In recent years, accountable care organizations (ACOs) have become more prevalent in the United States. This study describes the origins, implementation, and early results of Minnesota's Medicaid ACO payment model, the Integrated Health Partnership (IHP) demonstration project. We describe the structure of the program and present preliminary evaluation results to document the state's important work and to provide lessons for other states interested in implementing Medicaid ACOs. The IHP program has expanded in size over time, the state has reported significant savings, and evidence exists of capacity building among participating providers. We identify factors that may have contributed to the program's early success, but more work is needed to investigate the specific drivers of quality improvement and savings within Minnesota's ACO program and to compare the design and effects of the IHP with other Medicaid and Medicare ACO programs. We conclude with comments about the future of the state's Medicaid ACO program and situate Minnesota's findings within the context of the broader ACO movement.
The reach of accountable care organizations (ACOs) is growing in terms of numbers of covered lives and use by commercial and public payers. In January 2016 an estimated 838 ACOs in the United States provided services to 28.3 million individuals (Muhlestein and McClellan 2016). ACO payment arrangements are designed to transform care delivery through a model that contracts directly with health care providers who agree to accept financial risk and meet negotiated performance goals. Although Medicaid represents just 10 percent of the current ACO market in terms of covered lives, state Medicaid ACO models are growing.
In this article we describe the origins, implementation details, and early results of Minnesota's Medicaid ACO payment model, the Integrated Health Partnership (IHP) demonstration project. This new payment model, entering its fifth year of operation, is part of ongoing payment reform efforts in Minnesota. Minnesota reforms are being developed in the context of Minnesota's history of health care innovation, high health insurance coverage rates, and progressive public health care programs. We provide an overview of Minnesota's IHP program along with a preliminary look at evaluation results to document this important work and provide lessons for other states interested in implementing Medicaid ACOs.
We begin with an overview of Minnesota's political and health insurance coverage context, including the passage of state health reform legislation in 2008 that was critical to the launch of the IHP program. We then describe key operational elements of the IHP program and present preliminary evaluation outcomes. The evidence summarized in this article comes from program data and information publicly reported by the state as well as qualitative interviews reported in greater detail elsewhere (Dybdal et al. 2014; SHADAC 2016). We conclude with our thoughts about the future of the Medicaid ACO program in Minnesota and situate Minnesota's findings to date in the context of the broader ACO movement in the country.
Minnesota has historically had high rates of employer-based health insurance coverage, high quality health care, low costs, and comprehensive public programs that serve as a supplement to employer-based coverage (MDH 2014). The presence of seventeen Fortune 500 companies and a social commitment to the local community have created opportunities for public-private partnerships and innovations around health care financing and service delivery in the state.
The Minnesota model of care is based on a long history of relatively large multispecialty provider organizations and a mature managed care market with both staff-model managed care organizations (MCOs) and broader network preferred provider organizations. Minnesota's pre-ACA public programs included the Medicaid program (Minnesota's Medical Assistance program), the General Assistance Medical Care Program (GAMC, a state-funded program for uninsured, low-income childless adults), and MinnesotaCare (a state-funded health insurance program for uninsured families not poor enough to qualify for Medicaid).1 These programs have historically been delivered through MCOs with several plan options available to enrollees across the state. Until this year, Minnesota was the only state that did not allow for-profit MCOs, creating a unique, largely insulated managed care market dominated by four local health plans.
In 2011 Minnesota was one of six states that used the early Medicaid expansion option under the Affordable Care Act (ACA) to expand coverage to childless adults with incomes up to 75 percent of the federal poverty level (FPL) (SHADAC 2015). By 2014 Minnesota had a fully expanded Medicaid program under the ACA (to 138 percent FPL), and MinnesotaCare had been transitioned to become the first ACA Basic Health Plan (BHP) for individuals with incomes between 138 and 200 percent FPL. With the contribution of these programs, along with the state-based MNsure Health Insurance Marketplace, Minnesota now has an uninsurance rate of 4.3 percent, the lowest in state history (MDH 2016a). Figure 1 shows the current distribution of health insurance coverage by type for Minnesota.
Minnesota Health Care Reform Legislation
In the mid-2000s, policy makers and business leaders raised concerns about the rising costs of health care, the potential erosion of the private health insurance market, and the unprecedented increase in the uninsurance rate. There were additional worries about access to quality health care, increases in the obesity rate, and concerns about the health of Minnesota's population. (Health Care Transformation Task Force 2008). In response, the bipartisan Health Care Transformation Task Force was created in 2007 to “advise and assist the governor regarding activities to transform the health care system” and to develop “a statewide action plan for transforming the health care system to improve affordability, quality, access, and the health status of Minnesotans” (MDH 2008: n.p.). In addition, the Legislative Commission on Health Care Access was formed in 2008 to advance ideas to achieve universal health coverage. Both committees included public- and private-sector health care leaders, providers, and other stakeholders and made recommendations to move forward on payment reform in Minnesota's public programs (Health Care Transformation Task Force 2008; Legislative Commission on Health Care Access 2008).
Many of the specific recommendations of the task force, championed by Republican governor Tim Pawlenty, were passed into law in 2008. The 2008 Health Reform Law, detailed in Minnesota Statutes §256B.0755 (2008), included statewide quality reporting, payments for coordinated care services provided in health care homes (HCHs), and encouragement to providers to participate in bundled payment demonstrations, or baskets of care. The legislation has been modified over time, with the Democratic administration of Governor Mark Dayton, who took office on January 3, 2011, implementing the more recent reform strategies. The reform legislation was amended again in 2011 to require the Minnesota Department of Human Services (DHS) to develop a demonstration to “test alternative and innovative health care delivery systems, including ACOs that provide services to a specified patient population for an agreed-upon total cost of care or risk/gain sharing payment arrangement,” which led to the IHP demonstration (Minnesota Statutes 2008, §256B.0755 subdivision 1).
Legislation also included authorization for demonstrations in the urban counties of Hennepin and Ramsey, which led to the creation of the Hennepin Health program, a safety-net ACO to provide integrated health and social services to the county's low-income residents (with incomes less than 138 percent FPL). The IHP and the county-based demonstrations were intended to improve health care quality, lower public health care costs, and align with the opportunities available to states under the new ACA (SHADAC 2016). The demonstrations were also part of a larger legislative effort that included limits on managed care administrative costs, a competitive price bidding contracting pilot with MCOs, and other payment and care delivery reforms to reduce nonadministrative managed care payments.
CMS State Innovation Model Program
The federal Center for Medicare and Medicaid Innovation's (CMMI) State Innovation Models (SIM) initiative provided the next impetus in Minnesota's ACO initiatives. In February 2013 Minnesota was one of six states to receive a CMMI SIM “Model Test Award” designed to assist states ready to implement State Health Care Innovation Plans.2 States submitting plans were required to use multipayer strategies to transform the health care delivery system through payment reform. The DHS was awarded just over $45 million to improve health in communities, provide better care, and lower health care costs (DHS 2016b). This built on the state's existing initiatives (including the IHP demonstration, HCHs, and eHealth Initiatives) and provided additional resources needed to further develop the payment model.
Under the SIM initiative, the state has expanded the IHP program from six to twenty-one participating integrated provider systems of care delivery (DHS 2017). Other areas of SIM investment included the development of a data portal for IHP providers and grants and technical assistance to expand their data analytic capabilities.
Minnesota's Integrated Health Partnership Program
Minnesota's Medicaid ACO program is a voluntary program for provider-based systems of care that uses a shared savings/shared risk financing model. Provider performance is based on meeting negotiated total cost of care (TCOC) targets and quality measurement benchmarks (HCA 2016). The IHP program works alongside but separate from the Medicaid program's capitated payment arrangement with MCOs in the state, although contracted MCOs are required to participate as payers in the shared savings component of the program. IHPs agree to deliver the full scope of primary care services, coordinate access to specialty providers and hospitals, engage and partner with patients and families, and develop partnerships with community organizations and social service agencies that they will then integrate into care delivery (Zimmerman 2014).
Participating delivery systems have significant flexibility in the design of or modifications to their clinical models and innovations. To that end, the state provided flexibility for “integrated” delivery systems that are required to take on upside and downside financial risk and for smaller systems that participate as “virtual” partners with no downside risk. If costs are higher than anticipated for an IHP's attributed patient population and provider performance is subpar during a performance period, the provider system must reimburse the state the prenegotiated portion of the payment received. If the costs are lower and performance is positive, achieved savings are shared between the state and the provider per a prenegotiated agreement.
The first request for proposals (RFPs) for the Medicaid IHP program was issued in late 2011, with the first round of IHP contracts implemented in 2013. The state initially entered into contracts with a total of six provider systems, including four in the Minneapolis/Saint Paul metropolitan area (including a children's hospital and clinic system and a network of Federally Qualified Health Centers [FQHCs]) and two in other parts of the state.3 Between 2014 and 2016, thirteen more providers joined the demonstration for a total of nineteen participating IHPs. During 2016, an estimated 415 clinics representing 9,907 individual providers were participating in Minnesota's IHP program, providing care to an estimated 342,000 Medicaid enrollees (H. Peterman, personal communication, May 27, 2016). Minnesota issued an RFP in April 2016 for a fifth year of the IHP program, with two additional contracts awarded in January 2017 (HCA 2016; see fig. 2).
Patient Attribution Methodology
Fundamental to the IHP model is the assignment or attribution of public program enrollees to participating IHPs. Eligible enrollees include fee-for-service and managed care adults and children enrolled in Medicaid (including pregnant women, children under twenty-one, and adults without children) who have at least six months of continuous Medicaid enrollment or at least nine months of noncontinuous enrollment during the performance period and at least one evaluation and management or HCH visit as indicated by a payment claim during that period. All Minnesota Health Care Program (MHCP) enrollees (excluding those dually eligible for Medicare and Medicaid) can be potentially assigned to an IHP.
Minnesota assigns eligible Medicaid enrollees to IHPs using a retrospective attribution methodology based on patient claims. The specific claims include those for HCH visits, coordination of care activities, primary care evaluation and management visits (such as preventive or well-child services), and relevant specialist visits. The assignment follows the following hierarchy of decisions: During the first step of the process, enrollees with HCH claims are attributed to an IHP if (1) they only have HCH claims with a single IHP; (2) the majority of their HCH claims are with one IHP; or (3) they have a similar number of claims at multiple IHPs, in which case they are assigned to the IHP with the most recent HCH claim. During the second step of the process, claims for primary care physician visits are used to attribute additional eligible enrollees who lack an HCH claim. For those not attributed during the second step, claims for specialist visits are used to assign patients to an IHP.
During the performance year, the state provides each IHP with monthly reports on its attributed population's use of services and claims costs. Final performance is determined at the end of the performance year based on the final end-of-year attributed population. As a result, IHP providers have an approximate roster of attributed patients for whom they will face financial risk over the course of the year. Interestingly, only about 60 percent of IHP providers' patients are assigned or attributed to an IHP (Dybdal et al. 2014), either because a large proportion of members are not eligible or because the state's patient attribution methodology does not identify a provider who can be credibly assigned.
Total Cost of Care Calculation
A key component of the IHPs' performance assessment and the state's shared savings/loss calculation is the IHPs' TCOC. Minnesota's TCOC calculation is based on a subset of Medicaid services that providers can reasonably expect to control or coordinate. Core services included in the TCOC calculation are inpatient and outpatient hospital services, physician/professional clinic services, pharmacy services, and certain mental health and chemical health services. Dental, supplies, transportation, and long-term services and supports are generally excluded from the required core but can be optionally included by the IHP (Zimmerman 2014).
To determine the annual cost performance for an IHP, a per-member per-month (PMPM) target TCOC is compared with the observed PMPM TCOC for the IHP's performance period, adjusted for the size of the patient population, relative risk (based on the Adjusted Clinical Groups risk adjustment tool), and outliers. The target TCOC is based on a TCOC calculated for the attributed population using prior year claims and encounter data for the included services. The base period TCOC is then adjusted for the expected trend based on “the annual expected cost increases for the aggregate MHCP population, with appropriate adjustments for services excluded from the Base TCOC or other factors that are applicable to the total cost of care and goals of the program” (HCA 2017: 15). All the claims used to calculate the TCOC are attributed to one IHP, including all services from other providers.
Currently, potential financial gains for IHP providers are based on meeting performance targets for thirty-two measures of health care processes, health care outcomes, and patient experience. The amount of shared savings tied to quality performance increases over the three-year demonstration period. In year 1 of the demonstration, performance is based on measurement reporting only (not performance on quality measures) and has an impact on 25 percent of the potential financial gains. Performance targets are measured in all subsequent years with shared/savings adjustments based on 25 percent in year 2 and 50 percent in year 3 and beyond.
All quality metrics required of IHP providers are part of the Statewide Quality Reporting and Measurement System, an existing state measurement initiative administered by the Minnesota Department of Health (MDH 2016b). Therefore, IHP providers have no new or additional reporting requirements. The measures are standard across IHP providers, but providers may negotiate with the state to include additional measures or subsets of measures based on populations served. Each quality metric is assessed on whether a threshold of performance was achieved (at least in the 30th percentile of the previous measurement for the state) and whether there was demonstrated improvement over time (defined as the IHP's relative improvement of at least 5 percent since the last performance period).
Monitoring of IHP Performance and Settlement of Shared Savings/Losses
In order to provide critical information about performance and payment in a timely manner, the state provides both interim and final payments based on interim attribution and claims information. In addition, the state provides ready access to patient profiles, including utilization and risk for an IHP's attributed patient population through a data portal to help it better understand resource use and identify areas for targeted interventions. These data include provider alert reports (monthly reports of attributed patients with an emergency department visit or hospital admission in the prior month), care management reports (monthly patient-level clinical profiles for all attributed recipients), utilization detail files (monthly files containing select claims for attributed patients), and a TCOC package (quarterly reports on TCOC performance by provider and category of service). This information was initially provided to IHPs in a static report format but due to investments made by the state under SIM, is now available as patient- level microdata for download.
Throughout the SIM initiative in Minnesota, the IHP program has grown steadily, and expansion has brought greater diversity among the participating IHP providers. Contracted IHPs vary in terms of size, organizational structure, and geographic reach. Some new providers are beginning to test the inclusion of services not traditionally included (e.g., behavioral health) and other ACO innovations. This growth and the onboarding of different types of providers—especially those that treat populations with complex medical and social needs—have encouraged state officials. Minnesota is on track to meet its goal to include about 500,000 public program enrollees in the IHP program or similar value-based reforms by the end of 2018 (DHS 2016a).
Costs and Quality
Reported savings to the state have increased as more providers have joined the demonstration, from $14.8 million in 2013 to $76.6 million in 2015, for a total three-year savings of $157 million (DHS 2016c). Savings have also accrued to providers. In 2014, when 25 percent of shared savings was based on both quality reporting and performance for the six round one IHPs, all nine providers received a shared savings settlement, ranging from approximately $388,000 to $4.7 million and totaling $23 million.
A specific example of savings at the provider level is the work of the FQHC Urban Health Network. This IHP, made up of ten metro-area FQHCs, used e-health and data analytics for thirty thousand enrollees to provide coordinated care, emergency department utilization interventions, and effective clinical interventions. Program results showed a decrease in the growth of TCOC by 5 percent and a savings of $16.6 million over a three-year period through an 18 percent decline in emergency department use and an 8 percent decline in inpatient hospital use (Fournier 2016).
The development of data analytic capacities and on-site e-health infrastructures designed to promote better coordinated care is evident within the IHP program. Under the SIM initiative, Minnesota has made significant investments in the IHP data infrastructure and data analytic resources necessary to track metrics and provide reporting at the provider level. Many IHPs felt positive about the state IHP data and the reporting they received. Through the analysis and use of the state IHP data, IHPs report that they are learning more about the populations they serve, shaping clinical initiatives that will have an impact on both Medicaid and non-Medicaid patients, and supporting data infrastructure with a population health orientation. Some of the IHPs acknowledged that data analytics is a key area of growth for their systems and for their ability to focus on the population health components of care delivery (SHADAC 2016).
Significant advances have also taken place in the development of community partnerships and service integration, which is a requirement for IHP participation. Some IHPs had already forged community partnerships prior to the program, but others formed partnerships with community organizations under the incentives and requirements of the IHP program. Partnerships are evolving in a variety of ways and include referrals to long-term services and supports, coordination with mental health programs, and the use of food banks. To date, most of these partnerships include no financial commitment or binding language and are therefore informal (SHADAC 2016).
Interviews with the IHPs highlighted several insights related to the future of the IHP model (SHADAC 2016). First, IHPs expressed concern about a lack of clarity in the state's vision for the relationship between the IHP program and its managed care delivery system. IHPs also raised concerns about the continued ability of successful participating provider systems to generate shared savings over time, given that the base TCOC is derived from the provider's prior year claim totals, which is decreasing during program participation. Some IHPs called for a prospective attribution model (instead of a retrospective one) to allow for better targeting of clinical investments on the clearly identified patients. Some IHPs also indicated that although they were successful during the early years of the program and earned shared savings, they were not necessarily able to identify the key factors that led to these accomplishments, limiting their ability or others' ability to replicate these activities. Finally, some IHPs argued that a portion of the shared savings could be made on a prospective basis to help initiate and maintain the cost of care coordination, develop relationships with other providers, enhance their information/data sharing capabilities, and fund their population health efforts. These IHPs called attention to the significant investments that provider groups considering IHP participation must make.
In April 2016 the DHS issued a request for information to inform the program's design and future sustainability. Identified areas for potential improvement included more flexibility in the IHP payment model structure, to allow smaller or specialty-focused providers to participate, and adjustments to the member attribution, to stabilize the attributed population over the performance year.
In recent years, ACO membership has risen rapidly in the United States, with attributed patients increasing from 2.6 million in 2011 to 28.3 million in 2015 (Muhlestein and McClellan 2016). Medicare accounts for an estimated 30 percent of ACO-covered lives through its Medicare Pioneer and Shared Savings Programs. Evaluations to date have been mixed, with only modest reductions in per-beneficiary spending since the program started in 2012 (McWilliams et al. 2015; McWilliams et al. 2016). State Medicaid programs are moving forward with their own models of reform. Ten states have established ACO programs as of September 2016 (CHCS 2016b). As such, there is little evidence to date about their effect on costs and outcomes but more coming (Christensen and Payne 2016: 148–54; Gleeson, Kelleher, and Gardener 2016).
Medicaid populations may differ in important ways from attributed patients in Medicare or commercial ACOs, which may have implications for the design and the success of programs. First, Medicaid patients include a larger pediatric population with lower health care utilization than adults (and thus less potential cost savings), although there are chronic complex pediatric patients representing potential benefits from care coordination (Homer and Patel 2013). Second, due to Medicaid eligibility criteria, Medicaid populations are likely to be socioeconomically disadvantaged relative to beneficiaries attributed to Medicare or commercial ACOs and may also require nonmedical assistance, such as housing, employment counseling, or other social services (CHCS 2016a).
Given this context, the quality improvement and savings reported by the Minnesota IHP program are notable. Numerous factors may have contributed to the early success of the program, including a focus on data analytics and population health, the inclusion of behavioral health services, and the inspiration to establish partnerships with community providers. In addition, the large integrated delivery systems participating in the Minnesota IHP program may have a greater capacity to coordinate care across provider settings and other services for this patient population.
More work is needed to investigate the specific drivers of quality improvement and savings within Minnesota's ACO program and to compare the design and effects of the IHP with other Medicaid and Medicare ACO programs. These analyses may identify the specific factors associated with successful care coordination for Medicaid patient populations.
1. MinnesotaCare is funded by a 2 percent assessment on nonpublic provider revenue and a 1 percent premium tax on health care premiums in the commercial health insurance market.
2. Other “test” states included Oregon, Vermont, Massachusetts, Arkansas, and Maine.
3. The initial period of participation for each IHP is three years, but contracts are renewed annually. IHPs that have completed a three-year cycle can continue in a subsequent three-year cycle, with the expectation that their risk level will be similar to or greater than their third year of participation unless they take on additional partners or services.