Charles N. Kahn III President and CEO

October 3, 2016

Andy Slavitt, Acting Administrator Centers for Medicare & Medicaid Services Department of Health and Human Services Hubert H. Humphrey Building 200 Independence Avenue, S.W., Room 445-G Washington, DC 20201

Re: CMS-5519-P; 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)

Dear Mr. Slavitt:

The Federation of American Hospitals (“FAH”) is the national representative of more than 1,000 investor-owned or managed community hospitals and health systems throughout the United States. Our members include teaching and non-teaching, short-stay acute, inpatient rehabilitation, long-term acute care, psychiatric and cancer hospitals in urban and rural America, and provide a wide range of acute, post-acute and ambulatory services. The FAH appreciates the opportunity to provide comments to the Centers for Medicare and Medicaid Services (“CMS”) on the above notice of proposed rulemaking (“proposed rule”), published in the Federal Register (81 FR 50793-51040) on August 2, 2016.

GENERAL COMMENTS

I. The FAH Urges CMS to Delay the Start Date Until at Least January 1, 2018 As

CMS’s Current Pace of Implementing These Programs is Too Fast and Too Soon

FAH members appreciate the opportunity to test innovative care models developed by CMS’s Center for Medicare & Medicaid Innovation (“CMMI”), and believe that the three new EPMs ―acute myocardial infarction (“AMI”), coronary artery bypass graft (“CABG”), and

2

surgical hip/femur fracture treatment (“SHFFT”)―and the new cardiac rehabilitation (“CR”) and intensive cardiac rehabilitation (“ICR”) incentive payment model could hold significant promise for furthering the Triple Aim goals of providing high quality care at lower cost to produce better outcomes and advance population health. We also appreciate the critical and central role that hospitals will play in these models and look forward to working with CMS to help build and implement effective and successful payment models.

The FAH and its members, however, have become increasingly concerned about the pace of change proposed by CMS and the unreasonable expectations and burden that such rapid and multiple changes in the delivery system and related payment structure place on hospitals and their work forces. Simply put, this is too fast and too soon. We strongly believe that CMS first needs to evaluate and learn from hospitals’ Comprehensive Care for Joint Replacement Model (“CJR”) experience (a model with less than 6 months of history) and from the Bundled Payments for Care Improvement (“BPCI”) initiative Model 2 results. Establishing “proof of concept” is a very important tool to utilize before implementing new mandatory episode payment models that could affect large numbers of Medicare beneficiaries and potentially have significant and adverse unintended consequences if not implemented in a reasonable, thoughtful, and deliberate approach.

The BPCI Model 2 is most analogous to the EPM approach, as the BPCI Model 2 design

includes the anchor admitting hospital stay and all related professional services for a chosen episode length of 30, 60, or 90 days. CMS only recently released the BPCI Year 2 evaluation and monitoring report, which raises questions, in particular, about the cardiac models.1 Results from the evaluation of year 2 results showed no statistically significant difference in Medicare payments and an increase in mortality for the cardiovascular surgical episodes between the BPCI and comparison groups. This is a surprising result given that BPCI hospitals volunteered to participate in this program and should have been well-prepared to succeed and achieve cost savings and maintain quality. While CMS states that subsequent results do not show a statistically significant increase in mortality between the comparison and BPCI groups, the initial increased mortality findings are very concerning. Moreover, while there was a significant reduction in utilization of institutional post-acute care settings, there were instances where BPCI patients exhibited less functional improvement. This suggests that hospitals will need to work to develop effective collaborative relationships in determining optimal sites of post-acute care and establishing effective transitions to appropriate home health care.

FAH members believe that episode payment models such as those newly proposed, when realistically constructed with sufficient stakeholder preparation time, hold promise as part of CMS’s strategy to move from volume to value. FAH members also appreciate the opportunity to be involved with testing these innovative care models. However, given the challenges outlined and the lack of preparation time for hospitals, the FAH strongly recommends that CMS delay the start date of the proposed EPMs until no earlier than January 1, 2018. If the final rule is delayed beyond January 1, 2017, we urge CMS to provide hospitals with at least 12 months of preparation time from the date the final rule is finalized. We also believe hospitals should not be subject to downside risk for at least 12 months (as was provided in CJR) from our proposed delayed implementation timeline.

1 The Lewin Group. (August 2016). CMS Bundled Payments for Care Improvement Initiative Models 2-4: Year 2 Evaluation & Monitoring Annual Report. Prepared for the Centers for Medicare and Medicaid Services.

3

The FAH urges CMS not to implement the SHFFT model at this time. If CMS decides to move forward with implementing the SHFFT model, we urge, as discussed in more detail below, that CMS gradually phase-in these new models. Separately, the FAH supports the CMS proposal to establish a pathway for participants in the EPMs and CJR to be considered as participating in an Advanced Alternative Payment Model (“APM”) for the purpose of becoming a qualifying APM participant (“QP”), and support its implementation date of July 1, 2017.

Our members strongly believe that the start date of these EPMs should be delayed

at least 6 months, until no sooner than January 1, 2018. While we recognize that CMS has taken into account additional time with the proposed start date of July 1, 2017, we believe an additional 6 months is warranted given all of the other competing priorities for hospitals and the complexity of implementing these EPMs successfully.

Based on our members’ experience with implementing CJR, BPCI and the Medicare Shared Savings Program (“MSSP”) program, hospitals and their collaborating partners have consistently needed at least 12 months to transition to episode payment models. This time was needed to build the clinical, legal, financial and quality infrastructure, analyze and understand the clinical and cost data, educate providers and staff, and develop capabilities and networks required to successfully launch the model. In addition, and most importantly, our hospitals stressed that most of the time was needed to do the hard work necessary to redesign the clinical care patient care model in a manner that ensures patients receive the most appropriate and optimal care.

The BPCI initiative experience, in particular, is very informative to what is necessary

and what is a realistic implementation timeline to position hospitals and their patients for success, and this experience suggests that the proposed timeline will be very difficult to meet. BPCI participants were largely enthusiastic volunteers who seemingly were well positioned to adopt and adapt to the new model, yet many still found the program and timing demands very challenging, as evidenced by the significant departures from that program. Mandating this, especially for unprepared participants, brings even greater challenges, and increases the chance of failure and disruption of health care services for Medicare beneficiaries.

A six-month delay beyond CMS’s proposed July 1 start date would help ensure that

during the non-risk bearing performance year (“PY”1) of the EPM model, providers will remain focused on patient care and successful model implementation, rather than attempting to undertake and complete the necessary preparatory work required for successful implementation. By adopting this proposed timeline, EPM participants will be better prepared to succeed in the first risk-bearing year, PY2, of these models.

4

The exhibit below reflects the timeline of BPCI Round 2 participants.

Moreover, the June 9, 2015 MSSP final rule recognized that an even longer timeline may be required by providers to transition from traditional fee-for-service to a value-based alternative payment model (“APM”). This timeline allowed Track 1 Accountable Care Organizations (“ACOs”) to continue participating under Track 1 for an additional three-year agreement period. The Track 1 option is a one-sided model with no down-side risk, and the vast majority of MSSP ACOs are enrolled in Track 1. We believe the recognition that time is needed for providers to gain experience and manage and analyze claims data in the MSSP ACO context is also appropriate for these episode payments models, especially given it is a mandatory program, and with two-sided risk.

Given the experience with other models, as discussed above, it would be unreasonable to

expect that a large proportion acute-care hospitals mandated to participate in these new models would be fully prepared for a July 1, 2017 start date. For example, even the “simplest” scenario, in which a CJR participant hospital must newly undertake the SHFFT model, is in fact far from simple. SHFFT patients will be older and in poorer baseline health than their CJR counterparts. SHFFT patients are more likely to be unplanned and seen on an emergent basis, rather than being electively admitted during daylight hours on a weekday and postoperatively after total hip arthroplasty (“THA”). The most common treatment for hip fracture (internal fixation) is a materially different operation than THA, requiring different surgical implants and equipment, and in some communities the operations are performed by different surgeons (trauma versus total joint orthopedic subspecialists). Discharge planning for the patient who has sustained an unexpected major decrease in mobility (hip fracture) is distinctly different than for someone who has knowingly chosen THA and its associated postoperative rehabilitation. Other model implementation scenarios are even more complex, particularly for those hospitals without prior experience with CJR, BPCI, or other episode payment models, who will be now be mandated to implement de novo the AMI and AMI-Percutaneous Coronary Intervention (“PCI”) models. Some of these hospitals also will have the additional challenge of implementing the CABG model at the same time.

In addition, there are several other reasons that support the FAH’s request for a delayed

5

start date.

• The proposal to hold providers harmless against downside financial risk in the first nine months (Q3 and Q4 of 2017 and Q1 of 2018) is appreciated but unfortunately, is not long enough and does not protect hospitals against other significant legal and regulatory risks that will be present from inception of these models.

The proposed rule provides that if hospitals do not meet their target rates, they will not

be financially penalized in PY1 or the first quarter of PY2. We believe a longer delay of at least 12 months (as was provided in CJR) of downside risk from the delayed implementation timeline for the model as detailed above is required. It appears this delayed start of downside financial risk may be meant to recognize the challenges inherent in implementing these EPMs, including building out and operating an effective network for managing episodes of care. However, hospitals are exposed to more than financial risk for which the proposed rule does not provide relief.

The proposed rule does not offer protection from various legal and regulatory risks

that are inherent in developing coordinated care arrangements between hospitals, physicians and post-hospital providers. Notably, the proposed rule does not offer any waivers of program integrity laws, even though those types of waivers are available in the BPCI initiative. Similar relief is needed here, especially for a mandatory program. Specifically, all stakeholders would benefit greatly from waivers under the anti-kickback (“AKS”), physician self-referral (“Stark Law”), and civil monetary penalty (“CMP”) laws to ensure hospitals that are mandated to participate in these models are not forced into risky conduct in developing gainsharing and preferred provider network arrangements.

The proposed rule does offer certain regulatory waivers, which are important and

should be finalized with some modifications, and CMS should provide additional waivers, as discussed at length below. Further, waivers by CMS alone are insufficient; the HHS Office of Inspector General (“OIG”) needs to provide relief here too. In the absence of waivers, hospitals and their partners are exposed to significant risks, and law enforcement and whistleblowers are not likely to be swayed from taking action by the public policy goals of these models. If providers do not have lawful arrangements to share risk or reward with physicians and post-hospital suppliers, then lawsuits are a distinct possibility. More time is needed by hospitals to put into place the necessary arrangements.

In addition, revisions to program integrity activities carried out by Medicare

Administrative Contractors (“MACs”), Quality Improvement Organizations (“QIOs”), recovery audit contractors (“RACs”), and various other contractors pertaining to issues associated with Medicare’s coverage and medical necessity policies under the traditional fee-for-service framework, should be developed and made available under bundled payment programs such as CJR and EPM. Incentives under these “at risk” mandatory bundled payment programs such as CJR and EPM are different than those under the traditional fee-for-service environment, and are aimed at providing more efficient, value-based care. Simply stated, treatment decisions of “at risk” hospitals and their collaborating partners should not be second-guessed and denied on grounds of coverage and medical necessity, because hospitals

6

and their collaborating partners will be encouraged and incentivized to coordinate and align treatment decisions across providers, while ensuring that patients receive efficient, high-quality care.

• A hospital’s pre-implementation administrative work and data analysis for these

EPM models is very significant and vital to the success of the program.

Similar to our concerns with the implementation of the CJR model, the experience of our hospital members has shown that there will likely be considerable administrative burden and time necessary to establish physician arrangements, provider networks and other business arrangements to operationalize and train staff on how to handle patients in these new EPM models. Hospitals need sufficient time to conduct preparatory market analyses, understand the clinical and financial risk of their patient populations, form networks with select physicians and other providers, and establish the needed organizational capabilities to manage payment bundles.

• The variation in hospital preparedness and capabilities for managing health

care episodes means that many hospitals will not be prepared on the proposed timeline.

Given the mandatory nature of the proposal, as well as the number of and variation across hospitals that will be participating in this program, it is imperative that CMS proceed with caution. As discussed in more detail below, the FAH believes that these EPMs must be structured in a manner that adopts a more gradual and phased approach that facilitates success and rewards improvement. In other words, the EPMs should be implemented such that participating providers are allowed adequate time to learn about and improve their care delivery structures, and for CMS to measure the impact of the model on patient outcomes, program spending, and provider financial stability without unnecessarily causing broad systemic failure in transitioning to these models. This approach will help enable the health care industry to achieve the program goals while preserving access to care.

The EPMs, and in particular the cardiac models, will require sufficient lead time, broad-

based clinical experience with continuity-of-care across episodes, appropriate workforce capacity and technology infrastructure, and significant investment by both the public and private sectors in order to succeed. Many hospitals will be challenged significantly in developing these capabilities, such as small hospitals that often have limited financial resources, those that are located in lower income geographic regions, or that incur high amounts of uncompensated care, have low case volume on which to spread financial risk, do not yet have experience with episode-based payment, or lack existing networks with physicians and other providers. The tasks at hand are formidable.

Overall, the FAH believes it is imperative that CMS carefully consider the variability of preparedness of hospitals with different levels of experience, especially under a mandatory model. CMS should ensure that hospitals are fairly protected from severe financial dislocation and that patient access to care is preserved under these models.

7

• The proposed quality measurement framework is problematic and needs to be overhauled before these EPMs begin. As noted below in detail, there are significant problems with the proposed quality

infrastructure. We have serious concerns about the quality framework for the proposed EPMs. Our concerns relate to measure relevance, measure overlap, measure misalignment, measure gaps, composite scoring methodology flaws, risk adjustment and risk stratification, data availability, and attribution of responsibility for quality. Given these concerns, the FAH believes that major revisions to the quality framework and/or the EPM models themselves are needed before any of the EPMs are implemented.

For the foregoing reasons, the FAH strongly believes that Medicare beneficiaries, hospitals, physicians, post-hospital suppliers and the program itself would all benefit from a delayed start date of no sooner than January 1, 2018 for the three episode payments model (SHFFT, AMI/AMI-PCI, CABG). The FAH supports the implementation by CMS of the CJR model revisions of the proposed rule on July 1, 2017.

II. The FAH Urges CMS Not to Implement the SHFFT Model

The FAH urges CMS not to implement the SHFFT model at this time. As stated

above, CMS only has limited experience with CJR (not even a full year) and does not have sufficient evidence on which to expand this model. In addition, CMS has not provided any protections to beneficiaries with respect to monitoring quality as none of the proposed quality measure specifically target SHFFT models. Also, as we noted earlier, SHFFT patients will be older and in poorer baseline health than their CJR counterparts. SHFFT patients are more likely to be unplanned and seen on an emergent basis. The most common treatment for hip fracture (internal fixation) is a materially different operation than those provided as part of CJR. Finally, discharge planning for the patient who has sustained an unexpected major decrease in mobility (hip fracture) is distinctly different than for someone who has knowingly chosen THA and its associated postoperative rehabilitation.

Importantly and as described in greater detail below, CMS has not evaluated the CJR

model as required under section §1115A(b)(4) of the SSA or if it has, it has not made such evaluation available to the public as required under §1115A(b)(4)(B). Thus, we are concerned about the potential unintended consequences to Medicare beneficiaries.

If CMS decides to move forward with implementing the SHFFT model, the FAH

recommends that CMS phase-in the EPMs by first implementing the SHFFT model no sooner than January 1, 2018 and then implementing the cardiac EPM models six months later (no sooner than July 1, 2018). If the final rule is delayed beyond January 1, 2018, we urge CMS to provide hospitals with at least 12 months of preparation time from the date the final rule is finalized for implementing the SHFFT model and an additional six months to prepare for the cardiac EPMs. Of the models under consideration, SHFFT and CJR is the model pair with the most overlap and while we believe it is not advisable to proceed with SHFFT without better understanding the CJR outcomes, there are some operational advantages in beginning with this model. For instance, the Metropolitan Statistical Areas (“MSA”) for SHFFT

8

are already known, increasing the preparation lead time available to hospitals, whereas the AMI, AMI-PCI, and CABG MSAs will not be identified until the final rule is released.2 In addition, some of the patient care redesign process and clinical care pathways will be partially transferrable to some SHFFT patients. Moreover, some of the collaborator relationships already established may be applicable to the SHFFT model, and the selection of high-quality post-acute care (“PAC”) collaborators for SHFFT is more likely to be more informed by the CJR experience than selection of PAC providers for cardiac disease.

In contrast, hospitals could benefit greatly from having additional time to implement the cardiac models. Hospitals without prior experience with CJR, BPCI, or other episode payment models will need extra time to implement the cardiac episode models. As discussed previously, our members’ experience has shown that hospitals need sufficient time to conduct preparatory market analyses, understand the clinical and financial risk of their patient populations, form networks with select physicians and other providers, educate their staff, and establish the needed organizational capabilities to manage payment bundles.

III. The Statute Does Not Authorize CMS to Mandate Provider Participation in the

Proposed EPMs or Other CMMI Models

In the proposed rule, CMS cites its authority under §1115A of the Social Security Act (“SSA”) to implement its episode payment model proposals as well as to modify its existing CJR model. The FAH agrees that section 1115A authorizes demonstration projects carried out by the CMMI and that section 1115A sets forth the scope, parameters and requirements under which those projects are to be implemented. These include requirements for selection of models to be tested as well as a two-phase testing requirement for each model. Under Phase I (§1115A(b) of the SSA), a model is tested in certain geographic areas (but not on a nationwide basis) and under Phase II (§1115A(c) of the SSA), the model may be expanded, including on a nationwide basis, if certain criteria for quality and or efficiencies are met.

As discussed below, the FAH does not believe that section 1115A authorizes CMS to mandate provider participation in the proposed EPMs or other CMMI models. Further, to the extent that CMS has implemented mandatory models, and now proposes to implement the EPMs, under its CMMI authority, we do not believe that these models meet the requirements of section 1115A.

• CMS Lacks the Authority to Mandate Provider Participation in CMMI Models

The FAH has repeatedly expressed significant legal and policy concerns over any proposal to implement a CMMI model under which provider and supplier participation would be mandatory. Notwithstanding those concerns, CMS incorrectly believes that it may require mandatory participation of providers in a CMMI demonstration, as first evidenced by the CJR demonstration and now its current proposal for an EPM Demonstration. The FAH disagrees with CMS's interpretation that §1115A of the SSA provides the agency the authority to mandate

2 In the proposed rule, CMS notes that approximately 50 hospitals in MSAs selected for CJR were excluded from CJR participation due to BPCI overlaps. CMS proposes, however, to include these 50 hospitals in SHFFT.

9

provider and supplier participation in CMMI models. We also believe that any proposed or finalized requirement for such mandatory provider and supplier participation runs counter to both the letter and spirit of the law that established the CMMI and the scope of its authority to test models under section 1115A and make recommendations to Congress for permanent or mandatory changes to the Medicare program.

The purpose of the CMMI is to test innovative payment and service delivery models to reduce program expenditures while preserving or enhancing quality of care, with an emphasis on models that improve coordination, quality, and efficiency of health care furnished to Medicare and Medicaid beneficiaries (§1115A(a)(1) of the SSA). The statute directs the Secretary to select “from models where the Secretary determines that there is evidence that the model addresses a defined population for which there are deficits in care leading to poor clinical outcomes or potentially avoidable expenditures.” (§1115A(b)(1)(A) of the SSA). The law further directs CMS to evaluate each Phase I CMMI model, and only after taking into account this evaluation, if appropriate, the model may continue to be tested in Phase II to expand “the scope and duration”, provided certain requirements are met (§1115A(c) of the SSA), including a requirement for a separate notice and comment rulemaking for any expansion. CMS is required to report periodically to Congress on CMMI models and make proposals for legislative action on models it deems appropriate (§1115A(g) of the SSA).

The language, structure and requirements of section 1115A of the SSA clearly indicate that Congress did not delegate its lawmaking authority to CMS. Under section 1115A, any permanent or mandatory changes to Medicare payment systems must be enacted by Congress after taking into account results of models that have been tested. Congress is the branch of the Federal government responsible for enacting changes to Medicare payment systems through legislation; CMS is granted limited authority under specific provisions of law to make specific changes to those payment systems or to test new models. There is no language in the statute or any legislative history that supports the interpretation that Congress delegated its authority to make permanent changes to the program to the Secretary through the CMMI. In fact, the limited legislative history on this provision indicates the exact opposite. Notably, nowhere does the law expressly state that CMS can make models mandatory.

Again, mandates on providers of services and suppliers are made through individual legislative enactment; section 1115A of the SSA does not grant CMS the authority to usurp the role of Congress with respect to permanent or mandatory changes to the law. Because delegations of lawmaking authority to the agencies may be constitutionally suspect, Congress would have had to include specific statements in the legislation indicating that it both intended to and actually was delegating its lawmaking role to the agency. Any such delegation would have had to include clear standards for the administration of duties to limit the scope of agency discretion as well as procedural safeguards from arbitrariness or abuses. In other words, Congress would have had to specifically permit CMS to require participation of providers of services and suppliers in a model tested by the CMMI in the language of the authorizing statute. CMS may not impute that Congress granted the agency this authority.

The agency's aggressive and incorrect interpretation of the statute raises issues of impermissible delegation of lawmaking authority where none was intended. This is especially true because Congress precluded administrative or judicial review of a substantial number of

10

matters of CMMI demonstration authority under section 1115A(d)(2) of the SSA to permit the testing of models. The waivers of administrative or judicial review require that the scope of delegation to the agency be read in the narrowest terms, meaning that the agency may not infer additional grants of authority absent specific language in the statute. The agency's proposal to mandate participation of providers of services and/or suppliers is precisely the type of aggressive overreach in interpretation that both contradicts the statutory mandate and raises concerns about impermissible delegation of lawmaking authority to the executive branch. Absent specific language in section 1115A authorizing the mandatory participation or providers of suppliers, CMS may not implement a policy that requires such mandatory participation.

CMS has successfully demonstrated that it is fully capable of testing models under section 1115A solely through providers of services and suppliers that volunteer to participate in those models. Experience with the BPCI shows a substantial number and range of providers and suppliers willing to participate in carefully crafted models. Encouraging voluntary participation by providers and suppliers was the intent of Congress in enacting section 1115A and is the proper and appropriate use of legislatively granted demonstration authority. It was the manner in which previous demonstrations were conducted pursuant to section 402(a) of the Social Security Amendments of 1967 (P.L. 90–248), as amended by section 222(a) of the Social Security Amendments of 1972 (P.L. 92-603).

The inclusion of waiver authority under section 1115A(d)(1) of the SSA was intended to provide the agency some flexibility with respect to conflicts among the various provider and supplier payment systems and associated requirements for claims submissions; however, the presence of that waiver authority does not vest wholesale authority in the agency to make the type of policy decisions that are reserved to Congress, such as mandatory participation of all providers of services and/or suppliers in a model being tested in a particular geographic area. No reasonable reading of the statute vests in CMS any authority to take unilateral administrative action to make provider or supplier participation mandatory as part of its authority to test models, either at a regional or national level.

As noted above, CMS must periodically report to Congress on CMMI models and make proposals for legislative action on models the Congress determines to be appropriate using its lawmaking authority. (SSA §1115A(g).) The CJR model and, far more worrisome, the EPM proposal jumps over this process and imposes a mandatory program on affected hospitals. There was no Phase I or Phase II of testing CJR before the model became mandatory nor will there be any substantive testing or assessment by the agency or any review by Congress of the various EPM models that are to be tested in the future. Yet, these programs require mandatory hospital participation which effectively usurps Congress’s discretion to study the results of a demonstration project and use its deliberative lawmaking authority to decide if the model should become the basis for a mandatory change in Medicare policy. The FAH is very concerned with this approach to Medicare payment policymaking and believes that it is contrary to both the language and intent of section 1115A authority. The agency is granting to itself broad lawmaking authority; that authority was never granted to the agency.

Under the CJR and EPM models, hospitals are required to change the way they manage services and will be financially at risk for health care services they do not provide. The intent and impact of the CJR model was clear; it represented a major change in Medicare payment

11

policy. The EPM proposal presumably is a template for a vast number of future payment models (those in this proposed rule as well as others to follow). Thus, CMS has established as a de facto rule that participation of providers of services and suppliers in EPM models is mandatory. This policy mandate would be imposed on providers and suppliers without any testing; this is neither permitted nor contemplated by the language of the statute or any expression of congressional intent. It also fails to account for difference in types of providers or suppliers, or their particular circumstances (for example, size or ability to contract with other providers or suppliers). It also assumes all hospitals will be able properly to manage the risk and patient flow, despite that many hospitals may have little to no previous experience with respect to development of care re-design programs.

• The SHFFT EPM is a Prohibited Expansion of the CJR Model

As noted above, the Secretary must evaluate each model. The evaluation must address (i) the quality of care furnished under the model, including the measurement of patient-level outcomes and patient-centeredness criteria determined appropriate by the Secretary; and (ii) the changes in spending under Medicare and Medicaid due to the model. The Secretary must make the results of each evaluation available to the public and may establish requirements for States and other entities participating in model testing to collect and report information that the Secretary determines is necessary to monitor and evaluate the models.

Taking into account the required evaluation, the Secretary may expand the duration and the scope of a model through rulemaking if all of the following conditions are met:

(1) The Secretary determines that such expansion is expected to— (A) reduce spending under Medicare or Medicaid without reducing the quality of care;

or (B) improve the quality of patient care without increasing spending.

(2) The CMS Chief Actuary certifies that such expansion would reduce (or would not result in any increase in) net program spending under [Medicare or Medicaid]; and

(3) The Secretary determines that such expansion would not deny or limit the coverage or provision of benefits under Medicare or Medicaid. In determining which models or demonstration projects to expand, the Secretary is

directed to focus on models and demonstration projects that improve the quality of patient care and reduce spending.

The proposed SHFFT EPM is a prohibited expansion in scope of the CJR model, which appears to be a Phase I CMMI model, although as discussed above, we do not believe the CJR model has met the Phase I requirements. The SHFFT EPM is proposed to be tested in the same hospitals already chosen for the CJR model and is intended to permit all hip fracture surgical treatment options at the selected sites to be captured within a model. Procedures under the SHFFT EPM may involve many of the same surgeons, clinicians and provider participants as well as similar resource use for an expanded scope of hip fracture surgeries. More specifically, the models address the same patient populations with diseases limited to the hip/proximal femur region. The disease significantly restricts patient mobility and requires a major operative procedure performed only by orthopedic surgeons and usually under general anesthesia. The

12

procedure includes implantation of a medical device or medical hardware and requires inpatient hospitalization and associated post-acute care following hospital discharge, including inpatient and outpatient physical therapy services.

CMS has not evaluated the CJR model as required under section §1115A(b)(4) of the SSA or if it has, it has not made such evaluation available to the public as required under §1115A(b)(4)(B). Further, CMS has not made either of the requisite determinations (under paragraphs (1) and (3) of §1115A(c)) with respect to the CJR model, and the CMS Chief Actuary has not made the certification required under §1115A(c)(2) for savings under the CJR model. All of these steps are required before a model may be expanded in duration or scope, and none of these steps have been accomplished. Before implementing the SHFFT EPM, CMS must first complete the evaluation of the CJR model required under section 1115A(b)(4); make the determinations required under section 1115A(c); and receive the attestation from the Chief Actuary as required under 1115A(c)(2). Because CMS has not met any of the requirements for expansion under section 1115A, the agency should not implement the proposed SHFFT EPM.

• EPM and Other CMMI Models Must Address a Deficit in Care for a Distinct Population

As noted earlier, CMMI models are to be selected from among models “where the

Secretary determines that there is evidence that the model addresses a defined population for which there are deficits in care leading to poor clinical outcomes or potentially avoidable expenditures.” (§1115A(b)(1)(A)). CMS's recent proposal for a Part B Drug Demonstration failed to explain or cite any evidence that the Medicare reimbursement rate of Average Sales Price+6 for Part B covered drugs led to a deficit of care. Stakeholders noted that the demonstration will cause certain physician specialists to cease making Part B drugs available to their Medicare patients due to cost constraints. Those patients will have to seek alternative sites to receive their medicine, and in many cases causing beneficiaries to travel greater distances to access the services than they would otherwise have done absent the demonstration. In other words, for some Medicare patients, the Part B Drug Demonstration itself may result in a deficit of care that does not currently exist. It is also evident that the scope of the Part B Drug Demonstration applies to almost all Medicare beneficiaries, not a limited, defined population.

Similarly, CMS has not examined in any detail the impact mandatory provider participation will have on the hospitals upon which the requirement would be imposed. CMMI authority is intended to test the impact of various innovative payment models on quality and efficiency in the delivery of care for defined populations with deficits of care leading to poor outcomes or avoidable expenditures. Thus, CMMI must first conduct these tests to understand the impact of the models on beneficiary care for these populations as well as on those providers and suppliers that furnish the care. CMS must then report to Congress on the results and permit Congress to determine whether a permanent change to the Medicare program is warranted based on the findings of the model.

The EPM proposal establishes a set of rules and requirements without taking into consideration the special features of each model or whether the model will meet the statutory requirement to address the defined populations. In addition, as noted earlier, these EPM rules

13

and requirements may serve as the basis for an undetermined number of future proposals. Given the past experience of the Part B Drug Demonstration proposal, the FAH is extremely concerned that CMS will not adhere to the requirements of the statute or that it will, as it did in the Part B Drug Demonstration, avoid notice and comment rulemaking requirements.

For example, the Part B Drug Demo required that value-based purchasing (“VBP”) tools be used in 50 percent of the areas, representing roughly half the nation; those VBP tools were to be specified using a subregulatory process in lieu of notice and comment rulemaking. By failing to propose using the Administrative Procedure Act (“APA”) rulemaking before employing the VBP tools in the Part B Drug Demo to afford the public adequate notice and opportunity to comment on those tools, CMS ran afoul of meaningful engagement of the public for proposed regulatory action that will impact an enormous number of Medicare providers of services, physicians, and beneficiaries. Notice on the Website does not constitute adequate notice, and a 30-day comment period is inadequate given the fact many of the proposed tools had only recently been implemented on a small scale in the private sector and there is little practical experience with them.

CMS must adhere to the language and intent of section 1115A authority. That authority precludes requiring participation of providers or suppliers, but at the same time requires that models tested must be for certain defined populations and must satisfy meaningful APA notice and comment rulemaking requirements.

IV. Given the Complexities of These Models, CMS Should Seek Stakeholder Input and

Feedback, and Regularly Make Systemic and Hospital-Specific Mid- Course Program Adjustments As Necessary

As with any demonstration project, the need for constant monitoring, communication

and flexibility is a key to success. In our members’ experience, a collaborative public/private partnership is critical to making care coordination models work, as it is between providers and payers in the private insurance market. One benefit of the proposed EPMs is that they are part of a proposed rule that is subject to notice and comment rulemaking. This means that the public record will reflect, for all participants, as much information as possible to help those who are required to participate prepare and operate effectively within this model.

To foster an open dialogue, the FAH urges CMS to continue to seek stakeholder

input throughout the implementation of these models, using this public notice/rulemaking regularly (at least annually) so that updates, improvements and clarifications are clearly communicated to affected stakeholders. Like BPCI, these new episode payment models are complex and require that the Medicare program and its provider and practitioner partners learn, share experiences, and seek to improve the program over time. This can only happen through transparency and opportunities for public dialogue and feedback.

It is quite possible that these models will not be effective for all hospitals. Unlike the

BPCI program where providers can voluntarily decide to participate, and continue to participate or leave the program, no such opportunity is available in a mandatory program. A variety of fail-safe transition measures should be considered to prevent broad systemic failure in EPM implementation. As all acute care hospitals in the designated MSAs are mandated to participate

14

and be the ultimate bearer of financial risk, it is important to note their differences in being at different stages in care redesign, as well as their differences in size, financial resources, and existing infrastructure that could be tailored to support the bundling efforts.

Given the unique nature of this program, the FAH strongly urges CMS to allow

both CMS and hospitals the opportunities to assess hospitals’ mid-course performance and make corrections as necessary where these EPMs affect the overall provision of care and/or access to care for the geographic regions they serve. These assessments could happen on an annual basis upon having all the data collected and reconciled.

At its core, these EPMs are part of a demonstration project, and if a demonstration

project has negative financial effects that hinder a hospital’s ability to properly care for its community or major systemic design problems are detected after implementation, there should be triggers and steps in place to allow for the program to be suspended for modification or be discontinued. CMS should consider developing a contingency plan should the program not move forward according to plan.

SPECIFIC COMMENTS TO THE STRUCTURE OF EPISODE PAYMENT MODELS

I. Hospitals Need Timely and Relevant Historical Claims Data to Achieve Program Goals and Manage Financial Risk

Because these EPMs require acute care hospitals to be the ultimate bearers of

financial risk, hospitals must be given the tools needed to manage patient care and achieve program goals. Specifically, it is critical that hospitals receive relevant and timely historical claims data, be permitted enough time to analyze the data, and take appropriate action with participant partners on a timely basis. The data must be provided prior to the start of the program, and at regular intervals (e.g., monthly) throughout the program.

To successfully manage risk, hospitals must have sufficient time and data to analyze

and understand the composition, characteristics, and needs of their patient population, as well as the quality of local providers. As indicated by experience with the BPCI models and our members experience with CJR in its initial months, comprehensive management and analysis of data is the foundation for hospitals to redesign and coordinate care, select and form networks with the right partners, and establish the necessary organizational and technological infrastructure to manage bundled payments under these models.

Given our member hospital experience in receiving data from CMS on the CJR model, we have concerns about the timeliness of the data received and its quality. For example, the CJR Final Rule was announced in November of 2015, however, participant hospitals did not receive their performance year claims experience until September 2016. In many cases, our members did not find the data helpful, as it was produced in a “raw” format that was difficult for our smaller hospitals to analyze. Those hospitals that could analyze the data found the data to be incomplete in many cases and not consistent with the hospital’s own data. The FAH urges CMS to work more closely with hospitals to better define the data parameters and

15

the format(s) of episode data that would be most helpful to hospitals and its collaborators. This would allow them to more effectively examine their own cost and quality data, and act on these data to improve the care provided to beneficiaries in a cost-effective manner.

• Data Must Be Available Prior to the Start Date of the EPM Models

The experience of our members’ hospitals in CJR and the BPCI initiative

demonstrate that hospitals must put into place various organizational capabilities in order to manage payment bundles, including:

• Capabilities to manage care delivery and coordination, payments, and

financial risk; • Clinical and administrative infrastructure for care delivery and coordination; • Data analytic infrastructure to manage, analyze, and share claims data in

“real- time”; • Carefully developed affiliated networks of physicians and other providers; and, • Quality data collection and submission to CMS.

To achieve these goals, hospitals must be provided with historical claims data well

in advance of the start date. Consistent with its practice for the CJR model, CMS proposes to provide to an EPM participant, upon request, aggregate expenditure data available for all claims in which the EPM participant is located. Comprehensive analysis of claims data prior to the assumption of risk is a critical step in the preparation process. It serves as a foundation for hospitals to formulate processes and protocols to redesign care, develop networks with physicians, physician groups, and PAC providers, and establish necessary clinical and administrative infrastructure during the pre-implementation period.

Under BPCI and Medicare ACO programs, participants received both historical and monthly (with only a few months data lag) claims data feeds prior to start, and had approximately twelve months from receiving the data prior to enrollment in the program. Inadequate time for preparation and lack of data for preparatory analysis, prior to start, will hinder hospitals’ ability to effectively coordinate and ensure smooth transitions across the continuum of care for beneficiaries. EPM participants must be provided data with at least as much preparatory time as BPCI participants.

• Summary Claims Data Must Be Updated Monthly and Automatically

The BPCI initiative also indicates that ongoing data analysis is a crucial part of

hospitals’ ability to manage care under an APM. Having frequently updated data for analysis, such as trend monitoring and risk identification, lays the foundation for hospitals to understand and manage risk across the full continuum of care under these episodes.

CMS proposes that claims data for EPM hospitals be updated on a quarterly basis.

CMS states it has received requests in other initiatives to make data available on a more frequent basis, and proposes to eventually make these data available on as frequently as a monthly basis if practicable. We urge CMS to follow-through with this plan in an expedited manner. We continue to believe that a quarterly timeline would significantly delay hospitals in

16

identifying inefficiencies arising with regard to beneficiary utilization and spending or other issues that could occur in the continuum of care delivery and coordination. Such a delay in data analysis will hinder hospitals’ capacity to devise and implement strategies for continued process improvement. We appreciate the CMS proposal that hospitals only need to make an initial single request rather than multiple periodic requests for data as this will create less of an administrative burden on hospitals. The FAH urges CMS to provide disaggregated claims data updates on a monthly basis and automatically.

• Availability of Patient Claims Data Is Critical Due to Lack of Availability of Real-Time Clinical Data

Although claims data provided on a monthly basis is critical for hospitals to help

meet program goals of these EPMs, ultimately it can only provide a part of the data picture needed for hospitals to best manage patient care and financial risk. Real-time clinical data is the other part of the picture, and currently the technological capability to make this data available to providers is very limited, and takes significant financial investment. Due to these limitations, it is even more incumbent on CMS to provide hospitals with as much claims data as possible, and as quickly as possible, to help minimize the current gap between claims data and the availability of needed real-time clinical data.

Further, hospitals will face substantial challenges in being able to manage the data and exchange information with hospital partner providers and physicians, which is vital for hospitals in understanding and managing patient “pathways” and clinical/financial risk on a “real-time” basis. Experience under BPCI shows that many providers do not yet have the infrastructure to manage clinical data electronically. The establishment of electronic health records (“EHRs”) could be an important step toward allowing hospitals to manage, analyze, share, and interpret data in their current day-to-day operations in a timely manner—a capability that is essential for bundling success. According to the Medicare Payment Advisory Commission (“MedPAC”), the ability to track data on service use, costs, and payments over time and across settings is necessary for providers to implement bundled payment.3 CMS should recognize that these data challenges – including the significant financial investment, time and complexities involved in developing and using the necessary infrastructure to achieve these goals, along with substantial transaction fees for sharing health information – necessitate a delay in the start of the program.

Additionally, CMS must ensure that the EPMs support the adoption of EHRs, and facilitates making the use of EHRs sustainable for providers with regard to financial costs and administrative barriers currently borne by user organizations.

• The EPM Models Start Date Should Be Delayed Six Months to Allow Providers Time

to Receive and Analyze Claims Data Prior to the Start of the Program

CMS has proposed a start date of July 1, 2017 for the EPM models, which only allows about 8 months for hospitals in MSAs after publication of the final rule that are chosen to

3 Medicare Payment Advisory Commission (2010, March). Report to Congress: Medicare payment policy. (Washington, DC: MedPAC).

17

prepare for program implementation. Providers, particularly those without prior experience, need more time to prepare for implementation of EPM episode payment models. We believe that by delaying the start dates by six months to no sooner than January 1, 2018 instead of July 1, 2017 would allow hospitals sufficient time to conduct preparatory market analyses, understand the clinical and financial risk of their patient populations, form networks with select physicians and other providers, and establish the needed organizational capabilities to manage payment bundles. This would also give CMS time to develop the necessary policy and legal waivers, as discussed below, without which the program cannot succeed.

II. CMS Should Consider Reducing the Episode Period from 90 days

Our hospitals’ experience in both BPCI and CJR reflects that the 90-day episode period is

unmanageable and not feasible for hospitals as they follow the patient through the care continuum. In fact, our experience reflects that after 30 days it becomes difficult for the hospital to track the patient, contact the patient and impact patient behavior. Post 30 days, hospitals find themselves at risk for a great deal more than the anchor admission and often become responsible for chronic care management and for conditions unrelated to the episode of care. In mandating a 90-day episode period, CMS is, in effect, making hospitals managers of population health. Yet most hospitals, today, lack the resources, skill set and infrastructure to engage in the mission of managing population health, and the requirements are much different and much more complex and demanding than what is needed to implement reasonable episode-based payments.

In addition, CMS has also adopted quality and performance metrics that do not align

with the episode length. While CMS is requiring the episode length to extend for 90 days, the quality and performance metrics are based on the much more logical 30-day time period.

Given that CMS is now proposing to mandate an entirely new set of bundles on hospitals

reflecting a different patient population and different care management system, we believe the 90-day episode period is excessive and recommend that CMS reduce the length of the episode period to 30 days.

III. CMS Should Attribute Inpatient-to-Inpatient Transfer Episodes to the Receiving Hospital

CMS proposed special policies for hospital transfers of beneficiaries with AMI. CMS notes that nearly 20 percent of patients sustaining acute myocardial infarctions are transferred from the first facility to which they present for treatment, and most transfers are made to enhance patient access to coronary artery disease revascularization options. To address this issue, CMS outlines a framework within which to consider beneficiary transfers for advanced cardiac care.

• No transfer scenario • Outpatient-to-inpatient (o-i) transfer scenario: treatment as an outpatient (ED) at the first

hospital then transfer to another hospital (inpatient admission) • Inpatient-to-inpatient (i-i) scenario: admission to the first treating hospital with later

transfer to another hospital

18

The FAH is particularly concerned about the inpatient-to-inpatient scenario as CMS proposes an overarching policy in which every AMI or CABG episode would begin (be initiated) at the first AMI or CABG hospital to which the beneficiary is admitted. This initial (sending) hospital will be held responsible for costs and quality performance throughout the episode even after transferring the patient to another hospital. The FAH believe this inpatient-to-inpatient scenario is entirely too complex and unmanageable. Our members are concerned that the initial (sending) hospital may not know which MS-DRG has been assigned to the patient until three to four days after the patient is discharged from the hospital. This limits the opportunity to track the patient and it is very unlikely that the sending hospital has any control over the patient’s care or the resources being used. We also disagree with CMS’s reasoning that gives perhaps too much importance to the role of the local hospital and physicians associated with the initial treatment and too little importance to the role of the hospital providing the vast majority of the care. The receiving hospital from the transfer is much more likely to influence the post discharge care and the post-acute care the patient receives, and is in a much better position to retain financial responsibility for the patient and assume the associated risk. Further, this will cause confusion for beneficiaries who will be contacted by multiple providers trying to implement post-acute pathways which may be contradictory.

We also believe that attributing the episode to the initial (sending) hospital of the transfer could create some unintended and perverse incentives that may be harmful to beneficiaries. Attributing financial ownership and risk to the sending hospital in the transfer could potentially result in patients that present with AMI symptoms in the emergency department being transferred prematurely to another participant hospital or the receiving hospital may be more reluctant to transfer patients in order to retain more control of the episode and its associated costs. While this may be an appropriate response, it’s unclear whether this is the best approach for patients and whether long-term this type of model will reduce the capacity of small and rural hospitals to effectively manage care for cardiac patients and could create an over-reliance on larger hospitals. Consistent with one of the alternatives CMS considered, FAH strongly urges when both hospitals are AMI or CABG model participants: canceling the AMI episode initiated at the sending hospital, assigning the episode model based upon MS-DRGs determined by the receiving hospital, and attributing the episode to the receiving hospital.

While we disagree with CMS’s proposal to attribute the episode to the first participant

hospital in the case of a transfer, we do note CMS’s proposal that if the patient’s discharge MS-DRG from the receiving hospital is not one of the eligible cardiac model MS-DRGs, the episode would be cancelled. The FAH supports this proposal to cancel episodes that include a chained anchor stay, but have a final discharge MS-DRG that is ineligible for the cardiac model. While we support a more streamlined transfer policy as stated above, the experience of our BPCI participating hospitals is that while few episodes fall into this category, their spending is very high having severe negative impact on the first hospital to which the patient is admitted.

19

IV. Appropriate Risk Adjustment is Needed to Reflect Participant Hospital Characteristics and Socio-Demographic Conditions

The FAH continues to be concerned about the choice not to use a standard risk

adjustment approach to adjust for patient-specific clinical indicators or differentiation within a given DRG. While CMS has controlled some variation in target prices among hospitals by creating different target prices by DRG and condition, this may not be sufficient. CMS in its proposal continues to believe that the CMS Hierarchical Condition Categories (“HCC”) used to adjust for risk in the Medicare Advantage program would not be appropriate for risk adjusting EPM episodes. CMS is proposing, however, EPM quality measures that incorporate HCC risk scoring. The FAH has serious concerns about the failure to properly risk adjust and this could be a significant limitation of the proposed approach. These concerns are supported by a recent analysis published in Health Affairs in September 2016.4 The authors found that failure to risk adjust for comprehensive joint replacement episodes produces wide swings in reconciliation payments and that relying just on region-based target pricing led to reduced reconciliation payments to hospitals that treat medically complex patients. Using CMS-HCC risk scores, on the other hand, appears to have controlled for much of this variation and has certain advantages. Specifically, this approach currently is used in a number of other performance programs, can be computed from administrative claims with minimal burden, and factors that comprise the HCC-risk score have independently been shown to affect expenditures.

In addition, the FAH believes that CMS should explore and incorporate, as appropriate, additional risk adjustment to address socio-demographic factors, in order to reflect more accurately the level of financial risk that hospitals have to bear with regard to differences in the population socio-demographic status of the market areas where they deliver care. On hospital readmissions, for example, nearly 60 percent of the variation in national hospital readmission rates was found to be explained by the characteristics of the counties where hospitals are located. Local factors such as income, employment levels and nursing home quality were the major factors underlying county-level variation, or amounts of risk that hospitals could not mitigate in delivering and managing care. 5

Thus, the FAH urges CMS to incorporate a standard risk adjustment approach, such as the CMS-HCC risk scores, to risk adjust EPM target prices. The FAH also urges CMS to examine and consider incorporating other important risk-adjustment variables such as sociodemographic status, as appropriate.

V. CMS Should Use Metropolitan Statistical Areas Instead of Census Divisions to Establish Regional Prices

Further, while the transition from historical to regional prices is an important feature of

the EPM models, the use of the nine census divisions to establish regional prices is too broad, 4Ellimoottil C, Ryan AM, Hou H, Dupree J, Hallstrom B, and Miller DC. Medicare’s new bundled payment for joint replacement may penalize hospitals that treat medically complex patients. Health Aff (Millwood). 2016;35(9):1651-1657. 5 Herrin, J., St. Andre, J., Kenward, K., Joshi, M. S., Audet, A.-M. J. and Hines, S. C. (2015), Community Factors and Hospital Readmission Rates. Health Services Research, 50: 20–39. doi: 10.1111/1475-6773.12177

20

as there can be great variation across health care market areas and other sub-regions within the census divisions. Setting regional target prices by MSAs in which hospitals are located would better account for these differences. The census divisions are too large to allow for true differences across regions, and will reflect too wide a range of patient severity, practice patterns, and availability of specialized services (such as quaternary care), and risk the unintended consequence of over-penalizing hospitals for factors beyond their control. Using metropolitan statistical areas better reflects the health care provided in that area and the use of MSAs is already commonly used for other purposes, such as adjusting for differences in hospital wage levels.

In addition, CMS should consider allowing efficient hospitals the option to transition from historic to area-specific target prices at an accelerated rate. There is precedent for such an option in the Medicare program, as CMS provided Long-Term Care Hospitals (“LTCH”) with this level of flexibility when it transitioned providers to the LTCH Prospective Payment System.

VI. Limits or Adjustments to Participants’ Financial Responsibility Should Be Adjusted to Reflect Hospital Challenges and Specific Market Areas

CMS proposes to test the EPM models for five performance years during which

hospitals, and others providing services subject to the bundle, will continue to be paid according to the Medicare fee-for-service payment systems. However, after the completion of a performance year, the Medicare claims payments for services furnished to the beneficiary during the episode would be combined to calculate an actual episode payment. The actual episode payment then would be reconciled against an established EPM model target price, which is based on a 3 percent discount factor, which would be scaled downward from there to reflect high quality performance. The amount of this calculation, if positive, would be paid to the participant EPM hospital subject to satisfactory quality performance and stop-gain limits. This would apply for PYs 1 through 5. If negative, the participant hospital would be required to repay the difference, subject to stop-loss limits. In the proposal, CMS would be responsible for repaying Medicare when their actual EPM-episode payments exceed their quality-adjusted target prices beginning in the second quarter of PY 2 and extending through PYs 3 through 5 (a nine-month period given that PY1 is six-months.)

• The FAH recommends that a hospital participant should be held harmless against its

target prices for a full performance year of at least 12 months.

While we agree with CMS’s proposal that hospitals not experience down-side risk for a given period of time, we disagree with the length of the downside risk and how the target price is set. If CMS maintains its current schedule, we recommend that CMS hold hospitals harmless for all of PY 2—this would result in an 18 month hold harmless period. If CMS delays the implementation of these EPM models until January 1, 2018, as we recommend, we believe holding hospitals harmless for a 12 month PY 1 period would be sufficient.

• With respect to the discount factor applied to the target price, the FAH recommends that CMS apply a 2 percent discount factor instead of the proposed 3 percent and scale downward from there to reflect high quality performance.

21

As discussed extensively above, the EPM models are mandatory and we know from the experience of our member hospitals participating in CJR, BPCI, MSSP, and the Pioneer ACO Model that participation in the models requires significant and costly up-front investment to develop the legal, clinical, financial and quality infrastructure needed to achieve goals, including technology, data analyses, and development of provider networks. Further, these models will apply to many hospitals with little experience with episode-based payments, and there could be a great degree of variation in episode spending outside the control of the hospital, which is not adequately addressed through risk adjustment. CMS could monitor this program over time and re-propose a target price that reflects hospitals’ experience in meeting the many unique challenges they face in achieving the program’s goals.

• Regarding CMS’s proposed stop-gain and stop-loss limits policy, FAH urges CMS to

consider a different approach that would apply stop-loss limits at the episode level separately for low, medium, and high volume providers rather than at the model or program level.

CMS 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, CMS would determine stop-loss and stop-gain limits, in total, at the participant level by calculating a single weighted stop-loss/gain threshold based on the total spending under each model. Under the second option and the one CMS proposes, CMS would establish 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. In addition, consistent with its approach under CJR, CMS proposes applying a high-payment ceiling when calculating actual EPM-episode payment and when calculating historical EPM-episode payments used to set EPM-episodes benchmark and quality-adjusted target prices. A high payment episode would be an episode with payments two standard deviations or more above the mean calculated at the regional level.

In general, we prefer an option that is applied separately for each of the models rather

than calculating a single weighted-threshold. The FAH urges CMS, however, to consider an option in which CMS would apply stop-loss limits that would be applied at the episode level separately for low, medium, and high-volume providers rather than at the model or program level. The stop-loss limits would also incorporate the high payments episodes’ ceilings that CMS has separately proposed. An analysis performed for the FAH and others raises questions about applying the stop-loss thresholds uniformly across all hospitals, regardless of the number of episodes attributed to each hospital. The analysis shows that variation in spending is too great, particularly for low volume hospitals over a 90-day episode period; this has the effect of shifting health insurance risk to smaller hospitals, which are not in a position to manage such risk (see figure 1).

22

Figure 1: Variation in AMI spending of Low Volume Providers

Source: DHG Healthcare, September 2016.

Given the variation in spending, there is a great potential for “bundles busters”, defined as episodes with negative net payment reconciliation amount (NPRA) greater than 100 percent of the target price. For example, the analysis found that providers with less than 53 AMI episodes per year have significantly higher variability in spending as a percentage of target prices than high volume providers. Their analysis showed that the optimal episode-level stop-loss threshold (percent of target) would vary among low and high volume episodes for the AMI model. For example, the analysis shows that for hospitals with medium volume of episodes (>20 episodes and <=53 episodes), the estimated episode level stop-loss threshold (% of target) should be 175 percent. Likewise, for hospitals with greater than 53 episodes the stop-loss should be 200%. We recommend that CMS exclude hospitals with fewer than 20 episodes, but give these hospitals the option to participate in the program. For hospitals with fewer than 20 episodes that decide to participate, the stop-loss threshold should be 125 percent. The results shown are for illustrative purposes and CMS would need to calculate and stratify by number of episodes separately for each EPM, including the volume at which low volume hospitals should be excluded with a choice to participate. The FAH would be happy to discuss this proposal and analysis in more detail, at the convenience of CMS.

If CMS continues its current approach in applying stop-loss and stop-gain limits rather

than on an episode basis, we recommend that low volume hospital providers be treated in the same manner as rural hospitals and safety net providers. CMS proposes to apply a stop-loss limit of 3 percent in PY2 and a stop-gain limit of 5 percent of episode payments for PY3 through PY5 for rural hospitals and safety-net providers. In the absence of adopting the approach

23

described above, the FAH would support these stop-loss limits as proposed and further recommend that CMS consider extending that provision to small hospitals and those with low historic EPM-episode volume. CMS could provide additional protection for low volume hospitals by defining the measure of low volume to be consistent with the CMS proposal related to using only regional EPM-episode payments to calculate benchmark prices: fewer than 50 SHFFT model episodes, 75 AMI model episodes (MS-DRGs 280-282), 125 AMI model episodes (PCI MS-DRGs 246-251), and 50 CABG model episodes. These providers similarly may have lower risk tolerance, are subject to the results of wide variation in costs and acuity in small numbers of episodes, and are less capable of supporting needed infrastructure to achieve efficiency for high cost episodes. For the same reason, the FAH recommends that the mandate to participate not apply to hospitals that have low historic EPM-episode volume. These hospitals, however, should retain the option of participating in these models.

Moreover, given the differences in provider capacity to manage risk under episode-based payments in a relatively short period of time, as well as wide variation in episode spending that may be outside the control of the hospital, CMS should, after the period in which hospitals are held harmless, more gradually phase-in and reduce the stop-loss limit that is applied to all hospitals. If CMS continues its current approach in applying stop-loss limits rather than on an episode basis CMS should cap the stop-loss limit for hospitals at 10 percent by PY5 instead of the 20 percent proposed for PYs 4 and 5.

• FAH urges CMS to consider eliminating stop-gain limits.

Hospital experience with BPCI and CJR show the considerable investment hospitals

have to make in order to successfully implement these new bundled payment models. The EPMs proposed in this rule, will also require significant resource allocation. We appreciate CMS including a stop-loss limit in the proposed rule and encourage CMS to adjust their proposal with our recommendations, as outlined above, in mind. In addition, we encourage CMS to consider simply eliminating stop-gain thresholds. First, CMS has already built savings into the program by applying a discount factor into the target price. Second, as noted throughout this comment letter, the success hospitals achieve in implementing these bundles depends on making new, significant infrastructure investments, managing risk, and redesigning care delivery for these patients. The potential savings hospitals achieve are needed to help offset these considerable expenses, which will be incurred by providers across the continuum of care and to permit additional investments. CMS is proposing a “demonstration” program, and the incentives need to reflect these goals and help hospitals provide proof of concept. Limiting savings through the application of stop-gain thresholds, however, is counterproductive to the success of this program and the long-term goal of transforming health care delivery and advancing population health.

VII. A Shared Accountability Payment Model for IRFs Would Increase Efficiency and Competition for PAC Services

As CMS moves forward with its mandatory bundled payment programs which places

financial risk on acute care hospitals for PAC spending, it is important to provide payment flexibility to PAC hospitals in order to allow them to achieve efficiencies that inure to the benefit of acute care hospitals that are at financial risk under these bundled payment models. This is an

24

issue that the FAH brought to the attention of CMS in our comments to the CJR proposed rule and one which we reiterate and update here. It is disappointing that CMS missed the opportunity to address this issue as part of the CJR final rule as it would have undoubtedly strengthened the CJR program. Importantly, CMS has an opportunity in this rule-making to act on a recommendation that will help improve the odds of success for these models.

These additional mandatory bundled payment programs will encompass a larger portion of the patients that Inpatient Rehabilitation Facilities (“IRF”) currently treat. Also, acute care hospitals have historically relied on IRF care for a large portion of patients that will be in mandatory bundled payment programs (over 9 percent of CJR and CABG patients and more than 18 percent of SHFFT patients). Options for acute care hospitals to reduce PAC spending are currently limited to encouraging patients to receive PAC in settings that receive lower Medicare payments or encouraging PAC providers that have the ability to reduce payments through efficiencies to do so. Thus, providing payment flexibility to PAC hospitals is important to allow them to effectively compete in a changing environment and to continue to provide beneficiaries with PAC options that best meet their needs.

In this environment, PAC providers such as skilled nursing facilities (“SNF”) or home health agencies (“HHA”) have the ability under existing regulations to modify their practice or utilization patterns in a manner that produces lower Medicare payments for patient care. SNFs can reduce their Medicare payments within the current prospective payment rules by simply providing fewer days of care. In addition, SNFs can also reduce the level of therapies provided, which would put patients into lower-paid Resource Utilization Group categories. Similarly, HHAs can reduce the number of therapy encounters during a home health episode with the result of receiving less Medicare payment.

The second year evaluation of BPCI found that SNFs reduced the amount of Medicare spending for SNF services during an episode of care primarily through reduced length of stay (i.e., reducing the number of days patients were in SNFs). The study found a statistically significant reduction in SNF length of stay both when the SNF was an episode initiator itself as well as when the SNF was a downstream PAC provider for a BPCI participating acute care hospital.6

The purpose of the IRF shared accountability payment model is to provide a similar level of payment flexibility to IRFs in order to reduce Medicare spending for Medicare bundled payment patients, which is not available under the current Medicare IRF prospective payment system (“IRF PPS”). Since episode target prices and performance period spending in Medicare’s bundled payment programs are based on Medicare payments, and because Medicare payments to IRFs are per-discharge (not per diem) and diagnosis based (not therapy based), there is no flexibility for IRFs to reduce their Medicare payments for the benefit of hospitals participating in the bundled payment models, regardless of the cost-efficiencies an IRF may generate. The IRF 6 Dummit et.al., “CMS Bundled Payments for Care Improvement Initiative Models 2-4: Year 2 Evaluation & Monitoring Annual Report”, August 2016

25

shared accountability payment model would allow acute care hospitals to benefit from being able to maintain or enhance their relationships with IRFs under these programs by permitting IRFs to generate Medicare savings for patients attributed to the bundled payment programs.

The proposed approach would enable IRFs to more fully and robustly share in the potential risks and rewards of these bundled payment programs and allow hospitals participating in the bundled payment program to benefit from savings achieved by IRFs under the alternative payment model, which is similar to how acute care hospitals now benefit from SNFs’ reduced length of stay. Thus, this alternative voluntary payment model would permit greater accountability among and between acute care hospitals and IRFs. This approach directly aligns with CMS’s recognition of need for payment flexibility as Medicare reimbursement moves towards alternative payment models and away from fee-for-service.

We have attached to these comments a brief report prepared by Dobson-DaVanzo, which provides more details and analysis regarding a prototype version of such a “shared accountability” payment model.

VIII. CMS Should Provide Appropriate Waivers to Allow Hospitals the Needed Flexibility to Achieve Program Goals of the EPM Payment Models, While Managing Their Legal and Regulatory Risk

• Gainsharing Fraud and Abuse Waivers

As with its bundled payment predecessors, gainsharing stands at the heart of the EPM

proposed rule. It is, undoubtedly, the most critical component of the EPM model, serving to align participating providers’ otherwise disparate financial interests, and creating the potential to realize CMS’s Triple Aim.

Yet, and as CMS is assuredly aware, to facilitate such gainsharing arrangements under the EPM model, FAH members need legal certainty that such efforts will not run afoul of federal fraud and abuse laws. To date, no such legal certainty has been provided. EPM model fraud and abuse waivers of the Stark law and the AKS remain conspicuously absent.

FAH understands that the delay in waivers may simply be the result of CMS’s need to fine tune the EPM model, upon receipt of comments to the proposed rule. Such was the case for CJR. Nonetheless, the FAH would like to highlight for CMS the substantial and significant legal uncertainty this approach creates for its participant hospital members.

Indeed, gainsharing programs are not developed overnight. Rather, they take careful deliberation on the part of numerous stakeholders, involve painstaking drafting of sharing arrangements, and further entail drawn out negotiations with potential gainsharing partners. Simply put, the process takes a significant amount of time, time that CMS does not afford participants when waivers are issued upon release of the final rule.

Accordingly, the FAH urges CMS to put aside its current piecemeal approach to bundled payment fraud and abuse waivers and develop a single, overarching waiver applicable

26

to all gainsharing arrangements under a CMS-led bundled payment program. In the alternative, and as outlined below, the FAH requests that CMS consider a new, bundled payment program exception to the Stark law, or revisit and modify current Stark law exceptions to permit gainsharing under CMS-led bundled payment programs. We recognize that CMS has noted that certain Stark exceptions can apply to gainsharing arrangements, and we agree with CMS’s view. However, to remove any uncertainty for providers and incentivize continued development of innovative models, we encourage CMS to develop a specific Stark exception for CMS-led bundled payment programs.

Bundled Payment Program Waiver

The FAH asserts that CMS should develop a single, overarching waiver (“Bundled Payment Waiver”) of the Stark law and AKS, applicable to all gainsharing arrangements, developed and administered pursuant to the terms of any CMS-led bundled payment program (“Bundled Payment Program”). The Bundled Payment Waiver would apply to CJR, the EPM model, and any future CMS-led, bundled payment programs, with the understanding that CMS could issue program-specific waivers where circumstances warrant a different approach.

The FAH submits that its proposal, as outlined below, would not represent a dramatic overhaul of current bundled payment fraud and abuse waiver processes. Rather, the FAH believes that the development of a single waiver would simply: (a) streamline the process for both CMS and the Department of Health and Human Services, OIG; and (b) create additional legal certainty for program participants.

A. Considering ACO Fraud and Abuse Waivers as a Model

The FAH urges CMS to adopt a comprehensive Bundled Payment Waiver, and outlines potential waiver parameters, below.

The FAH first notes that, in developing its proposed Bundled Payment Waiver, we have drawn heavily upon existing BPCI Model 2, CJR, and proposed EPM model program safeguards. That being said, we also sought to incorporate CMS’s approach to, and the structure of, ACO fraud and abuse waivers. The FAH believes ACO fraud and abuse waivers have achieved a delicate and difficult balance: pairing critical program integrity safeguards with adequate flexibility for program participants.

Thus, the FAH’s Bundled Payment Waiver proposed below reflects an amalgam of what the FAH believes is the best of current CMS coordinated care models: ACO fraud and abuse waiver flexibility, paired with BPCI and CJR program-specific safeguards.

B. Bundled Payment Waiver: Proposed Requirements

The FAH proposes the following requirements for a new Bundled Payment Waiver:

• Any amounts gainshared by a participant hospital are earned by the participant hospital: (a) solely pursuant to the terms of the Bundled Payment Program; and (b) during the term of the Bundled Payment Program, even if the actual distribution or use of the gainsharing payments occur after the expiration of the Bundled Payment Program;

27

• The participant hospital has selected its collaborators: (a) based upon criteria related to the quality of care to be delivered to Bundled Payment Program beneficiaries; and (b) in a manner not related directly or indirectly to the volume or value of referrals or other business generated between the parties;

• The participant hospital’s sharing arrangement with each collaborator is set forth in writing, is signed by the parties, and specifies both the care redesign services to be provided by the collaborator and the Bundled Payment Program compliant gainsharing methodology;

• The participant hospital’s gainsharing methodology is set in advance of any earned amounts from CMS for that specific performance period;

• Any gainsharing payment made to a collaborator by the participant hospital is for actual care redesign services provided;

• The receipt or payment of a gainsharing payment between a participant hospital and its collaborator is not conditioned, directly or indirectly, on the volume or value of referrals or other business generated between the parties; and

• Any gainsharing payment made by a participant hospital to a collaborator is not knowingly made to induce the collaborator to reduce or limit medically necessary items or services to Bundled Payment Program patients under his or her care.

The FAH acknowledges that, and depending on the applicable Bundled Payment Program,

CMS may wish to add or subtract from the requirements of the above Bundled Payment Waiver. However, the FAH suggests for CMS’s consideration, that the core tenets of the waiver would remain the same across all such Bundled Payment Programs. In addition, as noted previously, CMS and the OIG would continue to have the ability to issue program specific waivers, where warranted.

An Alternative: Create a New Stark Exception or Revisit Existing Stark Exceptions

In the alternative to a Bundled Payment Waiver, the FAH suggests that CMS consider revising the Stark law exceptions in order to facilitate, with appropriate program oversight, CMS-led bundled payment gainsharing arrangements.

Accordingly, the FAH proposes a new bundled payment program (“BPM”) Stark law exception, and, in the alternative, a modification to the current Stark law risk sharing exception.

A. The Bundled Payment Program Exception

The FAH urges CMS to develop a new, BPM Stark law exception. The FAH previously proposed a similar exception in response to CMS’s request for comments to the 2016 Physician Fee Schedule Proposed Rule (referred to by FAH in its comments as an “Alternative Payment Exception”).

Pursuant to the BPM exception, to the extent that CMS leads and/or administers a Bundled Payment Program, the provision of direct or indirect monetary remuneration (“Incentive Payment”) by a designated health services (“DHS”) entity to a physician or physician practice group participating in the Bundled Payment Program (referred to collectively, as “Physician”) will be deemed protected by the BPM Exception, provided certain program and patient safeguards are met. FAH proposes such safeguards below. Notably, they are similar to FAH’s Bundled Payment Waiver program safeguards, as detailed earlier in this comment letter.

28

• The Incentive Payment arrangement is set forth in writing, is signed by the parties, and

specifies both the services to be provided and the Incentive Payment compensation methodology;

• Any Incentive Payments made to a Physician, by a DHS entity is for actual care redesign services provided;

• Only those Physicians who meet quality measures established by the DHS entity in advance of the Incentive Payment arrangement are eligible to receive an Incentive Payment; furthermore, such quality measures must be reasonably related to improving quality outcomes for the DHS’ entity’s patient population;

• The receipt or payment of any Incentive Payment is not conditioned by either party on the volume or value of referrals or other business generated;

• Any Incentive Payment made directly or indirectly from a DHS entity to a Physician must not be made knowingly to induce that Physician to reduce or limit medically necessary items or services to patients under the direct care of that Physician;

• The total amount of Incentive Payment that a Physician may receive is capped at 75 percent of the Medicare physician fee schedule for services provided by that Physician to applicable beneficiaries, for a given calendar year;

• The Incentive Payment methodology is set in advance; and • Irrespective of any care redesign measure undertaken, physicians retain the ability to make

decisions that are in the best interest of their patients.

The FAH believes that the scope of the above BPM Exception, the inherent protections that come with a CMS issued program, and the substantial program safeguards outlined above, will ensure Incentive Payment Arrangements evolve consistent with CMS’s program goals to promote transparency, improve quality, and safeguard against payments for referrals.

B. Revisit the Risk Sharing Exception

CMS may also wish to consider modifying the existing risk sharing exception to apply not only to compensation arrangements between a managed care organization or an independent physician’s association and a physician (either directly or indirectly through a subcontractor), but also to Bundled Payment Program arrangements. That is, compensation arrangements between CMS and a Bundled Payment Program participating physician (either directly or indirectly through a downstream contractor, like a hospital), would be protected under the Stark law, risk sharing exception. The FAH notes that it likewise made such a proposal in response to CMS’s 2016 Physician Fee Schedule Proposed Rule.

The FAH believes this proposed modification to the risk sharing exception aligns with prior

statements made by CMS in the preamble to the Stark Phase II regulations. Specifically, in response to a request for clarification relating to the definition of “managed care organization,” CMS stated that it “purposefully declined to define the term ‘managed care organization’ so as to create a broad exception with maximum flexibility.” 69 Fed. Reg. 16054, 16114 (March 26, 2005). It is the FAH’s contention that CMS’s statement regarding “maximum flexibility” serves as a natural springboard to allow for a more expansive risk sharing exception, one that encompasses CMS-led Bundled Payment Program arrangements. Also, and simply from an operational perspective, hospitals – as the leader and coordinator of any Bundled Payment Program – act similarly to that of their managed care

29

organization counterparts. We further note that here, as with the proposed BPM Exception, CMS may wish to consider

instituting specific safeguards to protect against patient and program abuse. We refer CMS to the “The Bundled Payment Program Exception” for our discussion of such program safeguards.

A Second Alternative: EPM Specific Fraud and Abuse Waivers

While urging CMS to develop a single Bundled Payment Waiver or developing a Bundled Payment Program specific exception, in the alternative, should CMS and the OIG issue EPM specific fraud and abuse waivers, the FAH seeks to ensure that any such waivers offer participants sufficient flexibility and are not based solely upon CJR.

The FAH believes that the CJR fraud and abuse waivers are highly technical in nature, and as a result, fail to appropriately incentivize gainsharing. For example, and because the CJR “Waiver for Distribution of Gainsharing Payments and Payment of Alignment Payments under Sharing Arrangements” requires that participant hospitals meet each and every requirement of 42 CFR § 510.500 in order to receive waiver protection, participant hospitals must adhere to the following, technical requirements (among others) in any gainsharing arrangement, or risk foregoing Stark law and AKS waiver protection:

• Collaborator agreements must include “management and staffing information, including type of personnel or contractors that will be primarily responsible for carrying out changes to care under the CJR model”;

• Participant hospitals must update their list of CJR collaborators on “at least a quarterly basis and publicly report the current and historical lists of CJR collaborators on a public-facing Web page on the participant hospital's Web site”; and

• Participant hospitals must keep records of, among other requirements, “information confirming the organizational readiness of the participant hospital to measure and track internal cost savings.”

42 CFR § 510.500. Further, to the extent that a participant hospital’s collaborator may inadvertently miss just one of the plethora of program requirements (e.g., a CJR collaborator fails to provide, in a single instance, a CJR beneficiary with the appropriate notice), the participant hospital likewise risks losing waiver protection and faces the potential for both criminal and massive financial penalties. See § 510.500(a)(4).

The FAH submits that the above requirements, while potentially of import from a program compliance perspective, are not appropriate to include in fraud and abuse waivers. That is, for example, were a collaborator agreement to not include management and staffing information, the FAH believes such an omission poses no fraud and abuse risk to any federal health care program, and accordingly, should not govern whether a participant hospital receives waiver protection.

In sum, the FAH asserts that the CJR fraud and abuse waivers have in fact hindered gainsharing arrangements. FAH members remain concerned that they may lose waiver protection as a result of minor technical infractions that, in reality, are not aimed at protecting against patient and program abuse. Indeed, the behemoth mountain of technical requirements that FAH members must wade through do not appropriately balance CMS’s program integrity interest with the need for

30

meaningful change.

As noted above, and consistent with CMS’s approach to waivers in the ACO context, the FAH urges CMS to take a different approach should it decide to develop EPM specific fraud and abuse waivers.

• Waivers Designed to Facilitate the Provision of Pre-Operative Care Management Items and Services

The FAH recognizes that CMS is not inclined to start the episode of care prior to the date of

the hospital admission. However, we strongly encourage CMS to evaluate the significant benefits to both patients and overall care redesign efforts that could result from more comprehensive pre-operative care management services. Specifically, CMS should consider extending any bundled payment fraud and abuse waivers (including any potential patient engagement incentive waiver) to permit participating hospitals to provide care management tools and services to beneficiaries and providers participating in care redesign efforts, prior to the start of the episode of care. For example, the following pre-episode services have proven to not only improve patient outcomes and satisfaction, but also result in the delivery of more efficient and higher quality care: comprehensive patient evaluations to assess a beneficiary’s overall condition and chronic comorbid conditions, patient education videos and materials, discharge planning review and counseling, home safety reviews, and patient and caregiver education. As further evidence of such services value, FAH notes that the aforementioned activities would be consistent with the activities contemplated by the Medicare Shared Savings Program, ACO participation waiver.

• Gainsharing

Allow Participant Hospitals to Increase the Frequency of Their Gainsharing Payments Under Both CJR and EPM

The FAH believes the current proposal that both CJR and EPM participant hospitals limit

gainsharing payments to “no more than once per calendar year” is too restrictive and creates an unintended advantage for BPCI program participants who distributed payments monthly and quarterly. 42 C.F.R. §§ 512.500(c)(1)(ii); 510.500 (c)(1)(ii). A practice that could adversely impact EPM and CJR hospital participants. Participant hospitals must be able to share savings with their collaborators on a more frequent schedule, such as quarterly.

The FAH acknowledges that CMS considered and ultimately rejected such an approach in the CJR final rule. However, in doing so, CMS placed significant emphasis on operational concerns – namely, that an annual reconciliation process is necessary to (a) limit the number of subsequent reconciliations and potential fluctuation in financial results for participants, (b) prevent the otherwise constant engagement of participants in the reconciliation and appeals process, and (c) align with the CMS-mandated composite quality score process. See 80 Fed. Reg. 73274, 73385 (Nov. 24, 2015).

The FAH does note dispute that the above pose real, operational challenges for CMS. However, they should not be resolved at the expense of an effective gainsharing program. In fact, some BPCI Models allow for monthly gainshare distribution.

Current CJR participant hospitals choosing to gainshare net payment reconciliation amounts

31

are prohibited from making any gainsharing payment until after the annual reconciliation process – a time-consuming process that may take up to 18 months from the start of a performance year. FAH members believe this lengthy process is stifling meaningful change and ultimately is reducing the quality and cost savings potential of the CJR model. Indeed, it is questionable whether any collaborator would be motivated to improve quality and reduce costs, when their potential financial reward is so far removed. Accordingly, the FAH urges CMS to revise both CJR and EPM to permit a quarterly gainsharing payment schedule, consistent with most BPCI models.

In the alternative, the FAH requests that CMS consider a modified gainsharing payment schedule, limiting gainsharing payments to no more than once per performance year for the initial performance year, and then thereafter allowing for quarterly payments. This may serve to alleviate CMS’s operational concerns, while also allowing participants the flexibility to create a more impactful, long-term gainsharing strategy.

Gainsharing Cap

The FAH acknowledges the importance of many of the conditions and restrictions concerning gainsharing payments, as proposed by CMS in CJR and EPM. In particular, the FAH agrees that the total amount of Gainsharing Payments for a calendar year provided to participating physicians should be subject to a cap.

That being said, we urge CMS to (1) reconsider its proposed stance as it pertains to the application of the gainsharing cap to physician group practices (“PGPs”) and (2) relax current gainsharing cap parameters.

A. Remove the PGP Gainsharing Cap

The FAH urges CMS to remove the proposed gainsharing cap for PGP collaborators under both CJR and EPM. While acknowledging that CMS rejected such a suggestion in its CJR final rule, the FAH believes that new facts warrant CMS’s reconsideration.

In the CJR final rule, CMS noted that a PGP gainsharing cap was necessary because CJR had simply one episode – LEJR procedures – versus the multitude of potential episodes in the BPCI Model 2 program. See 80 Fed. Reg. 73274, 73421 (Nov. 24, 2015). As such, CMS believed it was likely that most services furnished to CJR beneficiaries during an episode would be provided by an “identifiable subset of physicians and non-physician practitioners within a PGP.” Id.

The FAH respectfully believes that the above rationale is inapplicable to the EPM model, which applies to not just one episode, but three. Further, treatment of the three EPM episodes – AMI, CABG, and SHFFT – assuredly will not all draw on the same subset of physician expertise. Consequently, services furnished to EPM beneficiaries are likely to no longer be furnished by an “identifiable subset” of physicians, 80 Fed. Reg. at 73421, and participating multi-specialty PGPs will look to draw upon additional members to facilitate care redesign efforts.

As a result of this shift, the FAH believes it is appropriate to allow EPM participant hospitals, as in BPCI Model 2, the opportunity to gainshare with PGPs without gainshare cap limitations. The FAH submits this argument likewise now applies to CJR because (a) a large

32

number of hospitals will be participants in both CJR and EPM, and may seek to gainshare with a single, multispecialty PGP under both programs; and (b) given the inevitable (and, indeed, intentional) overlap across the two programs, uniformity will be imperative.

B. Increase the Gainsharing Cap

The FAH further requests that CMS consider increasing the total amount physicians and/or PGPs may be eligible to receive under both CJR and EPM. CMS has entrusted hospitals with the responsibility to oversee and implement care redesign. Accordingly, the FAH believes that CMS should likewise entrust, and grant hospitals increased flexibility in designing their respective gainsharing programs and determining the amount of savings to share with their collaborators.

That is, the FAH believes that CMS should consider allowing participant hospitals the opportunity to raise the gainsharing cap, i.e., increase the total amount of gainsharing dollars a physician or PGP is eligible to receive. This increase could be accomplished by applying the cap to the total episode savings up to 50% rather than limiting it only to the Medicare physician fee schedule payment. By doing so, the FAH believes CMS will enhance the effectiveness of any participant hospital’s gainsharing program and provide more meaningful financial incentives with limited additional fraud and abuse risk.

Additional Guidance is Needed for the Collaborator, Compliance Plan Requirement

As currently proposed, the FAH understands that all EPM collaborators, like their CJR collaborator counterparts, must have a compliance program that includes “oversight of the sharing arrangement and compliance with the requirements of the EPM.” 81 Fed Reg. 50921, 50794 (Aug. 2, 2016).

The FAH appreciates CMS’s comments that a collaborator’s compliance program need not take any one particular form and further, that there is no “one size fits all” compliance program. Id. That being said, the FAH believes additional guidance is needed.

A requirement that a collaborator include oversight of not only the sharing arrangement, but compliance with the requirements of the entire EPM or CJR program is assuredly a large undertaking for any one collaborator, let alone a solo practitioner. We also urge CMS to consider the practical implications of this compliance plan requirement in the event a participant hospital contracts with a physician individually, and that physician is also a member of a PGP not otherwise involved in the EPM or CJR program.

Remove Duplicative and Technical Requirements from CJR and EPM Sharing Arrangements

The FAH appreciates CMS’s efforts to streamline the CJR and EPM sharing arrangements and believes the proposed rule, if finalized, would reflect significant progress. That being said, and as alluded to in the FAH’s discussion of the CJR fraud and abuse waivers, we believe duplicative and technical requirements remain in sharing arrangements that warrant removal.

First, we note that the regulations lack a clear section laying out each and every requirement to be included in a sharing arrangement. For example, and while the proposed 42 C.F.R. § 512.500(b)(7)

33

(relating to EPM) and § 510.500(b)(7) (relating to CJR), states “The written agreement memorializing a sharing arrangement must specify the following . . . ,” the enumerated list does not capture every item that a participant hospital must ensure is included in the agreement. As a simple illustration of this fact, the FAH notes the following (non-exhaustive list of) sharing arrangement requirements scattered throughout the proposed, regulations:

• Sharing arrangements must require its collaborators and downstream contracts to comply with all applicable laws and beneficiary notification requirements. See §§ 512.500(b)(3), 510.500(b)(3);

• Sharing arrangements must require the collaborator to have an appropriate compliance program. See §§ 512.500(b)(4), 510.500(b)(4);

• Sharing arrangements must specify a participant hospital’s recoupment rights. See §§ 512.500(c)(9), 510.500(c)(9); and

• Participant hospitals must require each EPM collaborator (presumably, via the sharing arrangement) to maintain contemporaneous documentation with respect to the payment or receipt of any gainsharing payment or alignment payment. See §§ 512.500(d)(1)(iii), 510.500(d)(1)(iii).

A comprehensive list of the sharing arrangement requirements, as set forth clearly in any final

CJR and EPM sharing arrangement regulations, would assist participant hospitals in ensuring that they remain compliant with the EPM proposed rule.

Second, the required contents of the sharing arrangement may be overly inclusive. For example, we question whether it is necessary for CMS to mandate that a participant hospital’s sharing arrangement include “management and staffing information, including type of personnel or contractors that will be primarily responsible for carrying out EPM activities.” See proposed 42 § 512.500(b)(7)(iv). Similarly, and while the FAH understands the requirement that all gainsharing payments be only comprised of NPRA and/or actual ICS savings realized, without regard to any “paper savings” from accounting conventions, the FAH questions whether the requirement that gainsharing payments be administered in accordance with generally accepted accounting principles and Government Auditing Standards (“The Yellow Book”) is necessary. See proposed §§ 512.500(c)(15); 510.500(c)(15). (CJR has similar requirements.) The FAH believes such a requirement is overly technical, burdensome, and confusing for physicians, and does not lesson the fraud and abuse risk posed by any sharing arrangement. Finally, and as noted previously, failure of the hospital and/or its downstream contractors to fully satisfy these requirements results in the loss of waiver protection.

Finally, the FAH requests guidance from CMS regarding the impact of the proposed changes to the terminology under the CJR program. Recognizing that CMS’s efforts are aimed at providing consistency between the Bundled Payment Programs, it is unclear whether hospitals will need to revisit existing gainsharing arrangements and modify the terminology to reflect these changes. The FAH would appreciate clarification on this point from CMS and further notes any required revisions to existing arrangements would constitute a significant burden on hospitals.

34

• Patient Choice and Beneficiary Notification

The Development of Preferred Provider Networks Continues to be Hampered by Current Law

In recent years, CMS has increasingly recognized the importance and value of preferred provider networks in care coordinated models. Indeed, in the EPM proposed rule, CMS suggested preferred provider networks may help facilitate both the “coordination of care and optimization of care.” 81 Fed Reg. 50921, 50915 (Aug. 2, 2016).

Yet, and within all coordinated care model, CMS is continuing to require that participants develop preferred provider networks “within the constraints created by current law.” Id. The FAH urges that such patient choice constraints, specifically as set forth in SSA §1861(ee)(2)(H) and 42 C.F.R. §482.43(c), be waived to truly effect change under Bundled Payment Programs, like CJR and EPM.

While we believe that patient choice must continue to be respected, we also believe that CJR and EPM participant hospitals simply require additional flexibility above and beyond that currently permitted.

Presenting the full list of home health and/or skilled nursing facilities to a CJR or EPM beneficiary provides, in and of itself, little value. It is lengthy, offers no information to beneficiaries on the quality of care of such post-acute providers, and may ultimately serve to only confuse beneficiaries when paired with a “preferred provider” list. To that end, beneficiaries would benefit from receiving a “preferred provider” list only, provided that such list was based on objective, quality based metrics appropriately communicated to beneficiaries. In addition, beneficiaries could be informed that a full list of post-acute care providers is available upon request.

The FAH contends that this approach is consistent with and satisfies the statutory requirement found at Section 1861(ee) of the SSA. Specifically, Section 1861(ee) sets forth a hospital's obligations related to discharge planning, and specifically requires that hospitals may “not specify or otherwise limit the qualified provider which may provide post-hospital home health services, and identify any entity to whom the individual is referred in which the hospital has a disclosable financial interest or which has such an interest in the hospital.” SSA §1861(ee)(2)(H). The FAH urges CMS to adopt an interpretation of “qualified provider” that would permit hospitals to develop a more meaningful discharge planning process that will promote better care and patient experience.

In the alternative, and again in the interest of ensuring patients receive high quality care post-discharge, we urge that participant hospitals be afforded the opportunity to exclude from the full list of home health and/or skilled nursing facilities presented to a CJR or EPM beneficiary, certain post-acute care providers with objectively poor quality scores. In this circumstance, the patient’s choice would be respected if he or she expressly requested such a facility, but the hospital would not be required to include the facility in the full list of post-acute care providers in the first instance.

Lastly, the FAH notes that current patient choice requirements may not only serve to confuse beneficiaries, they also may hamper preferred provider network, continuity of care efforts. To the extent a CJR or EPM beneficiary selects a preferred post-acute care facility upon discharge, the discharging participant hospital will be limited in its ability to ensure the EPM beneficiary returns to the hospital in the event a readmission is deemed medically necessary. To facilitate data sharing,

35

improve patient outcomes, and reduce costs, such a process is necessary, yet is currently impeded by both patient choice limitations and AKS concerns.

Beneficiary notice

A. Current Beneficiary Notice Requirements Are Unnecessary

The FAH questions the necessity for a CJR and EPM hospital admission beneficiary notice.

We understand and appreciate CMS’s sentiment, as expressed in the CJR final rule, that “beneficiary notification and engagement is essential because there will be a change in the way participating hospitals are paid.” 80 Fed. Reg. 73517 (Nov. 24, 2015). Yet, historically, CMS has not always engaged beneficiaries in such a manner. Indeed, under the current DRG system – where hospitals are already motivated to contain costs during the hospital stay – CMS imposes no such notice requirements. The FAH believes beneficiary protections afforded by current law are sufficient and that the CJR and proposed EPM notice requirement serve only to burden participating providers and confuse beneficiaries.

B. In the Alternative, Beneficiary Notice Requirements are Duplicative at Best

As CMS continues to move towards its Triple Aim goal and unveil new coordinated care models, participating beneficiaries are being inundated with notices. Indeed, were a beneficiary to participate in BPCI Model 2, CJR, and EPM (a distinct possibility), they would receive at least three different notices from the participant hospital or Awardee, with the potential to receive additional notices from collaborators. Separate and apart from whether any of such notices are warranted in the first place, the sheer volume of notices is duplicative and may ultimately confuse beneficiaries (particularly those beneficiaries that are later determined by a participant hospital at the time of discharge to be outside of such programs).

Accordingly, the FAH urges CMS to develop one streamlined notice applicable to all beneficiaries that may participate in a Bundled Payment Program, like BPCI Model 2, CJR, or EPM, or incorporate this notice into existing CMS notices. For example, CMS could utilize the Important Message notice as the mechanism for conveying this information to beneficiaries. Further, we urge CMS to assume responsibility for such notice processes. CMS, as an objective and trusted voice, could detail the roles and potential conflicts of interest of the multiple Bundled Payment Program participants – e.g., participant hospitals, collaborators, and preferred providers.

C. Hospitals Require Additional Flexibility With Respect to the Timing of Notices

Provided CMS continues to place beneficiary admission notification requirements on participant hospitals, the FAH urges additional flexibility with respect to the timing of such notices is required. We appreciate the new standard proposed by CMS under both the CJR and EPM programs – which allows participant hospitals to provide patient notice as late as upon discharge where, due to the patient’s condition, it is not otherwise feasible to provide earlier notice. See proposed 42 C.F.R. §§ 512.450 (b)(1), 510.450(b)(1). However, and because a beneficiary’s DRG may not be assigned until three days post-discharge, there may be circumstances where a participant hospital is not able to identify a CJR or EPM beneficiary until after discharge. This will be a more likely occurrence in EPM, as it covers episodes of care that involve non-elective, unplanned treatments and transfers to other facilities for higher levels of care. The FAH urges CMS to design a beneficiary

36

notice process that protects the beneficiary without penalizing the hospital for clinical circumstances that are beyond the hospital’s control.

As a result, the FAH respectfully requests that CMS consider a revised timing standard for participant hospital notification, one that requires only “best efforts” prior to the time of discharge. In the alternative, the FAH would appreciate guidance in the event a participant hospital, despite best efforts, is unable to timely deliver notice to a CJR or EPM beneficiary prior to discharge.

• Other Waivers Are Needed to Level the Playing Field Among PAC Providers

Other waivers would remove barriers and help level the competitive playing field among PAC providers, and would furnish these providers with the incentives and tools needed to be able to offer PAC care in a manner that contributes to improved quality and efficiencies, while containing costs. EPM episode costs will vary dramatically depending on the PAC placement of the patient following the acute hospital stay. Many of these cost differences, for what could be essentially the same types of patients, may be due more to the “siloed” nature of Medicare’s PAC payment systems and conditions of participation (“COP”) requirements, rather than a reflection of efficient patient treatment rendered by providers.

As CMS moves forward with the EPM model and other APMs, it should provide strong

incentives in the form of regulatory waivers for the clinically appropriate and cost effective placement of patients into PAC settings, and allow PAC providers to compete fairly with one another on the basis of costs and quality. For example, as discussed earlier, CMS should provide waivers to allow pricing flexibility for IRFs, which are paid on a bundled payment basis, so that IRFs can compete on an even playing field with other PAC providers who are paid on a per diem basis. IRFs are well-suited to help acute care hospitals succeed under these models through their high-quality outcomes (including lower readmission rates compared to SNFs), ongoing medical management of patients, constant nursing care, and goal-oriented approach of restoring patients’ functional deficits or impairments

Further, existing COPs and other regulatory requirements restrict fair competition

across PAC providers. They would not only adversely impact hospitals’ ability to manage patient care in terms of spending and quality, but would negatively impact patient referral patterns and patient access to clinically appropriate care in certain types of settings.

One example, already noted, is the 3-Hour Rule for IRFs. Another, also described

above, is the 60% Rule, which should be modified for these models, especially for those IRFs that participate in the shared accountability payment model.

CMS needs to consider the effects of the 60% rule on IRFs in this environment, and

take appropriate steps to ensure that the program’s effects do not have unintended negative consequences for IRFs and patients who need their services. This rule has historically functioned to distinguish IRFs from acute care hospitals and other PAC providers in a fee-for-service environment where the post-acute care “payment silos” and the rules and regulations governing those silos function in isolation, without the dynamic effects of care

37

coordination/collaboration and accountability for expenditures incurred well beyond a prior hospital episode. However, as Medicare moves away from the traditional fee-for-service model and toward bundled payment and similar models, the IRF 60% rule needs to be appropriately modified –and ultimately, dispensed with – so that IRFs remain a viable component of our healthcare delivery system.

• CMS Should Expand Certain Proposed Waivers, Including the SNF 3-Day Rule

Proposed Waiver

The FAH recommends refinement and expansion of some of the EPM proposed waivers. For example, waiver of the SNF three-day rule only applies to the anchor hospitalization, is limited to episodes in the AMI model, and does not apply throughout the 90-day episode. Further, this waiver begins in the second quarter of PY2 (April 1, 2018), rather than in the first year. The SNF 3-day waiver should be expanded to incorporate all of the proposed EPM models including CABG and SHFFT, apply throughout the 90-day episode and apply at the inception of this program (i.e., as currently proposed this would include PY1 and first quarter of PY 2).

Hospitals and other providers need the tools necessary to maximize efficiencies and cost savings, as well as quality. Limiting the 3-day SNF waiver to the anchor episode is problematic if a patient is re-admitted to a hospital during the 90-day episode and subsequently needs SNF care. This could result in reduced quality of care or substantially increased costs for the patient if the patient cannot receive coverage for SNF care, or has to pay out-of-pocket for this care. We also do not believe it is necessary to limit the waiver only to the AMI models, as physicians should have the discretion to determine the most appropriate treatment and place of service for beneficiaries.

Further, the waiver should apply to the beginning of the program so that hospitals and other providers understand the applicable rules; otherwise, if the rules begin changing mid-stream, i.e., in PY2, this creates too much confusion. More importantly, although hospitals are not at risk for repayment during the first year, they do have the opportunity to share in any upside savings in PY1, and thus should have the full range of tools needed to create as much savings as possible during the first year, while also improving quality and efficiencies.

CMS Should Modify the SNF 3-day Waiver Beneficiary Protections to Ensure that Hospitals Receive Timely Information on SNF’s Quality Rating and Beneficiary Eligibility

CMS proposes that if a participant hospital discharges a beneficiary without a qualifying

3-day inpatient stay to a SNF that is not on the published list of SNFs that meet the CJR SNF waiver quality requirements as of the date of admission to the SNF, then the hospital will be financially liable for the SNF stay if no discharge planning notice is provided to the beneficiary, alerting them of potential financial liability. If the participant hospital provides a discharge planning notice in compliance with the revised requirements of §510.405(b)(4), the participant hospital will not be financially liable for the cost of the SNF stay and the normal Medicare FFS rules for coverage of SNF services will apply. CMS proposes to implement this provision for CJR episodes beginning on January 1, 2017 and April 1, 2018 for AMI episodes.

38

The FAH believes this proposal is unreasonable, as this puts too much of the burden on hospitals, with the potential for significant hospital penalties, despite that the information needed to identify beneficiaries to whom a notice must be provided may not be available.

Specifically, the proposal places too much burden on hospitals to continually check the list of qualifying SNFs to ensure that they meet the three stars or higher quality rating, as this list can change, and to ensure that the discharge planning notice distribution is targeted specifically to those patients in an EPM bundle. Further, it is extremely difficult for participant hospitals to timely identify bundled payment beneficiaries. Often, participating hospitals do not know that a beneficiary is in fact a bundled payment or shared savings beneficiary until post-discharge. Therefore, it would be extremely unfair to penalize participant hospitals for the cost of a particular beneficiary’s SNF stay. CMS is in a better position to update the list of eligible SNFs, and as such, we urge CMS to assume this responsibility for providing the beneficiary notice, as an objective, informed and trusted voice in this process.

If CMS moves forward with this proposal, at a minimum, we urge CMS to provide a list of eligible SNFs to hospitals on a quarterly or periodic basis, and not rely on hospitals to constantly check the website to ensure that the status of a particular SNF has changed. Because beneficiary eligibility also can change, we support the concept that participant hospital knowledge of beneficiary eligibility for a given EPM model should be determined by Medicare coverage status at the time the services under the waiver were furnished.

Finally, we reiterate our earlier comments regarding the need to revise patient choice requirements, as this would help address the proposed SNF 3-day waiver policy. With this proposal, CMS in essence is mandating that certain Medicare beneficiaries select high quality providers. Participant hospitals in bundled payment programs can more effectively assist CMS in this policy goal, were the patient choice rules revised.

IX. CMS Should Revise the Quality Framework Proposed under these EPM Models

The FAH has multiple, serious concerns with the quality framework for the

proposed Episode Payment Models (EPMs). The FAH concerns relate to quality measure relevance, measure overlap, measure misalignment, measure gaps, composite scoring methodology flaws, risk adjustment and risk stratification, data availability, and attribution of responsibility for quality.

• Measure relevance

In a value-based healthcare delivery model, payment is adjusted to reflect the quality of care delivered under the model. As such, the quality measures used for adjusting payments should have clear links to the condition or treatment upon which the model is focused. In the EPMs, however, CMS proposes to require the reporting of several measures with, at best, tenuous linkages to their associated models. First, CMS has proposed the Hospital Consumer Assessment of Healthcare Providers and Systems (“HCAPHS”) Survey as a required measure for all three new EPMs (AMI, CABG, and SHFFT). HCAPHS is very familiar to the provider community as a general measure of patient experience and satisfaction during an acute care hospitalization, but there is nothing about

39

HCAPHS that specifically addresses patient experience and satisfaction from the perspectives of patients having acute myocardial infarctions, undergoing coronary artery surgery, or being treated for hip fractures. These are the very patients, however, who would be receiving their care under the proposed EPMs. Further, unless these patients choose to seek care for these conditions well outside of their home communities, these patients will have no option but to receive care under the proposed mandatory MSA-based EPM models. Therefore, the FAH encourages CMS to consider substituting CG-CAHPS for HCAHPS in the EPM model. Experience within ACO programs indicates that CG-CAHPS better target attributed patients than HCAHPS. The CG-CAHPS are more targeted due to the connection between the patients’ care and the physicians and care intervention programs Secondly, one of the two clinical outcome measures proposed by CMS for the AMI model is AMI Excess Days (Excess Days in Acute Care after Hospitalization for Acute Myocardial Infarction). This measure is not NQF-endorsed. In the proposed rule discussion of this measure, CMS states “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”. Absent clear evidence as to what number of ED, observation, and/or readmission days are in fact “excess” and preventable rather than clinically justified as appropriate, inclusion of this measure seems to be all about costs of care rather than quality of care. In addition, the measures assess unplanned readmissions, emergency department services and observation days but does not look at the cost of care. The FAH does not support the use of this untested measure in this payment model. Thirdly, the sole clinical outcome measure in the CABG model, CABG mortality, (Hospital-level 30-Day Risk-Standardized Mortality Rate (“RSMR”) following Coronary Artery Bypass Graft Surgery), already is part of the Hospital Inpatient Quality Reporting program (“HIQR”). By using it in this program, a hospital will be measured on this measure twice in two different payment/reporting programs. The FAH strongly recommends that CMS refrain from duplication of measures in multiple programs. The FAH encourages CMS to rethink the measures proposed for new CABG model since the measures currently proposed do little to characterize the quality performance of CABG model participant hospitals beyond what it already being done through existing CMS programs. CMS is already measuring mortality after CABG, perhaps, CMS could consider developing a measure of morbidity in terms of major complications after CABG, analogous to the complications measure used in the CJR model. Utilizing one or more individual complication measures that are known, serious, potential complications of CABG operations, such as deep sternal wound infection or acute renal failure possibly could add value. Finally, neither the two required nor the one voluntary measures proposed for the SHFFT model have any specificity for the hip fracture patient population. While the FAH appreciates the administrative simplification offered by identical measure profiles for the CJR and SHFFT models, the trade-off of accepting measures for SHFFT that have nothing to do with the model under trial is of highly questionable worth to the Medicare program in its quest for value over volume.

40

• Measure overlap Several of the proposed EPM quality measures currently are utilized in one or more CMS hospital payment adjustment programs (e.g., see CABG mortality comments above). The FAH appreciates the attempt by CMS to utilize measures familiar to the provider community. However, there are potential unintended consequences of repeated use of the same measures in multiple Medicare hospital payment programs as reported last year in Health Affairs. 7

• Measure misalignment CMS proposes a bundled care episode that extends 90 days after hospital discharge for each of the new EPMs. The duration of this bundle presumably allows for capture of nearly all of the care typically delivered for the targeted disease/treatment of each model and incentivizes hospital-led collaboration across sites of service to reduce costs and to improve quality. However, the two clinical measures proposed for the AMI and CABG models, (AMI mortality, CABG mortality) are 30-day measures. Such misalignment creates potential issues such as how to generalize the results to the 90-day EPM episode when planning cost and quality initiatives and how to publicly explain the 30 day results for a 90-day episode in a way that helps inform beneficiary choices about their care.

• Measure gaps As noted above, major morbidity after CABG surgery is not addressed by the proposed CABG model. Similarly, major complications after AMI treatment (with or without percutaneous coronary intervention, PCI) are not addressed in the AMI model; instead, only mortality is measured, duplicating a measure already in use in other CMS hospital payment programs. None of the proposed measures for any of the new EPMs reflect post-acute care (PAC), despite the 90-day post-discharge episode duration during which a substantial majority of EPM patients will receive services from at least one type of PAC provider. The lack of attention to the interface between PAC and acute quality paints an incomplete picture of the services a patient receives, and does little to encourage provider collaboration across the care continuum.

• Composite scoring methodology flaws Hospital Compare data demonstrate that the overwhelming majority of hospitals (96-97%) perform at levels indistinguishable from national averages for all of the three NQF-endorsed clinical outcome measures proposed for EPM use (AMI mortality, CABG mortality, THA/TKA complications), so it is unclear that the proposed performance deciles describe importantly different quality achievement levels. A composite score that converts these deciles to points then uses very small incremental point differences to define four performance categories for payment adjustments to the effective discount factor (Below Acceptable, Acceptable, Good, and Excellent) as proposed by CMS for all three EPMs. This methodology

1Charles N. Kahn III, Thomas Ault, Lisa Potetz, Thomas Walke, Jayne Hart Chambers and Samantha Burch, “Assessing Medicare's Hospital Pay-For-

Performance Programs And Whether They Are Achieving Their Goals”; Health Affairs, 34, no.8 (2015):1281-1288.

41

imputes an unjustified precision to these measures, given their national performance distributions. Another flaw in the proposed methodology arises from the 3-year rolling average performance period used for these measures. It would not be until performance year 3 (PY3), well into the downside risk period of the proposed models, that the 3-year rolling average actually reflects performance under the model rather than historical, pre-EPM performance. Additionally, the use of a rolling average blurs recognition of improvement by a hospital from its baseline at the inception of the model, when summative, sustained improvement over the course of the model should be the desired goal. Further, the proposed reward for year-over-year improvement is very small -- at best, one point for the most heavily weighted measures -- and offers limited motivation for undertaking complex, costly care redesign. Finally, CMS proposes that additional composite score points can be accrued through submission of voluntary measure data. Both of the voluntary measures are hybrid measures that require sophisticated health information technology resources to document and to submit. These hybrid measures have not been proven to be effective, are not endorsed and are new to hospitals. The playing field for quality rewards should be level across EPM participants.

• Risk-adjustment and risk-stratification Risk adjustment and risk stratification in the proposed models are limited and inadequate. Risk adjustment is done only for age, sex, and comorbidities present at admission (as captured by CMS Hierarchical Condition Categories, HCC), by virtue of being built into the three NQF-endorsed EPM clinical outcome measures (AMI mortality, CABG mortality, THA/TKA complications). No adjustment reflects the population differences inherent between patients undergoing totally elective, discretionary operations (THA/TKA) and urgent operations for major illnesses that present acutely (hip fracture fixation, urgent CABG). For example, it seems intuitive that patients sustaining acute, major illnesses often associated with functional status declines (e.g., acute myocardial infarction, hip fracture) will assess their experiences of care quite differently from patients undergoing discretionary or planned treatment (e.g. THA/TKA, non-emergent CABG). Further, CMS once again fails to propose any adjustments for socioeconomic/demographic status (SDS factors). The FAH has commented to CMS on numerous occasions on the need for SDS adjustment and continues to strongly advocate for appropriate SDS adjustment for all outcomes measures. The FAH also notes that without robust risk adjustment, the proposed precedence for BPCI episodes over EPM episodes may incentivize steering of lower-risk patients to BPCI and higher-risk patients to EPMs for cardiac care and lower extremity joint replacement. Risk-stratification is captured only through MS-DRG assignment (i.e., with or without major complications), a methodology designed to reflect typical costs of care rather than quality, and no attention is given to individual, patient-specific characteristics (e.g., delay in seeking treatment, extent of myocardial infarction, pre-existing osteoporosis) that may significantly influence clinical outcomes. The financial pressures associated with new bundled payments may combine with this limited risk-stratification to create an unintended impetus to MS-DRG upcoding.

42

• Data availability Prior to EPM implementation, participant hospitals need full access to their historical quality data, some of which is available to them only through CMS (e.g., 90-day THA/TKA complications). In the proposal, CMS states that data release will precede EPM start dates. Meaningful, collaborative, quality improvement initiatives do not happen overnight, and EPM implementation should not be undertaken until hospitals have had sufficient time to analyze and act upon their data. Further, quality improvement programs are most likely to succeed when frequent, actionable feedback is provided to program participants. Participant hospitals should be provided with automatic performance updates at least quarterly. For hospital systems, data should be provided at the individual subunit and the system-wide levels. Additionally, CMS has proposed to use the HCAHPS Linear Mean Roll-up (“HLMR”) score for scoring hospitals’ performance. Currently the HLMR score is available to hospitals only through the annual preview reports released to hospitals on Quality Net (“QNet”). The HLMR score is not made publicly available. CMS must release this value publicly in order to allow hospitals to not only know their HLMR score, but that of other hospitals in order to understand their percentile levels. These data should also be released automatically and at least quarterly to facilitate hospitals’ ability to rapidly improve their performances and assess financial risks. Finally, the proposed rule says relatively little about public reporting of EPM-related performance data. The value of quality data is maximized when easily-understood data are made readily accessible to beneficiaries and their families as they make choices about where and by whom their health care will be delivered. The data also must be reliable, and there must be a transparent process through which hospitals can preview and offer corrections to CMS-provided data before the data are reported on Hospital Compare.

• Attribution of responsibility for quality CMS notes in the proposed rule that the asymmetrical distribution of cardiac care resources (e.g., cardiac catheterization suites, cardiac surgeons) is associated with patient transfers for care of acute myocardial infarction in about 20% of the cases. CMS terms AMI episodes that include inpatient-to-inpatient transfers as chained hospitalizations, describes a complicated system for episode attribution to a single hospital in the chain (most often the initial treating hospital), and proposes to include these transfer episodes in the AMI model, even though the discharge MS-DRG from the receiving hospital might be based on PCI or CABG rather than AMI. Attribution for quality performance appears to follow the rules for financial responsibility attribution, creating a situation in which a hospital with limited cardiac care resources potentially becomes responsible for quality throughout an episode of care, much of which was advanced cardiac care delivered elsewhere. While CMS notes that “in a chained anchor hospitalization... 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”, little operational detail is provided (e.g., to which hospital the AMI Excess Days will be attributed). CMS does go on to state that “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

43

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.” It seems highly counterintuitive for an initial treating hospital with limited cardiac care resources to be held accountable for a patient’s death when that patient may have subsequently undergone high-risk procedures and/or had a lengthy hospitalization at the receiving hospital. Further, episode attribution to the initial treating hospital has implications for quality improvement through collaboration with PAC providers. The initial treating hospital would be the entity permitted to establish “sharing arrangements” with PAC providers that could specify shared quality outcome goals, yet discharge planning for a chained hospitalization will be driven by the discharge planning process of the receiving hospital. The latter is likely to be unfamiliar with PAC providers with whom the initial treating hospital frequently interacts, and the proposed rule offers no incentive to the receiving hospital to make the extra effort to work with the initial treating hospital during discharge planning. Also, since the initial hospital has limited cardiac care resources, it seems likely that its local PAC providers are also less experienced in providing care for patients who have required more advanced inpatient cardiac care. Given the foregoing concerns, the FAH believes that major revisions to the quality framework and/or the EPM models themselves are needed before any of the EPMs are implemented. First, the time available before the proposed July 1, 2017 implementation of all of the new EPMs is clearly insufficient to resolve the identified quality framework issues through a process that appropriately takes into account the input of all stakeholders. The FAH believes that that quality issues outlined above support our recommendation that CMS delay EPM implementation until no earlier than January 1, 2018. While the FAH supports the evolution of the Medicare program from volume to value, the EPM quality framework as proposed is inconsistent with a true value-based payment system.

The FAH recommends that CMS consider an EPM approach similar to that taken previously by CMS to MSSP ACO quality reporting. In the first year of the EPMs, as was true for MSSP ACOs, there should be no penalty for quality performance, and EPM participants should only be accountable for reporting necessary quality data in a timely and accurate manner and establishing internal systems for analyzing quarterly claims reports to be received from CMS. Instead, EPM participants can use their PY1 quality results as an indication of whether they are at, below, or above the national average. If an EPM participant is below the national average, it could be required to work with its local quality improvement organization (“QIO”) to implement a plan of action for improvement. During PY2, results would be compared with PY1 and participant hospitals performing at or above the national average would be able to receive a full reconciliation payment. If a participant hospital that fell below the national average in PY1 followed its plan of action and improved, even though it did not reach the national average, it should be eligible to receive a portion of the reconciliation payment it otherwise earned. However, if a participant hospital fell below the national average in PY1, and did not follow the plan of correction established by their local QIO, its reconciliation payment would be reduced. This would allow PY1 and PY2 to serve as a testing ground for EPM participating hospitals and would encourage individual improvement. There is insufficient congruity between the existing CJR model (itself still in its first year of implementation) and the proposed cardiac care EPMs to think that preparations made by hospitals for the CJR will transfer seamlessly to the new EPMs and will be sufficient to enable

44

the likelihood of EPM success. An approach similar to that of the well-established MSSP ACO program offers a pragmatic solution to the proposed EPM composite scoring methodology issues.

The FAH strongly believes that CMS must address measure modification to

maximize the chances that the EPM models are successful. While, as stated above, the FAH does not support moving forward with the SHFFT model, should CMS choose to do so, to achieve face validity with the provider and beneficiary communities, the SHFFT model must incorporate one or more measures relevant to hip fracture patients. Modification to the 30-day duration of the AMI mortality and CABG mortality measures to align them with the 90-day EPM episode duration should be undertaken; alternatively, consideration should be given to shortening the AMI and CABG episodes. Measures should be sought to fill the identified gaps such as CABG complications and PAC functional outcomes. Measure overlap with other CMS hospital payment programs should be limited or avoided to reduce the resulting financial double jeopardy. If the HCAPHS Survey measure is to be used, CMS should filter the results by MS-DRGs or ICD-10 diagnoses and use only results matched to the EPM patient populations. Consideration could be given to using other CAPHS versions such as the surgical CAPHS (S-CAPHS). HCAPHS survey results should not be included for patients in the AMI model having chained hospitalizations, since patients’ responses are more likely to reflect care at the discharging hospital rather than at the initial treating hospital that bears the quality responsibility. In keeping with prior comments, the FAH recommends that the THA/TKA PRO measure should not be included in the EPM program until this complex and complicated measure has been significantly refined to streamline the associated reporting burden.

The FAH once again strongly endorses the application of risk adjustment for SDS

factors. Such adjustment is particularly important for small and rural hospitals and those serving vulnerable populations, the same group that is disadvantaged in reporting the proposed voluntary measures. Socio-demographic status (SDS) adjustment and stratification are also vitally important tools for accurately assessing health care provider performance for fair and transparent public reporting. Further, SDS adjustment is critical for measures that address readmissions, such as AMI Excess Days, as evidenced by a recent study reporting that nearly 60 percent of the variation in national hospital readmission rates was explained by the county in which the hospital is located rather than hospital characteristics. Local factors such as income, employment levels, access to services and nursing home quality were the major factors underlying county-level variation. 8

Attribution of hospital responsibility for quality performance during a chained hospitalization must be modified to exclude care outside of the responsible hospital’s control. The AMI mortality measure should not be utilized for chained hospitalizations. Alternatively, the AMI model should exclude inpatient-to-inpatient transfers, or episode initiation and attribution should be limited to the receiving hospital.

8 Herrin, J., St. Andre, J., Kenward, K., Joshi, M. S., Audet, A.-M. J. and Hines, S. C. (2015), Community Factors and Hospital Readmission Rates. Health Services

Research, 50: 20–39. doi: 10.1111/1475-6773.12177

45

X. CMS Should Ensure a Level Playing Field to Avoid Market Distortions and Inappropriate Patient Steering

CMS believes that it is important to simultaneously allow beneficiaries to participate in

broader population-based and total cost of care models, as well as models that target specific episodes of care. The proposed rule suggests that, given the overlap in CMS APMs that reward providers with shared savings, an order of precedence must be established to determine which program savings are attributed to which program participants (and to avoid double-counting). Entities that participate in shared savings programs and total cost of care models stand to benefit, at least in part, from the cost savings that accrue under these episode models.

BPCI is an example of a care redesign model that has the potential to overlap with these

models. According to CMS, ensuring that BPCI and EPM models do not overlap allows CMS to accurately apply the payment policies in both models. The FAH is concerned, however, that cases in which BPCI takes precedence over CJR and these EPMs, there exists the potential for patient steering. As CJR and these EPM models are mandatory for hospitals, giving precedence to voluntary BPCI participants, including physician group practices, could create market distortions by incentivizing BPCI participants to select lower severity cardiac cases to initiate in a BPCI episode while leaving high severity orthopedic and cardiac cases to be initiated under CJR and the EPM models in the hospital. Better risk adjustment, which we recommend, might be a mitigating factor, but would not eliminate the financial risk to hospitals participating in the mandatory models. In addition, such patient selection behaviors would distort an evaluation of the program and undermine its ultimate success. The FAH urges CMS to ensure that CJR and these EPM models are on a level playing field with BPCI and other similar programs, for example, by granting appropriate program waivers, as discussed above. Further, once a level playing field is established, these EPM models and CJR should take precedence over BPCI cases to avoid market distortions and patient steering.

Given CMS’s quick pace in developing and implementing new models of care, it is also

imperative that CMS understand the impacts the models have on each other. While it is understandable that CMS would venture to test varying care delivery models, the efficacy of each of those models will hinge on how the model is impacted by other CMS models. It is important that CMS not unintentionally undermine one model due to a lack of understanding of model overlap and impact.

To that end, the FAH believes that CMS could improve the success of its ACO models

by allowing ACO participating hospitals to opt-out of the mandatory bundled payment programs. Given the resources required to participate in any of CMS’s models, voluntary exclusion by the ACO participating hospitals would allow them to devote resources to success in the ACO model.

XI. Rural Hospitals Need Special Considerations

FAH urges CMS to take into account the unique needs of rural hospitals―for purposes of

this program this would include hospitals in a rural census tract within an MSA or one that has been reclassified to rural. Overall, rural hospitals serve 23 percent or 72 million of the US

46

population,9 who tend to be older, have lower incomes,10 and are more likely to suffer from chronic illnesses than the urban population.11 In addition to providing care and maintaining patient access to care, rural hospitals also play an important role as economic engines in rural areas, especially as a major source of employment, particularly highly-skilled jobs in in the communities.12 However, rural hospitals typically are much smaller and have less capital than urban hospitals. The constraints of being small in size, with a smaller patient base and sicker patients, limit rural hospitals’ ability to invest in infrastructure, technology, and workforce development, as well as weather financial fluctuations.

Given the challenges and difficulties that rural hospitals would face in payment reform as well as their important social and economic roles in the communities under the MSSP, a special program – the Advance Payment ACO Model – was designed to facilitate their participation in the reform program. This program offers special upfront payments and incentives to rural hospitals that have less capital or less access to capital, to address difficulties associated with the fixed and variable costs needed upfront for participating in an ACO. It is important that CMS consider similar incentives or programs to protect rural hospitals and facilitate their participation in the EPM models and others that may be introduced. These incentives could be in the form of upfront payments to offset partly their needed investment in infrastructures and technology.

In addition, we strongly suggest a broader waiver of the 3-day Rule for rural hospitals

than CMS has proposed. Operating in rural communities, in addition to serving a sicker, older, and smaller population, limited availability of PAC providers in these communities is another constraint faced by rural hospitals – for example the absence of a SNF with a 3-Star or above quality rating – which could further hinder their ability to move patients and structure care along the continuum. Waiving the 3-day SNF Rule also should be applied to small hospitals, as these hospitals face similar difficulties as rural hospitals with regards to patient mix, patient volume, and capital, and therefore would have lower risk tolerance and less infrastructure and support to increase efficiency under bundling.

XII. CMS Should Make Reconciliation Payments on a Quarterly Basis

CMS is proposing annual reconciliation payments to hospitals. The FAH recommends more frequent reconciliation payments, such as quarterly. This will provide hospitals with the necessary cash flow and legal footing to provide more frequent financial rewards to downstream providers, which, in turn, will maintain incentives on a more consistent basis to promote care coordination, with improved quality and efficiencies. An annual reconciliation payment schedule, as currently proposed, will hamper care redesign efforts and undermine the very purpose of gainsharing.

9 US Census Bureau. (2009) American Community Survey. 10 US Census Bureau. (2009). American Community Survey. 2008 and 2010 Annual Social and Economics Supplements. 11 Gamm, L.D., et al. (2010) Rural Healthy People 2010: A Companion Document to Healthy People 2010, Volume 1. 12 Doeksen, GA. And Schott, V. (2003) Economic Importance of the Health Care Sector in a Rural Economy. Rural and Remote Health.

47

XIII. Beneficiary Notification Should Be Permitted on an Electronic Basis

CMS proposes that upon an applicable beneficiary’s admission for an anchor hospitalization, a participant hospital must provide written notice to the beneficiary, which must include certain information including an explanation of the model, the beneficiary’s care choices, the right to retain freedom of choice of providers and services, and the hospital’s financial partners.

As with CJR, the FAH continues to believe that such beneficiary notification

should be permitted on an electronic basis, with proof of receipt by the beneficiary, rather than a paper process that requires a beneficiary’s signature. This would serve both the beneficiary and the hospital in terms of ensuring the beneficiary is engaged and receives all relevant information, while helping to minimize the administrative burden on the hospital. This approach also is consistent with current Medicare requirements that providers engage patients through health information technology portals.

XIV. CMS Should Continue to Evaluate Included and Excluded Services and Should

Use Notice and Comment Rulemaking When Updating Conditions and Services to be Excluded from EPMs

We urge CMS to continue to evaluate the list of services to be excluded from the

bundles. We encourage CMS to consider excluding hospital readmissions that were planned for the patient prior to the start of the episode. We also urge CMS to consider excluding ongoing care for patients’ chronic conditions, management of which is outside the scope of the bundled payment models.

In addition, under both the CJR and proposed SHFFT models, all post-acute care services have been included in episode costs without exclusion. In the circumstance when a readmission for an excluded MS-DRG occurs during the episode, the cost of the readmission is not counted toward the episode cost. However, costs for any post-acute care that follows the excluded readmission are included in the cost of the episode, because there is no exclusion for post-acute care providers. We urge CMS to study potential exclusions for post-acute care following an excluded readmission.

CMS proposes to update services to be excluded from these episode, on an annual basis, at a minimum, to reflect annual changes to ICD-CM coding and annual changes to the MS- DRGs under the IPPS, and to address any other issues brought to its attention. CMS would update these exclusions without a rulemaking, by using sub-regulatory guidance.

Because the EPM models are mandatory and affects a large number of hospitals

and providers, it is important that CMS implement this process, to update services to be excluded from these episodes, through notice and comment rulemaking. Provider feedback throughout the course of EPM model implementation should be reflected in CMS’s decision. Hospitals of different sizes, geographic locations, organizational capabilities, and socio- economic factors all have unique preferences, and their ideas and opinions should be accounted for when CMS makes changes to the list of conditions and services to be included and/or excluded from the episode.

48

XV. The FAH Supports the CMS Proposal to Establish a Pathway for Participants in

the CJR to Qualify as an Advanced APM

Under this proposed rule, CMS would establish a pathway for participants in the CJR to be considered as participating in an Advanced APM for the purpose of becoming a qualifying APM participant (QP). CMS proposes to establish two tracks for CJR participants. Those CJRs and their participating providers that meet the criteria for participating in an Advanced APM as described under the Quality Payment Program Proposed rule would be on Track 1. Other CJRs and their participants that do not meet such criteria would be on Track 2.

CMS asserts that the existing quality measures for the CJR model meet the criteria under the Quality Payment Proposed rule and starting in PY2 most participating hospitals will bear downside risk. CMS notes that certain other hospitals subject to special protections – rural hospitals, sole community hospitals, Medicare Dependent Hospitals, and Rural Referral Centers – have a lower stop-loss limit (3 percent in performance year 2) and would, therefore not meet the risk criteria for that year, but would in PY3 when these hospitals would be subject to a high stop loss limit. Therefore, CMS proposes that those hospitals with special protections would be in Track 2 for performance year 2, but seeks comment on whether protected hospitals should be able to elect a higher stop-loss limit for the second PY. CMS also proposes that a CJR model may participate in Track 1 for performance years 2 through 5 if they attest in a form and manner as required by CMS to their use of CEHRT to document and communicate clinical care with patients and other health professionals in accordance with 42 CFR 414.1305. Those choosing to require and attest to the use of CEHRT must also provide CMS with a list of clinician financial arrangements on no more than a quarterly basis.

The FAH supports the CMS proposal to establish a pathway for participants in the CJR to qualify as an advanced APM. We also support a modification to the proposal to allow hospitals subject to special protections – rural hospitals, sole community hospitals, Medicare Dependent Hospitals, and Rural Referral Centers -- to elect a higher stop-loss limit of the second PY in order to meet the qualifying risk criteria. With respect to the CEHRT requirements, the FAH urges CMS to make this attestation requirement administratively as simple as possible for participants. We recommend an annual attestation with electronic method of submission to minimize reporting burden while still providing the necessary information necessary for CMS to determine if a CJR model is an Advanced APM.

XVI. FAH Supports the CR/ICR Payment Model

CMS proposes a cardiac rehabilitation (CR) incentive payment model that integrates with

the proposed AMI and CABG EPMs. CMS believes the evidence supports that CR and intensive cardiac rehabilitation (ICR) can significantly improve long-term outcomes for patients following AMI or CABG. Specifically, CMS proposes a two-level, per service, CR incentive amount. The first (lower) level would support initial beneficiary engagement; the second (higher) level would foster continued adherence above a session utilization benchmark. CMS proposes to set the benchmark at 12 sessions based on evidence linking reduced mortality to increased session completion. CMS proposes to set the CR incentive payments at $25 per service for each of the first 11 CR/ICR services and $175 per service thereafter. CMS does not propose to cap the number of services counted toward the CR amount but notes that Medicare program coverage

49

limits already exist. For example, CR program sessions are limited to 2 one-hour sessions per day for up to 36 sessions over up to 36 weeks (with an option for 36 more sessions over an extended time period with MAC approval). CMS anticipates revisiting the incentive payment levels as experience is gained with the CR incentive payment program.

With respect to CR inventive payment model participants, CMS proposes to select 45 of the 98 MSAs selected form for participation in the AMI and CABG EPMs for inclusion in the CR model (termed EPM-CR MSAs), and select 45 additional MSAs (termed FFS-CR MSAs) for the CR model drawn from the 196 MSAs eligible but not selected for the EPMs.

Overall, the FAH supports the implementation of the CR incentive payment model.

*************************

We appreciate your consideration of our recommendations that are vital to ensuring this program provides hospitals the ability to achieve program goals, while managing their financial, legal and regulatory risk. If you have any questions about our comments or need further information, please contact me or Steve Speil, Katie Tenoever, or Paul Kidwell of my staff at (202) 624-1500.

Sincerely,

Shared Accountability Payment

Method for Inpatient Rehabilitation Facilities (IRFs) under the Medicare Bundled Payment Programs

Dobson DaVanzo & Associates, LLC Vienna, VA 703.260.1760 www.dobsondavanzo.com

Shared Accountability Payment Method for Inpatient Rehabilitation Facilities (IRFs) under the Medicare

Bundled Payment Programs

Submitted to:

Federation of American Hospitals (FAH)

Submitted by:

Dobson|DaVanzo Allen Dobson, Ph.D.

Randy Haught

Joan E. DaVanzo, Ph.D., M.S.W.

September 29, 2016 — Final Report

Table of Contents

Introduction ............................................................................................ 1

Rationale For Shared Accountability Payments .................................... 1

Use of IRF Care in Mandatory Bundled Payment Programs ................. 3

The Need for IRF Payment Flexibility .................................................... 4

Shared Accountability Payment Model’s Methodology ....................... 7

Regulatory Flexibility............................................................................ 12

Modifications to the 60% Rule for CJR and EPM Patients ............... 12

Modifications to the 3-Hour Rule for CJR and EPM Patients ........... 14

Summary ............................................................................................... 15

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 1

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

INTRODUCTION

This report describes a potential option for a shared accountability payment model

for Medicare payments to Inpatient Rehabilitation Facilities (IRFs) treating

patients attributed to Medicare’s bundled payment programs for the

Comprehensive Care for Joint Replacement (CJR) and the Episode Payment

Models (EPM). Specifically, this payment model would provide an optional,

voluntary discount from the standard payment amount for IRF admissions for

patients discharged from an acute care hospital that are attributed to one of the

current and future bundled payment models. This IRF “shared accountability

program” would be a voluntary payment policy alternative for IRFs and would

only apply to patients attributed to a bundled payment model episode and from

hospitals in which the IRF is a CJR or EPM Collaborator. In addition, flexibility

would be provided to further improve the efficiency of health care by

incorporating a per-diem payment method for shorter than average stay patients.

Finally, regulatory relief under the 60% Rule and 3-hour Rule would be required

to provide IRFs treating these patients, at payments below the current IRF PPS

rates, with the flexibility needed to participate in the program without

jeopardizing their Medicare status.

RATIONALE FOR SHARED ACCOUNTABILITY PAYMENTS

As CMS moves forward with its mandatory bundled payment programs which

places financial risk on acute care hospitals for post-acute care (PAC) spending, it

is important to provide payment flexibility to PAC hospitals in order to allow

them to achieve efficiencies that inure to the benefit of acute care hospitals that

are at financial risk under these bundled payment models. These additional

mandatory bundled payment programs will encompass a larger portion of the

patients that IRFs currently treat. Also, acute care hospitals have historically

relied on IRF care for a large portion of patients that will be in mandatory bundled

payment programs (over 9 percent of CJR and CABG patients and more than 18

percent of SHFFT patients). Options for acute care hospitals to reduce PAC

spending are currently limited to encouraging patients to receive PAC in settings

that receive lower Medicare payments or encouraging PAC providers that have

the ability to reduce payments through efficiencies to do so. Thus, providing

payment flexibility to PAC hospitals is important to allow them to effectively

compete in a changing environment and to continue to provide beneficiaries with

PAC options that best meet their needs.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 2

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

In this environment, PAC providers such as skilled nursing facilities (SNFs) or

home health agencies (HHAs) have the ability under existing regulations to

modify their practice or utilization patterns in a manner that produces lower

Medicare payments for patient care. SNFs can reduce their Medicare payments

within the current prospective payment rules by simply providing fewer days of

care. In addition, SNFs can also reduce the level of therapies provided, which

would put patients into lower-paid RUG categories. Similarly, HHAs can reduce

the number of therapy encounters during a home health episode with the result of

receiving less Medicare payment.

The second year evaluation of CMS’s Bundled Payment for Care Improvement

(BPCI) Demonstration found that SNFs reduced the amount of Medicare spending

for SNF services during an episode of care primarily through reduced length of

stay (i.e., reducing the number of days patients were in SNFs). The study found a

statistically significant reduction in SNF length of stay both when the SNF was an

episode initiator itself as well as when the SNF was a downstream PAC provider

for a BPCI participating acute care hospital.1

The purpose of the IRF shared accountability payment model is to provide a

similar level of payment flexibility to IRFs in order to reduce Medicare spending

for Medicare bundled payment patients, which is not available under the current

Medicare IRF prospective payment system (IRF PPS). Since episode target prices

and performance period spending in Medicare’s bundled payment programs are

based on Medicare payments, and because Medicare payments to IRFs are per-

discharge (not per diem) and diagnosis based (not therapy based), there is no

flexibility for IRFs to reduce their Medicare payments for the benefit of hospitals

participating in the bundled payment models, regardless of the cost-efficiencies an

IRF may generate. The IRF shared accountability payment model would allow

acute care hospitals to benefit from being able to maintain or enhance their

relationships with IRFs under these programs by permitting IRFs to generate

Medicare savings for patients attributed to the bundled payment programs.

The proposed approach would enable IRFs to more fully and robustly share in the

potential risks and rewards of these bundled payment programs and allow

hospitals participating in the bundled payment program to benefit from savings

achieved by IRFs under the alternative payment model, which is similar to how

1 Dummit et.al., “CMS Bundled Payments for Care Improvement Initiative Models 2-4: Year 2 Evaluation & Monitoring Annual Report”, August 2016

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 3

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

acute care hospitals now benefit from SNFs’ reduced length of stay. Thus, this

alternative voluntary payment model would permit greater accountability among

and between acute care hospitals and IRFs. This approach directly aligns with

CMS’ recognition of need for payment flexibility as Medicare reimbursement

moves towards alternative payment models and away from fee-for-service.

USE OF IRF CARE IN MANDATORY BUNDLED PAYMENT PROGRAMS

The procedures and conditions under both the CJR program and the proposed

EPM models accounted for more than 17 percent of IRF Medicare discharges in

2014 (Figure 1). Thus, as CMS moves forward with adding more procedures and

conditions to mandatory bundled payment programs it will encompass a larger

portion of the patients that IRFs treat.

Figure 1: Share of IRF Discharges in CY2014 for MS-DRGs in Mandatory Bundled Payment Programs 1

1/ CJR, SHFFT, AMI, and CABG patients treated in IRFs were identified based on the MS-DRG for patients from the prior acute care hospital stay that occurred within 30 days of the IRF stay. All IRF discharges were counted in this analysis including patients not discharged directly from an acute care hospital. Source: Dobson|DaVanzo analysis using the 100% Medicare Limited Dataset for 2014.

In addition, acute care hospitals have historically relied on IRF care for a large

portion of patients that will be in mandatory bundled payment programs. Between

2012 and 2014 acute care hospitals in the proposed bundled payment areas

discharged 9.4 percent of CJR patients to IRFs, 18.3 percent of SHFFT patients,

9.1 percent of CABG patients and 1.6 percent of AMI patients (Exhibit 1).

All Other Cases, 82.6%

CJR, 9.1%

SHFFT, 5.8%AMI, 1.1% CABG, 1.4%

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 4

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

Exhibit 1: Distribution of Acute Care Hospital Patients by Discharge Destination and Mandatory Bundled Payment Program (FY2012-2014)

Patient Discharge Destination CJR 1 SHFFT 1 CABG 2 AMI 2

All Models

HHA 36.4% 6.3% 36.6% 12.9% 22.9%

Home w/o HHA 14.0% 6.7% 34.2% 66.4% 38.9%

Hospice 0.0% 0.3% 0.0% 1.0% 0.5%

IRF 9.4% 18.3% 9.1% 1.6% 6.9%

LTCH 0.1% 0.6% 1.0% 0.6% 0.5%

Other Inpatient 1.1% 0.6% 1.3% 3.0% 1.9%

SNF 39.0% 67.2% 17.9% 14.4% 28.4%

Total 100.0% 100.0% 100.0% 100.0% 100.0% 1/ Includes hospitals in the 67 CJR MSAs that were not participating in BPCI for LEJR or hip fracture fixation procedures in July 2016. 2/ Includes hospitals in the proposed 294 AMI/CABG MSAs that were not participating in BPCI for AMI, CABG or PCI episodes in July 2016. Source: Dobson|DaVanzo analysis using the 100% Medicare Limited Dataset for 2011-2014.

As noted above, the recent Second Year BPCI Evaluation Report found that most

acute care hospitals focused on reducing PAC costs in their BPCI episodes. The

study found a statistically significant reduction in the use of institutional PAC

care (IRFs, SNFs and LTCHs) for BPCI patients relative to a comparison

population for both orthopedic and cardiac surgeries. However, this movement of

patients away from institutional PAC providers did not

result in a significant reduction in overall episode

spending relative to a comparison group. Therefore,

other options should be considered by CMS that would

provide bundled payment participants with the ability to

reduce episode spending while maintaining the services

that beneficiaries have historically required.

THE NEED FOR IRF PAYMENT FLEXIBILITY

The bundled payment programs as proposed are inherently and structurally

unfavorable to IRFs. Except in limited circumstances, the bundled payment

models do not risk-adjust payments nor stratify patients that need a higher level of

The movement of patients away from

institutional PAC providers did not

result in a significant reduction in

overall episode spending relative to

a comparison group

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 5

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

care into higher payment categories.2 Therefore, the payment and cost trends of

more complex Medicare patients will not be accounted for in these models. This

will lead to an emphasis on use of other post-acute care providers, such as SNFs

or HHAs that have the ability to modify their practice or utilization patterns in a

manner that produces lower Medicare payments for patient care.

IRFs do not currently have the flexibility to affect their Medicare payments

through the adoption of clinical or operational efficiencies, whereas SNFs have

substantial opportunities to reduce spending through reducing lengths of stay and

reducing therapy levels. As illustrated in Exhibit 2, we examined the variation in

SNF payments per user during bundled payment episodes of care compared to the

variation in IRF payments per user between October 2012 and September 2014.3

These data showed a greater variation in SNF payments per bundled payment

program user compared to IRF payments per user as indicated by the coefficient

of variation (CV). The differences in payment per user variation (SD) between

SNFs and IRFs were statistically significant using a Brown-Forsythe test.

2 CJR stratifies patients by MS-DRG and fracture status. The EPM CABG model stratifies patients by MS-DRG and with/without AMI. The EPM AMI model stratifies patients by MS-DRG and with/without a CABG readmission. 3 Both SNF and IRF payments were normalized to remove the effect of wage index differences. SNF payments were adjusted to remove special payments for HIV patients. IRF payments were standardized to remove special payment adjustments for rural location, low-income patient percentage, and teaching. However IRF outlier payments were not removed. These adjustments are consistent with the methodology specified in the CCJR NPRM.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 6

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

Exhibit 2: IRF and SNF Medicare Payments per User during Bundled Payment Episodes of Care by MS-DRG (2012-2014) 1

Model 2 MS-DRG

Mean IRF Episode

Payments Per User

Mean SNF

Episode Payments Per User

Standard Deviation of IRF Episode Payments Per User

Standard Deviation of SNF Episode

Payments Per User

P-Value of Variation

Difference

Coefficient of Variation for IRF Episode

Payments Per User 3

Coefficient of Variation for SNF Episode

Payments Per User 3

CJR 469 $18,109 $20,308 $7,266 $13,047 <0.0001 0.40 0.64

470 $14,770 $12,449 $5,582 $10,098 <0.0001 0.38 0.81

SHFFT

480 $19,948 $24,041 $7,790 $13,711 <0.0001 0.39 0.57

481 $18,518 $23,723 $5,872 $12,988 <0.0001 0.32 0.55

482 $17,470 $22,046 $5,446 $12,788 <0.0001 0.31 0.58

CABG

231 $17,392 $11,778 $7,315 $9,454 0.0003 0.42 0.80

232 $15,420 $9,980 $5,485 $7,951 0.0011 0.36 0.80

233 $17,786 $12,971 $7,808 $10,313 <0.0001 0.44 0.80

234 $15,880 $10,206 $6,367 $8,360 <0.0001 0.40 0.82

235 $17,717 $12,310 $8,153 $10,149 <0.0001 0.46 0.82

236 $15,469 $9,782 $6,143 $8,261 <0.0001 0.40 0.84

AMI

246 $17,784 $14,148 $8,570 $11,271 <0.0001 0.48 0.80

247 $16,838 $13,685 $8,265 $10,895 <0.0001 0.49 0.80

248 $17,986 $14,231 $10,058 $11,167 <0.0001 0.56 0.78

249 $16,548 $13,516 $7,352 $10,967 <0.0001 0.44 0.81

250 $17,384 $14,177 $8,844 $11,793 0.0002 0.51 0.83

251 $17,914 $13,924 $7,963 $11,216 0.0004 0.44 0.81

280 $18,141 $15,198 $8,770 $11,823 <0.0001 0.48 0.78

281 $17,404 $15,497 $8,000 $11,683 <0.0001 0.46 0.75

282 $16,617 $14,654 $7,400 $11,227 <0.0001 0.45 0.77 1/ Payments per IRF user is defined as total IRF payments (normalized for wage index difference and standardized to remove special payment adjustments specified under CJR) during the 90-day episode for patients that used an IRF following a CJR eligible hospitalization. This was similarly calculated for SNFs. 2/ Includes CJR and SHFFT simulated episodes for all acute care hospitals in the 67 specified CJR MSAs excluding BPCI participants selecting LEJR or hip fracture fixation episodes. Includes AMI and CABG simulated episodes for all acute care hospitals in 294 potential AMI/CABG MSAs excluding BPCI participants selecting AMI, CABG or PCI episodes. 3/ Coefficient of variation is calculated as the standard deviation divided by the mean. Source: Dobson | DaVanzo analysis using the 2011 – 2014 100% Medicare LDS files and includes episodes beginning between October 2012 and September 2014.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 7

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

The variation in IRF payments is primarily driven by payment differences in

patient case mix groups which are based on patient age, diagnoses and functional

status – all factors beyond the IRF’s control. However, variation in SNF

payments are driven by factors that include but are not limited to length of stay

and therapy levels – both of which can be controlled by the SNF. Thus, because

SNFs have a significantly greater variation in per-user spending across episodes

than IRFs, SNFs have greater opportunities to reduce costs in the immediate

future under the bundled payment programs. This opportunity is well known to

bundling conveners and other at-risk bundle-holders.

Accordingly, if IRFs are to be preserved and strengthened, there is a need for an

optional IRF payment policy specifically tailored for patients within the bundled

payment models that would allow IRFs to choose to receive lower reimbursement

rates from Medicare for these cases, relative to what otherwise would be paid

under the IRF PPS. This optional payment policy could be constructed in a

manner that is similar to how SNFs are reimbursed, on a per-diem basis. It is also

important to note that the IRF PPS already uses a per-diem methodology as a

payment unit for patients transferred from the IRF to another care setting whose

stay in the IRF is less than the average length of stay for non-transfer cases within

the same case-mix group.4

Currently the Center for Medicare & Medicaid Innovation (CMMI) is working

with its contractors to develop innovative payment systems under the Round 2

Health Care Innovation Awardee demonstrations. CMS is including payment

model development as an aspect of these specific CMMI demonstrations and

pilots. CMMI notes that “delivery system innovation is dependent upon payment

system innovation.” Thus, we believe that the shared accountability payment

system meets the intent of CMMI’s vision for payment innovation and should be

incorporated into the CJR and EPM models.

SHARED ACCOUNTABILITY PAYMENT MODEL’S METHODOLOGY

As mentioned above, the IRF shared accountability payment model would be a

voluntary payment policy alternative for IRFs and would only apply to patients

attributed to a bundled payment model episode and from general acute care

hospitals with which the IRF is a CJR or EPM Collaborator. The Second Year

BPCI Evaluation Report indicated that participating acute care hospitals tried to

4 See generally, 42 C.F.R. §412.624(f).

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 8

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

collaborate with area PAC providers in efforts to improve care coordination and

gain efficiencies across the entire episode of care. However, the evaluators found

few specific examples of successful collaborations and participants indicated that

it was challenging to establish relationships with other providers. This payment

alternative would help to provide greater incentives for acute care hospitals and

IRFs to collaborate in order to benefit from lower Medicare spending achieved

through efficiencies of the IRFs.

The shared accountability payment method would work within the payment

nomenclature of the existing IRF PPS payment system. We found that the

majority of CJR cases (94%) and SHFFT cases (90%) treated in IRFs would fall

under rehabilitation impairment category RIC 07 (fractures) and RIC 08 (joint

replacement).5 However, AMI and CABG patients tend to be categorized under a

variety of RICs including RIC 01 (stroke), RIC 03 (non-traumatic brain

dysfunction), RIC 06 (neurology), RIC 14 (cardiac disorders), and RIC 20

(miscellaneous). Thus, this payment method may apply to any CMG within which

a bundled payment patient is categorized. Exhibit 3 shows the distribution of IRF

patients that would meet the bundled payment model criteria across each RIC

category.

Under the shared accountability payment model, a qualifying bundled payment

patient discharged from a participating IRF would be paid a discounted per-diem

payment amount. Relative payment weights and average lengths of stay for each

of the tiered CMGs is published annually by CMS in the IRF PPS Final Rule.

Exhibit 4 provides an illustration of the payment calculation for an example case.

Steps 1 through 9 shows the current payment calculation under the IRF PPS,

before facility-specific adjustments are applied and excluding outliers. The

shared accountability payment model would continue to include IRF facility-

specific adjustments for rural location, LIP, and teaching. As specified in the CJR

Final Rule and the EPM Model NPRM, these special payments will not be

included in the target price calculation or performance period spending for

reconciliation purposes. Therefore, these payments are not discounted, but added

to the discounted rate.

5 Number of IRF discharges by CMG were identified from the 2014 Medicare Limited Dataset for IRF patients discharged from an acute care hospital with an MS-DRG included in the specific bundled payment models within 30 days of an IRF admission.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 9

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

Exhibit 3: Distribution of Bundled Payment Model Patients by Model and RIC in 2014

Distribution of Patients by RIC

RIC RIC Description CJR SHFFT AMI CABG

1 Stroke 0.7% 0.6% 16.5% 6.9%

2 Brain Dysfunction – Traumatic 0.1% 0.1% 0.8% 0.1%

3 Brain Dysfunction - Non Traumatic 0.3% 0.1% 5.8% 2.6%

4 Spinal Cord Dysfunction – Traumatic 0.0% 0.1% 0.1% 0.0%

5 Spinal Cord Dysfunction - Non Traumatic 0.0% 0.0% 0.8% 0.0%

6 Neurologic Conditions 0.6% 0.4% 14.0% 10.3%

7 Fractures 36.0% 90.3% 1.9% 0.1%

8 Joint Replacements 57.6% 0.4% 0.6% 0.1%

9 Other Orthopedic 0.5% 1.3% 1.1% 0.1%

10 Amputation - Lower Limb 0.0% 0.0% 0.5% 0.2%

11 Amputation - Upper Limb 0.0% 0.0% 0.0% 0.0%

12 Osteoarthritis 0.2% 0.0% 0.2% 0.0%

13 Rheumatoid Arthritis 0.1% 0.0% 0.2% 0.1%

14 Cardiac Disorders 0.1% 0.0% 41.1% 71.8%

15 Pulmonary Disorders 0.0% 0.0% 1.7% 0.1%

16 Pain Syndrome 0.0% 0.0% 0.1% 0.0%

17 Major Multiple Trauma w/o Brain/Spine Injury 1.0% 3.9% 0.2% 0.0%

18 Major Multiple Trauma with Brain/Spine Injury 0.0% 0.1% 0.1% 0.0%

19 Guillain-Barre Syndrome 0.0% 0.0% 0.0% 0.0%

20 Miscellaneous 0.2% 0.2% 7.7% 2.2%

21 Burns 0.0% 0.0% 0.0% 0.0%

50 Short Stay 2.5% 2.5% 6.2% 5.3%

51 Short Stay Deaths 0.0% 0.1% 0.2% 0.1%

Total 100.0% 100.0% 100.0% 100.0% 1/ CJR, SHFFT, AMI, and CABG patients treated in IRFs were identified based on the MS-DRG for patients from the prior acute care hospital stay that occurred within 30 days of the IRF stay. Source: Dobson|DaVanzo analysis using the 100% Medicare Limited Dataset for 2014.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 10

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

Exhibit 4: Example of Calculating the IRF Shared Accountability Payment for Participating IRFs under CJR for a Tier 2 CMG 804 Case with an 8 Day Length of Stay

Steps

1 Unadjusted Federal Standard Conversion Factor $15,478

2 Relative Weight (CMG 804 Tier 2) 1.0033

3 Federal Payment Amount (#1 x #2) $15,529

4 Labor Share of Federal Payment Amount 0.71

5 Labor Portion of Federal Payment (#3 x #4) $11,026

6 CBSA based wage index (illustrative) 0.8599

7 Wage Adjusted Labor Amount (#5 x #6) $9,481

8 Non-labor Share of Federal Payment (#3 - #5) $4,503

9 Wage Adjusted Federal Payment (#7 + #8) $13,984

Discounted Per-Diem Payment Calculation

10 Discount Factor -0.134

11 Discounted, Wage-Adjusted Federal Payment (#9 x (1 + #10)) $12,110

12 Average Length of Stay (CMG 804 Tier 2) 11

13 Per Diem Payment Rate (#15 x #16) $1,101

14 Medicare Covered Days for Discharge 8

15 Discounted Federal Prospective Payment (#17 x #18) $8,808

Facility Specific Adjustments (not counted toward CJR reconciliation)

16 Rural Adjustment (0.149 x #9) $2,084

17 LIP Adjustment (illustrative 0.015 x #9) $210

18 Teaching Adjustment (illustrative 0.050 x #9) $699

19 Total discounted Federal Prospective Payment (#15 + #16 + #17 + #18) $11,800

The shared accountability payment model would apply a discount factor (#10) to

the IRF PPS payment amount to determine a discounted federal payment amount

(#11). For the CJR and SHFFT models where collaborating IRFs electing to

receive a lower payment amount, we propose that a 13.4-percent reduction would

be applied to IRF PPS payment rates for IRFs that treat a CJR or SHFFT patient.

For AMI and CABG model collaborating IRFs that elect to receive a lower

payment amount, we propose that a 15.0-percent reduction would be applied to

IRF PPS payment rates for IRFs that treat a CABG or AMI patient. We arrived at

the level of discounts for CJR/SHFFT and AMI/CABG patients by reducing

current IRF payment levels to be closer to the cost of treating those patients.

Exhibit 5 shows our calculated payments and costs for IRF discharges with a

prior acute hospital stay under the MS-DRGs specified for each bundled payment

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 11

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

model for freestanding and hospital-based IRFs nationally. This would encourage

increased IRF efficiency within the bundled payment environment.

Exhibit 5: Average IRF Payments and Costs for LEJR Cases in 2016

Model and MS-DRG

Number of IRF

Discharges 1

Medicare Payment Amount 2

Estimated Cost Per

Discharge 3

Payment Reduction

Required to Equal Cost 4

CJR: MS-DRGs 469 and 470 34,579 $16,885 $14,497 14.1%

SHFFT: MS-DRGs 480-482 21,948 $20,708 $18,114 12.5%

SHFFT + CJR 56,527 $18,370 $15,901 13.4%

AMI: MS-DRGs 246-251 & 280-282 4,172 $19,137 $15,904 16.9%

CABG: MS-DRGs 231-236 5,281 $18,016 $15,616 13.3%

AMI + CABG 9,453 $18,511 $15,743 15.0% 1/ IRF discharges in calendar year 2014 for patients with a prior acute care hospitalization for MS-DRGs specified in each of the bundled payment models. 2/ IRF payments include adjustments for wage index, teaching, low-income patient share, rural location, and outliers calculated using 2014 claims data and 2016 payment rates 3/ Costs were estimated for each discharge as the sum of the following 2 factors: 1) facility-specific routine per-diem costs from the FY 2014 Medicare cost report multiplied by the number of days on the claim; 2) facility-specific ancillary cost to charge ratios multiplied by the ancillary revenue center charges on the claim. Costs for 2014 were inflated to 2016 using IRF market basket increases for each year. 4/ Reduction rate calculated as (payment – cost) / payment;

Source: Dobson | DaVanzo analysis using the 2014 Medicare 100% LDS and Medicare Hospital Cost Reports.

The discounted payment amount would be divided by the average length of stay

for the CMG to calculate a per diem payment rate (#13). Total discounted

payments for a stay prior to applying the facility-specific adjustments (#15) would

be calculated by multiplying the per-diem payment rate (#13) by the number of

Medicare covered days for the stay (#14), which would be capped to not exceed

the full discounted payment amount (#11). As described above, IRF facility-

specific adjustments for rural location (#16), LIP (#17), and teaching (#18) would

not be discounted, but added to the discounted amount to yield the total payment

for the IRF stay (#19).

As noted above, the per-diem approach is already part of the IRF PPS payment

system. Our approach is modeled after the existing transfer policy for short stay

IRF patients under the current IRF PPS system. Under the IRF transfer policy,

CMS provides the relative payment weight and IRF average length of stay for

each Tiered CMG that is already used to calculate a per-diem amount for certain

discharges.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 12

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

Because payments would be capped not to exceed the full discounted payment

amount, the shared accountability payment model would continue to pay for

extraordinarily high cost outlier cases using the same method as currently used

under the IRF PPS system. In addition, the shared accountability payment model

would not be used for very short stay patients (3 days or less). These cases would

continue to be categorized under CMG 5001 and paid at the current IRF PPS

payment rate.

REGULATORY FLEXIBILITY

The bundled payment programs represent a promising opportunity to reengineer

the American health care delivery system. As new ways of doing business are

under consideration, old regulatory constraints devised in support of siloed

prospective payment systems (PPS) systems must be reconsidered. The IRF 60%

Rule and so-called “3-Hour Rule” exemplify the need to reconsider how optimal

market flexibility can be built into the advanced payment policy environment.

For example, the Medicare Payment Advisory Commission (MedPAC) has

acknowledged that the effects of the 60% Rule and the 3-Hour Rule would need

to be addressed in the context of other similar payment policy modifications

affecting IRFs, such as their “site-neutral payment.”

An integral part of the shared accountability payment system should be a set of

modifications that would provide IRFs that choose to exercise

pricing/reimbursement flexibility, resulting in their receiving lower payment for

bundled payment MS-DRGs than what would otherwise be provided under IRF

PPS, relief from the restrictive effects of the 60% and 3-Hour Rule policies.

Modifications to the 60% Rule for CJR and EPM Patients

The 60% Rule exists as a mechanism to distinguish IRFs from acute hospitals and

to justify IRF PPS rates. By definition, if an IRF is not receiving full IRF PPS

rates for a particular type of case (in the case of the bundled payment programs,

joint replacement, hip fracture fixation, AMI and CABG patients), then the

constraints of the 60% Rule should not apply to any case that does not satisfy the

Rule (i.e., a “non-60% Rule case”) for which the IRF would receive the reduced

payment under the shared accountability payment model.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 13

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

Exhibit 6 shows the estimated percent of IRF bundled payment discharges that

are 60% Rule compliant (FY2015 compliance rules applied to CY2014 claims6).

More than 90 percent of SHFFT cases are 60% Rule compliant, more than 50

percent of CJR and AMI cases are 60% Rule compliant, and nearly 30 percent of

CABG cases are 60% Rule compliant.

Exhibit 6: Percent of LEJR Cases That Are 60% Rule Compliant

MS-DRG Number of IRF

Discharges

Number of Discharges That Are 60% Rule

Compliant

469 3,737 71.3%

470 30,842 48.1%

Total CJR 34,579 50.6%

480 3,434 87.7%

481 13,775 91.0%

482 4,739 93.1%

Total SHFFT 21,948 90.9%

246 809 54.5%

247 337 36.8%

248 334 53.0%

249 107 39.3%

250 298 46.0%

251 153 42.5%

280 1,507 53.7%

281 482 45.6%

282 145 50.3%

Total AMI 4,172 50.0%

231 151 35.8%

232 46 21.7%

233 1,712 37.3%

234 1,231 19.7%

235 1,071 36.7%

236 1,070 20.7%

Total CABG 5,281 29.5% Source: Medicare IRF Compliance Policies in effect for FY2015 applied to the Medicare LDS data for 2014.

6 The list of comorbid condition codes for 60% Rule compliance that became effective October 2015 are more restrictive than those for FY2015 used in this analysis. The FY 2016 comorbid condition list was not used because of the switch to ICD-10, whereas the claims data use ICD-9. The effect of these changes will make compliance with the 60% Rule more difficult for IRFs.

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 14

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

Under the shared accountability payment model, all bundled payment cases

accepted by IRFs that are not 60% Rule compliant cases would be removed from

the formula that is used to determine the Rule’s compliance threshold percentage.

Thus, any case that would satisfy the 60% Rule (for example, a bi-lateral joint

replacement, or a joint replacement with a qualifying comorbidity) would

continue to count toward the 60% Rule and would remain part of the formula.

In Exhibit 7 below, only 60%-Rule compliant bundled payment cases would

count toward determining the hospital’s ratio, but the non-compliant cases would

not be counted.

Exhibit 7: Impact of Excluding Only Non-Compliant Bundled Payment Cases from 60% Rule

Total Discharges Counted

Toward 60% Rule

Discharges That Are 60%

Rule Compliant

Discharges That Are Not

Compliant

Percent Compliant

Non-CJR or EPM Discharges 2014 313,852 208,212 105,640 66%

CJR and EPM Discharges 2014 41,090 41,090 24,890

62% (counted) (not counted)

Discharges Counted in 60%-Rule 354,942 249,302 105,640 70%

Source: Medicare IRF Compliance Policies in effect for FY2015 applied to the Medicare LDS data for 2014.

Modifications to the 3-Hour Rule for CJR and EPM Patients

Under the 3-Hour Rule, each patient treated by an IRF must be documented as

generally requiring and reasonably expected to participate in, and benefit from, an

intensive rehabilitation therapy program. Under the Rule, such an “intensive”

rehabilitation program is defined to mean that each patient must receive at least 3

hours of therapy per day for at least 5 days per week. If an IRF patient did not

receive therapy in amounts required under the 3-Hour Rule’s regulatory

framework, the IRF risks having the patient’s claim for services denied because it

failed to satisfy arbitrary process-specific requirements comprising the 3-Hour

Rule. The application of the 3-Hour Rule to bundled payment cases paid under a

discounted “shared accountability payment” will prevent IRFs from tailoring

therapy and other care delivery programs for their bundled payment patients.

For IRFs accepting reduced pricing/reimbursement for bundled payment patients,

it is essential that they have the flexibility to dose and provide therapy without the

requirement that the “preponderance” of such therapy be “one-on-one;” rather,

they should be able to provide concurrent and/or group therapy that meets the

SHARED ACCOUNTABILITY PAYMENT METHOD FOR IRFs UNDER BUNDLED PAYMENT FINAL REPORT 15-147 | 15

Dobson|DaVanzo

© 2016 Dobson DaVanzo & Associates, LLC. All Rights Reserved.

specific needs of the patient and reduces the costs of care. IRF PPS rates are

intended to capture the resources needed to satisfy the 3-Hour Rule policy.

Payment rates that are less than the full IRF PPS amount for bundled payment

cases will not adequately account for 3-Hour Rule-based resource use and

therefore the 3-Hour Rule should not apply to such cases.

Flexibility around the therapy policy should be a requirement. IRFs will continue

to be required to provide 3 hours of therapy per day. However, the Rule should

be modified to allow IRFs the flexibility to provide concurrent and/or group

therapy that meets the specific needs of the patient in addition to one-on-one

therapy.

SUMMARY

In summary, the voluntary IRF shared accountability payment model will provide

IRFs with an additional mechanism for contributing toward the goals of the

Medicare bundled payment programs for reducing Medicare spending and

improving patient care. The discounted IRF payment, as well as the ability to

further reduce Medicare payments for patients requiring a shorter than average

length of stay, will provide the flexibility required under the CJR and EPM

program, and any future mandatory bundled payment programs, to help IRFs

reduce Medicare spending inuring to the benefit of acute care hospitals with

which the IRF is a Collaborator. Along with these lower payments, the need to

modify the 3-Hour rule for bundled payment patients is essential to provide IRFs

with the flexibility to dose and provide therapy that meets the specific needs of

the patient at the lower cost. Finally, a modification to the 60% Rule is required

so that all bundled payment cases accepted by IRFs under a shared accountability

payment model that are not 60% Rule compliant cases are removed from the

formula that is used to determine the Rule’s compliance threshold percentage.

IRFs play an important role in rehabilitating CJR and EPM model patients and

have historically treated 9.4 percent of CJR patients, 18.3 percent of SHFFT

patients, 9.1 percent of CABG patients, and 1.6 percent of AMI patients. The

shared accountability payment model will provide acute care hospitals

participating in the bundled payment program with a powerful option for reducing

costs while placing patients in the most clinically appropriate setting for their

post-acute care needs to help meet the objectives of these programs.