This article presents a detailed descriptive analysis of how Massachusetts and Minnesota implemented Medicaid accountable care organization (ACO) models for their managed care population.
CMS aims to transition most Medicaid beneficiaries into accountable care relationships in the coming years. With more than 70% of Medicaid beneficiaries enrolled in managed care, CMS cannot reach this goal if Medicaid accountable care organizations (ACOs) focus exclusively on the fee-for-service population, as most Medicare ACOs do. We used a case study approach focused on the experiences of Minnesota and Massachusetts implementing a Medicaid ACO model that includes managed care, with details on stakeholder engagement, beneficiary attribution, and payment methodologies. Massachusetts took an approach that significantly changed the managed care organization (MCO)–provider–enrollee relationship by blurring the line between MCO and ACO, whereas in Minnesota the existing MCO structure remained largely unchanged for the beneficiary and provider. As Medicaid agencies increasingly adopt ACO models, the experiences of Minnesota and Massachusetts can help policy makers understand the effects of different implementation approaches.
Am J Accountable Care. 2022;10(3):28-33. https://doi.org/10.37765/ajac.2022.89233
As Medicaid costs continue to rise, accountable care organizations (ACOs) are gaining traction as Medicaid payment models. CMS aims to transition “the vast majority of Medicaid beneficiaries into accountable care relationships by 2030.”1 With more than 70% of Medicaid beneficiaries enrolled in managed care, CMS cannot reach this goal if Medicaid ACOs focus only on the fee-for-service (FFS) population, as most Medicare ACOs do.2 This necessitates understanding how ACOs can function in a Medicaid managed care population.
In 2019, Rutledge and colleagues summarized the implementation and early impacts of Medicaid ACOs in 4 states (Maine, Massachusetts, Minnesota, and Vermont), including implementation barriers and facilitators and statistical analyses suggesting that these models improved outcomes.3 Here we expand upon Rutledge et al by providing a more detailed description of the models of Massachusetts and Minnesota, 2 states with significant managed care penetration, that implemented provider-driven Medicaid ACO models for both their FFS and managed care populations. Specifically, we provide a detailed descriptive analysis of how their Medicaid ACO models intersected with their long-standing Medicaid managed care organizations (MCOs), including details on stakeholder engagement, infrastructure support, beneficiary attribution, and sharing of savings and losses.
In addition to providing information beyond what is available in the literature, we are the first to present information in a comparative context, better illuminating and contextualizing each state’s approach. Understanding differing approaches to Medicaid ACO implementation can facilitate development of these payment models in different state contexts.
We conducted a secondary analysis of qualitative data collected for the first State Innovation Models (SIM) Initiative Evaluation and used by Rutledge et al. The SIM Initiative provided significant funding to facilitate state-level payment and delivery system reform.4 The evaluators conducted interviews on health care transformation activities in Massachusetts and Minnesota, among other states, between 2013 and 2018. State-specific experts from the evaluation team analyzed relevant interview notes to understand beneficiary attribution, distribution of shared savings/losses, MCO procurement, and infrastructure support. We further reviewed notes for stakeholder perceptions of Medicaid ACO implementation. Where needed, we supplemented interview notes with notes from monthly calls with state officials and documentation review.
As payment models that are intended to increase coordination across providers and accountability for total costs, the goals of ACOs and MCOs overlap. Their organizational structures rely on different entities (insurance providers for MCOs vs provider organizations for ACOs), yet both Massachusetts and Minnesota used statewide health reform efforts to take advantage of both structures. Both Minnesota and Massachusetts had to make several key implementation decisions to blend their ACO model into their long-standing MCO model, including how to engage their MCOs and providers, how to provide infrastructure support, how beneficiaries would be attributed to an ACO, and how financial incentives would flow among the state, the MCO, and the ACO. After providing an overview of each state’s model, we describe the approach to each of those decisions.
In 2012, Minnesota launched its Medicaid ACO Model. In this model, the state contracts directly with the Medicaid ACOs, whose attributed beneficiaries may either be enrolled in MCOs or be part of the FFS population. Minnesota’s strategy was a “top-down” approach driven by the state that mostly maintained the existing relationships among Medicaid, the MCOs, and the enrollees. Its approach varied little over the course of our study. In 2016, Massachusetts launched the Primary Care Payment Reform Initiative (PCPRI), which was short-lived due to low MCO participation. The Medicaid ACO Model succeeded PCPRI and fundamentally changed the state’s MCO structure. It had 3 variants: a Partnership Model, a Primary Care Model, and an MCO-Administered Model. The Partnership Model is an exclusive contractual relationship between an MCO and an ACO with the entire ACO-attributed population covered by the same MCO. The Primary Care Model is a contractual relationship between an ACO and the state for the FFS population. The MCO-Administered ACO has nonexclusive contracts between ACOs and MCOs, which allows an ACO to contract with more than 1 MCO and vice versa. As in Minnesota, in the MCO-Administered Model an ACO’s beneficiaries may represent multiple MCOs. Massachusetts further contracted with 2 of its 5 MCOs to manage care for non–ACO-attributed managed care beneficiaries.5
Medicaid MCO and Provider Engagement
Minnesota’s ACO implementation coincided with Medicaid MCO reprocurement, and the state required MCO participation in its model. Providers, particularly those already in commercial or Medicare ACOs, were eager to take advantage of directly contracting with the state. ACO executives perceived this new relationship with the state as positive and valuable. MCO leaders, however, were disappointed that the state did not include them in model design discussions, given their experience in risk management and data analysis. MCO executives were concerned about managed care’s role in the context of a payment model in which the state would contract directly with providers, potentially blurring accountability. By the end of 2017 there were 21 Medicaid ACOs in Minnesota, reaching more than half of the state’s eligible Medicaid population.
Massachusetts initially asked for voluntary, rather than contractual, MCO participation in PCPRI. None participated, leading to minimal provider participation. Without the MCOs, providers did not have enough covered beneficiaries to implement the needed delivery system changes. MCOs cited competing priorities, such as reprocurement and pharmaceutical costs, as explanations for their lack of interest. In late 2016, Massachusetts shifted from PCPRI to pilot a new Medicaid ACO model. Prior to full implementation of this new ACO model, Massachusetts reprocured its MCO contracts and required 5 MCOs to contract with Medicaid ACOs for the next 5 years.
The new approach in Massachusetts, particularly the Partnership Model, represented a complete restructuring of how managed care would interact with its providers. Unlike those in Minnesota, Massachusetts MCOs had limited networks prior to ACO implementation and leveraged those relationships to create partnerships in the Partnership Model. Ultimately, 13 ACOs participated in the Partnership Model, 3 ACOs—which are among the state’s largest and served as pilot ACOs—participated in the Primary Care Model, and 1 ACO participated in the MCO-Administered Model. As of March 2018, ACOs covered more than half of the state’s Medicaid population.6
Data and Infrastructure Investment
Both states used a portion of their SIM Initiative funding to provide ACOs with technical assistance and infrastructure support, such as facilitating data analytics and connection to a health information exchange. These state investments were critical to encouraging providers to enter these new and sometimes unfamiliar payment arrangements.
Minnesota provided $4.2 million in grants directly to individual Medicaid ACOs to help them develop targeted analytic capacity, and the state hired a contractor to facilitate improvement in data analytics across ACOs. To create a cohesive population for each ACO to manage, the state Medicaid agency provided ACOs with encounter data for their entire panel of attributed beneficiaries—regardless of beneficiaries’ MCO enrollment. In other words, ACOs had the entirety of their Medicaid population’s encounter data in 1 place and in the same format, which would not be possible absent the state providing it. ACO leadership universally praised both the data and the support the state provided, stating that their ability to manage their Medicaid population using consistent data and metrics was invaluable.
Massachusetts supported infrastructure by renewing its Medicaid Section 1115 demonstration waiver, which includes a Delivery System Reform Incentive Program protocol.7 This helped ACOs connect to the state’s information exchange, develop data analytics, and fund community partners and other infrastructure development. Massachusetts ACO executives cited the generous infrastructure funding made available only to ACOs as a strong incentive to participate.
Although not yet implemented at the time of our study, Massachusetts planned to provide quarterly reports to ACOs with information on members with high costs, those with high emergency department and inpatient admission use, and those included in quality measures. In addition, Medicaid planned to provide monthly raw claims extracts, which, in combination with their clinical data, can inform ACOs’ population health management strategies in a timelier fashion.
In Minnesota, the state retrospectively attributes beneficiaries to an ACO based on where they received the plurality of their primary care. Because retrospective attribution identifies the beneficiaries upon which quality and cost outcomes will be determined after the end of the performance year, the state provided monthly attribution lists to facilitate care management. Minnesota MCOs typically have broad provider networks, so an ACO may have attributed beneficiaries across multiple MCOs. To help the MCOs understand what proportion of their covered lives were attributed to which ACO, the state provided regular attribution reports to the MCOs. The beneficiary MCO enrollment process was unchanged, with eligible beneficiaries either selecting their MCO during the enrollment period or the state automatically enrolling the beneficiary in an MCO.
Although specifics differ depending on the model, Massachusetts prospectively attributes beneficiaries to ACOs. For both the Partnership and Primary Care models, the ACO-attributed population represents a single payer (their exclusive MCO or the state). MCO-Administered ACOs’ attributed populations, such as Minnesota’s, may represent multiple MCOs. However, unlike Minnesota, the MCO—not the state—is responsible for the attribution.
Massachusetts required each primary care practice (PCP) to participate exclusively with 1 ACO. In the first year, Partnership Model ACO beneficiaries were assigned to their PCP’s ACO, which then enrolled them in the corresponding MCO. This preenrollment minimized disruption of established beneficiary and primary care relationships and encouraged MCO participation in the model. MCOs risked losing members if beneficiaries selected PCPs affiliated with ACOs that did not partner with the MCO (Partnership Model) or participate in its provider network (MCO-Administered Model). In subsequent years, beneficiaries selected both an MCO and a PCP, ensuring beneficiary awareness of the MCOs network restrictions. The Primary Care Model, which represents the FFS population, also attributed beneficiaries based on their selected PCP. The MCO-Administered Model used a prospective claims-based attribution algorithm.
Minnesota. Minnesota’s ACO providers received standard FFS or MCO-contracted payments, with their savings/losses dependent on how expenditures for the ACO-attributed population compared against a preestablished benchmark and on if the ACO met certain quality-of-care targets. Minnesota state staff performed the calculations and informed the MCOs what proportion of the savings/losses the MCO was responsible for. The MCO distributed the savings to (or recouped the losses from) applicable ACOs according to the state’s calculations, serving as a sort of “pass through.” The MCO shared in the proportional savings/losses for its enrolled population, whereas the state shared in the proportional savings/losses for the FFS population. MCOs continued to receive their contracted capitation payment from the state.
Centralizing financial reconciliation at the state level left MCOs perceiving a lack of transparency around the state’s methodology. The savings calculated on an ACO’s total Medicaid population may not reflect a single MCO’s experience. For example, one MCO described the state asking them to administer shared savings despite their beneficiaries exhibiting increased costs. Payers serving rural beneficiaries expressed particular concern that ACOs in larger health systems would overlook the rural MCO beneficiary populations and instead focus their improvement efforts on areas with the highest beneficiary concentrations, where the opportunity for earnings may be greatest.
With MCOs and ACOs having similar care management goals, it was difficult to tell whose efforts—the MCO’s or the ACO’s—contributed to quality and cost improvements. The MCOs felt strongly that the state should provide information to help differentiate and that the current structure financially disincentivized MCO care management programs. The state provided monitoring reports twice a year, which the MCOs considered valuable but insufficient.
Further complicating the relationship between MCOs and the state, many MCOs were implementing their own payment reforms separately from the state. MCOs worried that layering the state’s ACO model on top of an existing performance-based payment model had the effect of “double-paying” providers who met benchmarks in both. One MCO considered requiring providers to choose which model to participate in—the state’s or the MCO’s. Another MCO considered savings it paid for the state’s ACO as a “cost” in their MCO-specific model calculations.
Massachusetts.In Massachusetts Medicaid, most ACOs partner with a single MCO. As with attribution, Massachusetts’ payment approach varies by model. In the Partnership Model, the ACO/MCO partnership received a prospective capitated payment in lieu of the per-member per-month (PMPM) rate the state would have paid the MCO. The difference between the capitation payments and the ACO-attributed population expenditures is the ACO’s savings or losses, which the MCO also received by virtue of the partnership. This arrangement required creating new payment, contracting, and governing structures. In most cases, MCOs continued to pay ACO providers FFS, as providers already supported FFS payments and changes would have been administratively burdensome. Therefore, the Partnership Model MCOs still reconciled FFS costs relative to the capitated rate. The partnership further needed reinsurance or reserve funds to cover potential losses.
In the Primary Care Model, the state continued to reimburse FFS. At the end of the year, the state reconciled the ACO-attributed population expenditures against a preestablished benchmark, with the ACO able to share savings with the state. The MCO continued to receive its PMPM rate from the state, and the MCO paid the ACO providers their established contracted rate. At the end of the year, the MCO, not the state, reconciled the ACO-attributed population expenditures against a preestablished benchmark.
The Figure illustrates each state’s attribution and payment processes.8
Incorporating ACOs into existing MCOs allowed each state to continue paying for most Medicaid service delivery on a full capitation basis, which shifted the financial risk and unpredictability of enrollment and care volume to experienced MCOs. Given that both Minnesota and Massachusetts had been using the MCO model for decades, implementing reforms without maintaining MCOs as the risk-bearing organization would have been impractical.
As Minnesota and Massachusetts illustrate, there are different ways ACOs and MCOs can come together. Massachusetts’ approach significantly changed the MCO-provider-enrollee relationship by blurring the line between MCO and ACO. Conversely, in Minnesota the existing MCO structure remained largely unchanged for the beneficiary, with an ACO model layered on top of the state-insurer-enrollee relationship. Minnesota MCOs, however, had to adapt to a model in which the state—not solely the MCO—played a role in provider payment. Part of what enabled each state’s strategy is likely the MCO-enrollee relationship prior to ACO implementation: Massachusetts had MCOs with limited networks, meaning their beneficiaries were accustomed to associating a given provider with a given plan. Thus, transitioning to an ACO/MCO partnership model, although a significant change, was a much easier transition than it would have been for Minnesota, where most payers have statewide networks.
In both states, Medicaid used a significant lever—procurement of MCO contracts—to ensure participation by MCOs. Massachusetts’ PCPRI experience demonstrated the impact on provider participation if MCOs opted out. Large-scale changes to achieve improved cost and quality are difficult to implement if potential incentives apply to only a small portion of patients. Ultimately, MCO participation fostered high levels of provider participation, which in turn meant that the model reached more of the state’s beneficiaries. Importantly, each of these states had additional support from the federal government as part of the SIM Initiative, which allowed them to directly support ACOs with data analytics and other infrastructure, poising providers to succeed in these new payment arrangements.
Minnesota implemented a typical shared savings model, consistent with an approach that minimally disrupts the preexisting system. In exchange, Minnesota’s approach to ACO implementation caused tension with the long-standing MCOs. MCOs were excluded from the attribution and financial reconciliation processes but required to implement the incentives. In contrast, Massachusetts underwent a more radical transformation. MCOs determined how to administratively structure and implement MCO/ACO partnerships, yielding more operational transparency than Minnesota MCOs reported.
We describe only 2 states, each of which received federal support through the SIM Initiative. Other states may not have access to additional funding and may have different levels of state support. However, these are useful implementation examples.
The key factors in implementing Medicaid ACOs with MCOs were (1) anticipating the ACO design during MCO procurement, (2) incentivizing MCO participation, and (3) helping providers prepare for ACO contracts with data and infrastructure investments. Both states continued their Medicaid ACO models, with recent studies indicating positive outcomes—including in Minnesota.3,9,10 As these examples show, different states will need to approach accountable care differently. As Medicaid ACOs continue to mature and expand, states and policy makers can learn from the experiences of Minnesota and Massachusetts when developing new models or guidance.
The authors wish to acknowledge Jennifer Lloyd and Scott Holladay for their contributions to this work.
Author Affiliations: Abt Associates (CLH), Cambridge, MA; North Carolina State University (NW), Raleigh, NC; RTI International, Baltimore, MD (LMG), and Chicago, IL (SMK).
Source of Funding: This work was funded by CMS under contract number HHSM-500-2010-00021i; task order number HHSM-500-T0007. The statements contained herein are solely those of the authors and do not necessarily reflect the views or policies of CMS.
Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (CLH, LMG, SMK); acquisition of data (CLH, LMG, SMK); analysis and interpretation of data (CLH, NW, LMG, SMK); drafting of the manuscript (CLH, NW); critical revision of the manuscript for important intellectual content (CLH, NW, SMK); provision of study materials or patients (CLH); administrative, technical, or logistic support (CLH, SMK); and supervision (CLH, LMG, SMK).
Send Correspondence to: Catherine Hersey, MPH, Abt Associates, 70 Mount Vernon Ave, Melrose, MA 02176. Email: Catherine_hersey@abtassoc.com.
1. Brooks-LaSure C, Tsai D. A strategic vision for Medicaid and the Children’s Health Insurance Plan (CHIP). Health Affairs. November 16, 2021. Accessed August 26, 2022. https://www.healthaffairs.org/do/10.1377/forefront.20211115.537685/
2. Exhibit 30: percentage of Medicaid enrollees in managed care by state and eligibility group. Medicare and CHIP Payment and Access Commission. December 14, 2021. Accessed March 18, 2022. https://www.macpac.gov/
3. Rutledge RI, Romaire MA, Hersey CL, Parish WJ, Kissam SM, Lloyd JT. Medicaid accountable care organizations in four states: implementation and early impacts. Milbank Q. 2019;97(2):583-619. doi:10.1111/1468-0009.12386
4. State Innovation Models Initiative: model test awards round one. CMS. Accessed March 18, 2022. https://innovation.cms.gov/innovation-models/state-innovations-model-testing
5. MassHealth contracts with two managed care organizations. News release. Massachusetts Executive Office of Health and Human Services. October 3, 2017.
Accessed November 3, 2020. https://www.mass.gov/news/masshealth-contracts-with-two-managed-care-organizations
6. Medicaid enrollees by enrollment group. Kaiser Family Foundation. Accessed November 3, 2020. https://bit.ly/3TlJeLA
7. Massachusetts Executive Office of Health and Human Services. MassHealth and Health Connector request for federal flexibility. Mass.gov. August 2017. Accessed August 26, 2022. https://www.mass.gov/doc/masshealth-1115-waiver-hearing-slides-august-2017/download
8. RTI International. State Innovation Models (SIM) Initiative Evaluation: Model Test Year Four Annual Report. CMS; 2018. Accessed August 26, 2022. https://downloads.cms.gov/files/cmmi/sim-rd1-mt-fourthannrpt.pdf
9. Vickery KD, Shippee ND, Menk J, et al. Integrated, accountable care for Medicaid expansion enrollees: a comparative evaluation of Hennepin Health. Med Care Res Rev. 2020;77(1):46-59. doi:10.1177/1077558718769481
10. Blewett LA, Spencer D, Huckfeldt P. Minnesota Integrated Health Partnership demonstration: implementation of a Medicaid ACO model. J Health Polit Policy Law. 2017;42(6):1127-1142. doi:10.1215/03616878-4193666