Feasibility Study for ACO Pilot of Community Based Payment Reform: Summary of Objectives, Key Issues and Project Structure 8/15/08

Transcription

1 Feasibility Study for ACO Pilot of Community Based Payment Reform: Summary of Objectives, Key Issues and Project Structure 8/15/08 Health Care Reform Commission Vermont State Legislature Jim Hester PhD Director

2 Contents Page Summary 3 Introduction: Intent of the Feasibility Study 4 Community Based Payment Reform: Example of Accountable Care Organization 6 Design of a Vermont Pilot: Updated Major Tasks and Issues A. Scale and scope of the pilot 8 B. Responsibilities of the ACO and criteria for ACO participation 9 C. Design of the incentive system 11 D. Administration and funding of the pilot 13 Proposed Project Structure 14 Appendix: KNG Health Consulting letter of 7/3/08 Page 2 of 24

3 Feasibility Study for Pilot of Community Based Payment Reform Summary Changing the way that providers are paid for their services is a key component of health care reform. Vermont has made a significant beginning in this area through its reform of reimbursement for primary care physicians as part of the enhanced pilots in the Blueprint for Health chronic care initiative. This memo proposes building on that start by considering expanding the payment reform to a community level and including specialist physicians, the local acute care hospital, and other key providers. The conceptual model for this reform would be the Accountable Care Organization (ACO) as described by Elliot Fisher. The ACO would be a vehicle for enabling a regional health care delivery system composed of key providers of primary and secondary care within a community to foster a shared accountability for both the costs and quality of care of the population they serve. The ACO would have financial incentives to recapture part of the savings realized by operating more effectively, e.g. reducing the volume of questionable surgical procedures, and to avoid future capacity driven growth. This memo presents a working framework for the feasibility study based on an initial draft which was circulated in June and refined with the assistance of KNG Health Consulting, interviews and feedback from key stakeholders, and a public meeting. The memo summarizes the Accountable Care Organization concept as one example of community based payment reform and describes how the concept might be translated into a pilot in a small number of communities in Vermont. It lists key questions regarding the design and implementation of the Vermont pilot, organized into four clusters: - scale and scope of the pilot - responsibilities of the ACO and criteria for ACO participation - design of the incentive system - administration and funding of the pilot The product from the feasibility study will be an assessment of each of the key issues above to determine whether or not there are any show stoppers which would make it unlikely that the pilot could be implemented. This will require some preliminary work on the design of the pilot, but this study will not produce a detailed design. The product will include an overall recommendation of whether or not to proceed with a pilot at this time, the scope of the pilot, an estimate of the resources required to design and implement the pilot, and options for funding. The project will be staffed by commission staff and consultants and will use a workgroup of key stakeholders Separate study groups will be created as needed, for example to focus one of the clusters of major issues outlined above, or to refine the issues created by different provider structures (PHO vs. more integrated). The project will communicate with and seek input from a broad spectrum of interested parties through the dissemination of work papers and periodic public meeting, including progress reports to the commission. Page 3 of 24

4 Introduction: Intent of the Feasibility Study One of the most consistent themes of both the studies conducted by the commission and the testimony before committees during the last legislative session is that sustainable health care reform requires significant changes to the way providers are paid. The health reform legislation passed in the 2008 session (H.887) included the following provision: The commission on health care reform is charged with making recommendations to meet the goal of section 902 of Title 2 that by 2009, Vermont has an integrated system of care that provides all Vermonters access to affordable, high quality health care that is financed in a fair and equitable manner. Achieving this will require a series of fundamental changes which cumulatively will build a more integrated system with aligned financial incentives. The commission on health care reform should conduct studies to develop key building blocks for moving toward such a system in Vermont, to the extent that funds and staffing resources are available, including: (1) The feasibility of community-based payment reform and integration of care. This study should assess the feasibility of alternative designs for a pilot project to test using a systemwide budgeting initiative at the regional level within the state, including a design based on the accountable care organization model. The following language crafted by the House Healthcare Committee and included in the original House version of the bill provided additional insights into the intent and objectives of this initiative: Sec. 3. FEASIBILITY STUDY FOR COMMUNITY-BASED PAYMENT REFORM AND INTEGRATION OF CARE (a) The Health Care Affordability for Vermonters Act of 2006 expressly stated the general assembly s intention to ensure that all Vermonters receive affordable and appropriate health care, and that health care costs be contained over time. Continued increases in health care costs at unsustainable rates represent a threat to successful health care reform. Symptoms of this problem include: (1) High per-capita medical costs and an unsustainable rate of increase; (2) The failure of the existing health care system to deliver appropriate care in approximately 50 percent of all cases involving patients with chronic conditions and one to five percent of all common inpatient admissions; (3) Wide variations in performance and the counterintuitive finding that more care often results in poorer outcomes; (4) An estimated 25 to 40 percent of medical services provided unnecessarily due to preventable errors and no evidence of value; and (5) Avoidable administrative complexity and cost. (b) Despite the continued progress of cost control efforts enacted in 2006, it is clear that additional steps are necessary to address this complex issue. The commission on health care Page 4 of 24

5 reform shall facilitate a study to assess the feasibility of alternative designs for a pilot project to test using a system-wide budgeting initiative at the regional level within the state, including a design based on the accountable care organization model. The study should consider building on current significant efforts to build new forms of integrated community health systems, such as the FQHC model being implemented by Springfield Medical Care Systems and the medical staff reorganization being considered by Southwest Vermont Medical Center. (c) The goals of the pilot project are to achieve the Institute for Healthcare Improvement s Triple Aims of improving population health, reducing total per-capita costs, and enhancing the patient experience and to address cost controls via the following three strategic issues: (1) How to foster the development of local organizations which integrate the delivery of care at the local community level across the full spectrum of services; (2) How to counter the belief that more care is always better by providing balanced information on risks and benefits; and (3) How to reform the payment system to move away from the current system which rewards more care for high margin treatments and toward a shared risk and shared savings approach which rewards the Triple Aims. (d) The design of the pilot should build on the initiatives being developed in the Blueprint for Health, particularly the payment reform for primary care physicians, the adoption of the medical home model, and the development of community-based care coordination teams. This memo presents a working framework for the feasibility study based on an initial draft which was circulated in early June and refined with the assistance of KNG Health Consulting, input from a series of interviews with key stakeholders, written feedback from a number of parties including Elliot Fisher, and a public meeting. The memo summarizes the Accountable Care Organization concept as one example of community based payment reform and describes how the concept might be translated into a pilot in a small number of communities in Vermont. It concludes with a list of key questions regarding the design and implementation of the Vermont pilot which need to be addressed in the feasibility study, and the proposed initial project structure for the feasibility study. The report from KNG Health Consulting summarizing their detailed comments on the interviews and the issues is attached in the Appendix. Page 5 of 24

6 Community Based Payment Reform: Example of Accountable Care Organization In testimony before the House Health Care Committee, Elliott Fisher introduced the concept of an Accountable Care Organization. The ACO would be a vehicle for enabling a regional health care delivery system composed of key providers of primary and secondary care within a community to foster a shared accountability for both the costs and quality of care of the population they serve. The ACO would have financial incentives to recapture part of the savings realized by operating more effectively, e.g. reducing the volume of questionable surgical procedures, and to avoid future capacity driven growth. These incentives would reward effective population based care, rather than high margin specialized services. The ACO could use part of its shared savings to reallocate resources to under funded components such as support for chronic illness care or information technology. Fisher identified three key characteristics of an ACO. It should 1. be large enough to support comprehensive performance measurement of both cost and quality measures, and a population based budget. This is critical to being able to document for patients that less care can in fact be better care. The feasibility study will build on the Blueprint for Health s participation in the Institute for Healthcare Improvement s Triple Aim initiative and use the following three dimensions of performance: population health, total per capita cost and patient experience. 2. provide or effectively manage the full continuum of patient care, outpatient and inpatient to ensure accountability across sites of care. One of the challenges of the ACO is to integrate the delivery of care across this continuum at the community level in a much more effective manner than existing systems. 3. participate in shared-savings approaches to payment reform as an interim step toward fundamental payment reform. He provided several examples of possible ACO s including large, multi-specialty group practices which own their own hospitals (Lahey Clinic, Mayo Clinic), Physician Hospital Organizations, and Extended Hospital Medical Staff s. Recently, Shortell and Casalino expanded on Fisher original concept in calling for the creation of Accountable Care Systems and adding independent practice organizations and health plan-provider organization network to his list Page 6 of 24

7 (JAMA Vol 300, No 1, July 2, 2008 pp. 95-7). For the purposes of this memo, the ACO model is flexible enough that it can encompass an extremely wide range of payment models, provider structures and incentives which would create the system-wide budgeting initiative at the regional level called for in the legislative charge. At this stage, the ACO model will be used to identify the broad issues and concerns that have to be addressed in community based payment reform. If credible options emerge during the feasibility study that do not fit under the ACO umbrella, then they will be identified and assessed. One of the essential responsibilities of the ACO is to serve as a system integrator and reduce the fragmentation in the current system. The IHI Triple Aim project has identified multiple levels of integration, starting at the individual patient level where the medical home serves at the integrator, and building through the community, regional and state level. The feasibility study will explore alternative combinations of integrator role at the community level, building on the medical home model at the patient level, and supported in turn by the regional/state level integration structure which is being created by Vermont s health care reform initiatives. Individual payers in Vermont have experimented with a per capita medical cost incentive models for community based provider organizations such as Central Vermont PHO, the United Health Alliance and Vermont Managed Care for the last decade. These experiments have built a solid foundation for the proposed pilot, but also provide some important lessons. First, scale is clearly important. Like the payment reform for primary care physicians currently being implemented in the enhanced Blueprint pilots, any meaningful payment reform at the community level must also involve the broadest possible coalition of public and private payers. The eventual participation of Medicare will be essential to the long term success of the reform, but one of the design issues will be to assess whether or not a critical mass can be achieved through the combination of the three major commercial payers and Medicaid. Second, a broader range of incentive models needs to be explored for provider organizations that do not have either the scale or infrastructure of a large, mature PHO. If the ACO model only works for these larger systems, it will reduce its applicability to many parts of the country where such systems do not currently exist. Page 7 of 24

8 Design of a Vermont Pilot: Major Tasks and Issues The following is an initial list of the tasks and issues that need to be addressed in the feasibility study and initial pilot design, organized into four clusters: 1) the scale and scope of the pilot, 2) the responsibilities of the ACO and criteria for ACO participation, 3) the design of the incentive system and 4) the administration and funding of the pilot. A. Scale and scope of the pilot 1. Define the local community: - Generally defined as a hospital service area for an acute care hospital or hospital system. - The local delivery system is the hospital, medical staff affiliated with the hospital (primary care providers may not have admitting privileges), nursing homes, visiting nurse services, ancillary service providers, and possibly behavioral health. (The responsibilities of the delivery system in an ACO pilot and the criteria for qualifying as an ACO are discussed in the next section.) - Population defined in terms of the medical home model: people using the local primary care providers as their principal source of primary care. There will be some leakage of residents in the service area into and out of the local system, with some residents going to another community and some non residents receiving care in the local system. If people were required to designate a primary care physician, it would help, but implementing this will be problematic. The ACO pilot will benefit from the experience of the Blueprint enhanced pilots with the medical home concept, and may include one of those enhanced pilot communities. - Possible methods for attributing a population to the ACO: 1) possible enrollment or explicit buy in by patients, 2) use Medicare model of predominance of non inpatient visits, 3) other? - What scale of covered population is needed to be able to test the concept and what are the options for realizing that scale? What size of population would be a minimum threshold, and how does this threshold vary with the incentive model? Initially, a population of 30,000 was considered the minimum, but based on preliminary feedback, some pilot designs may require ,000 people. Achieving this scale in Vermont would require coordination across delivery systems, which would raise a host of new issues. If Medicare is not participating initially, can the combination of major commercial payers Page 8 of 24

9 and Medicaid have sufficient impact? Could Medicare be brought in through a MMA Section 646 waiver or demonstration? Will payers be able to include their self insured populations? A previous attempt by MVP to include IBM in such an incentive program had to be aborted for a variety of reasons. - What is the minimum time commitment required to provide a real incentive to drive change? Three years has been suggested as a working time frame often used in Medicare incentive demonstrations. 2. Define the scope of services included - Are there any options for carve outs from traditional comprehensive health insurance benefits, particularly behavioral health and outpatient prescription drugs? - Could the scope of services vary with pilot site, or is it one consistent broad package including most referral care? - Does the scope of services vary for different payers to reflect differences in benefit design? B. Responsibilities of the ACO and criteria for ACO participation: What range of potential ACO structures could be tested in the pilot? 1. Responsibilities of the ACO: The ACO must provide the infrastructure which will serve as a system integrator at the local community level across the full spectrum of services. The IHI Triple Aim project suggests that creating this system integrator role is a critical step in being able to reduce the fragmentation of payment and care and achieve a balance between the three aims of population health, total costs and patient experience. The potential functional roles of the community level system integrator include: o community based cost control platform for managing total per capita costs o care coordination across the full spectrum of care o community based public health initiatives to support population health o support of process improvement within and across providers 2. Establish criteria for local ACO infrastructure: - What are the minimum requirements for a pilot ACO, e.g. participation of the hospital, some percentage of primary care providers and some key specialist providers. Page 9 of 24

10 - Care/utilization management capabilities: what are the relative roles of the ACO and the payers? To what extent is it important to change the existing care management responsibilities. For example, should the ACO have a local Care Coordination Team like those in the enhanced Blueprint pilot? - What local data collection, management, analysis and reporting capabilities are essential? - How much experience with quality improvement and process redesign is necessary? - Is the successful implementation of a medical home model a pre-requisite for an ACO pilot? - Should preference be given to at least one community that already has a functioning Physician Hospital Organization (PHO) as a community level integrator which has had a risk sharing contract with MVP and/or Blue Cross for a number of years? Such organizations have demonstrated the capability to bring at least one hospital and local physicians to the table for meaningful conversations about resource use in the community. Potential candidates are a. Northwest Vermont: Vermont Managed Care composed of Fletcher Allen (Burlington), Porter Hospital (Middlebury), Northwest Vermont Medical Center (St Albans) and Copley Hospital (Morrisville) b. Bennington: United Health Alliance and Southwest Vermont Medical Center c. Barre: Central Vermont PHO and Central Vermont Medical Center - If the ACO is built around an existing structure, such as a PHO, to what extend does it need to expand to add key major local providers who are not currently participating? - What are the characteristics of the local structure where trade-offs could be made, e.g. risk sharing experience vs. degree of integration of the provider organization structure. A highly integrated structure may not need to have as much experience because they could be capable of building new capabilities more rapidly -.Several communities (Springfield, Bennington) are in the midst of significant restructuring to create much more highly integrated delivery systems. Considered collectively, what range of structures do we want to test in the pilot program in order to learn as much as possible about what characteristics are most important? Page 10 of 24

11 C. Design of incentive system 1. Define the performance measures for each of the three IHI Triple Aim dimensions. It might be valuable to consider using a common core set of measures that could be used by the Triple Aim participants and other pilots on measurement that NQF / AHRQ may end up helping with measure selection. - Population health: the health outcomes of the total population served by the care system - Total per capita costs for the population served - Patient experience: the experience of the individual as he/she moves through the care system 2. How does the incentive model vary with ACO scope and structure? Could it be quite different for a large scale, mature PHO vs. a smaller provider with a more integrated structure, e.g. Vermont Managed Care vs. Springfield Hospitals proposed FQHC model.? 3. An incentive model which is performance based is inherently retrospective it rewards providers for results that they have already achieved. Is there a need for a prospective component to the incentive model which will prime the pump and provide seed capital for investments in improvements? If so, what are possible sources of such funds, and what is the appropriate mix of prospective and retrospective incentives? 4. Define the payment model: - Begin with three major commercial insurers (Blue Cross, MVP, and CIGNA) and Medicaid, with intention of expanding to Medicare through a targeted demonstration project within two years. - How should the incentive model blend performance in the three dimensions of total per capita cost (the medical expense budget), population health and patient experience. What are the relative weights of the three dimensions? (See the Physician Group Practice Medicare demonstration) - How is the incentive pool funded? Is there a withhold on provider payments to help fund the pool and to create some degree of risk? - Should some of the shared savings be distributed to residents / enrollees through Healthy behavior incentives, etc.? 5. Global budget component of model Page 11 of 24

12 - Incentive based on a global budget developed from an actuarially based per member per month global medical expense target for an all inclusive menu of services, multiplied times an ACO population. - How does global budget for population covered by a specific payer reflect the provider payment contracts and/or utilization history of that payer, e.g. Medicaid? - Is there any risk adjustment for population segments, e.g. based on demographics or case mix? The problem of rewarding up-coding as a mechanism of demonstrating savings needs to be considered. - Services would continue to be paid on a fee for service basis using the current contracts for each payer, so there would not be a common fee schedule. - Financial settlement: periodically the total actual expenses of the ACO s population for all the services they have used would be compared to the global budget. Actual expenses would include care provided outside of the ACO by other providers, such as tertiary hospitals. If the actual expense were under budget, the payers participating in the pilot would share those savings with the ACO by making a lump sum payment to it. The specific s of how to share the savings are an art form, with possible variations ranging from the most simplistic (50/50 between payers and ACO) to complex options involving multiple corridors, caps on payments, etc. - The ACO would have the flexibility to allocate its share of the savings between the providers in the ACO and structural improvements such as EMR s. - ACO would have regular reports on total expenses compared to budget. Reporting could be based on compiling data from four payers, or possibly on the all payer claims data base being created by BISHCA 6. Establish risk sharing models and limits - This is NOT a capitation model, but it is essential to include limited risk if expenses exceed budget. What options should be available for risk sharing percentages and corridors? - How is the provider risk financed, e.g. withhold, promissory notes? - What are limits on risk sharing: aggregate for organization as well as individual stop loss? Who provides stop loss insurance and determines its costs? Page 12 of 24

13 D. Administration and Funding of the Pilot 1. Build capacity to design and implement the pilot - What organization is best suited for administering the pilot? Is this an existing public entity, or a contracted entity? - Who determines the budgets and performance criteria? How are they updated each year? From the feedback, it was clear that engaging an experienced, unbiased actuary who would use accurate data in a fair and transparent process was critical for the success of the pilot. - Who generates the periodic reports? Does each payer produce a report which is rolled up into a consolidated data? - Are accrual based reports which include IBNR available for the ACO? - What is the contracting framework for the payment reform? Does each payer have its own contract, or is there a common contract? 2. Funding of pilot: What additional resources will be required? - administrative costs of pilot manager and ACO s - what are the opportunities for external funding for design and/or implementation? 3. Evaluation: How will we know whether or not the pilot has worked? - Evaluation design and funding - Issue of confounding interventions: multiple interventions implemented concurrently. 4. What can we learn from other s experience? - Physician Group Practice Medicare demonstration: - Vermont specific data on population based cost and utilization - opportunities for savings/improvement? - Efforts in other states (New Jersey, Oregon) to implement the ACO or similar models - Should there be a learning collaborative established with others, perhaps with some funding. The product from the feasibility study will be an assessment of each of the key issues above to determine whether or not there are any show stoppers which would make it unlikely that the pilot could be implemented. This will require some preliminary work on the design of the pilot, Page 13 of 24

14 but this study will not produce a detailed design. For example, it could determine that reasonable data sources exist to create reliable measures for the three performance dimensions, but not specify exactly which measures to use. The product will include an overall recommendation of whether or not to proceed with a pilot at this time, the scope of the pilot, an estimate of the resources required to design and implement the pilot, and options for funding the pilot. In the appendix, KNG Health Consulting suggests using an RFP process as a tool for building and selecting successful pilot sites. The process would differentiate between 1) critical elements of a successful pilot which would be required of all sites, 2) criteria which might be negotiable within certain ranges, and finally, 3) elements where we would be looking for creative responses. In particular, see their Table 1 for an example of how provider systems with different characteristics might try to qualify as a pilot ACO. Proposed Project Structure The project will be led by the Director of the Health Care Reform Commission, Jim Hester, who will be assisted by commission consultants. Elliot Fisher has agreed to serve as an advisor to the project, and the staff of the Engelberg Center for Health Reform at the Brookings Institution will also be resources for the study. The initial composition for the Workgroup of key stakeholders is listed in Table 1, but is expected to expand. Separate study groups will be created as needed, for example to focus one of the clusters of major issues outlined above, or to refine the issues created by different provider structures (PHO vs. more integrated). The project will communicate with and seek input from a broad spectrum of interested parties through the dissemination of work papers and periodic public meeting, including progress reports to the commission. Page 14 of 24

15 Table 1 Feasibility Study Workgroup Initial List of Organizations Vermont Managed Care Central Vermont PHO United Health Alliance Springfield Hospital Vermont Association of Hospitals and Health Systems Vermont Medical Society Blue Cross Blue Shield of Vermont CIGNA MVP Healthcare Office of Vermont Health Access BISHCA Blueprint for Health Page 15 of 24

16 Appendix: KNG Health Letter July 3, 2008 Mr. Jim Hester Director Health Care Reform Commission Baldwin St Montpelier VT Dear Jim: This letter provides our comments on the feasibility assessment being conducted by the Health Care Reform Commission to determine whether the Accountable Care Organization (ACO) model can be pilot tested in Vermont. Our thinking on the ACO pilots is evolving as we learn more and get our heads wrapped around the issues. We hope that our perspective is helpful in further shaping a direction for the ACO pilots. We present our comments in five areas. First, we summarize initial provider and payer feedback. Second, we suggest some additional areas for inquiry with stakeholders in future communications. The remaining sections discuss considerations for a process going forward. Specifically, we find it useful to think in terms of an end product, which we view as being a request for proposal (RFP). In addition to identifying the goals of the project, the RFP would need to specify the critical or required elements as well as those elements for which the sites may have more flexibility in determining (perhaps within a range). In this context, it is also important to consider how sites will be selected and how the pilots will be evaluated. We consider these issues in the section Building Successful Models to Pilot Test. Understanding of the Issue Under legislation H.887, The Vermont Health Care Reform Commission (hereafter, the Commission) has been charged to perform a feasibility study of developing a viable payment model for ACOs to be tested in the State of Vermont. The study is intended to test the ACO concept developed by Elliott Fisher and colleagues by creating a viable incentive system on a pilot basis, which would include Vermont primary care physicians (PCPs), specialists, the local community hospital(s) and other key providers as well as payers. The legislation states that the design of the pilot should build on the Blueprint for Health pilots, particularly the payment reform for primary care physicians, the adoption of the medical home model, and the development of community-based care coordination teams. Summary of Provider and Payer Feedback to Date To begin to understand the issues around developing an ACO pilot, we participated in three provider interviews and two payer interviews during which feedback was sought on the proposed ACO outline. In general, the interviewees were very positive about potentially participating in an ACO pilot project. Not yet knowing the details, the pilot seems to promise opportunities to help providers focus more on measuring clinical efficiency, reducing clinical variation, and expanding their abilities to align incentives across provider types. Page 16 of 24

17 Project Size/Scale Interviewees indicated a concern about the need to have a critical mass of patients to ensure the financial incentives were significant enough to impact utilization trends. While we began talking about 30,000 covered lives, it seems as the interviews occurred the numbers discussed continually became higher, going up to 50,000 and then 80,000. One provider indicated that in order to affect change in practice patterns of specialists, it would be necessary to include a population closer to 100,000 enrollees, which might not be feasible without possible collaboration of multiple providers or physician hospital organizations (PHOs) across a region. The advantage of this kind of collaboration is that it might also allow for those more sophisticated at medical and care management as well as in the use of information technology to improve clinical care to share these skills across a broader network of providers. Inclusion of Medicare and Employee Retirement Income Security Act (ERISA) Exempt Organizations Another question raised was the State s ability to engage Medicare and the other large payers in this project. It was speculated that for some providers, without the involvement of Medicare enrollees, it would be difficult to expect physicians to change behavior patterns given the importance of this population. This, however, did not seem to be something that would prevent at least two of the three providers organizations from considering pilot project participation. 1 In setting up the Blueprint pilot projects, CIGNA initially wanted to exclude its ERISA-exempt members (i.e., those belonging to self-insured organizations). This population includes state employees and large employers including IBM. Although CIGNA has relented on its resistance to including state employees in the pilots, it is currently maintaining that members from ERISA-exempt private industry be excluded. This same issue may arise in development of the ACO pilots, although we suspect there may be other and different challenges with securing payer participation (see discussion below). Risk Assumptions There were a number of questions about risk assumptions, with the general sentiment that the decisions made in this area would be critical to any ACO pilot s success. All those interviewed agreed that reaching an agreement on how the global budgets are established will be a critical component moving forward. In addition to this necessitating the engagement of an experienced, unbiased actuary for global budget development, the pilot as envisioned would require a separate entity, perhaps hired by the state, to administer the program. Issues raised as concerns associated with global budget development included the need for accurate data, a transparent and fair method that does not hold providers responsible for factor beyond their control, and a method for accounting and adjusting for differences in benefit packages and geographic regions. From the interviews, it became clear that there seem to be varying levels of sophistication in providers ability to effectively do medical management. One provider indicated they would not be willing to assume a global budget without taking on all aspects of medical management, and questioned the payers willingness to fully delegate this management to them. Another provider indicated they would not be 1 It is important to note that one provider group chose not to participate in the medical home pilot due to the inability to include Medicare. While the medical home pilot was originally designed not to include prospective payments (PPPM dollars for recognized medical homes) for Medicare enrollees, after pressure from both participating payers and providers, the Blueprint agreed to foot the bill for PPPM dollars to the medical home pilots to cover their Medicare patients. Page 17 of 24

18 surprised if payers were uncomfortable delegating all medical management to provider and PHO groups that have yet to demonstrate their effectiveness in this area. An ACO pilot could result in a significant increase in the share of dollars that are at risk. Providers mentioned the importance of maintaining financial incentives so that under any ACO pilot they would be at least as good as what the providers currently experience under risk agreements (some providers are apparently doing quite well and view this new income sources as important). Scope of Services Each provider was asked their thoughts concerning which, if any, services should be excluded from the ACO pilots. All agreed that nurse practitioners and physician s assistants should be recognized as primary care providers for the pilots, given current physician shortages in the State. In terms of behavioral health, several providers were quite hesitant to consider taking on the management of this area, given their limited experience and the shortage of adequate mental health providers. One provider indicated a willingness to take on this risk and then contract out for such services. While both behavioral health and drug management were recognized as important aspects of managing medical expenses, providers were ambivalent about taking on drug management at this point in time, because of a lack of experience in managing these expenses. There was also some discussion about chiropractors, nurse midwives, and other alternative medicine providers who are likely to pressure the state to be included in any pilot, but at this point not typically included in provider networks. Payers, too, were skeptical about including prescription drugs in the global budget, at least initially, given that currently providers have limited experience in that area. One payer was willing to consider including behavioral health in the global budget given the expectation that the ACOs would most likely subcontract out that service for all its members, offering continuity across payers. Another payer was skeptical, indicating that this had been previously attempted unsuccessfully and questioned the ability of the new ACO to manage the challenge of contracting for behavioral health services in the current environment in Vermont. Other Concerns Mentioned Questions were raised about how the medical home pilots and/or the concepts incorporated into those pilots might be worked into the ACO pilots, and both payers and providers were interested in hearing more about this as the ACO model is further developed. Several providers indicated the pilots would provide the opportunity to become more creative in their efforts to redistribute bonuses and any shared savings across providers in such a way to better align incentives between the primary, specialty and hospital providers. One payer asked the Commission to carefully consider the implication of proposing an ACO model that provided no up-front resources to organizations that may be needed to successfully function as an ACO. The ACO model discussed until that time presumed all savings and or rewards would be experienced at the end of each pilot year. This is different than the medical home pilot, which is designed to encourage infrastructure building by paying a PPPM for those practices meeting certain recognition criteria. While it might be difficult or even impossible to provide too much financial support on the front end of much larger pilots, the potential advantages of providing at least some up-front investments in the ACO pilots with less experience might be worthy of consideration. Finally, several of those interviewed mentioned there will be a number of challenges in trying to blend both the medical home and ACO pilots, given the potential differences in goals and design. They indicated an interest in hearing how the Commission intends to address these issues. This will be a Page 18 of 24

19 challenge in terms of determining which inputs have resulted in change, making evaluation a challenge. None, however, suggested that those involved in the medical home pilots should be automatically excluded from participating in an ACO. While all those interviewed agreed that the implementation of ACOs in Vermont would significant require political and administrative support, they seemed to welcome the opportunity to further explore participation in ACO pilot projects. Additional Questions for Potential Sites and Payers Sites: Data capabilities? Experience with quality, cost, and outcome measurement? Areas for improvement that they would target in an ACO pilot (e.g., low-hanging fruit)? Barriers to achieving improvement/efficiencies? Payers: Experience with capitated payment in the state? Lessons learned and concerns? What features should the ACO model include to address those concerns? Building a Successful Model(s) for Pilots The goal of the feasibility assessment should be to determine whether or not an ACO pilot would be successful in Vermont. We believe there are two key components of success. First, the project needs to be acceptable to providers and payers so that they will want participate. This means there needs to be sufficient upside to all stakeholders. It may also mean that, given the Blueprint experience of the medical home pilots, a legislative mandate may be necessary to ensure final payer cooperation. As you may know, full cooperation of all the major payers in the Blue Print pilots only occurred after the mandate was in place. Second, the pilots should be structured so that the pilot sites could demonstrate the value of the ACO model (whether or not they actually do). This means barriers to success should be minimized to the extent they are in the control of the Commission, as designers of the study. Toward this end, we find it useful to think in terms of developing an RFP for the ACO pilots. The RFP would state the goals and objectives of the pilots, qualifications of pilot sites, elements that the responders to the RFP must include in their proposals, and the criteria for selecting a site. The document you developed, which summarized the objectives and key issues of the study, provides a solid foundation for the issues that need to be resolved prior to the issuance of an RFP. Another perspective that might prove helpful is to identify those RFP elements of the ACO that would be required of pilot sites, those elements where some flexibility would be allowed, and those elements where you are looking for creative proposals from responders. This way of approaching the issues would allow the Commission to focus on those parameters of the study that would need to be finalized prior to issuance of an RFP. We consider each of these RFP parameters (using, to a large extent, the questions and issues you present in your discussion document) and make some suggestions. Page 19 of 24

20 (It is worth noting that we understand the importance of advance planning and coordinating with such a large and complex pilot. So much so, that the issuance of an RFP might be a formality, when it comes time to make the pilot site award.) Critical or Required Elements These are elements that must be in place before an ACO pilot could take place. Therefore, they must either be acquired or established by the Commission or be part of the minimum requirements for a site to be considered a viable pilot candidate. The following discusses potential elements that may be required. Length of pilot The length of the pilot should be predetermined by the Commission. This decision needs to balance the need for sufficient time for the model to affect provider behavior and yield results with the recognition that a pilot that does not produce results becomes unsustainable. The Centers for Medicare and Medicaid Services often choose a 3-year period to conduct their demonstration projects, which is believed to provide sufficient time to test new payment or care models. Site requirements Minimum site requirements would need to be established well in advance of a pilot. There are three areas we have discussed, in this respect. 1. Provider and care settings: The Commission should establish provider and care setting requirements, but may want to consider establishing minimums, for example, requiring primary care and specialist participation and ambulatory and inpatient settings. By not exactly specifying the features of the delivery system for the pilot, this may allow some organizations to participate who otherwise would not be eligible. Moreover, it would allow potential sites to focus on areas where they feel they could have the most impact on utilization. 2. Patient populations: The Commission should specify those populations that would be included in the pilots and these should be consistent across sites. As you ve noted, to have any chance of succeeding, this means all enrollees of the current major payers with the exception of Medicare. Further work is required to understand how and if new entrants into the market (e.g., United) should be involved. The Commission should also establish a patient attribution approach to be implemented consistently across sites. In the Blueprint pilots, an attribution approach has been agreed upon among payers and providers that assigns patients to a medical home. This approach could serve as a model for the ACO pilots. 3. Data Capabilities: Quality measurement would be an important component of an ACO pilot. Sites should have sufficient IT infrastructure that would allow the development of measures across providers and settings. If the Commission chooses to encourage organizations with less experience in assuming financial risk to participate in the pilot, it may well need to develop some aspect of prospective financing to provide adequate up-front capital to ensure necessary infrastructure capacity is promptly part of any new ACO. Page 20 of 24

21 Capacity to design and implement pilot: The issue of pilot design and implementation abilities will vary significantly depending on the nature of the sites sought and selected. See Table 1 for more details. Performance measures: These are likely to vary by site. It is recommended that a core set of measures, perhaps linked to those already agreed to by the medical home pilots, and endorsed by a national quality organization be carefully considered. A core set of measures should be established as a way to evaluate the effectiveness of the ACO model to affect utilization, costs, and outcomes. This core set should be standardized across pilot sites to allow for transparency and a robust evaluation of the program. The set of measures needs to balance the need for information with health IT limitations and reporting burden. Therefore, it becomes important to assess the level of data capabilities potential pilot sites have as well as performance measurement experience. Sites may choose to expand the total number of measures to be considered beyond those identified in the core set if there are areas that they are focusing on that are not sufficiently reflected in them. For the Blueprint pilots, a committee consisting of representatives for the state, payers, and providers have work collaboratively to determine the measures to assess the success of the pilots. A similar approach may be useful here to ensure payer and provider buy in. Some of the same measures may be useful for assessing the ACO pilot. The advantage of adopting a similar approach and measures is that they have been vetted and agreed upon already. Method to establish the global budget The ability to establish a fair and transparent process and method for developing the global budget is going to be critical to determining whether or not an ACO pilot is likely to be successful. The basic method for developing the global budget needs to be uniform across settings and agreed to by providers and payers. However, the level of risk and rewards may vary by site. To get provider participation, there would need to be sufficient upside and a limit on the level of risk they would bear. On the provider side, acceptance will depend on whether or not the pilots are designed to be payment neutral, the extent to which they share in any savings, and whether and how much they will be asked to pass on medical management and other services typically provided by the health plans (such as care and disease management) to providers who are less experienced. Again we remind you that while many of the aspects of the Blue Print pilots were discussed in advanced with both payers and providers, most payers were unwilling to make the financial commitment to get on board until the legislature mandated their participation. Negotiable within Range Scale As noted above, interviewees indicated a concern about the need to have enough participation in the program to ensure the financial incentives are significant enough to impact utilization trends. Larger pilot sites also provide several technical advantages with respect to performance measurement. With respect to financial incentives, sufficient scale to affect utilization trends could be achieved through the collaboration of PHOs, through collaboration between PHOs and its broader community, or, possibly, through the inclusion of complete primary care and specialist physician practices. The Commission may want consider having some flexibility in the scale and organization of pilot sites and allow the sites to suggest approaches to ensure the financial incentives are significant enough to affect provider behavior. Page 21 of 24

22 Site experience assuming risk You have expressed an interest in exploring whether sites with a range of structures and experiences could serve as pilot sites. One aspect of this is whether a site would be required to have a certain amount of experience assuming risk. We believe that flexibility in this respect could be acceptable, if there was also flexibility in the risk sharing arrangements and terms between payers and providers as well as in the scope of services. Scope of services Some flexibility here may be good to account for different experience levels. A site with limited experience assuming risk may be more inclined to participate if the scope of services were somewhat limited and related only to those factors they felt they could impact. Establishment of risk sharing arrangements The establishment of risk sharing arrangements could vary by site, depending on the scope of services and sites experiences with assuming risk. Those sites willing to assume more risk should be eligible for the greater rewards. The use of risk corridors and amount of withholds could be negotiated between the Commission, pilot sites, and payers depending on characteristics of the ACO model being tested. The Commission should establish the basic structure of the reward. For example, it could require that rewards be based on the IHI Triple Aim and specify the core set of measures and standards. Payment model/financial redistribution of savings Medical management, care mgt, disease management community care team model Financial distribution of incentives between specialists, PCPs Examples of ACO Models for Pilot Sites In the table below, we present some examples of how the ACO pilot models might be structured depending on the type of care delivery organization and experience in assuming risk, performance measurement, and quality improvement. The table is structured around the primary RFP elements discussed in the section above. We exclude some elements that we feel should be standardized across pilot sites, including the length of the pilot, the capacity and infrastructure to design and implement the pilots, the payer populations covered, minimum data capability standards, and the method to develop the global budgets. Page 22 of 24