ACOs leaders say downside risk is hurting Shared Savings Program

Accountable care represents an attempt to transform how health care is delivered and reimbursed. Its core premise is to assign responsibility for a population of patients to health care providers, with payments depending on the cost and quality outcomes for that population. These providers form accountable care organizations (ACOs) to collaborate with the intent to deliver high-quality, cost-effective care across the continuum of services for their covered populations (Tu 2015). For policymakers, the ACO model (being one permutation of what the authors refer to as “accountable care” more generally) represents a change in how health insurance payers pay for care. Payment reform is only the first half of the equation, however, because delivery reform drives benefits to patients by more effectively managing the care of the covered population. As the ACO model continues to proliferate (Muhlestein 2017) and increase its impact on patients, there is a growing need to better understand the delivery aspect of accountable care as it is broadly construed across Medicare, Medicaid, and commercial payers.

Managing a population requires ACOs to oversee care that is delivered across multiple locations and provided by multiple clinicians (IOM 2001). To be successful, ACOs must coordinate care delivered by the various providers to address the population’s clinical—and in some cases, social—needs that affect the cost of health care services. At times, entities seeking to become ACOs already will have all the providers necessary for population management within their organizations, as is often the case for large integrated health systems.

More often, however, prospective ACOs will not have providers representing the entire care continuum, requiring them to find new partners, including personnel outside the traditional provider categories. These relationships could be established through mergers and acquisitions or could also come about through virtual integration, where providers work together, sharing aligned incentives, while maintaining separate ownership and control of their organizations (Kreindler 2012).

In either case, these new partnerships, which include a variety of arrangements from novel contractual service relationships to much less formal “quality compacts,” may allow providers to expand their influence throughout the delivery system as they begin to work across location and episodes of care to manage population health.

While new partnerships will be necessary for ACOs to succeed in their population-based contracts, not every partnership will be ideal or successful. Provider partnerships are becoming increasingly important, but little research has been done to identify best practices for organizations to follow when selecting partners. This paper provides an overview of how health care providers can prioritize and evaluate potential partnerships by identifying characteristics of high-value providers. ACOs, prospective ACOs, vendors that support ACOs, and policymakers can use these findings to improve the practices and policies that drive the creation of new health care provider partnerships.

This project was funded by a grant from the Robert Wood Johnson Foundation to identify a framework for assessing potential provider partners for ACOs. Data were collected from three sources: (1) an indepth literature review of materials describing high-value health care organizations as a foundation for draft framework development, (2) an expert panel convened to evaluate the framework and help prioritize provider types to evaluate, and (3) interviews with ACOs and entities representing various types of health care providers. Data collection, analysis, and evaluation were conducted from Spring 2015 through Fall 2016. Targeted topic briefs centering on different types of provider partnerships were published online (de Lisle 2016).

Literature review

The first step in creating this framework was to review any works that treated the topic of high-value characteristics at either the system or organizational level. We used only

materials that addressed organizational competencies holistically rather than individually (e.g., patient centeredness as one of multiple proposed items rather than on its own) in order to maintain the appropriate depth that would be necessary for an all-encompassing framework. This criterion led us to a modest list of prominent works (see “Past work,” below) that served as the foundation for a significant amount of subsequent research. Through this process, we developed a draft framework that guided a detailed review of specific competencies organized into categories.

Expert panel

Next, we convened a group of expert panelists to review and finalize the draft assessment framework and establish a list of provider categories to which the framework would be applied. In order to include the perspectives of multiple stakeholders, we selected panelists from a variety of health care industry segments, including providers, payers (public and private), purchasers, consumer advocates, and academics. The expert panelists (Appendix A) were convened in Salt Lake City in July 2015 for a daylong meeting to discuss the research objectives, framework creation, and provider types for evaluation. Group discussion and subsequent communications led to the final frameworks, provider list, and interview guide.


To test the assessment framework in the field, we conducted 26, 60-minute interviews with provider associations (7), ACOs (16), and researchers (3). See Appendix B for a complete list of interviewees and their organizations. We selected provider associations to represent the provider categories set by the expert panel (i.e., primary care, hospital, specialty, postacute care, behavioral health, and pharmacy). All the ACOs interviewed were actively engaged in provider partnerships and represented diverse organizational structures, market factors, and contracting experience. Interviews were conducted between the fall of 2015 and the summer of 2016. All interviews were transcribed and coded using the Dedoose platform to enable qualitative analysis.

Past work

Our review considered past works from academic institutions and researchers, industry experts, and associations with a high degree of experience in provider operations. “High-value” is often a very ambiguous term in health care, but Porter and Teisberg offered the most concrete definition in their seminal work, Redefining Health Care,

as “health outcomes achieved per dollar spent.” Most other versions of this concept in the literature are variations or expansions of their work (Porter 2010).

The various frameworks that fit the scope of this project ranged from those addressing provider characteristics at the level of the system, such as a nation’s health care system of providers (IOM2001, Schoenbaum 2008, Smith 2012) to those of an individual provider organization, such as a medical group or hospital (AMGA 2012, CMS 2011, Kabcenell 2010). Some frameworks took a recipe-like approach in enumerating the “ingredients” and steps to creating a high-value health care system while others used cross-industry comparisons to show what American health care could be were it to imitate other industries that have undergone transformations (Coye 2001). Several works identified concrete characteristics such as health information technology capabilities, while others identified more conceptual features such as “patient centeredness.” Surprisingly little research took a comprehensive view of health care

provider competencies and characteristics, a blind spot that reaffirms the need for further exploration of this topic. The concepts embodied in the ACO approach represent a new emphasis on holistic patient treatment that has little precedent in the literature.

Framework overlap was a key point of interest in our review. We found that despite differing terminology, most frameworks had nearly complete overlap, with the most significant differences being the level at which the topics were treated (e.g., health information technology [HIT] at the national level versus the level of individual provider organizations).

By deconstructing the various framework categories, we were able to map the individual characteristics and domains of each framework to create one overarching framework. Our expert panel also contributed several new characteristics to each

domain and helped to rank domains by level of importance for partnership evaluation (Table).


ACOs will need to prioritize their unique challenges and opportunities to successfully manage their patient populations. A critical selfexamination can help ACOs identify the needs, gaps, and opportunities to help their populations. Self-examination can also help them to decide which providers they should approach about establishing partnerships. After identifying the types of providers with which to partner, ACOs must then assess which specific providers may be ideal partners. Based on our research, we have identified two frameworks to help with this process. The first lays out a process for identifying which types of providers ACOs should work with. The second provides a method for assessing individual partners.

Framework to identify provider Partners

ACOs come in many varieties with different organizational structures, capabilities, and levels of experience. Organizational structures range from integrated delivery systems that already provide the spectrum of health care services under common ownership to newly created physician groups that only provide outpatient medical care under a virtually integrated arrangement (Muhlestein 2014). Covered populations vary based on the payers that ACOs work with and may include Medicare, Medicaid, or commercial populations. Market and regional factors will also vary. Some ACOs cover rural populations, others are concentrated around a small urban region, and still others span both (Muhlestein 2017). Because of the variety of entities, populations, and regional factors involved, each ACO must assess its ability to appropriately manage its population and that will, in turn, help inform decisions about which partnerships it should prioritize (Fisher 2012).

Our framework for identifying needed provider partners is summarized in Figure 1. ACOs should first assess their covered population needs; second, identify and assess opportunities to intervene and eliminate or address risks the population faces; and third, evaluate what level of sophistication is needed in a partner while assessing what is available in the market. By conducting a meaningful self-assessment, ACOs will be positioned to identify areas where they can improve by either internally developing the capability or by partnering with other organizations that can help.

  1. Population needs

The first step is assessing the population for which the ACO is responsible. Under this framework, population assessments should address three important areas: evaluation by payer, demographics, and clinical needs. First, the payer (Medicare, Medicaid, or commercial) that an ACO works with will define who the ACO is responsible for, the breadth of services covered under the agreement, and the terms of the patients’ individual financial responsibility for care, such as copayments and deductibles.

Second, the demographics of a population—including factors such as age, race, sex, economic status, physical location, and social needs— dictate many potential interventions to improve care for the population.

Finally, the clinical needs of the population include the prevalence of disease states, common diagnoses, and utilization patterns. The assessment for each area can be as simple or in-depth as the ACO has the ability to perform.

Organizations performing this assessment should stratify their populations based on expected risk, including common disease states, financial and logistical barriers to care, and challenges unique to the served population, such as language and cultural barriers.

Following that assessment, ACOs should be able to list the major challenges that may influence the volume (including related access issues), cost, and quality of care their population needs. In a more thorough evaluation, potential costs could be estimated for each challenge, including associated patient financial ability.

  1. Intervention opportunities

The next step is to identify and assess opportunities for intervention to eliminate or address major risks facing the population. Evaluating the risks facing a population differs from evaluating the needs of a population. Risk is what may happen if a need is not met; for example, a patient who needs a doctor’s appointment is a risk for an unnecessary emergency department visit. An ACO must also look at the entire continuum of care and evaluate nonclinical intervention opportunities and resources that will affect health outcomes such as social support services. Interventions can address countless concerns. Examples include diabetes case management programs, transportation assistance, and implementing medication reconciliation protocols to help avoid polypharmacy issues. The process of identifying interventions will allow ACOs to determine which skill sets, capabilities, and resources will be necessary to address the identified concerns in an effective way.

  1. Partner needs

After intervention opportunities are identified, the final step is to determine the type of partner that could best fill those gaps and review the available potential partners within the region. Potential partners come in many forms. Figure 2 is a non-exhaustive list. This figure was derived, in part, from an article in Health Affairs in which the authors outline several “domains of influence” over a patient’s health that are not currently receiving sufficient attention due to America’s overemphasis on institutional (medical) health care (McGinnis 2002). In addition to medical care, those categories include genetics, social circumstances, environmental conditions, and behavioral choices.

Acknowledging, as the authors do, that little can be done from a health system standpoint about genetics, we set that domain aside. However, we created two categories, preventive care and behavioral health, that can encompass social circumstances, environmental conditions, and behavioral choices. Some aspects of those two categories involve the individual and fall within the purview of conventional providers. Others, though, would mean extending the current health system beyond its traditional role of providing medical diagnosis and treatment.

These potential partners have the ability to fill multiple gaps and require varying amounts of effort from the ACO, both in terms of the evaluation and the actual partnership activities. For example, partnering with a skilled nursing facility network that covers the entire footprint of an ACO’s market would require less effort from the ACO than partnering with a series of independent skilled nursing facilities. By estimating the managerial and financial impact of partnering with different provider types, an ACO will be able to identify which providers represent the highest opportunity for cost reduction and quality improvement for the population. The assessment outlined by this framework is directional rather than conclusive as ACOs begin to identify the specific providers with whom they wish to partner.

Framework to assess individual potential provider partners

After creating a list of potential partners, ACOs must evaluate each one to assess whether it is likely to provide high-value care. High-value provider partners must be able to effectively work with the ACO, communicate and share information, and provide appropriate interventions to lower the cost and improve the quality of care for the patient population.

Based on our literature review and expert panel, we have defined seven domains that give ACOs a framework for vetting potential partners. Framework categories include: patient centeredness, high-value culture, system and public accountability, team-based care, HIT systems, performance improvement, and financial readiness. An overview of these domains is provided in the Table (page 42).

While these domains are relevant, they are not equally important for all ACOs and all provider types. When evaluating potential partners, ACOs should decide which domains are most relevant for the opportunities they have identified based on (1) the strengths of the ACO, and (2) their ability to accurately evaluate the potential partners (some aspects of a provider partner may be more difficult to measure and are therefore reprioritized for the sake of practicality). When an ACO has a particular strength, a relative weakness in a particular partner may not matter, such as when an ACO has an established HIT platform that could easily be shared with a new partner.

Evaluating a partner can be a challenge as accessible data are not readily available and metrics to define different capabilities are not always clear. When initially assessing multiple providers, whatever information is at an ACO’s disposal should be used, including government data and qualitative information about potential partners. When meeting with potential partners, open conversation is necessary for a candid discussion of the relative maturity of the ACO and its partners. An agreement to work together toward a common objective is of paramount importance.

Provider evaluations

Based on our interviews, we created qualitative rubrics which suggest important factors on which to evaluate the organizational characteristics of six provider types: primary care providers, specialists, hospitals, postacute care providers, behavioral health providers, and pharmacists. These evaluation rubrics were previously released as two-page applied briefs (de Lisle 2016). A fuller rubric for each of the provider types has been published and can be accessed at


The goal of this project was to create a framework to help risk-bearing ACO providers identify high-value partners for long-term success. Because of that narrow focus, this work has several limitations, including limitations related to implementation, market factors, legal concerns, and metrics. We created frameworks and identified important factors for ACOs to evaluate when choosing providers. We did not, though, develop an implementation guide outlining the detailed steps on how to engage providers and work together. There is much work to be done to enhance the practical application of this framework. Significant customization will be necessary for individual organizations.

Additionally, we did not explicitly address how to choose providers when market conditions or a limited pool of potential partners largely dictate who ACOs must partner with. In cases where the partner is not at the level or organizational readiness that the ACO is looking for, there are opportunities for the ACO to help the provider partner become better at offering high-value care or for the organization to improve its own care delivery using this framework to identify areas for improvement. But this framework is not meant to show the ordered steps that providers should pursue to improve their ability to provide high-value care, and any approach will need to be customized on a case-by-case basis.

Whether partnering through ownership or affiliation, there are legal issues, including Stark and anti-kickback laws, that must be acknowledged by the participating providers. This work has not addressed those concerns, but providers considering partnerships must address them. While there are safe harbors (CMS/OIG 2015a) and other legal options (CMS/OIG 2015b), advice for specific situations should be sought from qualified legal counsel.

As referenced earlier, the underlying research for this article was carried out between 2015 and 2016 when the industry was rapidly evolving. Even so, the conclusions and recommendations presented here will remain relevant as long as the U.S. health care system remains fragmented to any substantive degree.

Finally, while this framework provides guidance on how to evaluate multiple provider types, it does not provide a comprehensive set of metrics for assessing the providers. Meaningful quality measures that validly quantify providers’ performance need to be collected from existing sources or newly created and customized for each provider type.

Much more work is necessary in this area and could be an opportunity for trade associations, medical societies, or specialized academic researchers to further the field of research.


Sustained adoption of the ACO model has revealed the challenges many organizations face in managing the entire continuum of care. This significant shortcoming has led some ACOs to virtually integrate with providers from across the care continuum, but the partnering dynamic introduces a whole new set of challenges.

This paper attempts to give guidance to ACO leaders who are embarking on or expanding their care continuum management capabilities through partnerships. Continued development of a “partnering science” will be essential for health care in the U.S. to operate more like other industries where strategic partnerships are critical for economic survival and add value for the consumer.

The principal consideration in an ACO’s partnership strategy will be striking the balance between finding partners that can fulfill the immediate needs of a dysfunctional American health care system (the status quo) versus those that will bring the most value in a quickly evolving system (the future state). The expert panel assembled for this research repeatedly emphasized the need to balance an ACO’s aspirations against the realities of the current system. While the evolution of payment structures continues to keep health care providers in two different worlds, partnership building may also need to happen on two tracks to make the transition possible.

A related challenge for ACOs is to prioritize their partner selection efforts. It is possible that an ACO centered on primary care will decide that it will eventually need to partner with all six provider types highlighted in this research, but that current limitations in staffing and resources make it impossible to carry out due diligence on more than one partnership at a time. A thorough evaluation of the population’s needs will lead to a natural prioritization of which providers will make the largest contribution toward fulfilling those needs. This, combined with an assessment of intervention opportunities and a determination of market availability, will lead to a clear sequencing strategy.

Another important issue in the movement to partner is the effect that novel successful business relationships could have on the broader health care market. Some ACOs will use partnerships to develop an internal competency that will then obviate the need for a continued partnership, while others will use newly acquired competencies together with a trusted business partner to increase market share through their combined efforts. In other cases, successful partnerships could lead to acquisitions, mergers, or the creation of narrower networks and therefore increased consolidation in health care.

Policy recommendations

While we believe the frameworks set forth above will provide guidance for ACO leaders, many outside influences may hinder successful partnerships. These obstacles could be addressed with a combination of state and federal policies in addition to industry developments already underway.

Our first recommendation is that state and federal payers continue providing ACOs and the public with tools based on government data to help shed light on the quality performance of potential partners. Various ACO leaders in our interviews cited the benefits of the Medicare Hospital and Nursing Home Compare datasets. Despite widespread recognition that these datasets are not perfect and do not adequately communicate the quality (or lack thereof in some instances) of a provider, they have given the industry a starting point and will surely form the basis for some important partnering discussions.

Our second recommendation is to create payment models that specifically incent behavioral health providers, postacute care providers, and pharmacists to work with ACOs. Creating an advanced alternative payment model (APM) under the Medicare Access and CHIP Reauthorization Act (MACRA) specifically for those provider types could encourage them to be proactive about working with physicians and hospitals in ACOs. Even a basic model that covers only a portion of services rendered by these providers will send a clear signal to the industry about the expectation of collaboration (through virtual integration) and simultaneously spur industry imitation and improvements for commercial arrangements.

Finally, a fundamental element for advances in partnership development will be the availability of practical, user-friendly population-assessment tools that will enable quicker identification of population needs. As previously discussed, this step should be the initial impetus in partnership decisions, but is currently not in the skill set of many providers. Development of population assessment tools has been an industry focus and continues to improve, but it is currently out of the price range of many smaller physician groups—a sizeable portion of the early ACO contingent. Furthermore, state and federal governments

could encourage the availability of such tools by subsidizing or offering such tools in concert with ACO program participation, as has been done in Colorado under its Accountable Care Collaborative, the state’s version of a Medicaid ACO.

Next steps

The two frameworks set forth in this paper can serve as an initial draft to be used and adapted to a variety of market scenarios and provider configurations. Further development is needed to identify and define high-value characteristics as they relate to different provider configurations. The process surrounding a partnership with a hospital will be very different for a hybrid ACO with physician–hospital collaboration than for a physician ACO that has no prior experience with an acute care partnership. More development is also needed to build out and refine the high-value characteristics of provider types not highlighted in the previously referenced brief series or this paper. The metrics for assessing these partnerships are underdeveloped due to a dearth of experience, so they will require a deliberate—and transparent—effort among the various purchasers of health care to improve for industry-wide use. An effective way to spur this development is to tie payments to patient outcomes rather than provider inputs (fee-for-service), which forces the issue of outcomes measurement. The journey toward an accountable health care system will of necessity represent an iterative process where the only way to “learn” is to “do.”


Accountable care offers incentives for entities to improve the cost and quality of health care. To accomplish this in an effective way, ACOs must recognize the needs of their assigned populations and work to provide comprehensive care management across the spectrum of provider types and sites of care. Accomplishing this will require many ACOs to create novel partnership arrangements. The frameworks discussed in this paper can assist ACOs in setting up a process for evaluating potential provider partners that will help them achieve the goal of providing high-value health care.