.756.20
≥50th and <60th7.005.60
≥40th and <50th6.255.00
≥30th and <40th5.504.40
<30th0.000.00

Given the current national distribution of subsection (d) hospitals that are eligible for payment under the IPPS performance on these measures, we believe that small point increments related to higher measure performance deciles would be the most appropriate way to assign more points to reflect meaningfully higher quality performance on the measures. The absolute differences for each decile among the three measures reflect the intended weight of the measure in the SHFFT model composite quality score. These two measures are well-established measures in use under CMS hospital programs, so we do not believe that scores below the 30th percentile reflect quality performance such that they should be assigned any individual quality measure score points under the SHFFT model.

As in the CJR model, we propose to assign a measure quality score of 2 points for SHFFT model participants that successfully submit THA/TKA voluntary PRO and limited risk variable data and 0 points for participants that do not successfully submit these data (80 FR 73376).

Finally, we would award improvement scores on a measure-by-measure basis to those SHFFT model participants that demonstrate improvement on the measure (defined as year-over-year improvement of 2 or more deciles in the performance distribution); improvement points would be awarded for up to 10 percent of the maximum measure performance points available, with the total SHFFT model composite quality score capped at 20. Thus, improvement scores would be up to 1.0 points for the Hip/Knee Complications (NQF #1550) measure; and up to 0.8 points for the HCAHPS Survey (NQF #0166) measure.

We would sum the performance and improvement scores on the two required quality measures and the score on successful submission of THA/TKA voluntary PRO and limited risk variable data to calculate a SHFFT model composite quality score for each SHFFT model participant. For those CJR model participants (the majority of SHFFT model participants), the SHFFT model composite quality score would be the same as the participant's score for the CJR model.

The proposal for the methodology to calculate the SHFFT model composite quality score is included in § 512.315(d)(1) through (4). We seek comment on our proposed methodology to calculate the SHFFT model composite quality score.

f. EPM Pay-for-Performance Methodologies To Link Quality and Payment

(1) Overview of Pay-for-Performance Proposals Applicable to the EPMs

As in the CJR model, we propose that the maximum effective discount factor for all EPM participants that could be incorporated in quality-adjusted target prices would be 3.0 percent (80 FR 733760). We refer to section III.D.4.b.(10) of this proposed rule for further discussion of the application of the effective discount factor to EPM-episode benchmark prices in calculating quality-adjusted target prices. EPM participants that provide high-quality episode care would have the opportunity to reduce the effective discount factor used to calculate their quality-adjusted prices at reconciliation. The effective discount factors are displayed in tables in the following EPM-specific sections, based on the EPM-specific composite quality score that would place each EPM participant into one of four quality categories, specifically “Below Acceptable,” “Acceptable,” “Good,” and “Excellent,” for each EPM performance year. Three tables are required to display the proposed effective discount factor and applicable discount factor (the discount Start Printed Page 50888factor that represents the phase-in of repayment responsibility in performance years 2 (DR) and 3) for each quality category due to the phase-in of EPM participant repayment responsibility from no responsibility in performance year 1 and performance year 2 (NDR), to partial responsibility in performance years 2 (DR) and 3, and finally full responsibility in performance years 4 and 5 as discussed in section III.D.2.c. Note that the applicable discount factor only applies to EPM performance years 2 (DR) and 3.

(2) AMI and CABG Model Pay-for-Performance Methodologies

(a) AMI Model Pay-for-Performance Methodology

We propose to incorporate the AMI model composite quality score in the AMI model payment methodology by (1) requiring a minimum AMI model composite quality score for reconciliation payment eligibility if the AMI model participant's actual episode payments are less than the quality-adjusted target price and (2) determining the effective discount factor included in the quality-adjusted target price experienced by the AMI model participant in the reconciliation process. The payment policies we would apply are displayed in Tables 20, 21, and 22 for the performance years of the AMI model. Under the AMI model as proposed, there is no AMI model participant repayment responsibility in performance year 1 and performance year 2 (NDR) and this responsibility begins to be phased-in in performance year 2 (DR), with full implementation in performance year 4. Because repayment responsibility is phased-in, in performance years 2 (DR) and 3, repayment responsibility only applies to a portion of the amount of excess AMI model episode spending that results from the quality-adjusted target prices that include the AMI model participant's effective discount factor. We, therefore, refer in the repayment column to the applicable discount factor for repayment amount in performance years 2 (DR) and 3. The effective discount factor applies to both the reconciliation payment and the repayment amount in performance years 4 and 5. We note that the average Medicare payment for historical AMI episodes beginning in CYs 2012 to 2014 was $24,200.[70]

Table 20—Performance Year 1 and Performance Year 2 (NDR): Relationship of AMI Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

AMI model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation payment %Effective discount factor for repayment amount
<3.6No3.0Not applicable,
>=3.6 and <6.9Yes3.0Not applicable.
>=6.9 and <=14.8Yes2.0Not applicable.
>14.8Yes1.5Not applicable.

Table 21—Performance Years 2 (DR) and 3: Relationship of AMI Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

AMI model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation payment %Applicable discount factor for repayment amount* %
<3.6No3.02.0
>=3.6 and <6.9Yes3.02.0
>=6.9 and <=14.8Yes2.01.0
>14.8Yes1.50.5
* The applicable discount factor for the repayment amount only applies in performance years 2 (DR) and 3 when repayment responsibility is being phased-in.

Table 22—Performance Years 4 and 5: Relationship of AMI Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

AMI model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation paymentEffective discount factor for repayment amount
<3.6No3.03.0
>=3.6 and <6.9Yes3.03.0
>=6.9 and <=14.8Yes2.02.0
>14.8Yes1.51.5
Start Printed Page 50889

Under this approach, the maximum AMI model effective discount factor included in the quality-adjusted target price would be 3.0 percent, consistent with the CJR model (80 FR 73365). We believe that a maximum effective discount factor of 3.0 percent is reasonable as it is within the range of discount percentages included in the ACE demonstration and it is the Model 2 BPCI discount factor for 30- and 60-day episodes, where BPCI participants are testing AMI episodes subject to the 3.0 percent discount factor. AMI model participants that provide high quality episode care would have the opportunity for a lower effective discount factor to be included in their quality-adjusted target prices at reconciliation as displayed in Tables 20, 21, and 22.

Under this methodology, we would require AMI model participants to achieve a minimum AMI model composite quality score of >=3.6 to be eligible for a reconciliation payment if actual episode payments were less than the quality-adjusted target price based on the 3.0 percent maximum effective discount factor. Participants with below acceptable quality performance reflected in an AMI model composite quality score <3.6 would not be eligible for a reconciliation payment if actual AMI model episode payments were less than the quality-adjusted target price. A level of quality performance that is below acceptable would not affect AMI model participants' repayment responsibility if actual AMI model episode payments exceed the quality-adjusted target price. We believe that excessive reductions in utilization that lead to low actual AMI model episode payments and that could result from the financial incentives of an EPM would be limited by a requirement that this minimum level of AMI model episode quality be achieved for reconciliation payments to be made. This policy would encourage AMI model participants to focus on appropriate reductions or changes in utilization to achieve high quality care in a more efficient manner. Therefore, these participants would be ineligible to receive a reconciliation payment if actual AMI model episode payments were less than the quality-adjusted target price.

AMI model participants with an acceptable AMI model composite quality score of >=3.6 and <6.9 would be eligible for a reconciliation payment if actual AMI model episode payments were less than the quality-adjusted target price based on a 3.0 percent effective discount factor because their quality performance was at the acceptable level established for the AMI model. Therefore, these AMI model participants would be eligible to receive a reconciliation payment if actual AMI model episode payments were less than the quality-adjusted target price.

AMI model participants with a good AMI model composite quality score of >=6.9 and <=14.8 would be eligible for a reconciliation payment if actual AMI model episode payments were less than the quality-adjusted target price based on a 2.0 percent effective discount factor that reflects their good quality performance. Thus, participants achieving this level of quality for AMI episodes under the AMI model would either have less repayment responsibility (that is, the reduced effective discount factor would offset a portion of their repayment responsibility) or receive a higher reconciliation payment (that is, the reduced effective discount factor would increase the reconciliation payment) at reconciliation than they would have otherwise based on a comparison of actual AMI model episode payments to quality-adjusted target prices that include the maximum 3.0 percent effective discount factor.

Finally, AMI model participants with an excellent AMI model composite score quality score of >=14.8 would be eligible to receive a reconciliation payment if actual AMI model episode payments were less than the quality-adjusted target price based on a 1.5 percent effective discount factor that reflects their excellent performance. Thus, participants achieving this level of quality for AMI episodes under the AMI model would either have less repayment responsibility (that is, the reduced effective discount factor would offset a portion of their repayment responsibility) or receive a higher reconciliation payment (that is, the reduced effective discount factor would increase the reconciliation payment) at reconciliation than they would have otherwise based on a comparison of actual AMI model episode payments to quality-adjusted target prices that include the maximum 3.0 percent effective discount factor.

Under this methodology, the proposed stop-loss and stop-gain limits discussed in section III.D.7.b. of this proposed rule would not change. We believe this approach to quality incentive payments based on the AMI model composite quality score could have the effect of increasing the alignment of the financial and quality performance incentives under the AMI model to the potential benefit of AMI model participants and their collaborators as well as CMS, and would be consistent with the CJR model methodology linking quality and payment.

The proposal to link quality to payment in the AMI model pay-for-performance methodology is included in § 512.315(b)(5). We seek comment on our proposal to link quality to payment in the AMI model pay-for-performance methodology.

(b) CABG Model Pay-for-Performance Methodology

We propose to incorporate the CABG model composite quality score in the CABG model payment methodology by—(1) requiring a minimum CABG model composite quality score for reconciliation payment eligibility if the CABG model participant's actual episode payments are less than the quality-adjusted target price; and (2) determining the effective discount factor included in the quality-adjusted target price experienced by the CABG model participant in the reconciliation process. The payment policies we would apply are displayed in Tables 23, 24, and 25 for the performance years of the CABG model. Under the CABG model as proposed, there is no CABG model participant repayment responsibility in performance year 1 and performance year 2 (NDR) and this responsibility begins to be phased-in in performance year 2 (DR), with full implementation in performance year 4. Because repayment responsibility is phased-in, in performance years 2 (DR) and 3, repayment responsibility only applies to a portion of the amount of excess CABG model episode spending that results from the quality-adjusted target prices that include the CABG model participant's effective discount factor. We, therefore, refer in the repayment column to the applicable discount factor for repayment amount in performance years 2 (DR) and 3. The effective discount factor applies to both the reconciliation payment and the repayment amount in performance years 4 and 5. We note that the average Medicare payment for historical CABG episodes beginning in CYs 2012 to 2014 was $47,000.[71]

Start Printed Page 50890

Table 23—Performance Year 1 and Performance Year 2 (NDR): Relationship of CABG Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

CABG model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation payment %Effective discount factor for repayment amount %
<2.8No3.0Not applicable.
>=2.8 and <4.8Yes3.0Not applicable.
>=4.8 and <=17.5Yes2.0Not applicable.
>17.5Yes1.5Not applicable.

Table 24—Performance Years 2 (DR) and 3: Relationship of CABG Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

CABG model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation payment %Applicable discount factor for repayment amount * %
<2.8No3.02.0
>=2.8 and <4.8Yes3.02.0
>=4.8 and <=17.5Yes2.01.0
>17.5Yes1.50.5
* The applicable discount factor for the repayment amount only applies in performance years (DR) and 3 when repayment responsibility is being phased-in.

Table 25—Performance Years 4 and 5: Relationship of CABG Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

CABG model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation payment %Effective discount factor for repayment amount %
<2.8No3.03.0
>=2.8 and <4.8Yes3.03.0
>=4.8 and <=17.5Yes2.02.0
>17.5Yes1.51.5

Under this approach, the maximum CABG model effective discount factor included in the quality-adjusted target price would be 3.0 percent, consistent with the CJR model (80 FR 73365). We believe that a maximum effective discount factor of 3.0 percent is reasonable as it is within the range of discount percentages included in the Medicare Acute Care Episode (ACE) demonstration and it is the Model 2 BPCI discount factor for 30 and 60 day episodes, where BPCI participants are testing CABG episodes subject to the 3.0 percent discount factor. CABG model participants that provide high quality episode care would have the opportunity for a lower effective discount factor to be included in their quality-adjusted target prices at reconciliation as displayed in Tables 23, 24, and 25.

Under this methodology, we would require CABG model participants to achieve a minimum CABG model composite quality score of >=2.8 to be eligible for a reconciliation payment if actual episode payments were less than the quality-adjusted target price based on the 3.0 percent maximum effective discount factor. Participants with below acceptable quality performance reflected in an CABG model composite quality score <2.8 would not be eligible for a reconciliation payment if actual CABG model episode payments were less than the quality-adjusted target price. A level of quality performance that is below acceptable would not affect participants' repayment responsibility if actual CABG model episode payments exceed the quality-adjusted target price. We believe that excessive reductions in utilization that lead to low actual CABG model episode payments and that could result from the financial incentives of an EPM would be limited by a requirement that this minimum level of CABG model episode quality be achieved for reconciliation payments to be made. This policy would encourage CABG model participants to focus on appropriate reductions or changes in utilization to achieve high quality care in a more efficient manner. Therefore, these participants would be ineligible to receive a reconciliation payment if actual CABG model episode payments were less than the quality-adjusted target price.

CABG model participants with an acceptable CABG model composite quality score of >=2.8 and <4.8 would be eligible for a reconciliation payment if actual CABG model episode payments were less than the quality-adjusted target price based on a 3.0 percent effective discount factor because their quality performance was at the acceptable level established for the CABG model. Therefore, these CABG model participants would be eligible to Start Printed Page 50891receive a reconciliation payment if actual CABG model episode payments were less than the quality-adjusted target price.

CABG model participants with a good CABG model composite quality score >=4.8 and <=17.5 would be eligible for a reconciliation payment if actual CABG model episode payments were less than the quality-adjusted target price based on a 2.0 percent effective discount factor that reflects their good quality performance. Thus, participants achieving this level of quality for CABG episodes under the CABG model would either have less repayment responsibility (that is, the reduced effective discount factor would offset a portion of their repayment responsibility) or receive a higher reconciliation payment (that is, the reduced effective discount factor would increase the reconciliation payment) at reconciliation than they would have otherwise based on a comparison of actual CABG model episode payments to quality-adjusted target prices that include the maximum 3.0 percent effective discount factor.

Finally, CABG model participants with an excellent CABG model composite score quality score of >17.5 would be eligible to receive a reconciliation payment if actual CABG model episode payments were less than the quality-adjusted target price based on a 1.5 percent effective discount factor that reflects their excellent performance. Thus, participants achieving this level of quality for CABG model episodes would either have less repayment responsibility (that is, the reduced effective discount factor would offset a portion of their repayment responsibility) or receive a higher reconciliation payment (that is, the reduced effective discount factor would increase the reconciliation payment) at reconciliation than they would have otherwise based on a comparison of actual CABG model episode payments to quality-adjusted target prices that include the maximum 3.0 percent effective discount factor.

Under this methodology, the proposed stop-loss and stop-gain limits discussed in section III.D.7.b. of this proposed rule would not change. We believe this approach to quality incentive payments based on the CABG model composite quality score could have the effect of increasing the alignment of the financial and quality performance incentives under the CABG model to the potential benefit of CABG model participants and their collaborators as well as CMS, and would be consistent with the CJR model methodology linking quality and payment.

The proposal to link quality to payment in the CABG model pay-for-performance methodology is included in § 512.315(c)(5). We seek comment on our proposal to link quality to payment in the CABG model pay-for-performance methodology.

(c) Alignment Between the AMI and CABG Model Methodologies

The AMI and CABG models are closely related, given that they both are based on a significant event or procedure for a beneficiary with CAD. As discussed in sections III.D.2.b. and c. of this proposed rule, we propose the use of a 30-day mortality measure in both models, specifically MORT-30-AMI (NQF #0230) with a weight of 50 percent in the AMI model composite quality score and MORT-30-CABG (NQF #2558) with a weight of 75 percent in the CABG model quality score. The beneficiaries included in the measure have some overlap, because some beneficiaries with AMI will have a CABG during their hospitalization that begins an episode. Analysis of both the MORT-30-AMI (NQF #0230) and MORT-30-CABG (NQF #2558) measure national distributions suggests that improving from the 25th percentile to 75th percentile represents roughly a 1 percentage point decrease in mortality rates for both measures.

In addition, we note that for historical episodes beginning in 2012 to 2014, the average Medicare spending for an AMI episode that extends 90 days post-hospital discharge was approximately $24,200 and for a CABG episode was approximately $47,000.[72] However, because we propose the same 1.5 percent to 3.0 percent effective discount factor range based on quality performance and improvement for the AMI and CABG models (and, to a lesser degree, because of the modestly lower weight assigned to the mortality measure under the AMI model), the absolute dollar amounts tied to changes in AMI or CABG mortality rates are different in the two models. A larger absolute financial incentive is associated with improvement in CABG mortality than AMI mortality under our proposal. We recognize that mortality is a serious outcome for beneficiaries with CAD who have a significant event or procedure, and we considered setting a wider effective discount factor range based on quality in the AMI model than the CABG model to align the absolute financial incentives to improve mortality under both models. For example, to create a more similar absolute financial incentive between the lowest and highest effective discount percentages in the AMI and CABG models, we could set the effective discount factor range for the AMI model at 0.75 percent to 3.75 percent and the CABG model range at 1.5 percent to 3 percent. Alternatively, we could set the AMI model effective discount factor range at 1.5 percent to 3 percent and compress the CABG effective discount factor range. While we do not propose different effective discount factor ranges for the AMI and CABG models in order to retain consistency with the CJR model and the BPCI initiative, we seek comments about the potential benefits and drawbacks of establishing the same absolute dollar incentive for similar improvements in quality across different models that have similar measures but vary in average episode cost. This feedback will be useful as we consider future episode payment models and candidate quality measures for potential new and existing models, as well as consider future refinements to the pay-for-performance methodologies under the models. Our goal in all of our episode payment models is to create strong financial incentives for quality performance and improvement for participants at all level of current quality performance and to rationalize the strength of the financial incentives in the context of the specificity and importance of the quality measures used under the models.

(3) SHFFT Model Pay-for-Performance Methodology

We propose to incorporate the SHFFT model composite quality score in the SHFFT model payment methodology by (1) requiring a minimum SHFFT model composite quality score for reconciliation payment eligibility if the SHFFT model participant's actual episode payments are less than the quality-adjusted target price and (2) determining the effective discount factor included in the quality-adjusted target price experienced by the SHFFT model participant in the reconciliation process. The payment policies we would apply are displayed in Tables 26, 27, and 28 for the performance years of the SHFFT model. Under the SHFFT model as proposed, there is no SHFFT model participant repayment responsibility in performance year 1 and performance year 2 (NDR) and this responsibility begins to be phased-in in performance year 2 (DR), with full implementation in performance year 4. Because repayment Start Printed Page 50892responsibility is phased-in, in performance years 2 (DR) and 3, repayment responsibility only applies to a portion of the amount of excess SHFFT model episode spending that results from the quality-adjusted target prices that include the SHFFT model participant's effective discount factor. We, therefore, refer in the repayment column to the applicable discount factor for repayment amount in performance years 2 (DR) and 3. The effective discount factor applies to both the reconciliation payment and the repayment amount in performance years 4 and 5. We note that the average Medicare payment for historical SHFFT episodes beginning in CYs 2012 to 2014 was $43,000.[73]

We refer to section V.E. of this proposed rule for discussion of the correction to the composite quality score ranges for the four quality categories from what was presented in the CJR final rule (80 FR 73378). The SHFFT model composite quality score ranges displayed in Tables 26 through 28 are the corrected ranges that also apply to the CJR model.

Table 26—Performance Year 1 and Performance Year 2 (NDR): Relationship of SHFFT Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

SHFFT model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation payment %Effective discount factor for repayment amount
<5.0No3.0Not applicable.
>=5.0 and <6.9Yes3.0Not applicable.
>=6.9 and <=15.0Yes2.0Not applicable.
>15.0Yes1.5Not applicable.

Table 27—Performance Years 2 (DR) and 3: Relationship of SHFFT Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

SHFFT Model Composite Quality ScoreEligible for Reconciliation PaymentEffective Discount Factor for Reconciliation Payment %Applicable Discount Factor for Repayment Amount* %
<5.0No3.02.0
>=5.0 and <6.9Yes3.02.0
>=6.9 and <=15.0Yes2.01.0
>15.0Yes1.50.5
*The applicable discount factor for the repayment amount only applies in performance years 2 (DR) and 3 when repayment responsibility is being phased-in.

Table 28—Performance Years 4 and 5: Relationship of SHFFT Model Composite Quality Score to Reconciliation Payment Eligibility and the Effective Discount Factor Experienced at Reconciliation

SHFFT model composite quality scoreEligible for reconciliation paymentEffective discount factor for reconciliation payment %Effective discount factor for repayment amount %
<5.0No3.03.0
>=5.0 and <6.9Yes3.03.0
>=6.9 and <=15.0Yes2.02.0
>15.0Yes1.51.5

Under this methodology, we would require SHFFT model participants to achieve a minimum SHFFT model composite quality score of >=5.0 to be eligible for a reconciliation payment if actual episode payments were less than the quality-adjusted target price based on the 3.0 percent maximum effective discount factor. Participants with below acceptable quality performance reflected in a SHFFT model composite quality score <5.0 would not be eligible for a reconciliation payment if actual SHFFT model episode payments were less than the quality-adjusted target price. A level of quality performance that is below acceptable would not affect participants' repayment responsibility if actual SHFFT model episode payments exceed the quality-adjusted target price. We believe that excessive reductions in utilization that lead to low actual SHFFT model episode payments and that could result from the financial incentives of an EPM would be limited by a requirement that this minimum level of SHFFT model episode quality be achieved for reconciliation payments to be made. This policy would encourage SHFFT model participants to Start Printed Page 50893focus on appropriate reductions or changes in utilization to achieve high quality care in a more efficient manner. Therefore, these participants would be ineligible to receive a reconciliation payment if actual SHFFT model episode payments were less than the quality-adjusted target price.

SHFFT model participants with an acceptable SHFFT model composite quality score of >=5.0 and <6.9 would be eligible for a reconciliation payment if actual SHFFT model episode payments were less than the quality-adjusted target price based on a 3.0 percent effective discount factor because their quality performance was at the acceptable level established for the SHFFT model. Therefore, these SHFFT model participants would be eligible to receive a reconciliation payment if actual SHFFT model episode payments were less than the quality-adjusted target price.

SHFFT model participants with a good SHFFT model composite quality score of >=6.9 and <=15.0 would be eligible for a reconciliation payment if actual SHFFT model episode payments were less than the quality-adjusted target price based on a 2.0 percent effective discount factor that reflects their good quality performance. Thus, participants achieving this level of quality for SHFFT model episodes under the SHFFT model would either have less repayment responsibility (that is, the reduced effective discount factor would offset a portion of their repayment responsibility) or receive a higher reconciliation payment (that is, the reduced effective discount factor would increase the reconciliation payment) at reconciliation than they would have otherwise based on a comparison of actual SHFFT model episode payments to quality-adjusted target prices that include the maximum 3.0 percent effective discount factor.

Finally, SHFFT model participants with an excellent SHFFT model composite score quality score of >15.0 would be eligible to receive a reconciliation payment if actual SHFFT model episode spending was less than the quality-adjusted target price based on a 1.5 percent effective discount factor that reflects their excellent performance. Thus, participants achieving this level of quality for SHFFT model episodes would either have less repayment responsibility (that is, the reduced effective discount factor would offset a portion of their repayment responsibility) or receive a higher reconciliation payment (that is, the reduced effective discount factor would increase the reconciliation payment) at reconciliation than they would have otherwise based on a comparison of actual SHFFT model episode payments to quality-adjusted target prices that include the maximum 3.0 percent effective discount factor.

Under this methodology, the proposed stop-loss and stop-gain limits discussed in section III.D.7.b. of this proposed rule would not change. We believe this approach to quality incentive payments based on the SHFFT model composite quality score could have the effect of increasing the alignment of the financial and quality performance incentives under the SHFFT model to the potential benefit of SHFFT model participants and their collaborators as well as CMS, and would be consistent with the CJR model methodology linking quality and payment.

The proposal to link quality to payment in the SHFFT model pay-for-performance methodology is included in § 512.315(d)(5). We seek comment on our proposal to link quality to payment in the SHFFT model pay-for-performance methodology.

4. Details on Quality Measures for the EPMs

a. AMI Model-Specific Measures

(1) Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate Following Acute Myocardial Infarction (AMI) Hospitalization (NQF #0230) (MORT-30-AMI)

(a) Background

AMI is one of the most common principal hospital discharge diagnoses among older adults and is associated with high mortality. AMI was the tenth most common principal discharge diagnosis among patients with Medicare in 2012.[74] Each year, over 600,000 Americans will experience an AMI. Despite improvements in treatments, 30-day mortality rates following AMI exceed 7 percent. CMS pays approximately $11.7 billion annually for in-hospital costs for Medicare beneficiaries with coronary heart disease, of which AMI is a major contributor. The high prevalence and considerable morbidity and mortality associated with AMI create an economic burden on the healthcare system.[75]

Hospital mortality is an outcome that is likely attributable to care processes and is an important outcome for patients. Complex and critical aspects of care, such as communication between providers, prevention of and response to complications, patient safety, and coordinated transitions to the outpatient environment, all contribute to patient outcomes. Many current hospital interventions are known to decrease the risk of death within 30 days of hospital admission.[76 77] We believe it is important to assess the quality of care provided to Medicare beneficiaries who are hospitalized for AMI.

The measure developed by CMS, and currently implemented in the HIQR and HVBP Programs, assesses a hospital's risk-standardized mortality rate, which is the rate of death after admission to a hospital with a principal diagnosis of AMI. The measure outcome is the rate of mortality occurring after admission with a principal diagnosis of AMI for patients 65 and older during a 30-day period that begins with the date of the index admission for the specific hospital. An index admission is the hospitalization which is included in the measure cohort because it meets all inclusion criteria and does not meet any exclusion criteria. The index admission is the hospitalization to which the mortality outcome is attributed. The median hospital-level risk-standardized mortality rate for 2016 public reporting on Hospital Compare was 14.2 percent, with a interquartile range from 13.7 percent to 14.6 percent in hospitals. The variation in mortality rates suggests that important differences in the quality of care delivered across hospitals exist, and there is room for quality improvement.

We developed the measure of hospital-level risk-standardized mortality rate (RSMR) following AMI hospitalization, which was later endorsed by the NQF (NQF #0230). The measure has been publicly reported on Hospital Compare since FY 2007, and was incorporated into what is now the HIQR Program since FY 2008 (FY 2008 IPPS/LTCH final rule 71 FR 67960), and the HVBP Program since FY 2014 (FY 2011 IPPS/LTCH final rule 76 FR 26510).

(b) Data Sources

We propose to use Medicare Part A and Part B FFS claims submitted by the Start Printed Page 50894AMI model participant as the data source for calculation of the MORT-30-AMI (NQF #0230) measure. Index admission diagnoses and in-hospital comorbidities are assessed using Medicare Part A claims. Additional comorbidities prior to the index admission are assessed as Part A inpatient, outpatient, and Part B office visit Medicare claims in the 12 months prior to the index (initial) admission. Enrollment and post-discharge mortality status are obtained from Medicare's enrollment database which contains beneficiary demographics, benefits/coverage, and vital status information.

(c) Cohort

The MORT-30-AMI (NQF #0230) measure includes Medicare FFS beneficiaries, aged 65 years or older, discharged from non-federal acute care hospitals with a principal discharge diagnosis of AMI and with a complete claims history for the 12 months prior to admission. Eligible hospitalizations are defined using the following ICD-10-CM codes: I2109, I2119, I2111, I2119, I2129, I214, and I213.

We propose that the measure will include index admissions to all non-federal acute care hospitals, which includes all AMI model participants. Hospital performance will only be publically reported for hospitals with 25 or more index admissions in the 3-year measurement period. The AMI model cohort would differ from the hospital cohort that is currently captured in the measure through the HIQR Program. Although performance on the measure will not be publically reported for hospitals with fewer than 25 cases, they will receive information about their performance. We refer readers to section III.B.5. of this proposed rule for participant selection for the AMI model. For eligible hospitalizations defined using ICD-9-CM codes, we refer readers to the CMS Web site at: http://cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​HospitalQualityInits/​Measure-Methodology.html.

(d) Inclusion and Exclusion Criteria

We propose that an index admission is the hospitalization to which the mortality outcome is attributed. We note that for purposes of the EPMs where we need to identify episodes that are included in the EPMs, we use the terms anchor and chained anchor hospitalization to identify hospitalizations that initiate EPM episodes for beneficiaries whose care is included in the EPMs. In describing the quality measures themselves in detail in section III.E.4. of this proposed rule, we use the term index hospitalization to identify hospitalizations of beneficiaries whose outcomes are included in the measures. Thus, anchor hospitalizations and index hospitalizations would have varying degrees of overlap depending on the specific quality measure. The measure includes the following index admissions for patients:

  • Having a principal discharge diagnosis of AMI.
  • Enrolled in Medicare FFS.
  • Aged 65 or over.
  • Not transferred from another acute care facility.
  • Enrolled in Medicare Part A and Part B for the 12 months prior to the date of index admission, and enrolled in Part A during the index admission.

This measure excludes the following index admissions for patients:

  • Discharged alive on the day of admission or the following day who were not transferred to another acute care facility.
  • With inconsistent or unknown vital status or other unreliable demographic (age and gender) data;
  • Enrolled in the Medicare hospice program any time in the 12 months prior to the index admission, including the first day of the index admission;
  • Discharged against medical advice American Medical Association (AMA); or
  • Without at least 30 days of post-discharge enrollment in FFS Medicare as the 30-day mortality outcome cannot be assessed for these patients.

Finally, for the purpose of this measure, admissions within 30 days of discharge from an index admission are not eligible to also be index admissions. Thus, only one index admission for AMI per beneficiary is randomly selected for inclusion in the cohort.

(e) Risk-Adjustment

We note that this measure is aligned with the risk-adjustment methodologies adopted for the MORT-30-AMI (NQF #0230) measure under the HIQR Program in accordance with section 1886(b)(3)(B)(viii)(VIII) of the Act, as finalized in FY 2008 IPPS/LTCH final rule (2008 IPPS/LTCH final rule 71 FR 67960). We also note that the measure risk adjustment takes into account patient age, sex, and comorbidities to allow a fair assessment of hospital performance. The measure defines the patient risk factors for mortality using diagnosis codes collected from all patient claims 1 year prior to patient index hospitalization for AMI. As previously noted in the MORT-30-AMI measure (NQF #0230), ICD-10-CM codes on Medicare Parts A and B administrative claims are used to inform the risk prediction for each patient; diagnostic codes from post-acute care settings are included in the measure, but this information is only used to identify a hospital's patient case mix in order to adequately adjust for differences in case mix across hospitals. Use of Parts A and B data does not mean the measure is applicable to post-acute care settings, only that it uses comprehensive data to predict the risk of the outcome and adjust for hospital patient case mix. We note that the patient diagnosis codes are grouped using Hierarchical Condition Categories (HCCs), which are clinically relevant diagnostic groups of codes. The CCs used in the risk-adjustment model for this measure are provided on the CMS QualityNet Web site at: https://www.qualitynet.org/​dcs/​ContentServer?​c=​Page&​pagename=​QnetPublic%2FPage%2FQnetTier4&​cid=​1219069856694.

In summary, age, sex, and comorbidities present at the time of admission are adjusted for differences in hospital case mix (patient risk factors). The measure uses the hierarchical logistic regression model (HLM) statistical methodology for risk adjustment.

(f) Calculating the Risk-Standardized Mortality Ratio (RSMR) and Performance Period

We propose to calculate hospital 30-day, all-cause, risk-standardized mortality rates consistent with the methodology used to risk standardize all readmission and mortality measures used in CMS hospital quality programs. Using HLM, we calculate the hospital-level risk-standardized mortality rate following AMI hospitalization by producing a ratio of the number of “predicted” deaths (that is, the adjusted number of deaths at a specific hospital) to the number of “expected” deaths (that is, the number of deaths if an average quality hospital treated the same patients) for each hospital and then multiplying the ratio by the national raw mortality rate.

A 3-year rolling period for calculating measure results would be consistent with the time frame used for the HIQR Program (FY 2008 IPPS/LTCH final rule 71 FR 67960). Section III.E.5. of this proposed rule, Form, Manner, and Timing of Quality Measure Submission, summarizes the proposed measure performance periods for AMI model performance years 1 through 5. We note that, for each performance year, improvement on the MORT-30-AMI (NQF #0230) measure would be determined by comparing measure results from that performance year to results in the 3-year rolling Start Printed Page 50895measurement period immediately preceding each AMI model performance year to results from the 3-year period from July 1, 2014 through June 30, 2017, for performance year 2 by comparing measure results in this year to results from the 3-year period from July 1, 2015 through June 30, 2018, in performance year 3 by comparing measure results in this year to results from the 3-year period from July 1, 2016 and June 30, 2019, in performance year 4 by comparing measure results in this year to results from the 3-year period from July 1, 2017 and June 30, 2020, and in performance year 5 by comparing measure results in this year to results from the 3-year period from July 1, 2018 and June 30 2021.

The proposal to include Hospital-level 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) following AMI hospitalization (NQF #0230) measure in the AMI model is included in § 512.411(a)(1). We seek comment on this proposal to include Hospital-level 30-Day, All-Cause, Risk-Standardized Mortality Rate (RSMR) following AMI hospitalization (NQF #0230) measure in the AMI model to assess quality performance.

(2) Excess Days in Acute Care after Hospitalization for Acute Myocardial Infarction (AMI Excess Days)

(a) Background

The Excess Days in Acute Care after Hospitalization for Acute Myocardial Infarction (AMI) measure (AMI Excess Days) is a risk-standardized outcome measure that compares the number of days that patients are predicted to spend in acute care across the full spectrum of possible acute care events (hospital readmissions, observation stays, and ED visits) after discharge from a hospital for AMI, to the days patients are expected to spend in acute care based on their degree of illness.

Some of the costs for AMI can be attributed to high acute care utilization for post-discharge AMI patients in the form of readmissions, observation stays, and emergency department (ED) visits. We note that patients admitted for AMI have disproportionately high readmission rates, and that readmission rates following discharge for AMI are highly variable across hospitals in the United States.[78,79]

For the previously adopted HIQR Program measure, Hospital 30-Day, All-Cause Risk- Standardized Readmission Rate (RSRR) following Acute Myocardial Infarction (AMI) Hospitalization (NQF #0505) (CY 2009 OPPS/ASC final rule with comment period; 73 FR 68780 through 68781), publicly reported 30-day risk-standardized readmission rates for AMI ranged from 17.5 percent to 30.3 percent for the time period between July 2011 and June 2012.[80] However, in addition to an increased risk of requiring readmission in the post-discharge period, patients are also at risk of returning to the hospital for both observation stays and ED visits which also characterize potentially preventable acute care. ED visits represent a significant proportion of post-discharge acute care utilization for all conditions, including patients with AMI. Two recent studies conducted in patients of all ages showed that 9.5 percent of patients return to the ED within 30 days of hospital discharge; additionally, about 12 percent of these patients are initially discharged from the ED and are not captured by the previously adopted HIQR Program readmission measures.8.9 The rising use of observation stays among Medicare beneficiaries between 2001 and 2008 sparked concern among patients, providers, and policymakers that the AMI 30-day Readmission (NQF #0505) measure does not capture the full range of unplanned acute care events that occur in the post-discharge period. In order to address the rising use of observation stays amongst Medicare beneficiaries CMS is proposing the Excess Days in Acute Care after Hospitalization for AMI (AMI Excess Days) measure for use in the AMI model. The AMI Excess Days measure comprehensively captures all post-discharge, unplanned acute care events as a count of the excess days a hospital's patients spent as inpatients, in observation, or in the ED over a 3-year measurement period.

In 2014, we developed the proposed measure of excess days in acute care following AMI hospitalization, supported for use in the Hospital Quality Reporting Program by the MAP and submitted to the NQF for endorsement. We note that this measure was submitted for endorsement to the NQF All-Cause Admissions and Readmissions Committee in January 2016 with appropriate consideration for sociodemographic status. The measure was finalized for the HIQR Program FY 2018 payment determination (FY 2016 IPPS/LTCH final rule 80 FR 49690).

(b) Data Sources

We propose to use Medicare Part A and Part B FFS claims submitted by the AMI model participant as the data source for calculation of the AMI Excess Days measure as harmonized with the MORT-30-AMI(NQF #0230) and READM-30-AMI(NQF #0505) measures. Index admission diagnoses and in-hospital comorbidities are assessed using Medicare Part A claims. Additional comorbidities prior to the index admission are assessed as Part A inpatient, outpatient, and Part B office visit Medicare claims in the 12 months prior to the index (initial) admission. Enrollment and post-discharge mortality status are obtained from Medicare's enrollment database which contains beneficiary demographic, benefits/coverage, and vital status information.

(c) Cohort

The AMI Excess Days measure includes Medicare FFS beneficiaries, aged 65 years or older, discharged from non-federal acute care hospitals with a principal discharge diagnosis of AMI and with a complete claims history for the 12 months prior to index admission. Eligible hospitalizations are defined using the following ICD-10-CM codes: I2109, I2111, I2119, I2129, I214, and I213.

We propose that the measure will include index admissions to all non-federal acute care hospitals, which includes all participants in the AMI model. Hospital performance will only be publically reported for hospitals with 25 or more index admissions in the 3-year measurement period. The AMI model cohort would differ from the hospital cohort that is currently captured in the measure through the HIQR Program. Although performance on the measure will not be publically reported for hospitals with fewer than 25 cases, such hospitals will receive information about their performance on the measure. We refer readers to section III.B.5. of this proposed rule for a discussion of AMI model participant selection.

(d) Inclusion and Exclusion Criteria

We propose that an index admission is the hospitalization to which the excess days in acute care outcome is attributed. We note that for purposes of Start Printed Page 50896the EPMs where we need to identify episodes that are included in the EPMs, we use the terms anchor and chained anchor hospitalization to identify hospitalizations that initiate EPM episodes for beneficiaries whose care is included in the EPMs. In describing the quality measures themselves in detail in section III.E.4. of this proposed rule, we use the term index hospitalization to identify hospitalizations of beneficiaries whose outcomes are included in the measures. Thus, anchor hospitalizations and index hospitalizations would have varying degrees of overlap depending on the specific quality measure. The measure includes the following index admissions for patients:

  • Having a principal discharge diagnosis of AMI.
  • Enrolled in Medicare FFS.
  • Aged 65 or over.
  • Not transferred from another acute care facility.
  • Enrolled in Medicare Part A and Part B for the 12 months prior to the date of index admission, and enrolled in Part A during the index admission.

The measure excludes the following index admissions for patients:

  • Discharged alive on the day of index admission or the following day who were not transferred to another acute care facility.
  • With inconsistent or unknown vital status or other unreliable demographic (age & gender) data.
  • Enrolled in the Medicare hospice program any time in the 12 months prior to the index admission, including the first day of the index admission.
  • Discharged AMA.
  • Without at least 30 days of post-discharge enrollment in FFS Medicare as the 30-day excess days outcome cannot be assessed for these patients.

Finally, for the purpose of this measure, hospitalizations that occur within 30 days of discharge from an index admission are not eligible to also be index admission. Thus, only one index admission for AMI per beneficiary is randomly selected for inclusion in the cohort.

(e) Risk-Adjustment

We propose for the AMI model to align this measure with the risk-adjustment methodologies adopted for the AMI Excess Days measure under the HIQR Program in accordance with section 1886(b)(3)(B)(viii)(VIII) of the Act, as finalized in the FY 2016 IPPS/LTCH final rule (80 FR 49682). We also note that the measure risk adjustment takes into account patient age, sex, and comorbidities to allow a fair assessment of hospital performance. The measure defines the patient risk factors for excess days using diagnosis codes collected from all patient claims 1 year prior to a patient's index hospitalization for AMI. Accordingly, only comorbidities that convey information about the patient at the time of index admission or in the 12 months prior, and not complications that arise during the course of the index hospitalization, are included in the risk-adjustment model. The measure does not adjust for patients' index admission source or their discharge disposition (for example, SNF) because these factors are associated with the structure of the healthcare system, not solely patients' clinical comorbidities. Regional differences in the availability of post-acute care providers and practice patterns might also exert undue influence on measure results. In addition, data fields that capture discharge disposition, for example to post-acute care settings, on inpatient claims are not audited and are not as reliable as diagnosis codes.

As previously noted in the AMI Excess Days measure, ICD-10-CM diagnosis codes present on Parts A and B administrative claims are used to inform the risk prediction for each patient. Diagnostic codes from post-acute care settings are utilized in the measure calculation, but this information is only used to identify a hospital's patient case mix in order to adequately adjust for differences in case mix across hospitals. We note that the patient diagnosis codes are grouped using HCCs, which are clinically relevant diagnostic groups of codes. The CCs used in the risk-adjustment model for this measure are provided on the CMS QualityNet Web site: https://www.qualitynet.org/​dcs/​ContentServer?​c=​Page&​pagename=​QnetPublic%2FPage%2FQnetTier4&​cid=​1219069856694.

In summary, age, sex, and comorbidities present at the time of index admission are adjusted for differences in hospital case mix (patient risk factors). The measure uses the HLM statistical methodology for risk adjustment.

(f) Calculating the Rate and Performance Period

We propose to calculate hospital 30-day excess days in acute care with the methodology used to risk standardize all excess days measures used in CMS hospital quality programs. The outcome of the measure is a count of the number of days the patient spends in acute care within 30 days of discharge. We define days in acute care as days spent in an ED, admitted to an observation unit, or admitted as an unplanned readmission for any cause within 30 days from the date of discharge from the index AMI hospitalization. Each ED treat-and-release visit is counted as 1 half-day (0.5 days). Observation stays are recorded in terms of hours and are rounded up to the nearest half-day. Each readmission day is counted as 1 full day (1 day). We count all eligible outcomes occurring in the 30-day period, even if they are repeat occurrences. The measure incorporates “exposure time” (the number of days each patient survives after discharge, up to 30). This exposure time is included to account for differential risk for excess days in acute care after discharge among those patients who do not survive the full post-discharge period. If a readmission or observation stay extends beyond the 30-day window, only those days within the 30-day window are counted.

Using a two-part random effects model, or “hurdle” model, we account for the structure of the data (patients clustered within hospitals) and the observed distribution of the outcome. Specifically, we model the number of acute care days for each patient as: (a) The probability that the patient will have a non-zero number of days in post-discharge acute care; and (b) the number of days the patient is predicted to spend given that this number is non-zero. The first part is specified as a legit model, and the second part is specified as a Poisson model, with both parts having the same risk-adjustment variables and each part having a random effect. This model is used to calculate the predicted (including random effects) and expected (assuming random effects are zero) number of days in post-discharge acute care for each patient. The average difference between patients' predicted and expected estimates for each hospital is used to construct the risk-standardized excess days outcome. The excess days outcome is reported at the hospital-level per 100 discharges.

We define the time period for the measure as within 30 days of the date of discharge of the index AMI hospitalization. The 30-day post-discharge window for assessing the outcome is consistent with the claims-based MORT-30-AMI (NQF #0230) and Hybrid AMI Mortality (NQF #2473) measures as noted in this proposed rule.

A 3-year rolling performance period would be consistent with that used for the HIQR Program (FY 2016 IPPS/LTCH final rule 80 FR 49681). Section III.E.5., Form, Manner, and Timing of Quality Measure Data Submission, of this proposed rule summarizes the proposed measure performance periods for AMI model performance years 1 through 5. We note that improvement on the AMI Start Printed Page 50897Excess Days measure would be determined from the immediate 3-year rolling performance period available for the year preceding the AMI model performance year as explained in Table 30.

The proposal to include the Excess Days in Acute Care after Hospitalization for AMI measure in the AMI model is included in § 512.411(a)(2). We seek comment on this proposal to include the Excess Days in Acute Care after Hospitalization for AMI measure in the AMI model to assess quality performance.

(3) Hybrid Hospital 30-Day, All-Cause, Risk-Standardized Mortality Rate Following Acute Myocardial Infarction (AMI) Hospitalization (NQF# 2473)(Hybrid AMI Mortality)

(a) Background

In keeping with our goal to move toward the use of EHRs, and in response to stakeholder feedback to include clinical data in outcome measures, we have developed the hospital 30-day risk-standardized acute myocardial infarction (AMI) mortality eMeasure (NQF #2473) (herein after referred to as Hybrid AMI Mortality measure). This measure will incorporate a combination of claims data and EHR data submitted by hospitals, and because of these combined data sources, it is referred to as a hybrid measure. The Hybrid AMI Mortality (NQF #2473) measure cohort and outcome are identical to those in the hospital 30-day, all-cause, risk-standardized mortality rate (RSMR) following acute myocardial infarction (AMI) (NQF #0230), measure which is also being proposed in the AMI model.

In contrast to the claims-only MORT-30-AMI (NQF #0230) measure, the proposed Hybrid AMI Mortality (NQF #2473) measure utilizes five core clinical data elements (age; heart rate; systolic blood pressure; troponin; creatinine) in the risk-adjustment methodology that are obtainable through EHR data. These five core clinical data elements are intended to reflect patients' clinical status when they first present to an acute care hospital for treatment of AMI. The clinical data elements include age at the time of admission, first-captured vital signs (heart rate, systolic blood pressure) collected within 2 hours of the patient first presenting to the hospital, and the first captured laboratory values (troponin, creatinine) collected within 24 hours of the patient first presenting to the hospital to which they are subsequently admitted. We note that these five data elements are routinely collected on hospitalized adults with AMI upon presentation to the hospital, consistently captured in medical records under current clinical practice, and can be feasibly electronically extracted from hospital EHRs.

In order to prepare for future reporting of the Hybrid AMI Mortality (NQF #2473) measure, we are proposing to seek and reward voluntary data submission of the five core clinical data elements included in the risk model for the Hybrid AMI mortality (NQF #2473) measure. We are also proposing to require submission of six additional linking variables (CCN, HIC Number, date of birth, sex, admission date, and discharge date) to ensure that the datasets containing administrative claims data are correctly linked with EHR datasets containing the core clinical data elements for proper risk adjustment. The voluntary data submission initiative will allow AMI model participants to build processes to extract and report the EHR data elements, as well as support CMS testing of systems required for Hybrid AMI Mortality measure (NQF #2473) production including data receiving and auditing, the merging EHR and claims data, calculation and production of measure results.

Finally, we are considering using the Hybrid AMI Mortality (NQF #2473) measure as a replacement for the current publicly reported MORT-30-AMI (NQF #0230) measure in CMS models or programs when appropriate. In future years CMS may implement the Hybrid AMI Mortality (NQF #2473) measure in models and/or programs, such as in the AMI model or HIQR Program. In that event, we would propose to adopt the measure through notice and comment rulemaking. We refer readers to more detailed information on the measure specifications in this proposed rule and to the CMS Web site at: http://cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​HospitalQualityInits/​Measure-Methodology.html.

(b) Data Sources

We propose to use two sources of data submitted by AMI model participants to calculate the Hybrid AMI Mortality (NQF #2473) measure: Medicare Part A and Part B (FFS claims to identify index admission diagnoses; and EHR-captured clinical information collected at presentation for risk-adjustment of patients' severity of illness. Deaths are identified using the Medicare Enrollment Database which contains beneficiary demographic, benefits/coverage, and vital status information.

For the voluntary data submission initiative, EHR data submission will align with existing Electronic Clinical Quality Measure (eCQM) standards and data reporting procedures for hospitals. In alignment with these standards, we are posting the electronic specifications for the Hybrid AMI Mortality (NQF #2473) measure, which include the Measure Authoring Tool (MAT) output and value sets for all included data elements, on the CMS Web site at: http://cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​HospitalQualityInits/​Measure-Methodology.html.

The Office of the National Coordinator for Health Information Technology (ONC) adopted quality reporting document architecture (QRDA) as the standard to support both QRDA Category I (individual patient) and QRDA Category III (aggregate) data submission approaches for Meaningful Use Stage 2 in the Health Information Technology: Standards, Implementation Specifications, and Certification Criteria for Electronic Health Record Technology, 2014 Edition; Revisions to the Permanent Certification Program for Health Information Technology rule (77 FR 54163 through 54292). We intend to provide AMI model participants with information about how many qualifying admissions are submitted successfully. We refer readers to the definition of “successful data submission” in section III. E.4.a.(3)(vii) of this proposed rule.

We seek comment on our proposal to use the following reporting mechanisms in performance year 1: QRDA, a simpler mechanism such as a spreadsheet, or both. We propose using QRDA in AMI model performance years 2 through 5. The purpose of the use of a simpler reporting format in the first performance year reporting format in the first performance year would be to allow hospitals to perfect data extraction with the 2017 data and postpone mastery of reporting in the QRDA format to the following year.

(c) Cohort

The Hybrid AMI Mortality (NQF #2473) measure includes Medicare FFS beneficiaries, aged 65 years or older, discharged from non-federal acute care hospitals with a principal discharge diagnosis of AMI. Eligible hospitalizations are defined using the following ICD-10-CM codes: I2109, I2111, I2119, I2129, I214, and I213.

Hospital performance for the Hybrid AMI Mortality (NQF #2473) measure will not be publicly reported. However, AMI model participants will receive hospital-specific reports for each performance year with information about the success of their voluntary submission of EHR data.Start Printed Page 50898

(d) Inclusion and Exclusion Criteria

We propose that an index admission is the hospitalization to which the mortality outcome is attributed. The Hybrid AMI mortality (NQF #2473) measure includes the following index admissions for patients:

  • Having a principal discharge diagnosis of AMI.
  • Enrolled in Medicare FFS.
  • Aged 65 or over.
  • Not transferred from another acute care facility.
  • Enrolled in Medicare Part A and Part B for the 12 months prior to the date of index admission, and enrolled in Part A during the index admission.

This measure excludes the following index admissions for patients:

  • Discharged alive on the day of admission or the following day who were not transferred to another acute care facility.
  • With inconsistent or unknown vital status or other unreliable demographic (age & gender) data.
  • Enrolled in the Medicare hospice program any time in the 12 months prior to the index admission, including the first day of the index admission.
  • Discharged AMA.
  • Without at least 30 days of post-discharge enrollment in FFS Medicare as the 30-day mortality outcome cannot be assessed for these patients.

Finally, for the purpose of this measure, only one index admission per patient for AMI is randomly selected for inclusion in the cohort.

(e) Risk-Adjustment

We note that this measure is aligned with the methodology approach adopted for the MORT-30-AMI (NQF #0230) measure under the HIQR Program in accordance with section 1886(b)(3)(B)(viii)(VIII) of the Act, as finalized in FY 2008 IPPS/LTCH final rule (2008 IPPS/LTCH final rule 71 FR 67960). The Hybrid AMI Mortality (NQF #2473) measure uses EHR data and not administrative claims data to adjust for differences across hospitals in how at-risk their patients are for death, relative to patients cared for by other hospitals. The risk model was developed with input from the literature, clinical and EHR experts, and Health Information Technology vendors. In order to be included as risk variables, clinical data elements had to be—(1) consistently obtained in the target population (Medicare FFS AMI patients) based on current clinical practice; (2) captured with a standard definition and recorded in a standard format within the EHR; and (3) entered in structured fields that are feasibly retrieved from current EHR systems. The final measure includes five variables that meet these feasibility criteria, are present for most patients at the time of clinical presentation to the hospital, are clinically relevant to patients with AMI, and demonstrate a strong statistical association with 30-day mortality. Hospitals will submit the first-captured data values of each of the five core clinical data elements upon patient presentation to the hospital. They are: Age; the first-captured heart rate and systolic blood pressure measured within 2 hours of a patient presenting to the hospital; and the first captured troponin and creatinine values within 24 hours of a patient presenting to the hospital. Although EHRs likely will ultimately link across clinical episodes of care and contain historical patient data, given the EHR environment at the time of measure development and inability to reliably obtain data from the outpatient setting prior to admission, we only considered for inclusion those measure variables that would be available and consistently collected at first presentation to the hospital.

The overall performance of the model was comparable with or better than that of current publicly reported outcome measures.[81] We tested measure score validity by correlating the RSMR with that of the previously validated, publicly reported, administrative claims-based MORT-30-AMI (NQF #0230) measure. For more detailed information on the model performance, we refer readers to the CMS Web site at: http://cms.gov/​Medicare/​Quality-Initiatives-Patient-Assessment-Instruments/​HospitalQualityInits/​Measure-Methodology.html.

(f) Calculating the Risk-Standardized Mortality Ratio (RSMR) and Performance Period

We calculate hospital 30-day, all-cause, risk-standardized mortality rates consistent with the methodology used to risk standardize all readmission and mortality measures used in CMS hospital quality programs. Using an HLM statistical methodology for risk adjustment, we calculate the hospital-level risk-standardized mortality rate following AMI hospitalizations by producing a ratio of the number of “predicted” deaths (that is, the adjusted number of deaths at a specific hospital) to the number of “expected” deaths (that is, the number of deaths if an average quality hospital treated the same patients) for each hospital and then multiplying the ratio by the national observed mortality rate.

We propose defining AMI model performance years as outlined in section III.E.5. of this proposed rule. A performance period for the voluntary data submission are those timeframes in which a hospital discharge occurs for an eligible AMI index hospitalization. For performance year 1 of the AMI model, participants voluntarily submitting data will only be asked to submit data for a 2-month period. The 2-month period for AMI voluntary data reporting was identified due to data processing and coordination with other proposed timelines for this model. Data submitted for the first year would be for cases that fulfill the measure specifications described in section III.E.4.a.(3) of this proposed rule, and would be restricted to the data elements from eligible AMI index hospitalizations with discharges occurring between July 1, 2017 and August 31, 2017.

For performance year 2 of the AMI model, AMI voluntary data reporting would be 10 months of data for discharges from eligible AMI hospitalizations occurring between September 1, 2017 and June 30, 2018. For subsequent years of the model, the performance periods for submission of voluntary data will consist of discharges within calendar-year 12-month time periods from July 1 through June 30. The proposed performance periods would enable AMI model participants to receive points toward the AMI model composite quality score for data submission starting in performance year 1. We seek comment on our proposal for defining the data reporting period for performance year 1 episodes for an AMI model participant as eligible AMI index hospitalizations with discharges occurring between July 1, 2017 and August 31, 2017, and for performance year 2 as eligible AMI hospitalizations with discharges occurring between September 1, 2017 and June 30, 2018, with subsequent performance year data reporting periods each being calendar-year 12 month periods and starting every July 1st. Refer to Table 30 for summary of proposed performance periods.

(g) Requirements for Successful Submission of AMI Voluntary Data

In order for CMS to assess if AMI model participants that submit the AMI voluntary data are eligible for points toward the hospital's AMI model composite quality score, we propose to use the following criteria to determine if a participant has successfully submitted AMI voluntary data: Start Printed Page 50899Submission of the first-captured data values for the five core clinical data elements (age; first-captured heart rate and systolic blood pressure measured within 2 hours of a patient presenting to the hospital; and first-captured troponin and creatinine values measured within 24 hours of a patient presenting to the hospitals), and six linking variables required to merge with the CMS claims data CCN, HIC Number, date of birth, sex, admission date, and discharge date).

All of these data elements must be submitted for each qualifying AMI hospitalization as described in section III.E.5. of this proposed rule. If troponin was not measured in the patient within 24 hours of presentation to the hospital, the hospital will still receive credit for successful data submission if all other clinical data elements (age, heart rate, systolic blood pressure, and creatinine) as well as the six linking variables are all reported in the submission. We recognize that some patients may have clinical signs or symptoms that require emergent treatment; and that in such cases treatment might proceed without first obtaining a troponin level. However hospitals are required to report troponin values on all patients in whom a troponin test was performed within the first 24 hours of presenting to the hospital and to indicate in their data submission each instance in which a troponin value was not measured and therefore not available for a patient.

AMI voluntary data submission must occur within 60 days of the end of the most recent data collection period as described in the listing of reporting periods for all 5 model performance years in section III.E.5. of this proposed rule.

To fulfill AMI voluntary data collection criteria for model performance year 1, hospitals must submit valid data on 50 percent of qualifying AMI hospitalizations (identified by the denominator in the measure authorizing tool (MAT) output). To successfully submit AMI voluntary data for performance years 2 through 5, hospitals must submit valid data for all five core clinical data elements on over 90 percent of qualifying AMI patients (with the exception for troponin values described in this section). Further details on scoring of the voluntary data submission are discussed in section III.E.3.e.(1) of this proposed rule.

Each year, AMI model participants voluntarily submitting data for this measure will receive hospital-specific reports that detail submission results from the most recent performance period. The reports will include the match rate between the hospital's submitted EHR data and corresponding claims data, as well as the proportion of patient data submitted relative to all qualifying AMI admissions with all five core clinical data elements. As the initiative seeks to test and reward hospitals' ability to submit data, hospitals will not be penalized for missing troponin values for patients in whom these values were not measured at the time clinical treatment was provided. If hospitals successfully submit the remaining four clinical data elements and all of the linking variables, a missing troponin value which is due to troponin having not been measured in that patient will not result in an unsuccessful submission as long as hospitals indicate that the troponin value was not measured and therefore not available for that patient. Hospitals will still be rewarded for successfully submitting data in these cases.

We previously described a qualifying AMI patient in section III.E.4.a.(3)(iii) of this proposed rule. This description is important, as these patients are those for whom we seek submission of voluntary data from AMI model participants. We selected the requirement of submitting 90 percent of qualifying AMI patients' data for performance years 2 through 5 because this volume of cases will result in a high probability that we will have a national sample of AMI patient data representative of each hospital's patient case mix. Having 90 percent of the data for qualifying AMI patients in performance years 2 through 5 will enable an accurate and reliable assessment of the potential implementation of a Hybrid AMI mortality (NQF #2473) measure that utilizes EHR data. In addition, the testing we have performed in hospitals' EHR data indicate that these data elements are captured in over 90 percent of Medicare FFS patients who are 65 years or older and admitted to acute care hospi