Trust, Money, and Power: Life Cycle Dynamics in Alliances Between Management Partners and Accountable Care Organizations

Abstract

Policy Points.

• Accountable care organizations (ACOs) form alliances with management partners to access financial, technical, and managerial support.

• Alliances between ACOs and management partners are subject to destabilizing tension around decision‐making authority, distribution of shared savings, and conflicting goals and values.

• Management partners may serve either as trainers, ultimately breaking off from the ACO, or as central drivers of the ACO.

• Management partner participation in ACOs is currently unregulated, and management partners may receive a significant portion (in some cases, majority) of shared savings.

Context

Accountable care organizations (ACOs) are a prominent payment and delivery model. Though ACOs are often described as groups of health care providers, nearly 4 in 10 ACOs partner with a management company for services such as financial investment, contracting, data analytics, and care management, according to recent research. However, we know little about how and why these partnerships form. This article aims to understand the reasons providers seek partners, the nature of these relationships, and factors critical to the success or failure of these alliances.

Methods

We used qualitative data collected longitudinally from 2012 to 2017 at 2 ACOs to understand relationships between management partners and ACO providers. The data include 115 semistructured interviews and observational data from 7 site visits. Two coders applied 48 codes to the data. We reviewed coded data for emergent themes in the context of alliance life cycle theory.

Findings

Qualitative data revealed that management partners brought specific skills and services and also gave providers confidence in pursuing an ACO. Over time, tension between providers and management partners arose around decision‐making authority, distribution of shared savings, and conflicting goals and values. We observed 2 outcomes of partnerships: cemented partnerships and dissolution. Key factors distinguishing alliance outcome in these 2 cases include degree of trust between organizations in the alliance; approach to conflict resolution; distribution of power in the alliance; skills and confidence acquired by the ACO over the life of the alliance; continuity of management partner delivery on promised resources; and proportion of savings going to the management partner.

Conclusions

The diverging paths for ACOs with management partners suggest 2 different roles that management partners may play in ACO development. In some cases, management partners may serve as trainers, with the partnership dissolving once the ACO gains skills and confidence to work alone. In other cases, the management partner is a central driver of the ACO and unlikely to break off.

Literature on accountable care organizations (acos) typically defines and discusses them as provider organizations,1 implicitly assuming that ACOs are provider‐driven. However, recent research suggests this notion may be incorrect: more than one‐third of ACOs have a management partner.2 Some well‐known management partners include Collaborative Health Systems (Universal American), Evolent Health, Aledade, and Caravan Health. Survey data suggest these partners often play a central role in ACO development. For example, in two‐thirds of ACOs with partners, the management partner shares financial risk or reward; in half, the management partner provides a combination of potentially critical ACO services, including care coordination, data analytics, education, and administrative and contracting services.2

Despite this evidence on the prevalence of partners, we know little about why providers work with management partners, the nature and dynamics of these partnerships, and the extent to which management partners promote ACO performance. Without understanding more in depth the role management partners play, research on ACO implementation and performance may be laden with errors in conclusions—for example, overlooking major factors predicting performance, misattributing effects of particular capabilities, or misunderstanding challenges providers face in pursuing ACO contracts. In addition, without a fuller understanding of ACO partnerships, policymakers and stakeholders will be unable to develop meaningful policy, programs, or supports to better aid providers who wish to pursue ACO contracts. For example, the need for up‐front financing and the need for expertise in data analytics present 2 quite different challenges.

In this paper, we provide the first indepth examination of relationships between ACOs and management partners. Our contribution is to identify ACOs’ motivations to partner with management companies and to characterize the nature and dynamics of these partnerships for 2 ACOs, while also examining the costs of such partnerships. We use existing literature and theory on strategic alliances in health care, paired with in‐depth qualitative data from 2 ACOs with management partners, to understand how and why these partnerships come about; tensions that arise; and the long‐term stability of these partnerships.

Theoretical Framework

Salk and Vora define strategic alliances as “any goal‐oriented contractual cooperation between two or more partner organizations, involving some asset‐specific investment by one or more of the partners.”3 Drawing on this definition, we conceive of ACOs as strategic alliances when 2 or more independent organizations, including health care providers and management partners, voluntarily pool resources to achieve mutual goals. In our view, alliances have a life cycle, including partnership formation, growth accompanied by a rise of tension, and maturation or dissolution of these partnerships.^3,4 In this analysis, we focus particularly on (1) why alliances form; (2) sources of tension; and (3) how the management of tension and conflict affects the stability and success of alliances. The alliance literature provides a useful framework to situate our study of management partnerships with health care providers in ACOs for 3 compelling reasons.

First, though the Centers for Medicare and Medicaid Services (CMS) intended ACOs to have tight organization and governance, empirical work on ACOs shows this typically is not the case (eg, D'Aunno et al.4 and Lewis et al.5). We argue that a major reason many ACOs are loosely organized and governed is that they are “start‐ups” with no widely accepted blueprint with fine details for how to manage or govern themselves. Though the applications that organizations file to participate in CMS ACO programs may look solid on paper, arrangements on the ground often are fluid and not well defined. Indeed, in our view, an important contribution of our paper is to show how the dynamics of ACO formation and development unfold over time.

Second, an alliance conceptual framework is useful for our study because it emphasizes the life cycle of interorganizational arrangements. Neither individual organizations nor alliances are static in their form or operations. Rather, organizations and alliances typically change over time, often in unplanned ways. Alliance conceptual models aim to analyze these changes.

Last, alliance models can accommodate the use of other conceptual lenses from organizational theory to analyze relationships among ACO partners (see the special issue on ACOs published in Medical Care Research and Review, 2016). D'Aunno and Zuckerman,6 for example, used multiple conceptual models (transaction costs, resource dependence, institutional theory) to analyze the development of hospital alliances. On the one hand, our paper follows in this vein by using an alliance conceptual framework that has the benefits noted earlier, especially its focus on life cycle dynamics, while, on the other hand, we draw on other important concepts (trust, power) from the organizational theory literature.

Emergence of Alliances

Health care providers may be prompted to seek partners for a variety of reasons; these reasons in turn can influence which partners the providers choose. Research suggests that threats from their external environments (eg, a turbulent health care market), opportunities (eg, new Medicare programs), and uncertainty motivate providers to seek partners.5, 7, 8 The current health care landscape may provide both uncertainty about health care payment and new opportunities in the form of initiatives from payers (such as ACO programs). Additionally, in some regions, independent providers may experience threats such as a heavily consolidated local market that prompt them to seek out partners to form an alliance.9 Research on ACO alliances suggests that many new partnerships between providers have formed in response to Medicare and Medicaid ACO programs.5

Partnership selection is often key to a successful alliance. Complementarity, commitment, and trust are 3 major factors cited in how partners are selected.10 Alliances are often most successful when partners bring unique and complementary resources and expertise, rather than similar (or competing) expertise.10 Among ACO alliances between provider organizations, providers sought partners largely to decrease real or perceived risk associated with pursuing an ACO and to provide expertise or resources.5, 11, 12 Trust is also a key element of partnership selection and long‐term success of alliances.13, 14 Among ACO provider partnerships, partners often had a history of working together that provided a foundation of trust.5, 11, 12 Additional factors that influence trust include degree of fit between organizational cultures and for‐profit versus not‐for‐profit status.

In ACOs, providers may seek partners to fill perceived gaps in their skills, expertise, or capacity rather than attempt to develop these skills or capacities internally. In some cases, expert partners serve as trainers for organizational learning, enabling transfer of expertise from the management partner to the ACO.15, 16, 17 Research on partnership ACOs suggests that some providers in ACOs look to hospital partners to provide expertise, data, and capital that physician practices lack.5, 11 However, relationships with hospitals bring challenges as well, such as the difficulty of aligning the hospital business model with ACO payment, which often encourages reduced use of hospitals.11, 18 These challenging relationships may push some providers to look for these capabilities elsewhere when pursuing accountable care.

Partnership Instability and Dissolution

While a large body of literature has examined factors predicting alliance formation and ultimate success, a smaller body has focused on alliance stability, tension, and conflict (but see Hearld et al. on fragility of community health alliances19). This translates to a great deal of work on the formation stage, but less research has focused on the growth and associated tension phase and the resulting maturation or dissolution of alliances, particularly using in‐depth data.3 Some work shows that alliance instability is often a result of internal tension and conflict among partners. As partners experience lack of trust and conflict, alliances become unstable and at risk of dissolving, particularly if there is an external shock.20 Other factors internal to an alliance that increase the risk of dissolution include power imbalances among members, status loss of a member, and failure to achieve alliance goals.20, 21, 22, 23, 24

To our knowledge, no research has yet studied the dissolution of ACO alliances. Work on ACOs shows that some provider partnerships under ACOs experience conflict and distrust,5 but this research is thin and has not studied dissolution. The larger literature on alliance dissolution predicts several conditions that may prompt ACO alliances between providers and management organizations to dissolve. ACO alliances may fail due to significant inequality between providers and their management partners in making key decisions about ACO strategy or operations. Additionally, failure of the ACO to achieve goals such as attaining financial bonuses may precipitate alliance instability. Some have predicted that ACOs may face similar challenges as the integrated delivery networks of the 1990s,25 and inability to overcome these challenges may lead to the dissolution of ACOs or alliances between ACOs and management partners, akin to the unraveling of integrated delivery networks in the prior era.25, 26

In sum, this paper fills a gap in current literature by examining in depth how and why providers partner with management organizations in ACO contracts; the nature of these partnerships, such as the level of power or control the management partners exert compared to providers in the ACO; tension or conflicts that arise; and the long‐term stability of these partnerships.

Methods

For the analysis in this paper, we use qualitative data from 2 Medicaid ACOs; the composition of each ACO is described in Table 1. These 2 ACOs are drawn from a larger, comprehensive array of qualitative data we collected that includes 66 ACOs. At these 66 ACOs, we spoke with 475 individuals between 2012 and 2017 through a combination of phone interviews (n = 122) and site visits (n = 15) (during which we collected both interview [n = 170] and observational data). For this analysis, we chose to compare these 2 ACOs for 4 reasons. First, each ACO began as a partnership between local providers and a management partner. Second, at each ACO we have robust data collected over multiple years, including interviews with members of the ACO and the management partner as well as observational data, representing deep data on alliances between management partners and ACOs. While other ACOs in our larger data also have strategic alliances with management partners, our data on the alliances themselves are not comparably robust, with some sites limited to semistructured phone interviews. Third, as Medicaid ACOs, the sites serve similar patient populations, making them an informative comparison by holding patient population and payer constant, while differing in key ways useful for alliance comparison, as highlighted in the results. Finally, based on our knowledge of the larger set of ACOs with management partners in our broader qualitative data, our assessment is that these ACOs are not outliers. For example, the needs that drive partnership formation as well as the key issues that impact trust and alliance stability resonate with other such partnerships in our larger data set.

Table 1.

Features of Management Partner Alliances

	Collaborative ACO	Access ACO
ACO composition
Approximate number of attributed patients	28,500	30,000
Includes a hospital	No	Yes
Includes a physician group	Yes	Yes
Includes behavioral health providers	Yes	Yes
Includes safety net providers (eg, public hospital, FQHC)	Yes	Yes
ACO board membership
Includes management partner	No	Yes
Includes safety net providers	Yes	Yes
Includes physician group	Yes	Yes
Includes hospital systems	No	Yes
Includes community organizations	No	Yes
ACO contract	Medicaid	Medicaid
Approximate shared savings earned
Year 1	$2 m	$1.5 m
Year 2	$3 m	$5 m
Year 3	$3.5 m	$5 m
Year 4		$5.5 m
Management partner
For profit	Yes	No
Alliance initiated by	Providers	Management partner
Role on board	No vote	2 votes, with veto power
Partnership status at last data collection	Dissolved	Cemented
Services provided by management partner
Financial investment	Yes	Yes
Administrative services	Yes	Yes
Data analytics	Yes	Yes
Care coordination and/or quality improvement	Yes	Yes

Abbreviations: ACO, accountable care organization; FQHC, federally qualified health center.

From these 2 ACOs, we collected data from 2012 to 2017 consisting of 11 semistructured phone interviews as well as observational data and 104 semistructured interviews from 7 site visits. (See Table 2 for a breakdown of data sources.) Site visits lasted 3 to 5 days, and 2 to 4 members of the research team went on‐site for data collection. The research team created tailored interview guides for each site visit interview. Typical topics included daily work; internal and external communication and collaboration; responses to the changing health care market; work on cost reduction and quality improvement; data collection, analysis, and sharing; and provider engagement. Sample interview guides are available upon request. We spoke with a range of individuals drawn from different categories, including ACO leadership; management partner personnel; ACO board members; ACO staff; leadership at participating organizations; providers; and patient support personnel. All interviews (phone or in‐person) were recorded and transcribed.

Table 2.

Data Included in Analysis

	Collaborative ACO	Access ACO
Number of individuals spoken with	112	87
Number of site visits	5	2
Years of data collection	2012; 2013; 2014; 2016; 2017	2015; 2016; 2017
Number of interviews	59 in‐person; 8 by phone	45 in‐person; 3 by phone
Meeting observations	19	4
Types of meetings observed	Board; finance committee; quality committee; ACO–management partner meetings	Board meeting; ACO–management partner meetings; clinical huddle
Types of documents collected	Committee meeting minutes; job descriptions and advertisements; strategy documents; performance data; quality plans and protocols; grant reports	Job descriptions and advertisements; strategy documents; performance data; quality plans and protocols; project reports
Quantitative data accessed	Cost and quality performance	Cost and quality performance

Abbreviation: ACO, accountable care organization.

The research team has a strong relationship with each site, with both ACOs granting us a notable degree of access. The research team regularly observed a variety of meetings, including internal clinical meetings; meetings between ACO leadership and providers; meetings of ACO leadership, providers, and management partner staff; quality and finance committee meetings; and board meetings. Additionally, our observational data include natural interactions between each ACO and its management partners, ranging from uncontentious (eg, care management workflows) to contentious (eg, negotiation of shared savings distributions) interactions. Observers on the team are trained to diagram the space and participants, take near verbatim notes, and catalog nonverbal cues.

Team members read transcripts multiple times to identify emergent themes. A preliminary codebook was developed, first deductively drawing on relevant literature and then expanded inductively based on the language and themes that emerged in the data.27 Codes included conceptual, relational, characteristic, and setting codes. Two members of the study team double coded 10 interview transcripts to assess the adequacy of the codebook and establish agreement on meanings and appropriate application of codes. Agreement on application of codes was strong, with the majority of disagreements concerning extent of the surrounding text coded, not whether or not material met the coding definition. Following this, minor changes to the codebook were made, including additions, deletions, and refinement of definitions. The 2 coders then split the remaining transcripts and individually coded, bringing uncertainties up for group discussion to maintain agreement on interpretations. The final codebook used in this analysis consisted of 48 codes and is available upon request.

We reviewed coded data and analyzed for emergent themes. Particularly salient data were reviewed as a group during analytic meetings and discussed in relation to our evolving understanding of the management partnerships in our data.

Throughout, we use pseudonyms and avoid identifiable details to protect the sites’ anonymity.

Results

Our comparative data from 2 cases, Access ACO and Collaborative ACO, illustrate key themes that may shape management partnerships across the life cycle of the alliance. Table 1 shows characteristics of both ACOs. Both hold Medicaid contracts; have a similar number of attributed patients; include a physician group, behavioral health providers, and safety net providers; and have earned shared savings in each year of their ACO contracts. They also differ in key ways: Access ACO includes a hospital and has a more diverse board than does Collaborative ACO, including management partner representation on the board.

Throughout the alliance life cycle, we identified themes particularly salient to management partnerships that cut across phases. These key themes include money (where it is coming from and how it is being spent); power dynamics between the ACO and its management partner; and expertise of the management partner. We find that how these elements are handled in the context of organizational culture fundamentally shapes trust between the ACO and the management partner in these 2 cases, and it critically shapes stability and outcome of the alliance. In the following sections, we explore the life cycle of these alliances for each of the 2 ACOs, beginning with partnership formation, to growth and the emergence of tension, to maturation or dissolution of partnerships. These 2 cases provide fine‐grained data to compare how key issues around money, expertise, and power influence trust and ultimately the outcome of the alliance. We begin by examining the case of Access ACO, followed by the case of Collaborative ACO.

Access ACO

Partnership Formation

Upon formation, Access sought financial, data analytic, risk‐based contracting, and administrative support from a partner in order to pursue an ACO contract. Access ACO chose to form an alliance with a management partner that worked with several ACOs in its region and that had courted the ACO as part of a self‐described “speed dating” process. As described in a report, the ACO “entered into a management services agreement with [the management partner] to provide staffing support for finance, information services, customer service, governance, and other [ACO] administrative functions.” In addition, the management partner agreed to provide “strategic guidance and key operations infrastructure for [Access ACO]” as well as an executive to manage the ACO and key operational staff. The management partner had historically served as a health plan and was focused on redefining its work in the context of health care reform. As a health plan, the management partner had a long‐standing prior relationship with some of the participating primary care practices in Access, with leadership crossing over from the boards of participating practices to Access's board and to the management partner's board, building on and cementing long‐standing collaborative relationships and trust. Moreover, both Access and its management partner were nonprofit organizations with experience working with Medicaid populations; both ACO and management partner staff talked openly about their shared organizational culture, described as “mission, vision, values” in their terms, focused on serving vulnerable populations, fostering community ties, and building robust relationships.

These strong personal and governance ties were an important factor in early partnership building. In a joint meeting between Access and the management partner, the management partner said that, from the start, they were interested in:

really trying to blend the best of all worlds, trying to have a local board, a local community‐led effort, which is really in the spirit of what [ACOs] were trying to do, but also have [management partner] as the engine for it, so that from a financial backing standpoint, from a services standpoint and hopefully bringing our expertise and systems and such to really blend those two things together to create [an ACO].

At the same time, the management partner provided significant, no‐strings, up‐front funding for the ACO to administer in pursuit of health care transformation. The management partner fostered independent, local, ACO‐level decision making about the use of investment funds. In an interview with the management partner leadership, they provided history on the partnership:

We made it very clear that those were community dollars for the community and that board to decide how to spend, not that if they [the ACO] thought of a good idea then [the management partner] would say, “No, we're not going to do that one.” That really I think built the trust.

Moreover, the ACO and management partner agreed on a governance structure that was designed to foster democratic engagement, with each organizational representative having, in the words of the ACO executive, “the same vote and the same voice” to level the playing field among hospitals, community health centers, and individual members. Additionally, from its inception the ACO board deliberately opened meetings with declarations of conflicts of interest and discussions of these conflicts in a move to, in their words, “surface” latent conflicts to openly address and resolve. We observed this same language of surfacing and consistent effort to openly discuss and resolve conflict outside of board meetings as well, including working meetings within the ACO as well as meetings between the ACO and the management partner.

While other Access board members have equal representation on the board, the board includes 2 representatives of the management partner. Despite this structural inequality, at this early stage, their partner was not interested in dominating the board. Instead, in an explicit move to foster trust and the board's development, the management partner “really moved to the backseat intentionally and took a large step back and provided information and shared data such as the financials … but didn't come with very many recommendations.” A power‐sharing approach and consistent investment in the community with support for local ACO‐level decision making fostered trust at the leadership level between the ACO and the management partner, as did the shared emphasis on community engagement. At the same time, the management partner provided critical support to the ACO in terms of finance, staffing, network development, data analytics, and reporting.

Partnership Growth: Rising Tension

Our longitudinal data show an evolution in the relationship between Access and its management partner. The management partner initially gained Access's trust through financial transparency, consistent follow‐through on plans and promises, and no‐strings investments in the local community described earlier. Two and a half years into the partnership, leadership at Access's management partner believed that “increasingly, though, we are seeing ourselves as all a team, and there is not a[n] [Access] team and a [management partner] team, but we all work together to sort of define what the projects are, how we implement them, where we prioritize.” For Access, no significant conflicts arose around the management partner's capacity to deliver on technical expertise or agreed‐on resources, which they consistently provided per their agreement. In part, ACO‐level satisfaction with technical expertise was supported by the ACO executive's choice to take a complementary skills approach to the alliance. She recalls making a strategic choice to encourage her ACO staff to view the management partner as expert in areas of risk‐based contracting, claims, and utilization management. Although certain participating providers expressed frustrations with perceived technical gaps that arguably derived from the management partner, these gaps in technical expertise were less visible to Access ACO staff and hence not a source of conflict.

Despite the management partner's optimism, additional elements of friction developed as well. Tension arose around issues of central versus local identity and decision‐making power. Not long after hearing the management partner's perspective on unity as a team, we observed friction at a staff meeting at Access. Access leadership pushed ACO staff to do strategic planning for the upcoming calendar year. Staff requested more time because “it depends on what [Access] wants to prioritize and what [management partner] wants to prioritize.” When ACO leadership instructed staff to start with local Access planning, ACO staff who worked closely with the management partner interjected that the partner was making decisions and setting priorities without including or informing Access. When we asked a clinical leader at Access about this tension, she cast the partnership in slightly more measured terms, as becoming “more of a we than it was” but still “I don't think it's as strong as [management partner] wants it to be.”

We also observed this tension between the power of the ACO versus the management partner on the Access board, particularly around the question of whether the board or the management partner held final authority in decision making. One provider board member acknowledged that “we're trying to decide if we're an advisory board or a governance board.” Two years into the partnership, the Access board had a retreat expressly focused on defining Access's and the management partner's roles, particularly with respect to decision making. While the retreat resulted in greater clarity about mechanisms for decision making, months after the retreat Access's CEO noted that the question remains “exactly what is it over which [the board is] governing and where are those lines drawn?”

The retreat clarified that, while the management partner has only 2 voting representatives on the board, those representatives hold veto power over Access board decisions. In addition, although the Access board approves financial decisions, “99% of the dollar is delegated out to entities [by the management partner] and we [the ACO board] don't approve financials on that portion.” Despite efforts at clarification, board members remained uncertain after the retreat regarding the nature of the board's authority in relation to the management partner. It remained unclear to what extent board members’ uncertainty was a product of their disinterest in recognizing the extent of the power imbalance between the local ACO board and its management partner.

Outcome: Mature Alliance

Although Access ACO experienced potentially destabilizing tensions around local expertise and decision making, both organizations used open communication to identify and address tensions, and the ACO explicitly chose to remain in an alliance with the management partner. In an interview we conducted jointly with Access and its management partner, the leaders at the management partner explained the complicated dynamics around continuation or dissolution of the partnership:

We've always kind of put out there would [Access] want to be its own [separate organization]? Two of the [ACOs] we're involved with are separate organizations.… I don't know where it'll end up but I do think that for now it's really in a balance where we're really trying to find that sweet spot, or the best of both worlds.

In this case, separation would be achieved by the ACO buying out its management partner in the joint venture. External reports verify this statement of openness to separation. An early report from 2013 describing the “partnership” between the 2 organizations establishes that the management partner “is explicitly committed to having the [ACO] be locally accountable, and will facilitate the transition to an independent, community‐based organization, when appropriate.” Statements from Access's CEO mesh with these descriptions of the partnership:

That was a very live question: “Do we go independent? Do we separate out into our own [organization]?” and [we] are very comfortable now saying “no,” that this is the best model for us and that we work best when we are locally developed strategies in alignment with [management partner]’s strategies as well. And that the depth of thinking and ability to implement some of the transformation is magnified tremendously by being closely connected to them.

When asked in a follow‐up interview a year later if they had revisited the discussion about separating from their management partner, Access remained strong in its commitment to the partnership, stating, “No, we're done with that. Done.”

Access ACO Analysis

In our estimation, several key factors fostered and protected the ACO's trust in its management partner and the long‐term vitality of the strategic alliance: the financial investment of the management partner; shared organizational culture, including the commitment to openly addressing and resolving tension; and the cultivation of complementary skills. On the financial side, the management partner's continued financial investment in the ACO was coupled with the partner taking a moderate portion (<30%) of savings earned, promoting trust from the ACO that the management partner was committed to Access for more than just profit.

As nonprofit organizations currently and historically pursuing Medicaid contracts, Access and its management partner's shared organizational culture included a focus on mission, rather than profit, and, critically, a shared commitment to openly addressing and resolving sources of tension in the alliance. Both organizations identified and worked to resolve tension through formal meeting structures, such as statements of conflicts of interest at the start of board meetings and adapting a meeting agenda to add time for discussion aimed at resolving conflicts when they were raised. Moreover, the language used for such discussions was marked by more measured tones and word choice, with elements like raised voices, assertions of blame, and other more aggressive communication approaches absent. Finally, the ACO and the management partner cultivated a complementary skills approach to the alliance, in which Access's CEO encouraged her staff to conceive of the management partner as holding specialized technical expertise in contracting, claims, and utilization management that dovetailed with the ACO's skills. This is in contrast to an approach where the ACO could have encouraged the development of comparable internal expertise to that held by the management partner.

These issues are critical throughout the alliance's life cycle and were clear in all stages of our data collection; ultimately, they work together to produce an alliance that has remained stable thus far. We turn now to comparative data from Collaborative ACO, which illustrate sources of instability in the ACO's alliance with its management partner.

Collaborative ACO

Partnership Formation

In contrast to Access ACO, Collaborative ACO and its national, for‐profit management partner had no prior history of working together before the establishment of the partnership. Initially, the providers in Collaborative ACO determined they needed an external partner to provide financing, data analytic support, project management, and contracting support. At this stage, in addition to lacking adequate financing, the ACO felt it lacked the necessary skills in the aforementioned areas to pursue an ACO contract on its own. Collaborative ACO initiated a competitive bidding process that involved multiple potential management partners. The ACO ultimately chose a national health services firm with expertise in data analytics, care management, and other administrative services. Their partner was “instrumental in helping us, guiding us,” according to ACO leadership, and provided crucial up‐front financial support to cover “direct costs associated with all the tools and systems and personnel” necessary to succeed as an ACO. In the words of one Collaborative provider, “I think anyone else working at [a safety net] level is going to need a partner like this who is willing to step up and put some skin in the game and take some risk with us.” The management partner was willing to take financial risk, investing at a level that the ACO could not afford on its own. Collaborative leaders were clear from the start that their management partner was critical: “They have enabled us to do something we otherwise could not have done.”

From the start, however, there was an awareness of differing agendas and the need to protect each organization's self‐interest. As a group of safety net providers, Collaborative ACO and its management partner approached the partnership with different goals and values. Collaborative was aware that the management partner had an agenda to use the ACO as a laboratory to gain knowledge that “benefits them and their business models into the future.… I think we understand that and are fine with that, because we feel it's a mutually beneficial arrangement.” Through the development of the partnership, “you have gotten the Kumbaya group with the business group, and we're meeting somewhere in the middle.” Despite this stated acceptance, Collaborative leaders shared that the “board also wanted to have some staffing and legal services to be sure that the interests of the [providers] was balanced against the interests of [the management partner].”

Partnership Growth: Rising Tension

Despite initial enthusiasm, as the partnership progressed, tensions arose around expertise, power, and money. Collaborative leaders expressed doubt about their partner's capacity to fulfill their commitments on health information technology and data analytics. One provider felt that the management partner (as well as others who had participated in the competitive bidding) had exaggerated their expertise, saying, “‘Oh yeah, the interfaces won't be any problem, the data mining will be easy.’ That hasn't been. So I think they really oversold themselves and I think any of the other organizations it would have been the same story.”

Both Collaborative ACO respondents and management partner respondents identified leadership as an area of contention, with management partner personnel challenging providers to lead the ACO's work: “You guys now own this; now what do you need from us to keep this moving?” Some Collaborative providers “were getting a little frustrated” to hear their management partner push them to take the lead when their partner had yet to follow through on staffing promises such as personnel to focus on quality and process improvements. Although Collaborative leaders did rise to their partner's challenge, taking greater ownership of the work, tension remained. However, Collaborative leaders thought these kinds of tensions were part of a “normal process” in partnerships as organizations attempt “a balancing of the power.” Moreover, the push‐pull around leadership resulted in a highly engaged quality committee in which ACO and clinic personnel worked alongside management partner personnel, gradually acquiring some of the technical skills the providers had originally sought from their management partner along with a degree of confidence in their technical capacity that was originally lacking.

While tensions around leadership were understood to be a normal part of partnership development, when financial pressures intensified, underlying tensions escalated. Partway through the ACO contract it appeared that the ACO would not achieve savings. From one Collaborative provider's perspective, “suddenly, in a one‐week period, we went from being in the money to not in the money and the pressure at [the management partner] built so quickly around that point.” It seemed as if their partner “retrenched back into their health plan mentality.” Leaders on the management partner side were frustrated that “we haven't made any money at this.” Although they had previously stated that they were banking on gains in learning and experience more than financial gains, when push came to shove, finances mattered: we “can't continue to do this. We're not a nonprofit.” It was at this point, when Collaborative's management partner began to decrease financial investment in the ACO, reducing staffing levels that were originally planned under the partnership agreement, that the possibility of dissolution was first raised.

Fundamentally, as one of Collaborative's leaders regularly said, “We're trying to work together here, but if it doesn't work out … then you've [management partner] got to do what you've got to do, and we've got to do what we've got to do.” In the face of rising tensions and financial pressures, Collaborative consistently maintained the distinction between the ACO and the management partner with an eye to the possibility of separation. These factors led to a marked degree of instability in Collaborative's alliance with its management partner.

Outcome: Conflict‐Driven Dissolution of Alliance

Unlike Access ACO, Collaborative ACO's alliance ultimately dissolved. Also unlike Access ACO, Collaborative provider leaders had used the alliance as a learning opportunity, gaining technical skills and confidence in their capacity to lead cost and quality work internally. Because of this, they set a course to dissolve this alliance and remain independent, rather than seek to replace their management partner.

The management partner's reduced financial investment in the ACO, including lower staffing levels than initially planned, may have initiated the possibility of separation; however, dissolution was ultimately carried out by a push from Collaborative ACO providers, prompted by dissatisfaction with not only the management partner's withdrawal of resources but also the high price for its services. When Collaborative ACO did eventually earn savings under its ACO contract (see Table 1 for approximate amount per year), a significant portion went to the management partner. The combination of the management partner's withdrawal of resources and receipt of a significant portion of shared savings once they were earned proved fatal to the alliance.

At this point, the ACO determined to work toward total dissolution of the alliance, with a stepwise process to shift the management partner to a vendor relationship, while the ACO further built up internal capacity with the ultimate goal of severing the relationship entirely. Delays in developing internal data analytics capacity meant that the ACO maintained a vendor relationship with its management partner longer than planned, but the ACO remained committed to cutting that tie as soon as it was able. As one Collaborative leader put it: “I say we're in a separation agreement; we're divorcing.”

Collaborative ACO Analysis

In our analysis, several key factors undermined trust and destabilized the alliance between Collaborative ACO and its management partner, including conflicting goals and organizational cultures; differences over relative power and decision making; and failure of the management partner to maintain investments while charging high fees, in the form of taking a significant portion of shared savings. From the start, there was a significant divide between Collaborative's goals and organizational culture as a mission‐driven, nonprofit safety net provider and those of its management partner as a for‐profit, national firm. Moreover, the 2 organizations had neither prior working relationships nor personal history to draw on to help bridge their differences. These challenges could have been overcome; however, rising tension over decision‐making power and limited tactics to openly address and resolve these tensions eroded trust and alliance stability.

Finally, failure of the management partner to maintain financial investments in the alliance (eg, promised staffing levels) while receiving a significant portion of savings constituted a critical challenge to trust and commitment to the alliance. These factors created a relationship characterized by initial wariness that ultimately resulted in distrust and a conflict‐driven dissolution of the alliance. While it may be tempting to see the dissolution of Collaborative's alliance as inevitable given the mission and cultural mismatch and wariness that characterized the formation stage of the alliance, in our analysis these early challenges were surmountable. In contrast, the removal of promised staff whilst taking a high share of shared savings proved more difficult to overcome. Critically, though, the ACO providers had used the strategic alliance as an opportunity to learn from the management partner during the course of the alliance, thereby gaining both technical capacity and the confidence derived therefrom to not only dissolve their alliance but choose not to seek another management partner.

Conclusion

These 2 cases illustrate key facets of strategic alliances between 2 Medicaid ACOs and their management partners. The most important commonality we find in their partnership formation phases is the drive to partner because providers felt they lacked necessary funding, infrastructure, and expertise to successfully pursue an ACO on their own. Critical issues in the formation stage included degree of investment; whether or not the management partner was represented on the board; and approach to relationship building. For both alliances, during the partnership growth phase, tensions arose around issues of money, expertise, and decision‐making power. For Access ACO, these tensions were resolved through active conflict resolution, and the alliance remained intact throughout our research period. In contrast, for Collaborative ACO growing distrust paired with the management partner's decreased investment and high fees prompted dissolution; because the ACO had gained expertise through the alliance, it did not seek an alternate management partner.

Our in‐depth examination of 2 cases in which ACOs worked with management partners suggests that these partners may play a central role in ACO formation, allowing providers to participate in ACO programs when they otherwise likely would not have joined. Money, power, and expertise were key issues for the 2 partnerships examined here, both as the fundamental matters spurring partnership formation and the key flash points for conflict and dissolution. Together, how these issues were handled by partners influenced the level of trust between partners and ultimately the fate of these alliances. Where tensions were ignored and left to fester, trust eroded, thereby fundamentally weakening the partnership. In contrast, in the alliance in which tensions and their sources were explicitly addressed, there was an opportunity to strengthen trust, with implications for the longevity of the alliance.

Our qualitative data on these 2 alliances identify reasons why health care providers may seek management partners, finding that management partners often provided both tangible resources and services (such as funding, data analytics, and care management services) while also inspiring a more generalized confidence among providers about pursuing an ACO. As the 2 ACOs moved into performance periods of contracts, tensions between ACOs and their management partners surfaced. These tensions arose around issues of who received and controlled money, often at a point where the ACO was achieving savings and health care providers in the ACO became uneasy with the financial cost of partnering with a management organization. These tensions are related to questions over the power that the ACO and the management partner hold and whether or not the ACO perceives the management partner as delivering the level of expertise and services promised at the beginning of the partnership. It is worth noting that most management partners may by unable to fully deliver on promised technical expertise insofar as health information technology, predictive analytics, and care coordination remain challenging areas still under development. In our data, however, inadequate delivery of expertise alone prompted frustration but not outright conflict. When this occurred during the formation phase for Collaborative ACO, participants were able to move past this frustration with an understanding that these are challenging areas for any firm to deliver on. In contrast, reneging on promised staff constituted a more thorough disruption to the stability of the alliance.

Several years into ACOs, data here and elsewhere suggest diverging paths for ACOs with management partners.28 Some alliances between ACOs and management partners mature and persist, while others dissolve. In our in‐depth analysis of 2 alliances, we identified 2 key factors that influenced the decision to continue or dissolve partnerships. First, the level of internal capacity an ACO develops is important: ACOs must feel they can take over the services previously delivered by their management partners in order to dissolve the partnership. ACOs that have developed enough internal capacity are likely candidates for dissolving partnerships. Second, whether or not ACOs and their management partners are able to overcome inevitable tension that arises can influence whether a partnership dissolves, as well as the terms on which the partnership dissolves.

The diverging paths for these 2 ACOs with management partners suggest 2 different roles that management partners may play in ACO development. In some cases, management partners may serve as training partners, assisting providers with little experience with new payment models and population health to develop capacity and confidence internally while working alongside a partner, until the ACO feels confident to run the ACO alone. This mirrors existing literature on alliances as conduits for organizational learning between participants.15, 16, 17 At this point, the partnership dissolves and the ACO may retain a vendor relationship with the partner in a reduced capacity. In other cases, the management partner is a central driver of the ACO and unlikely to break off.

Our work has some important policy implications. First, in line with prior research, 2, 5, 11 this investigation suggests that providers who feel a need for financial resources and technical expertise may turn to partners to fill this gap. While some providers turn to hospitals for these resources, others turn to management partners. These findings suggest that ACOs lacking capital or technical expertise (eg, physician‐group ACOs or safety net ACOs) require support that is currently lacking from payers. Programs that can provide more advance financing, such as expanding programs like Medicare's Advance Payment program, could alleviate this need. In addition, payers providing more data analytics may also provide an alternative to private sector management partners. In contrast, another way stakeholders or researchers may support providers is by developing more transparency in the types of management partners and models of partnership, such as ways to structure governance across ACOs and management partners, information on models of fee structures, and characteristics or capabilities of partners that may be meaningful. With this type of information, providers seeking to start ACOs may learn from the experience of existing ACOs to best assess whether potential partners will be a good fit with a given ACO's needs and goals. Finally, payers or policymakers may consider whether they want to regulate some aspects of management partnerships with ACOs, such as the amount of savings derived from tax dollars (Medicaid and Medicare) that can flow to management companies, akin to some states requiring Medicaid health plans to be nonprofits, or the federal government limiting profits by commercial plans sold on insurance exchanges.

Our analysis has several limitations. First, we draw on data from just 2 ACOs; our results are not representative or generalizable. Studies using broader, more representative data would be needed to determine, for example, the overall extent to which these partnerships are maturing and dissolving as well as the generalizability of the mechanisms we identified for these 2 alliances. However, we believe our results are still instructive to the field. The use of narrow, but deep, data allows us to understand in a much richer and more nuanced way how partnerships emerge and function. Second, our analysis focused exclusively on alliances with management partners, precluding work comparing why providers choose to ally with management partners compared to other types of organizations, such as hospitals. Third, our analysis focuses on 2 ACOs that are intrinsically strong organizations on their own. As such, their experiences are unlikely to reflect those of weaker ACOs that may be more likely to be on the losing end of alliances with management partners. Finally, our study focused exclusively on Medicaid ACOs, which means we may miss dynamics specific to other payers, such as Medicare ACOs. Other research, however, suggests there may be much commonality across cases.2, 4

There are several avenues where additional research would significantly further our understanding of this field. First, it would be productive to explore dynamics around trust, money, power, and expertise in a larger, comparably rich, qualitative sample. Second, this paper focused on alliances with management partners; the ACOs described, however, also involve new partnerships among health care provider organizations. The literature on alliances and partnerships has not focused in depth on highly complex alliances, such as multiple types of new partners. Given that our data suggest such complex alliances may be common as well as significant to understanding accountable care, future work exploring this topic would be important for understanding both ACOs and broader trends in health care. Third, the literature on alliances would benefit from further study of conflict in alliances, or how such conflicts are resolved.