The Affordable Care Act (ACA) called. An Examination Of Medicaid Delivery System Reform Incentive Payment Initiatives Under Way In Six States

Paying For Care Origins And Context Two major factors provide a backdrop for the emergence of DSRIP waivers in the post-aca period. The first is the growing importance, over the past two decades, of demonstration waivers as a policy tool. Before 1993, concerns that waivers have highly developed research designs and that they strictly adhere to a requirement that they be budget-neutral undercut federal willingness to approve them. Federal officials had endorsed about fifty Medicaid demonstration waivers from the program s birth in 1965 to 1993 and seldom renewed them. 2 The Clinton administradoi: 10.1377/hlthaff.2015.0165 HEALTH AFFAIRS 34, NO. 7 (2015): 1162 1169 2015 Project HOPE The People-to-People Health Foundation, Inc. By Michael K. Gusmano and Frank J. Thompson An Examination Of Medicaid Delivery System Reform Incentive Payment Initiatives Under Way In Six States Michael K. Gusmano (gusmanom@thehastings center.org) is a research scholar and codirector of the Yale-Hastings Program at the Hastings Center, in Garrison, New York. Frank J. Thompson is a professor at the Rutgers School of Public Affairs and Administration, in Newark, New Jersey, and at the Center for State Health Policy, in New Brunswick, New Jersey. ABSTRACT Medicaid waivers for Delivery System Reform Incentive Payment (DSRIP) seek to hold hospitals and other providers accountable for measureable improvements in health care delivery. We explore the policy context giving rise to these waivers in six states, with particular attention to the interplay among the financial needs of hospitals; the rise of managed care; and federal interest in replacing an unconditional Medicaid funding stream, the upper payment limit, with one rooted in pay-for-performance. Key characteristics of these main DSRIP waivers are compared with a particular focus on the establishment of project menus, performance metrics, and pay-for-performance processes. Concluding sections discuss the potential and limits of the waivers as vehicles for health care reform. The long-term durability and impact of DSRIP remains unclear. But federal and state officials have made considerable headway in planning and otherwise laying the groundwork for it. The Affordable Care Act (ACA) called for transforming the health care delivery system. In contrast to ACA initiatives, the adoption of Delivery System Reform Incentive Payments (DSRIP) reflects the traditional exercise of federal administrative discretion under demonstration waiver authority dating back to 1962 (section 1115 of the Social Security Act). After negotiations with the states, the Centers for Medicare and Medicaid Services (CMS) had by the end of 2014 granted Medicaid DSRIP waivers to six states California, Kansas, Massachusetts, New Jersey, New York, and Texas and look-alike waivers to two others. 1 These waivers are incorporated into much larger Medicaid demonstrations that seek to transform health services (for example, by shifting the elderly and disabled into managed care). DSRIP waivers provide Medicaid funds to hospitals and certain other providers if they achieve performance gains on an array of metrics linked to such targets as system redesign, clinical improvements, and enhancements of population health. This article explores the context giving rise to DSRIP waivers. We also compare key features of the waivers, focusing in particular on the establishment of project menus, performance metrics, and pay-for-performance processes. Finally, we assess the waivers potential and limits in promoting delivery system reform. 1162 Health Affairs July 2015 34:7

tion opened the gates to states initiatives to transform their Medicaid programs via waivers. Under President Bill Clinton, administrative measures 2 were initiated that made it much easier for states to launch demonstrations, and the George W. Bush administration followed suit. A flood of waivers occurred, with a substantial majority of states operating some facet of their Medicaid program as a demonstration by 2008. 2 The administration of President Barack Obama has continued employing demonstration waivers to promote its policy goals. 3 Each presidential administration places its own thematic stamp on Medicaid demonstration waivers, at times negotiating modifications in waivers approved by a predecessor. 2 This propensity creates uncertainty for the DSRIP waivers since all six will likely extend into 2017, when the Obama administration leaves office. A second key contextual factor promoting DSRIP involved the interplay among the financial needs of hospitals, the rise of Medicaid managed care, and CMS interest in moving toward performance-based accountability. Medicaid had for decades provided subsidies to hospitals under various guises. The disproportionateshare hospital (DSH) program, for instance, directs monies to hospitals that serve higher percentages of Medicaid enrollees and the uninsured. However, it was the possible loss of another funding stream, upper payment limits (UPL), that prompted states to pursue DSRIP. This funding provision sets an upper bound on what the federal government will pay for different classes of Medicaid services and can be no greater than what Medicare would pay. The provision reflects the aggregate amount that can be paid to an entire class of providers if every provider in that class were paid the Medicare rate. 4 The federal UPL often exceeds what states actually pay for the services. On the basis of the gap between the UPL and regular Medicaid payments, states can create funding pools to subsidize hospitals. By 2011, supplemental payments to providers through this provision exceeded those of the disproportionate-share program. 5 The threat to UPL subsidies stemmed from the response of CMS to the rise of managed care. Many of the waivers approved in the 1990s shifted Medicaid beneficiaries into a managed care arrangement. By 2010 nearly 70 percent of Medicaid enrollees (mostly children and ablebodied adults) were in some form of managed care. 6 More recently, CMS has granted waivers to several states (including the six DSRIP jurisdictions) to enroll the elderly and people younger than age sixty-five with disabilities in managed care. The difficulty for the states was that CMS had by administrative regulation decreed that no payments under the UPL provision could go to hospitals for services rendered under contract with managed care organizations. 7 In part, the CMS stance reflected the technical complexities of applying a payment method linked to fee-forservice to capitated payments under managed care. But it also stemmed from a desire to transition relatively unconditional UPL subsidies into funding based on the achievement of measurable delivery system reform. UPL allocations to the states lack transparency. Unlike DSH subsidies, they are not reported to CMS at the provider level. Nor are they directly tied to specific services and enrollees. This makes it difficult to assess their effects on payment methods and delivery models. 7 Thus, DSRIP allowed CMS to replace poorly understood subsidies to providers with a system of performance-based accountability. In turn, the states and providers had strong incentives to cooperate with CMS in order to replace the UPL subsidies with another funding stream. CMS further induced state participation by permitting them to use intergovernmental transfers to fund their share of DSRIP costs (Exhibit 1). This approach involves having local public hospitals, and other public entities, transfer dollars to a state Medicaid agency, which spends the sum on health services. This action leverages federal matching payments at least equivalent to the amount the state spent. The contributing hospitals can typically count on being reimbursed for the transfers when the state receives the federal Medicaid match for DSRIP or a related initiative. CMS receptivity to intergovernmental transfers as a funding source reflects a historical departure. In the past, CMS had seen intergovernmental transfers as part of a strategy used by states to maximize their Medicaid reimbursement from the federal government without actually providing a real match. The agency viewed it as cost shifting to the federal government and, during the second term of the George W. Bush administration, had discouraged use of this funding source in Medicaid waivers. 4 Now, CMS interest in encouraging states to pursue DSRIP made the agency more receptive to intergovernmental transfers. Comparison Of DSRIP Waivers A core conceptual underpinning of DSRIP waivers is the Triple Aim approach. Developed by Donald Berwick (acting administrator of CMS from July 2010 until December 2011) and associates, the approach stresses the balanced pursuit of three interdependent goals: improving the individual experience of care; improving the July 2015 34:7 Health Affairs 1163

Paying For Care Exhibit 1 Key Characteristics Of Delivery System Reform Incentive Payment Waivers State Eligible providers Number of participating providers Time frame Federal funds State funding source Hospital states CA Public hospitals with heavy Medicaid volume 21 hospitals 2010 15 $3.4 billion Intergovernmental transfers (IGTs) KS A large public teaching hospital and a border city children s hospital 2 hospitals 2013 17 $60 million General revenue and IGTs MA Hospitals with a high Medicaid volume 7 hospitals 2011 14 Renewed a 2015 19 $690 million State general funds and IGTs NJ All acute care hospitals 51 hospitals 2014 17 $583.1 million General revenue and hospital assessments Regional network states NY Safety-net hospital-led performing provider systems 25 PPS: networks of hospitals and community providers 2014 19 $6.42 billion IGTs TX Public hospital-led regional health care partnership 20 RHP: networks of hospitals and community providers 2012 16 $6.6 billion State general revenue funds and IGTs SOURCES Coughlin TA, Long SK, Peters R, Arguello R. Strategies in 4 safety-net hospitals to adapt to the ACA. Washington (DC): Kaiser Commission on Medicaid and the Uninsured; 2014 Jun 11. Shearer C, Kennedy-Shaffer L, Myers N. Performing provider system projects: tackling the health needs of communities. New York (NY): Medicaid Institute at United Hospital Fund; 2015 Jan. Aragón S. Intergovernmental transfers and the Texas Medicaid transformation waiver. Austin (TX): Texas HealthandHuman ServiceCommission;2013Aug2. a Waivers are time limited, but states may persuade the Centers for Medicare and Medicaid Services to renew them for a new period. health of populations; and reducing the per capita costs of care for populations. 8(p760) The Triple Aim envisions a major role for an integrator that promotes all three components for a specified population. Berwick underscored that managed care had by itself failed to serve this integrative role and that new reform initiatives were needed. 8 Overview Exhibit 1 summarizes key characteristics of the six DSRIP waivers. 9 Based on the providers they target, waiver states fall into two basic categories: those focused on hospitals and those creating regional networks. The four hospital-oriented states vary considerably in the number and kinds of facilities targeted. New Jersey stands out as the most participative, with close to 70 percent of the state s acute care hospitals involved in DSRIP. Kansas ranks as the least, with two hospitals (one in Missouri that serves Kansas Medicaid enrollees) out of about sixty engaged. In turn, California, the first to use the DSRIP initiative, targeted twenty-one public hospitals a little more than 5 percent of the state s acute care facilities. The twenty-one hospitals are principal providers for about half of all uninsured individuals and about one-third of all Medicaid enrollees in the state. 9 Like other DSRIP states, Massachusetts targeted safety-net hospitals. The seven participating facilities account for about 9 percent of all acute care hospitals in the state. Massachusetts recently renewed its waiver, 10 and California is applying for a renewal. The regional network approaches of New York and Texas reflect a different assumption about promoting vertical integration in the health care system. The hospital-oriented DSRIP model assumes that performance payments to these facilities will prompt them to form various relationships with community providers that are not formally specified in waiver documents. The regional model requires that a lead organization (typically a hospital) in an area explicitly create formal partnerships among hospitals and community providers. Texas created twenty regional health care partnerships that cover the entire state and are coordinated by anchoring entities thirteen hospital districts, three academic health science centers, three county governments, and one state-owned hospital. The more than 300 formal partners in the regions include public and private hospitals, community mental health centers, academic health science centers, local health departments, and physician practice plans. A Texas regional health care partnership identifies projects for its area, which some participating providers may choose to implement. DSRIP payments become available to those involved in the specific project. New York established twenty-five performing provider systems, led by public or other safetynet hospitals, which cover a portion of the state. The geographic boundaries of the payment systems emerged from providers in various areas banding together to submit proposals that won approval. Hospitals become eligible for membership in the system if they passed one of three 1164 Health Affairs July 2015 34:7

A core conceptual underpinning of DSRIP waivers is the Triple Aim approach. tests. The facility must either be a public, critical access, or sole community hospital; have at least 35 percent of its outpatient volume be uninsured or enrolled in Medicaid, and at least 30 percent of its inpatient treatment associated with these groups; or serve at least 30 percent of all Medicaid and uninsured patients in the proposed geographic area of the performing provider system. Community providers qualify for participation if 35 percent of their patients are Medicaid beneficiaries or uninsured. Such providers include health homes, skilled nursing facilities, community health centers, and behavioral health providers. To a much greater degree than the regional plans in Texas, the New York performing provider systems resemble accountable care organizations in that they are networks of providers jointly accountable for improving the quality of health care services and reducing costs for a defined patient population. 11 Each network is responsible for most or all Medicaid beneficiaries in a given geographic area, with each enrollee attributed to a particular provider system. 12 New York allocates DSRIP monies to the systems, which then apportion payments among participating providers. Three Key Waiver Components The processes establishing DSRIP involved negotiation among CMS, state Medicaid agencies, and providers to determine three primary DSRIP components: the project menu, performance indicators and metrics, and pay-for-performance processes. We sought to illuminate these components by drawing on examples rooted in more intensive analysis of two hospital-oriented states (Kansas and New Jersey) 13 and both regional network jurisdictions (New York and Texas). PROJECT MENUS: In negotiating with CMS to obtain DSRIP approval, states develop project menus to reform care. The menu specifies the nature of the reform projects and the performance metrics associated with each. Participating providers in a state typically pick projects from the menu. In New Jersey, for instance, state officials offered its hospitals a menu listing seventeen alternative projects sorted under eight health conditions: HIV/AIDS, cardiovascular disease, asthma, diabetes, obesity, pneumonia, behavioral disorders, and substance abuse. 14 Kansas offered its two participating hospitals a shorter menu of six projects. New York listed forty-four menu items, with each performing provider system required to conduct five to ten projects. 15 In turn, the Texas Health and Human Services Commission offered its providers a choice among more than 100 projects. Lengthier project menus give hospitals and the regional networks greater latitude to shape the targets of DSRIP reform. Certain health conditions tend to attract more attention than others. For instance, in picking from among eight health conditions, thirty-six of the fifty-one New Jersey hospitals chose projects related to cardiac or diabetes care. No hospital opted for initiatives aimed at HIV/AIDS; obesity and pneumonia attracted only one hospital each. PERFORMANCE INDICATORS AND METRICS: The states, with CMS concurrence, establish myriad performance indicators and metrics for each menu item. DSRIP nomenclature sorts performance indicators into four categories or stages. The first is infrastructure development, which encompasses process measures such as expanding the electronic record keeping system. The second stage focuses on process indicators associated with system redesign. Clinical outcome improvements comprise the third category of metrics and population focused improvements the fourth. Exhibit 2 illustrates the indicators that Kansas and New Jersey use for each of the four stages. For Kansas, we present some of the nineteen metrics for a menu project focused on the expansion of patient-centered medical homes and neighborhoods. The New Jersey examples come from a menu item aimed at improving the quality of care for diabetes and hypertension. DSRIP providers receive performance pay for activities related to infrastructure development and system redesign early in the life of the waiver. Clinical and population health outcome measures weigh more heavily in subsequent years. DSRIP states typically distinguish between pay-for reporting and pay-for-performance metrics. The former category incorporates measures a hospital must present to be eligible for DSRIP (for example, the percentage of people with asthma receiving appropriate medication). But the level of performance on the metric does not affect DSRIP payment. In contrast, pay-forperformance measures require the provider to achieve a more specific target to receive payment (for example, reduce the number of patients with emergency department visits for asthma within the last six months by some percentage). Pay-forreporting reflects the desire of CMS and some July 2015 34:7 Health Affairs 1165

Paying For Care Exhibit 2 Delivery System Reform Incentive Payment Performance Metrics: Examples From Kansas And New Jersey Category/stage Kansas: patient-centered medical home New Jersey: improve quality of diabetes care Infrastructure development System redesign Clinical outcome improvement Population-focused improvement Identify team members Develop a work plan Expand ambulatory clinic office hours to include service on nights and the weekends Locate a site for urgent care in close proximity to the emergency department Percent of primary care patients with an assigned medical home Percent of patients who have been recommended for immunizations Percent of children who have height and weight documented at each visit Reductions in emergency department visits for medical home patients Decreases in thirty-day hospital readmission rate Assemble and list all project staff members Complete quality improvement plan Document total number of participating patients Document by describing initiation of a pilot program including any start-up challenges Lower uncontrolled diabetes admission rate Lower hypertension admission rate Lower long-term diabetes complications Reductions in postoperative sepsis Reductions in central line associated bloodstream infection events SOURCES Kansas Department of Health and Environment. KanCare program protocols [Internet]. Topeka (KS): KDHE; 2013 May 31 [cited 2015 May 26]. Available from: http://www.kancare.ks.gov/download/kansas_dsrip_protocols.pdf. State of New Jersey Department of Health. Delivery System Reform Incentive Payment (DSRIP) program planning protocol [Internet]. Trenton (NJ): The Department; 2013 [cited on 2015 Jun 9]. Available from: https://dsrip.nj.gov/documents/nj%20dsrip %20PLANNING%20PROTOCOL_v1_08-09-2013.pdf. states to bolster the capacity of hospitals and other providers to collect, use, and share performance data. 16 PAYING PROVIDERS: Payment-for-performance processes vary widely across DSRIP states. Most incentive payments are project specific. Providers receive payments if they meet standards embedded in the metrics associated with each project they have selected. In states, such as New Jersey, some metrics are scored as one (achieved) or zero (not achieved). In other instances, the state establishes a baseline for providers that must then achieve a certain percentage improvement to receive performance pay (for example, reducing thirty-day readmission rates for patients hospitalized with heart failure). Some states permit partial payment for less than full achievement of a metric. A given project on the state menu typically has a defined set of metrics that apply to all providers implementing it. Texas, however, permits the metrics associated with a project to vary with providers selecting from an approved list. The various projects have different monetary valuations associated with them. Texas has given participating providers leverage in negotiating these valuations with state officials. In contrast, New York relies on a complex formula to value the achievement of various project metrics. In turn, the performing provider systems established in New York assume responsibility for allocating these monies within the provider networks. 12 In dealing with this internal allocation, the provider systems must deal with tensions over the relative compensation of network participants. In New York, for instance, the association representing community health centers has voiced concerns that the performing provider systems are driven by hospitals ; it has called for protections to ensure that funds flowing to primary care providers fairly reflect their contribution to achieving performance goals. The association has also expressed concern about the adequacy of compensation that primary providers receive from participating managed care organizations. 17 Some DSRIP states also distribute funds related to universal metrics those applicable to all participating providers. New Jersey, for example, has identified twelve stage-four metrics for a universal performance pool. (For example, all participating hospitals that achieve certain reductions in postoperative sepsis and in admissions rates for diabetics with short-term complications receive DSRIP payments from this pool.) New Jersey also illustrates how hospitals may divide funds among project-specific and universal measures. The proportion of DSRIP monies allocated to the universal performance pool is slated to grow from 10 percent to 25 percent. New York has taken the imposition of universal metrics to an even more challenging level that goes beyond the achievements of its twenty-five performing provider systems. CMS will withhold federal monies from the state s DSRIP pool if providers across the entire state fail to achieve a 25 percent reduction in avoidable hospital admissions over five years. Should this occur, the incentive payments received by the performing provider systems would decline even if they score well on project-specific metrics. 1166 Health Affairs July 2015 34:7

DSRIP waivers have reinforced efforts to develop information systems that produce valid and timely feedback on performance. Discussion The post-aca period has witnessed a panoply of initiatives to galvanize delivery system reform. Against this backdrop, DSRIP represents a significant CMS effort to replace one form of hospital subsidies, upper payment limits, which lacked transparency and were hard to track, with financing rooted in performance-based accountability. Performance-based systems inexorably present implementation challenges. A fully functioning performance model depends on establishing and communicating reform goals and objectives, the selection of indicators and metrics that accurately reflect the goals, the development of an information system that produces valid and timely reports about performance metrics, and the use of the performance results in administrative and policy decisions. Although it is still early in the implementation process, DSRIP states in varying degrees show signs of addressing these four challenges. Goals Under the banner of the Triple Aim, considerable decentralization characterizes the DSRIP approach to reform goals. 8 CMS has been assertive in negotiations with the states over DSRIP waiver proposals as well as project menus and metrics. But, in fact, states, by establishing the lengthy menu options, have given local providers considerable leeway to determine which health challenges receive priority for delivery system reform, unlike a top-down system where the options are chosen by a federal authority. DSRIP is in this sense locally responsive and reflects the expertise of frontline providers as to what can be feasibly accomplished. Performance Indicators And Metrics Inevitably, performance-based accountability systems raise issues of whether key metrics distort providers behavior, causing them to ignore important but less measurable quality dimensions. Some contend that designers of such systems often rely on metrics readily available from existing data sets, instead of developing more valid measures. 18 The failure to appropriately riskadjust measures can also lead to problems. For instance, some suggest that a focus on reducing thirty-day hospital readmissions may unfairly penalize safety-net hospitals because it fails to consider their patients illness severity and social challenges. 19 Whether DSRIP states can risk adjust this and other measures to address such distortions remains to be seen. While performance metrics have limits, 20 those employed in the DSRIP waivers tend to reflect the expertise and recommendations of major professional associations and agencies, such as the National Committee for Quality Assurance, the American Academy of Pediatrics, and the Agency for Healthcare Research and Quality. 14 Information Systems DSRIP waivers have also reinforced efforts to develop information systems that produce valid and timely feedback on performance. DSRIP states have generally identified plausible sources of performance information. In New Jersey, for instance, much of the information on the metrics will come from the state s Medicaid Management Information System and hospital electronic health records. Still, challenges exist. Some state data systems are not sufficiently developed to provide the information required for the DSRIP programs at all or in a timely fashion. Incompatibilities between the electronic records of hospitals and collaborating community physicians can impede information sharing. In some instances, hospitals are unable to rely on electronic health records and have to extract pertinent information from medical charts. At a more basic level, many Medicaid providers still lack electronic records. Performance Consequences Performancebased accountability rests on the assumption that success or failure has concrete implications for those charged with attaining specified objectives. In this regard, officials often face challenges in tying the performance of an agent to positive or negative consequences. 21 In the case of DSRIP, however, CMS and the waiver states appear poised to reward providers that perform well on the metrics and withhold funding from those that do not. Providers that do poorly may well point to mitigating circumstances and request a reconsideration of a negative ruling on DSRIP pay. But CMS, not just the state, must approve the incentive awards heightening prospects that payment for performance will in fact occur. Policy Questions The variation in DSRIP programs among the waiver states raises policy questions about the efficacy of various ap- July 2015 34:7 Health Affairs 1167

Paying For Care proaches. One issue revolves around the advantages of a broadly participative New Jersey model compared to the more common pattern of targeting system reform on hospitals that disproportionately serve Medicaid enrollees and the uninsured. New Jersey has engaged nearly 70 percent of the state s hospitals in DSRIP reform. Attractive as broad provider participation is, however, it could divert funds from institutions that primarily serve the disadvantaged. Hence, the New Jersey model has less appeal in states where key safety-net hospitals are experiencing high levels of fiscal stress. A second policy question concerns the degree to which CMS should encourage the regional network model over the hospital-oriented one. CMS increasingly appears to favor the regional approach, but it is premature at present to judge the relative efficacy of the two models in promoting the Triple Aim. The degree to which either model will provide sufficient inducements to overcome the organizational and financial barriers to delivery system reform remains to be seen. It is hoped, however, that federalism will provide laboratories for policy learning. The relative degree to which providers in the regional and hospital-oriented DSRIP states attain their performance metrics will cast some light on the plusses and minuses of the two models. So, too, will the formal evaluations being conducted of each demonstration. Durability Whether hospital-focused or regional, DSRIP initiatives face an array of factors that will affect their long-term durability and impact. One important factor is the extent to which states and providers find that the rewards of DSRIP participation outweigh its substantial administrative costs. To the degree that providers fail to meet the often demanding performance or budget-neutrality targets and suffer penalties, they will likely join with state officials in pressuring federal policy makers to modify DSRIP or replace it with alternative Medicaid funding. More broadly, developments in the political stream will shape the future of DSRIP. Without obtaining any changes in law, the new presidential administration will in 2017 have great leeway to refashion DSRIP implementation. Consider issues of budget-neutrality.while demonstration waivers are supposed to be budget-neutral, CMS has vast discretion to interpret this standard. 22 If the agency wishes to encourage a waiver, it can approve more optimistic state estimates of an initiative s costs. In this vein, CMS s receptivity to cost calculations that Texas officials used Most DSRIP states have made considerable headway in planning and otherwise laying the groundwork for their initiatives. to obtain approval of its DSRIP waiver drew criticism from the Government Accountability Office for failing to uphold the budget-neutrality standard. 7 CMS also allowed New York to use banked savings from its existing managed care waiver to support its claims of budget-neutrality. 23 An incoming administration committed to a more stringent methodology for assessing budget-neutrality could impede the renewal of DSRIP waivers and the propensity of other states to pursue them. Regardless of whether this occurs, the example reinforces a more general point. As a product of executive branch discretion, DSRIP s durability depends heavily on CMS values and priorities, as shaped by a presidential administration. Conclusion The reform efforts authorized by the ACA and through waivers, such as DSRIP, have been called a spaghetti approach to enhancing the health care system: throw a bunch of strands against the wall and see what sticks. 24 It is too soon to judge whether DSRIP will be among the reforms that stick. The degree to which DSRIP states will achieve their performance objectives is unclear. Previous CMS efforts to use pay-forperformance have generated mixed results. 25 Still, CMS and the state Medicaid agencies have paid careful attention to implementation details, and most DSRIP states have made considerable headway in planning and otherwise laying the groundwork for their initiatives. The replacement of unconditional upper payment limit funding with financial assistance rooted in performance-based accountability may well be a substantial step forward. 1168 Health Affairs July 2015 34:7

The authors thank Scott Brunner, Joel Cantor, and anonymous reviewers for their helpful comments on a prior draft. NOTES 1 New Mexico and Oregon have incentive waivers that are less comprehensive and do not include funding for specific projects. 2 The second Bush administration, for example, frequently negotiated changes in waivers originally approved by the Clinton administration. Thompson FJ. Medicaid politics: federalism, policy durability, and health reform. Washington (DC): Georgetown University Press; 2012. p. 135 8. 3 Thompson FJ, Gusmano MK. The administrative presidency and fractious federalism: the case of Obamacare. Publius. 2014;44(3): 426 50. 4 Coughlin TA, Zuckerman S. States strategies for tapping federal revenues: implications and consequences of Medicaid maximization. In: Holahan J, Weil A, Weiner JM, editors. Federalism and health policy. Washington (DC): Urban Institute Press; 2003. p. 145 78. 5 Government Accountability Office. Medicaid: more transparency of and accountability for supplemental payments are needed. Washington (DC): GAO; 2012. p. 5. 6 Sparer M. Medicaid managed care: costs, access, and quality of care. Princeton (NJ): Robert Wood Johnson Foundation; 2012 Sep. 7 Government Accountability Office. Medicaid demonstration waivers: approval process raises cost concerns and lacks transparency. Washington (DC): GAO; 2013. p. 7 30. 8 Berwick DM, Nolan TW, Whittington J. The Triple Aim: care, health, and cost. Health Aff (Millwood). 2008; 27(3):759 69. 9 Harbage P, Happoldt IA, Ryan J. A review of select state DSRIP programs and implications for California s Medi-Cal 1115 waiver renewal. Sacramento (CA): Harbage Consulting; 2014. 10 Seifert RW, Grenier M, Sullivan JC. The MassHealth waiver extension for state fiscal years 2015 2019: foundation for coverage, engine for innovation. Charlestown (MA): Center for Health Law and Economics, University of Massachusetts Medical School; 2015 Feb. 11 Summer L. Integration, concentration, and competition in the provider Marketplace. Washington (DC): AcademyHealth; 2010 Dec. 12 New York State Department of Health. NY DSRIP program funding and mechanics protocol [Internet]. Albany (NY): The Department; [updated 2014 Apr 14; cited 2015 May 26]. Available from: https:// www.health.ny.gov/health_care/ medicaid/redesign/docs/program_ funding_and_mechanics.pdf 13 We chose Kansas and New Jersey partly because they ranged from lowest to highest in hospital participation. 14 Department of Human Services. Delivery System Reform Incentive Payment (DSRIP) program funding and mechanics protocol [Internet]. Trenton (NJ): The Department; 2013 [cited 2015 May 26]. Available from: http://www.medicaid.gov/ Medicaid-CHIP-Program- Information/By-Topics/Waivers/ 1115/downloads/nj/Comprehensive- Waiver/nj-1115-request-dsrip- funding-mechanics-protocol- 07302013.pdf 15 A performing provider system with ten projects can opt to pursue an eleventh. 16 Brennan N, Oelschlaeger A, Cox C, Tavenner M. Leveraging the big-data revolution: CMS is expanding capabilities to spur health system transformation. Health Aff (Millwood). 2014;33(7):1195 202. 17 Community Health Care Association of New York State. Comments on the funding mechanics protocol and the strategies menu and metrics for Delivery System Reform Incentive Payment program [Internet]. Albany (NY): CHCANYS; 2014 May 14 [cited 2015 May 26]. Available from: http://www.chcanys.org/client uploads/2014_policy/dsrip/ CHCANYS_DSRIP_Funding_ Mechanics_Protocol_Comments_ 051414_FINAL.pdf 18 Smith PC. Performance measurement in health care: history, challenges, and prospects. Public Money Manage. 2005;25(4):213 20. 19 Gilman M, Adams EK, Hockenberry JM, Wilson IB, Millstein AS, Becker ER. California safety-net hospitals likely to be penalized by ACA value, readmission, and meaningful-use programs. Health Aff (Millwood). 2014;33(8):1314 22. 20 Coleman K, Reiter KL, Fulwiler D. The impact of pay-for-performance on diabetes care in a large network of community health centers. J Health Care Poor Underserved. 2007;18(4): 966 83. 21 At times, for instance, CMS has done little to assess whether demonstrations ultimately achieve the performance and cost projections embedded in the original waiver proposals. 22 The budget-neutrality standard applies to the entire demonstration of which the Delivery System Reform Incentive Payment program is one part. 23 Abrams M. Putting New York DSRIP in context: what we can learn from other states? New York (NY): Commonwealth Fund; 2014 Nov 20. 24 Gruber J. Massachusetts points the way to successful health care reform. J Policy Anal Manage. 2011;30(1): 184 92. 25 Werner RM, Kolstad JT, Stuart EA, Polsky D. The effect of pay-forperformance in hospitals: lessons for quality improvement. Health Aff (Millwood). 2011;30(4):690 8. July 2015 34:7 Health Affairs 1169