SHARED-SAVINGS PAYMENT ARRANGEMENTS IN HEALTH CARE SIX CASE STUDIES Michael Bailit, Christine Hughes, Megan Burns, and David H. Freedman august 2012 Bailit Health Purchasing, LLC The Commonwealth Fund, among the first private foundations started by a woman philanthropist—Anna M. Harkness—was established in 1918 with the broad charge to enhance the common good. The mission of The Commonwealth Fund is to promote a high performing health care system that achieves better access, improved quality, and greater efficiency, particularly for society’s most vulnerable, including low-income people, the uninsured, minority Americans, young children, and elderly adults. The Fund carries out this mandate by supporting independent research on health care issues and making grants to improve health care practice and policy. An international program in health policy is designed to stimulate innovative policies and practices in the United States and other industrialized countries. SHARED-SAVINGS PAYMENT ARRANGEMENTS IN HEALTH CARE SIX CASE STUDIES Michael Bailit, Christine Hughes, Megan Burns, and David H. Freedman Bailit Health Purchasing, LLC August 2012 Abstract: Driven by widespread interest in improving health care quality and reducing costs and by the Affordable Care Act’s “accountable care” provisions, shared-savings programs are gaining traction as an alter- native approach to paying health care providers. Providers receive a share of the savings they achieve by reducing the growth in costs for delivering care to a defined patient population. This report presents six case studies of pilot shared-savings programs across the country. The cases reveal program variation in the patient populations subject to shared-savings approaches, the health care services covered, the determina- tion of cost savings and payouts to providers, the use of performance targets, and performance measurement. Early results from the pilot programs also vary. Exploring differences in shared-savings approaches, and the outcomes they achieve, will be essential to determining whether they work, how to improve them, and whether and how to diffuse them. Support for this research was provided by The Commonwealth Fund. The views presented here are those of the authors and not necessarily those of The Commonwealth Fund or its directors, officers, or staff. To learn more about new publications when they become available, visit the Fund’s Web site and register to receive e-mail alerts. Commonwealth Fund pub. no. 1624. CONTENTS ABOUT THE AUTHORS................................................................................................................................................................ 6 EXECUTIVE SUMMARY.................................................................................................................................................................7 INTRODUCTION.........................................................................................................................................................................10 Maryland Multi-Payer Patient-Centered Medical Home Program .................................................................................... 11 Medica and Fairview Health Services .............................................................................................................................. 13 Health Care Incentives Improvement Institute (Prometheus Payment)........................................................................... 15 Blue Cross Blue Shield of Illinois and Advocate Health Care............................................................................................ 17 HealthPartners................................................................................................................................................................. 18 Harvard Pilgrim Health Care........................................................................................................................................... 20 CONCLUSION............................................................................................................................................................................. 21 LIST OF EXHIBITS EXHIBIT ES-1 KEY DESIGN FEATURES OF SHARED-SAVINGS PILOT PROGRAMS ABOUT THE AUTHORS Michael Bailit, M.B.A., is president of Bailit Health Purchasing, LLC. For the past 16 years, Mr. Bailit has worked extensively with public agencies, purchaser coalitions, and employers to advance the effectiveness of their health care purchasing activities. He previously served as assistant commissioner for the Massachusetts Division of Medical Assistance and as a benefits manager for Digital Equipment Corporation. He also has experience working in the health insurance industry. Mr. Bailit received a master of business administration degree from the Kellogg Graduate School of Management at Northwestern University. Christine Hughes, M.P.H., has been a senior consultant with Bailit Health Purchasing for more than 11 years. Ms. Hughes’ work at Bailit Health includes research on best practices regarding: payment reform, patient-centered medical homes, pay-for-performance, and value-based insurance design. She previously served as deputy director for the Medicaid Managed Care Program at the Massachusetts Division of Medical Assistance. She also has experience in network contracting with a large integrated delivery system (Partners Community Health Care) and a staff model HMO (Health Care Plan). Ms. Hughes received a master of public health degree from the Boston University School of Public Health. Megan Burns, M.P.P., is a senior consultant with Bailit Health Purchasing. Ms. Burns has done extensive work on payment and delivery system reform, including research and design work involving bundled payment arrangements, integration of behavioral health within patient-centered medical homes, and researching strategies to reduce variation in provider prices. She previously served as a hospital administrator within the UPMC Health System and worked for a Washington, D.C.–based think tank researching and analyzing health, social, and economic policy issues. Ms. Burns received a master of public policy degree from the Terry Sanford Institute at Duke University. David Freedman is a science journalist and author who contributes to the Atlantic, the New York Times, Scientific American, and Discover, among other publications. He is a consulting writer at Johns Hopkins Medicine International working on projects related to global collaborative health care, and a writer in residence at McGill University working on projects related to obesity, nutrition, and preventive medicine. He is the author of five books, the most recent being Wrong: Why Experts Keep Failing Us— and How to Know When Not to Trust Them. Editorial support was provided by Sandra Hackman. 6 Shared-Savings Payment Arrangements in Health Care: Six Case Studies EXECUTIVE SUMMARY • A third initiative requires providers to serve a One of the most talked-about new ideas in health care minimum number of patients to participate, uses is rewarding providers for reducing medical spending by the average per-patient cost of health care in a giving them a share of the net cost savings. Driven by metropolitan area as a benchmark, and has paid out an interest in seeing medical homes and other providers up to 75 percent of shared savings to a provider. shift to some form of performance-based payment, as well as by the Affordable Care Act’s push for “account- Despite these variations, the case studies reveal able care,” shared-savings approaches are currently consistent themes regarding shared-savings approaches being tested by numerous payer and provider organiza- to payment for health care services. These common tions across the United States. elements include a willingness among most payers to Given uncertainties regarding an ideal shared- absorb many of the costs entailed in setting up and savings approach and how best to implement it, we sharing tools for measuring health care performance examined six shared-savings pilot initiatives. For each and cost savings. Overall themes also include a belief case study, we interviewed leaders at payer and provider that shared-saving programs must evolve to include organizations and state agencies about their attempts to shared risk, and a conviction that even when pilot pro- design and implement shared-savings programs. grams fail to achieve savings, they are moving in the These pilot programs, we found, vary consid- right direction. erably on several dimensions (Exhibit ES-1). These We do not yet know whether this approach include the patient populations subject to shared- is a long-term strategy for promoting better health savings arrangements, the health care services those care while lowering costs, or a transitional strategy to arrangements cover, how payers determine cost savings some other model, such as global payments for which and payouts to providers, whether the model incorpo- the provider also assumes risk if spending is higher rates performance targets, and how it measures per- than a budget target. Exploring the organizational and formance. These pilot projects also vary in their early environmental differences in how participants pursue impact on health care costs and payouts to providers. shared-savings approaches, and the outcomes they For example: achieve, will be key to determining whether they work, how to improve them, and whether and how to diffuse • One initiative measures cost savings related to them. preventable complications from specific procedures, and is on track to make a substantial payout to providers. • A second initiative does not require providers to serve a minimum number of patients to participate, uses a control group and 21 quality measures to determine payouts to participating providers, but has not yet demonstrated cost savings. www.commonwealthfund.org7 EXHIBIT ES-1. KEY DESIGN FEATURES OF SHARED-SAVINGS PILOT PROGRAMS MARYLAND MULTI-PAYER HEALTH CARE INCENTIVES MEDICA AND FAIRVIEW PATIENT-CENTERED MEDICAL IMPROVEMENT INSTITUTE HEALTH SERVICES HOME PROGRAM (MMPP) (PROMETHEUS PAYMENT) Patients, Services, and Payments Commercial; Medicaid Patient population(s) Commercial Commercial; Medicare Advantage managed care 12-month claims history Johns Hopkins ACG software Methodology for attributing patients and a 12-month look-back to Not required to provider groups Most frequently visited provider determine site of at least (most recent if tied) 50% primary care spending Per-patient, per-month Semiannual lump-sum payment management fee for disease Shared savings for reducing care coordinator potentially avoidable complications Type of payments Incentive payment based on (PAC) associated with treating a cost savings Shared savings based on chronic condition comparison with a control group Special Adjustments to Shared-Savings Calculations ACG grouper Stop-loss provision Adjustments for patient risk None at the time High-cost claims truncated at High-cost claims (episodes $250,000 to $500,000, depending exceeding two standard deviations) on contract truncated With state funding, University Data support and analysis Tools and services to help payers of Maryland and Johns Hopkins and providers use the Prometheus Supports for providers School of Medicine provide Some care management and health Payment software and share training on care coordination and coaching may be available best practices evidence-based medicine, etc. Inputs for Calculating Savings Minimum number of patients per Minimum number of patients with None 15,000–20,000 member-months practice group certain conditions and treatments Retention of initial percent of savings Payer takes upfront share None None by payer (discounted PAC allowance) Method for Calculating Savings Comparison with: External benchmark (TBD) Control group Budget Method for Distributing Savings Up to 75%, split among providers: Percent cost savings eligible 30%–50%, depending one-third to hospital; one-third to Varies by payer, but PAC allowances to providers on performance care management; one-third to must be at least 50% of budget physician groups 21 quality measures; and reductions Must achieve net 6% reduction in in use of high-cost services, such Minimum quality gate, then cost of PAC to receive payout Performance targets as emergency department and confidential algorithm hospital readmissions Quality scorecard available for use 8 Shared-Savings Payment Arrangements in Health Care: Six Case Studies EXHIBIT ES-1. KEY DESIGN FEATURES OF SHARED-SAVINGS PILOT PROGRAMS (CONTINUED) BLUE CROSS BLUE SHIELD OF ILLINOIS AND HEALTHPARTNERS HARVARD PILGRIM HEALTH CARE ADVOCATE HEALTH CARE Patients, Services, and Payments Commercial; Medicaid (two of the latter Commercial providers opted to participate, Commercial but with separate calculations) Jointly developed confidential Retrospectively attributed using Prospectively assigned using internally methodology with a two-year look-back internally developed algorithm developed algorithm and tie-breaking algorithm Shared savings based on comparison Shared savings based on comparison Shared savings compared with a budget with a control group with a negotiated target Special Adjustments to Shared-Savings Calculations DxCG software ACG software DxCG software High-cost claims (>$50,000) truncated High-cost claims truncated Services excluded: transplantation Services excluded: transplantation, (amount confidential) behavioral health, and out-of-area services Detailed analysis of performance, Basic reporting tools for tracking Reports, data feeds, and software including cost and use of services, performance; access to best-practice tools to download data directly into location of services, medical examples electronic medical records systems conditions, and specialty Care management tools, hospital Analytical support for targeted Some grant money to help build utilization review, and in-depth reports high-risk members needed infrastructure available for extra fee Inputs for Calculating Savings None, but only provider groups with None, but only large provider groups 3,000–5,000 patients >100,000 members are participating are now participating Yes (amount confidential) None 2% Method for Calculating Savings Control group: Jointly negotiated target Budget nonparticipating provider network Method for Distributing Savings 50/50 split of savings up to 6% of total 50% (commercial and Medicaid savings Up to 50% budget, after plan takes 2% (3% max for calculated separately) provider group) Minimum quality gate includes 12 measures; No, but separate pay-for-performance None Year 1: maintain performance program Years 2 & 3: negotiated targets www.commonwealthfund.org9 SHARED-SAVINGS PAYMENT ARRANGEMENTS Given these uncertainties regarding an ideal IN HEALTH CARE: SIX CASE STUDIES shared-savings approach and how best to implement it, we chose six initiatives from among the many we cited in that earlier issue brief to examine more closely. For Introduction these case studies, we interviewed leaders at payer and One of the most talked-about new ideas in health care provider organizations and state agencies regarding is rewarding providers for reducing medical spending their attempts to design and implement shared-savings by giving them a share of net cost savings. Driven by programs. widespread interest in seeing medical homes and other We found that these pilot programs vary con- providers shift to some form of performance-based siderably on several dimensions (see Exhibit ES-1). payment, as well as by the Affordable Care Act’s push These include the patient populations subject to for “accountable care,” and shared-savings approaches shared-savings arrangements, the health care services involving numerous payer and provider organizations those arrangements cover, how payers determine cost are emerging. The structure and performance of these savings and payouts to providers, whether the model pilot projects can shed light on what to expect from the incorporates performance targets, and how it measures Medicare Shared Savings Program for accountable care performance. organizations and other initiatives—and on how payers On one idea, however, we found complete and providers can engage in these efforts. agreement: These programs have the potential to spur In a previous Commonwealth Fund policy essential changes in the delivery and cost of patient care, brief, we summarized 27 examples of shared-savings given an existing system widely recognized as untenable. initiatives.1 We found wide variation in critical details in how participants implemented such initiatives. These variations included how payers assign patients to pro- viders to evaluate cost savings, how payers adjust the risk profiles of pools of patients based on their health care needs, and how payers actually calculate and dis- tribute savings. That earlier study also made clear that payers and providers must resolve a number of key challenges if shared-savings plans are to realize their promise. These include ensuring that identified savings do not merely reflect random variations in health care costs— which is particularly important if providers incur no explicit penalty if they fail to achieve savings. Other key challenges include selecting measures that focus provid- ers on improving performance as well as reducing costs, and equipping them with the data and tools they need to improve their effectiveness and efficiency. 1 M.H. Bailit and C. Hughes, Key Design Elements of Shared- Savings Payment Arrangements (New York: The Commonwealth Fund, Aug. 2011). 10 Shared-Savings Payment Arrangements in Health Care: Six Case Studies MARYLAND MULTI-PAYER PATIENT-CENTERED the event of a tie, the practice that the patient visited MEDICAL HOME PROGRAM most recently is deemed responsible for the patient. The Maryland Health Care Commission launched its For patients who have no claims history for the past 12 Multi-Payer Patient-Centered Medical Home Program months, staff members use the prior 12 months to iden- (MMPP) in April 2011. This three-year pilot aims to tify a practice that is responsible for those patients. provide 200,000 patients with access to high-quality, patient-centered primary care. To support that goal, the Determining and Distributing Savings MMPP is providing training and support to encourage The first of the two extra payments to providers is a 200 primary care providers working at 50 practices to semiannual lump sum designed to cover the extra cost coordinate care, use evidence-based medicine, improve of high-value efforts that historically have been unre- the quality and safety of care, and strengthen communi- imbursed. These include coordinating and managing cation with patients. care and using information systems. Provider groups The Maryland legislature required the five larg- are automatically eligible for this payment if they fulfill est health plans in the state—CareFirst, Aetna, Cigna, requirements for participating in the MMPP, includ- UnitedHealthcare, and Coventry—to participate in the ing reporting their scores on specific measures of health MMPP.2 Medicaid managed care plans have also joined care quality. the pilot program. These private and public payers have The payments are based on projected cost sav- agreed to award participating practices two extra pay- ings from improving the quality of care, especially for ments beyond the usual fee-for-service or capitation patients with chronic health conditions. The MMPP payments. uses a set of budgeting and evaluation tools made available by the Health Care Incentives Improvement Institute to estimate potential savings from avoiding Why Shared Savings? complications for seven common chronic illnesses. The commission did not expect that providing funding The commission bases these payments on for better care management alone would reduce costs. expected savings rather than the actual investments The commission is therefore pursuing a shared-savings needed to improve care for two reasons. First, pay- model to promote and support the concept of a medical ers and providers alike were unwilling to invest “new home, including among self-insured employers. money” into improving the health care system. Second, the commission expects costs to vary depending on Shared-Savings Methodology which systems a medical practice already has in place, Patient Population: Minimum Size and Attribution and how efficiently it transforms itself into a patient- A practice does not need to serve a minimum num- centered medical home. ber of patients to qualify for the incentive payment. This payment varies with the size of a practice’s Participating practices average about 1,850 patients. patient population, whether it has achieved certification The Maryland Health Care Commission assigned from the National Committee for Quality Assurance patients to a given practice at the outset of the program, (NCQA), and the level of that certification.3 Payments and reattributes patients every six months. also vary depending on whether the payer is a commer- To do this, the commission uses a two-step cial health plan or Medicaid. process based on 24 months of claims information from all payers. First, staff members identify the prac- 3 Practices must achieve NCQA recognition as a Level 1 patient- tice each patient has visited most often in the past 12 centered medical home within the first six months of the pro- months, and attributes the patient to that practice. In gram, and Level 2 recognition within 18 months. For more information on NCQA standards, see https://inetshop01.pub. 2 The state excluded Kaiser from the mandate because of the ncqa.org/publications/product.asp?dept%5Fid=2&pf%5Fid=30 nature of its affiliated delivery system. 004%2D301%2D11. www.commonwealthfund.org11 The second extra payment to providers, known Both Group One and Group Two criteria are as an incentive payment, rewards them for control- tiered, so practices receive different percentages of cost ling health care costs during a given year, as well as savings based on their performance. For example, prac- for meeting certain performance targets. The share of tices in the lowest tier are eligible for a 30 percent share cost savings that a practice may retain rises with its of savings, while those in the highest tier are eligible performance. for 50 percent. Health plans and self-funded employers An incentive payment is available only to keep the rest of the cost savings. practices that reduce medical expenses compared with practices that are not participating in the pilot program. Adjusting for Patient Risk The commission uses the Maryland all-payer claims The MMPP does not have a formal process for adjust- database to calculate cost savings. The commission used ing the results reported by each practice based on the 2009 and 2010 data to perform test calculations in May risk profile of its patient population. However, the 2012 and reported results to participating primary care commission plans to decide whether to pursue some practices. The first shared-savings payments will be form of risk adjustment after gaining experience with made in late 2012, based on data for 2010 and 2011. attributing patients to practices and awarding incentive The MMPP relies on two sets of performance payments. criteria to distribute incentive payments to practices. The shared-savings method adopted by the The first, Group One, consists of 21 measures of health commission controls for risk by calculating costs only care quality. In the first two years, practices must simply for patients attributed to the primary care practice in report their performance on some of these measures. In both the baseline and performance years. There is no the third year, practices must meet quality thresholds additional risk adjustment. However, the commission (e.g., HbA1c (blood sugar) controlled to a level of less does adjust patient-cost calculations to mitigate the than 8 percent). impact of random events, such as by excluding cata- The second set of criteria, Group Two, measures strophic cases. reductions in the use of high-cost health care services, such as emergency department visits and admissions for Technical Assistance ambulatory care–sensitive conditions. The latter stan- The state is funding the University of Maryland and dards apply only in the second and third years. Practices the Johns Hopkins School of Medicine to provide must improve their performance each year to receive technical support and training to help practices partici- incentive payments based on these criteria. pating in the MMPP implement the patient-centered medical home. Maryland Health Care Commission’s Experience with the Shared-Savings Model The commission began distributing MMPP payments in July 2011. It has not yet evaluated the results of the shared-savings model. 12 Shared-Savings Payment Arrangements in Health Care: Six Case Studies MEDICA AND FAIRVIEW HEALTH SERVICES Minneapolis/St. Paul, where most of Medica’s members Medica provides health coverage to 1.6 million mem- are also concentrated. bers, including individuals as well as through employers, Medica and Fairview developed their approach third-party administrators, and government programs to shared savings jointly. Both see the move away from in Minnesota and select counties in Wisconsin, North fee-for-service payment as permanent, but acknowledge Dakota, and South Dakota. More than 96 percent of that this model is transitional, expecting it to last two providers in Minnesota participate in Medica’s network. to three years. What the next model will be is unclear, Some 600 primary care providers representing about but they have discussed relying on a comprehensive or 20 clinics in Medica’s network are participating in its global payment per patient.4 shared-savings arrangement. The largest of these is Fairview Health Shared-Savings Methodology Services. Fairview is a nonprofit health care system Patient Population: Attribution and Minimum Size based in Minneapolis with more than 40 primary care Medica’s shared-savings pilot covers 800,000 to clinics, a wide range of specialty services, inpatient care, 900,000 individuals. These include members enrolled in home care, and senior services. The integrated medical fully insured and self-insured preferred provider orga- practice includes more than 700 Fairview-employed nizations (PPOs), as well as small numbers of members physicians and more than 700 physicians affiliated enrolled in commercially insured health maintenance through University of Minnesota Physicians, the mul- organizations (HMOs). tispecialty practice of the university’s medical school Medica uses the John Hopkins Ambulatory faculty. Care Groups (ACG) software to attribute members to Fairview began a shared-savings arrangement participating practices. This software uses a 12-month with Medica in June 2009, with 300 of its primary look-back to identify the medical practice where each care providers participating. Fairview signed shared- patient has incurred 50 percent or more of his or her savings agreements with two other commercial pay- primary care dollars. ers, Blue Cross and Blue Shield of Minnesota and Medica requires participating providers to have HealthPartners, in January 2011. a minimum of 15,000 to 20,000 member-months, or Medica added five other provider groups in 1,250 to 1,667 patients (the range allows for changes in 2010. More than 70 percent of Medica’s members are a practice’s patient population). Medica used actuarial now covered under some type of shared-savings model. testing to develop this requirement, which is designed to ensure that the plan can gather enough accurate data Why Shared Savings? to calculate cost savings. Medica chose shared savings because employers Fairview wanted to include low-volume provid- could not afford escalating health insurance costs. ers in the pilot. According to a Fairview interviewee, “It Furthermore, in Minnesota’s not-for-profit insurer envi- was very clear that the small practices add real value to ronment—where provider payments account for 90 per- the system; some are very efficient, so excluding them cent of costs—shared savings was a promising strategy is not smart.” Fairview therefore uses its own method- to control these costs. ology to pool providers so they can participate in the Medica chose Fairview as a partner for sev- pilot. eral reasons. It was not only the largest practice in the insurer’s network, but appeared to be the most ready to participate in a shared-savings approach, having imple- 4 CMS accepted Fairview as one of 32 Pioneer accountable care mented a chronic care management model for one year organizations, effective January 1, 2012. Under this program, before the pilot began. Fairview’s providers are also in Fairview will contract with CMS with the possibility of both shared savings and shared losses. www.commonwealthfund.org13 Determining and Distributing Savings Adjusting for Patient Risk Medica and Fairview continue to rely on fee-for-service Medica uses the ACG grouper to account for variations claims to award base payments to provider groups. and changes in the risk profiles of each practice’s patient They also award a per-member, per-month (PMPM) population. It also truncates the annual health care costs management fee to participating practices, which are of an individual at $250,000 or $500,000, depending expected to hire a disease care coordinator. Medica does on the contract. That is, Medica does not include per- not net out any portion of these payments when calcu- patient costs beyond that amount when calculating cost lating practice savings to account for the support it pro- savings. vides to practices. According to a Medica interviewee, “We know that we are incurring costs, but hoping that Technical Assistance our share of the savings will cover those upfront costs.” Medica relies on a handful of full-time analysts to Medica and Fairview also award shared savings present a robust analysis of the costs incurred by the to each practice based on its performance on all services patients of each provider group each month. For exam- compared with a control group: the average perfor- ple, Medica might identify spine surgery as a high-cost, mance of other large-group practices in the metropoli- high-volume procedure, collect and analyze data on tan area. They base this comparison on quality measures the use of that surgery during a given month, and then that serve as a qualifying gate that enables practices to discuss the results with leaders and physicians in each receive savings. A confidential algorithm then deter- provider group. Medica also provides dedicated support mines the shared-savings pool based on health care to each group through a care manager or health coach, quality, cost, efficiency, patient experience, and provider depending on the group’s needs. collaboration among qualifying practices. Fairview appreciates the commitment of Fairview takes 75 percent of the calculated Medica and other payers to sharing information, but savings, while Medica receives the other 25 percent. prefers to receive raw data in real time rather than Fairview then distributes its share as follows: one-third reports. Fairview prefers to perform its own analysis to to the hospital, one-third to care management infra- help providers manage their patient population most structure, and one-third to providers in the system. efficiently. Fairview adjusts the management fee to include the self-insured population. Medica and Fairview calculate Medica’s Experience with the these shared savings annually. Shared-Savings Model Medica and Fairview have also identified an Medica has been evaluating and tweaking its model, “optimal group” that delivers the most cost-efficient and and is encouraged by preliminary drops in the cost of high-quality care, to help practices identify opportuni- treating patients with chronic diseases. ties for improvement and monitor performance. 14 Shared-Savings Payment Arrangements in Health Care: Six Case Studies HEALTH CARE INCENTIVES IMPROVEMENT To qualify for shared savings, practices must INSTITUTE (PROMETHEUS PAYMENT) perform at least 30 of each included procedure each The Health Care Incentives Improvement Institute year, and must serve at least 100 chronic-care patients (HCI3), a nonprofit think tank that develops programs with each included condition. HCI3 acknowledges that and tools to facilitate health care reform, used a three- these minimums are low by some standards. However, year grant to pilot a shared-savings program applied to it contends that a sensitivity analysis of the algorithms the Prometheus Payment bundled payment methodol- used to calculate health care budgets (see below) sup- ogy. HCI3 has piloted this approach with three payers, ports those minimums. which were free to develop variations. HealthPartners, a nonprofit HMO in Determining and Distributing Savings Minnesota that also operates multipayer clinics, ran HCI3 uses Prometheus Payment to set a budget for the first shared-savings program with four provider the expected costs associated with a procedure, or treat- networks in 2009. That pilot included only health care ment for a chronic condition. The budget incentivizes services related to acute myocardial infarction. providers to reduce potentially avoidable complications Employers’ Coalition on Health, a nonprofit (PACs), and therefore health care costs. A retrospective PPO headquartered in Rockford, Illinois, began the calculation assesses whether providers’ actual spending second pilot in 2010. Intended to run through 2012, fell below or exceeded the budget. that program is focusing on health care services related Like most other shared-savings programs, to diabetes, hypertension, and coronary artery disease, at HCI3’s does require providers, particularly hospitals, least initially. to assume some risk, in that they must invest in chang- Independence Blue Cross and Crozer- ing approaches that affect all patients—not only those Keystone—the latter a nonprofit integrated provider— covered by the shared-savings agreement. These shifts began the third pilot in Pennsylvania in the first quarter may mean that providers lose some fee-for-service of 2010. That program has focused on procedures revenue for patients not covered by the shared-savings related to hip and knee replacements, at least initially. agreement. Prometheus uses an algorithm that calculates a risk-adjusted, per-case PAC budget based on “evidence- Why Shared Savings? informed case rates” for each covered treatment or con- HCI3 has implemented Prometheus Payment with a dition. (The algorithm is based on a multivariate regres- shared-savings approach as a transition to a more com- sion model developed by HCI3.) Providers are account- plex, risk-sharing system that might otherwise repre- able for the costs of all care for all episodes related to sent a daunting leap for providers. HCI3 worried that contracted conditions and treatments. providers might fear that such risk-sharing approaches HCI3 suggests that payers ask providers to would leave them owing money to payers at the end of reduce PAC costs by 50 percent compared with bud- the year, or even push them into insolvency. geted costs, and further suggests that payers give pro- viders a PAC allowance—50 percent of the per-case Shared-Savings Methodology PAC budget—up front. Unlike many other shared- Patient Population: Attribution and Minimum Size savings arrangements, that essentially means that pay- Prometheus-based shared-savings pilot programs have ers prospectively withhold a portion of the budget. not required health care plans to attribute patients to However, providers never have to return any of the specific provider groups. However, the programs apply allowance to payers. only to particular medical procedures and chronic At the end of the year, Prometheus scores pro- health conditions. viders on a 100-point scale measuring their success in www.commonwealthfund.org15 reducing PAC costs. That score determines their end- Technical Assistance of-year payouts from the withheld portion of the bud- HCI3 provides all services and tools needed to imple- get. To receive any payout, providers must reduce PAC ment the methodology. These include budgeting PAC costs compared with the budget by at least 6 percent. costs and determining PAC savings, scoring providers, Payers can also make payouts contingent on a provider’s and producing quarterly reports that allow providers score on a quality scorecard developed by HCI3 using to assess their performance compared with the budget. external benchmark data, though none of the pilot pro- HCI3 also provides advice on improving health care grams are implementing that option. processes, and enables payers and providers to share Payers are free to set PAC allowances higher budgeting and performance data and other information. than 50 percent of the budget—that is, to withhold less or even none of the PAC budget. However, HCI3 HCI3’s Experience with the Shared-Savings Model strongly discourages payers from setting PAC allow- In the first pilot program, participating providers came ances lower than 50 percent of the budget, to ensure in 1 percent over budget, and therefore did not receive that the target is not too daunting, and that providers any payout. As of the date of the interview, HCI3 had have some financial cushion against unexpectedly high not yet calculated savings for the other two pilots, but PAC costs. HCI3 says it might lower suggested allow- the highest score among providers so far was 72 out of ances in the future. 100. If providers sustain that score, they would receive HCI3 reports providers are likely to prefer 50 percent of the withheld PAC budget—or 25 percent this overall approach to a program that measures their of the total PAC budget. performance against a control group, because the bud- HCI3 says it is satisfied with these results, get gives them a stationary target and allows them to on balance, because the program has spurred desired continually assess where they stand. HCI3 also believes behavioral shifts among providers. HCI3 emphasizes that provider groups will consider the shared-savings that transitioning to a shared-risk approach will further program a success even if they fail to earn a payout, if it fuel behavioral change and ensure the financial viability leads them to reduce PAC costs. of the payment model. HCI3 advises other organiza- HCI3 acknowledges, however, that payers tions considering a shared-savings model to spell out might reasonably object to sharing savings based on a the “glide path”—an explicit timeline—for moving to fixed budget during a period when non-program pro- sharing downside risk with payers and providers. viders could achieve the same savings. In that case, a control-group approach would seem fairer to payers. Adjusting for Patient Risk HCI3 recommends protecting providers against excep- tional costs related to any one medical episode by employing a stop-loss provision, and truncating costs for episodes that exceed two standard deviations of the expected cost. 16 Shared-Savings Payment Arrangements in Health Care: Six Case Studies BLUE CROSS BLUE SHIELD OF ILLINOIS AND would deliver high-quality care even with incentives ADVOCATE HEALTH CARE to reduce costs. Third, BCBS IL and Advocate have Blue Cross and Blue Shield of Illinois (BCBS IL) is the several years of successful experience with a shared-risk largest health insurance company in the state, insuring arrangement BCBS IL uses with HMOs. more than 6.5 million members. BCBS IL offers a full range of products for individuals, families, employer Shared-Savings Methodology groups, and seniors, including a shared-risk HMO Patient Population: Attribution and Minimum Size with more than 800,000 members served by some 75 The shared-savings arrangement applies to members physician groups. BCBS IL has received accreditation enrolled in fully insured and self-insured commercial from NCQA and the Utilization Review Accreditation PPOs.5 The pilot excludes members insured under the Commission. Federal Employees Health Benefits Program. Advocate Health Care, a faith-based, not-for- BCBS IL uses a confidential methodology, profit system based in Oak Brook, Illinois, is the larg- developed with Advocate, to attribute patients to a est integrated health care system in the state, and one provider group based on a two-year look-back and a tie- of BCBS IL’s largest networks of provider practices. breaking algorithm. BCBS IL shares reports of attrib- Advocate encompasses more than 250 care sites, includ- uted populations on a monthly basis with Advocate, ing 10 acute care hospitals and two integrated children’s and the December 2011 attributed population was to be hospitals. used for its initial shared savings calculation. BCBS IL began a shared-savings program in BCBS IL has set no minimum size for the 2011 with Advocate for services delivered to members patient population of participating provider groups. of a PPO, most in a seven-county Chicago area, plus However, BCBS IL sought a provider with a substantial one hospital in Bloomington, Illinois. Two groups of population for this first arrangement with an account- Advocate primary care physicians are participating in able care organization. this program: employed physicians, and 900 doctors in a physician hospital organization, for a total of 3,500 Determining and Distributing Savings physicians. Advocate and its physicians must fulfill two require- ments to receive shared savings: they must reduce costs Why Shared Savings? and achieve quality outcomes. For the first require- BCBS IL and Advocate both recognized that the ment, BCBS IL compares baseline medical costs for the current health care cost structure was unsustainable. attributed population to the medical costs of patients According to an Advocate interviewee, “We wanted to served by the rest of BCBS IL’s PPO network. To systematically eliminate waste through various initia- qualify, Advocate’s trend must be lower than the risk- tives (decreasing ER visits, shortening length of stay, adjusted trend of the nonparticipating PPO network by managing ambulatory-sensitive conditions to avoid more than a predefined amount. admissions). We believed that by working collabora- After Advocate has fulfilled this requirement, tively we could develop a creative approach and step participating physicians must also meet minimum outside the usual adversarial relationship between pro- quality thresholds to receive shared savings. In the first viders and payers.” year of the new approach, Advocate physicians had BCBS IL cited several reasons for choosing to maintain performance on 12 quality measures, and Advocate as a partner. First, it was both a high-quality faced penalties if performance declined. In the sec- and a high-cost provider. Second, Advocate’s historic ond and third years, Advocate and BCBS IL planned commitment to improving health care quality and 5 Commercially insured HMO members are in a different type infrastructure reassured BCBS IL that the organization of full-risk contract with BCBS IL. www.commonwealthfund.org17 to negotiate targeted increases in an expanded set of HEALTHPARTNERS measures. HealthPartners is the largest consumer-governed, non- If Advocate achieves overall savings and par- profit health care organization in the nation, covering ticipating physicians pass the quality “gate,” they are more than 1 million members in Minnesota, western eligible to share up to about 50 percent of the cost sav- Wisconsin, North Dakota, and South Dakota. The ings—minus a confidential percentage set aside to offset regional network includes more than 38,000 doc- BCBS IL’s investments in infrastructure. To ensure a tors and other care providers. NCQA has awarded transparent process, BCBS IL has given Advocate the HealthPartners it highest rating, and ranked it among right to audit its calculations. the top 10 percent of commercial health plans in the nation in 2010–2011. The National Business Coalition Adjusting for Patient Risk on Health has also ranked HealthPartners as the best in BCBS IL uses DxCG software to account for variations the nation on its eValue8 assessment. and changes in the risk profile of Advocate’s patient In 2009, during its contract renewal process, population. However, BCBS IL has chosen not to limit HealthPartners began developing shared-savings the risk that providers face for patients with extremely arrangements with all major care systems in its net- high costs, because the number of patients covered by work that were subject to total-cost-of-care approaches the shared-savings arrangement is large, and BCBS IL to payment. (These systems include Fairview Health assumes that a normal distribution of patients will occur Services—see the Medica–Fairview case study.) across providers. However, BCBS IL excludes the costs of a few Why Shared Savings? high-cost services, such as transplantation, from the HealthPartners chose shared savings as an expres- savings calculations. When determining the cost trend sion of its dedication to the Institute for Healthcare for the nonparticipating BCBS IL network, to compare Improvement’s triple aim: to improve the quality of with Advocate’s costs, BCBS IL also excludes very low- care, the health status and care experience of each indi- cost claimants whom it could not attribute to a provider. vidual, and the affordability of care. HealthPartners views shared savings as an incremental step toward Technical Assistance more comprehensive risk-sharing with providers. Participating practices receive reports, data feeds, and access to software tools that allow them to download Shared-Savings Methodology data directly into their electronic medical record sys- Patient Population: Attribution and Minimum Size tems. BCBS IL also provides analytical support on HealthPartners’ shared-savings arrangement applies treating targeted groups of high-risk members. to members enrolled in commercial, fully insured, and self-insured PPOs and HMOs. Two care systems also BCBS IL’s Experience with the opted to include their Medicaid-insured members. Shared-Savings Model HealthPartners uses a confidential internal methodol- While the model is still relatively new, BCBS IL ogy to retrospectively attribute members to a primary believes that its basic framework—especially the quality care provider. targets—is sound. Although it notes that the methodol- HealthPartners has set no minimum size ogy may evolve, BCBS IL hopes that participants in for the patient population of participating provid- other markets can use it as a template. ers, because it now contracts only with large groups of physicians. However, if HealthPartners begins to contract with smaller physician groups, it would look to the Centers for Medicare and Medicaid Services 18 Shared-Savings Payment Arrangements in Health Care: Six Case Studies for guidance on an appropriate minimum patient HealthPartner’s Experience with the population. Shared-Savings Model HealthPartners had just begun calculating shared sav- Determining and Distributing Savings ings when we interviewed plan officials, and did not yet HealthPartners compares each care system’s perfor- have findings. A Health Partners interviewee noted that mance to a jointly negotiated target. The average per- “Some [providers] initially opposed [the arrangement], patient cost of care in a metropolitan area is the negoti- but we are having a lot more success over time.” ated target for one system. HealthPartners calculates The interviewee also observed that transpar- savings by comparing the performance of participating ency in all aspects of the approach is critical, and rec- practices to the costs of all health care services in the ommended that organizations strive to ensure that all target. The organization calculates shared savings for participants benefit. HealthPartners offered to share its commercially insured and Medicaid-insured patients experience with other payers and providers interested in separately. pursuing a shared-savings arrangement. Providers and the plan share any cost savings equally. Unlike some shared-savings arrangements, HealthPartners does not recapture plan costs associated with the arrangement, require reinsurance to protect against high-cost outliers, or have a minimum savings threshold. Adjusting for Patient Risk HealthPartners uses the ACG software to account for variations and changes in the risk profile of the patient population of each provider group. Health Partners has also chosen to truncate the costs of high-cost patients, although the dollar level is confidential. Technical Assistance HealthPartners gives each provider a sophisticated report analyzing health care use and costs by condi- tion, place of service, and specialties. The report also compares these results to those of other providers in the region, to suggest opportunities for improvement. HealthPartners meets with providers to review these reports. HealthPartners also provides some grant funds to help physician groups build the capacity to succeed in this shared-savings arrangement, and to cover their costs of implementing it. www.commonwealthfund.org19 HARVARD PILGRIM HEALTH CARE HCHP acknowledges that this figure is lower Harvard Pilgrim Health Care (HPHC) is an indepen- than what actuaries might desire. However, the plan dent, nonprofit health plan with some 1 million mem- wants to offer the approach to many providers, and bers in New England, and the second-largest health believes that any risk it would incur is worth taking plan in Massachusetts. HPHC offers a variety of plan relative to the fee-for-service status quo. HPHC also options for employers, families, and individuals, includ- believes that payers need to afford providers a fair ing fully insured and self-insured plans and traditional amount of leeway in the early years of shared-savings HMOs. A subsidiary of HPHC, Health Plans, Inc., is arrangements. the largest third-party administrator in New England. In the first quarter of 2010, HPHC adopted a Determining and Distributing Savings shared-savings model for its HMO members—a large HPHC builds a health care budget for assigned patients percentage of its total membership—with a large pro- based on the two-year average costs of a provider group. vider network affiliated with an integrated health care The plan also considers a multiyear cost trend for the delivery system. HPHC has since made the shared- network as a whole, and negotiations with each provider savings model available to other providers. It is a stand- group, when creating this budget. ing option in an array of approaches to payment used by HPHC assesses providers’ performance every HPHC, which include full-risk options and traditional 12 months by comparing it to the average costs of that fee-for-service with pay-for-performance incentives. medical practice over the previous two years. HPHC uses this average to take into account random varia- Why Shared Savings? tion in the use and cost of care among the patients of a HPHC introduced shared-savings contracts with pro- practice. vider groups for three reasons. First, the health plan If the group reduces health care costs compared wanted to focus the attention of smaller and midsized with the budget, HPHC retains the first 2 percent medical groups on managing the use and cost of health of those savings to account for variation in medical care services without requiring them to assume financial expenses across time, and for the absence of downside risk, which can be hazardous for small groups because risk. HPHC then gives 50 percent of any remaining of the volatility of medical costs. savings to the provider group, up to a maximum of 6 percent of the target. That means that a provider can Second, the plan believed that shared savings receive up to 3 percent of the budget target as shared could serve as a bridge to a shared-risk model for larger savings. medical groups. Finally, HPHC was concerned that Unlike many other shared-savings arrange- practices would agree to shared-risk or full-risk arrange- ments, this approach does not tie payments to any ments before they were prepared to manage them. performance targets. HPHC maintains a preexisting pay-for-performance program that coexists with the shared-savings model to provide separate bonuses for Shared-Savings Methodology high-quality performance. Patient Population: Attribution and Minimum Size HCHP uses an internally developed methodology to Adjusting for Patient Risk assign patients to provider groups. Providers must serve HPHC uses DxCG software to account for variations at least 3,000 patients to participate in the shared-sav- and changes in the risk profile of patient populations. ings arrangement. HPHC excludes patients with costs above $100,000 20 Shared-Savings Payment Arrangements in Health Care: Six Case Studies from the provider’s population when calculating savings. CONCLUSION The plan also excludes certain services, such as behav- These six case studies reveal consistent themes regard- ioral health and transplantation, from the calculations. ing shared-savings approaches to payment for health Under the shared-savings model, HPHC assumes all care services. These common elements include a will- reinsurance costs. ingness among most payers to absorb many of the costs entailed in setting up and sharing tools for measur- Technical Assistance ing health care performance and cost savings. Overall Before HPHC begins to share savings with a provider themes also include a belief that shared-saving pro- group, it assesses its readiness to handle this new pay- grams must evolve to include shared risk, and a convic- ment model. HPHC uses an internally developed tion that even when pilot programs fail to achieve sav- assessment tool to determine whether the practice has ings, they are moving in the right direction. the clinical expertise and leadership to manage care for We do not yet know whether this approach is its overall patient population. a long-term strategy for promoting better health care After the assessment, HPHC offers techni- while lowering costs, or a transitional strategy to some cal assistance to further prepare practices for this new other model, such as global payments involving shared model. HPHC gives them access to basic reporting risk. Exploring the organizational and environmental tools to track their own performance, examples of best differences in how participants pursue shared-savings practices, and medical leadership, all at no cost. For a approaches, and the outcomes they achieve, will be key fee, practices also have access to other tools and techni- to determining whether they work, how to improve cal assistance, such as care management, hospital uti- them, and whether and how to diffuse them. lization review, and the production and distribution of more in-depth reports. HPHC’s Experience with the Shared-Savings Model HPHC believes that its shared savings model will remain a long-term option for some provider groups. HPHC believes in building a collaborative relationship with providers, and has no intention of forcing groups into this or a full-risk payment model. In fact, HPHC believes that provider groups with historical efficiencies can remain in the fee-for- service model while receiving pay-for-performance bonuses. HPHC’s collaborative approach also includes negotiating with providers so they have a fair and rea- sonable say in setting budgets and basic fees for service. www.commonwealthfund.org21 One East 75th Street 1150 17th Street NW New York, NY 10021 Suite 600 Tel 212.606.3800 Washington, DC 20036 Tel 202.292.6700 www.commonwealthfund.org