How "accountable" are accountable care organizations?

270

OctoberYDecember & 2014

How ‘‘accountable’’ are accountable care organizations? Rachael Addicott Stephen M. Shortell Background: The establishment of accountable care organizations (ACOs) in the Affordable Care Act (ACA) was intended to support both cost savings and high-quality care. However, a key challenge will be to ensure that governance and accountability mechanisms are sufficient to support those twin ambitions. Purpose: This exploratory study considers how recently developed ACOs have established governance structures and accountability mechanisms, particularly focusing on attempts at collaborative accountability and shared governance arrangements. Methodology: Four case studies of ACOs across the United States were undertaken, with data collected throughout 2012. These involved 34 semistructured interviews with ACO administrative and clinical leaders, observation of nine meetings, and a review of documentary materials from each ACO. Findings: We identified very few examples of physicians being held to account as a collective and therefore only limited evidence of collaborative accountability impacting on behavior change. However, ACO leaders do have many mechanisms available to stimulate change across physicians. The challenge is to determine governance structure(s) and accountability mechanisms that facilitate the most effective combination of approaches, measures, incentives, and sanctions to achieve the goals of more accountable care. Practice Implications: Accountability structures and processes will need to be tailored to local membership composition, historical evolution, and current stage of development. There are also some common lessons to be drawn. Shared goals and incentives should be reflected through performance criteria. It is important to align measures and thresholds across payers to ensure ACOs are not unnecessarily burdened or compromised by reporting on different and potentially disjointed measures. Finally, emphasis needs to be placed on the importance of credible, transparent data. This exploratory study provides early evidence regarding how ACOs are establishing their governance and accountability arrangements and provides a foundation for future research and theory-building in this area.

Key words: Accountability, accountable care organization, collaboration, governance Rachael Addicott, PhD, is Senior Research Fellow, King’s Fund, London, United Kingdom. E-mail: [email protected]. Stephen M. Shortell, PhD, MPH, MBA, is Dean Emeritus, School of Public Health, Blue Cross of California Distinguished Professor of Health Policy and Management, and Professor of Organization Behavior, University of California, Berkeley. E-mail: [email protected]. Rachael Addicott received funding from a Harkness Fellowship in Health Care Policy and Practice from the Commonwealth Fund to undertake this research. The authors have disclosed that they have no significant relationship with, or financial interest in, any commercial companies pertaining to this article. DOI: 10.1097/HMR.0000000000000002 Health Care Manage Rev, 2014, 39(4), 270Y278 Copyright B 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins

O

ne of the 2010 Patient Protection and Affordable Care Act (ACA) ambitions was to reform the systems of health care payment and provision to encourage cost savings and high-quality care. One means promoted to achieve these goals was the development of accountable care organizations (ACOs; Berwick, 2011). These organizations are to be held accountable for both the cost and quality of care for a defined group of patients. A key challenge for ACOs will be to ensure that governance and accountability mechanisms are sufficient to support their goals (Fisher & Shortell, 2010). Although there has been considerable discussion regarding high-level payment reforms and regulations for ACO development (Berwick, 2011; Delbanco et al., 2011; Guterman et al., 2011; Higgins, Stewart, Dawson, & Bocchino, 2011; Lieberman & Bertko, 2011) and what is necessary to put ACOs into practice (Larson et al., 2012), very little is

Copyright © 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

271

How ‘‘accountable’’ are accountable care organizations?

understood about the governance structures and accountability mechanisms that have been established by provider networks to accompany these developments. The Centers for Medicare and Medicaid Innovation has established several ACO initiatives across the United States to test their efficacy for the management of Medicare populations. The Shared Savings Program allows ACOs to develop from a one-sided risk model (where they do not take on downside risk and share in a limited percentage of savings) to a two-sided risk model (where the ACO shares in both downside risk and a greater percentage of savings). The Pioneer model involves provider organizations with a greater degree of experience in care coordination that can apply to an accelerated population-based payment arrangement. Third, the Advance Payment model provides additional upfront support to developing ACOs, which is then recovered from any savings generated. There is also considerable activity in the commercial sector, with some provider networks having considerable experience in taking on at-risk contracts with commercial health plans (Higgins et al., 2011). At present, there are over 400 ACOs operating in the United States with nearly all states having at least one (Lewis, Colla, Carluzzo, Kler, & Fisher, 2013). To achieve the intended cost savings, ACO providers typically collaborate to develop a coordinated care management approach targeted at patients who are at risk of potentially avoidable hospital admissions or emergency department visits. This requires the development of team-based care involving primary care physicians, specialists, nurse practitioners, pharmacists, and other health professionals. The locus of decision-making, planning, and accountability begins to shift from the individual to the team and the larger organization or network of which the team is a part. In a climate where the financial success of an entire provider network is dependent upon individual physicians and others pulling their weight and collaborating, there is considerable pressure to perform (Blackmore, Mecklenburg, & Kaplan, 2011; Mechanic, Santos, Landon, & Chernew, 2011; Tollen 2008). Thus, there is great need to build structures and processes to effectively monitor and ensure this performance. The ACO model suggests a governance structure and processes of accountability where providers (individuals and organizations) operate collaboratively within the overarching ACO boundary. For the purposes of this article, governance is defined as the policies and decision-making arrangements that support the organizational goals and objectivesVsuch as the boards and other forums where decisions are made, members of these forums, and the processes for decision-making (Lynn, Heinrich, & Hill, 1999; Reed 2003). Within this overarching structure, accountability is defined as the processes by which one party reports to another on its actions or performanceVeither with or without consequence (Day & Klein 1987; Emmanuel, 2012). In the ACO context, governance structures provide the forums

necessary to meet the demands of external bodies such as payers and accrediting agencies as well as to make decisions and take corrective actions to improve internal performance. Thus, the ACO model is built on the implicit assumption that governance and accountability operate in tandem to achieve the shared aims of improving quality and containing costs. Similar approaches in the past, such as integrated delivery networks and systems of the 1990s designed to improve quality and constrain costs, frequently lacked the clinical integration capabilities to be successful (Shortell, Gillies, Anderson, Erickson, & Mitchell, 2000). Under capitated contracts, they also had limited success in shifting accountability for costs onto providers (Goldsmith, 2012). Burns and Pauly (2012) argue that there is even less consensus regarding the ambitions or strategies of ACOs. As such, there is understandable skepticism whether the collaborative accountability espoused through ACOs will have a positive impact on either cost or quality. Drawing on observations from four commercial ACOs across the United States, we undertook an exploratory study of the governance structures and accountability mechanisms that ACOs are establishing, particularly focusing on the extent to which collaborative accountability approaches are being implemented by ACO membersVincluding shared governance structures, accountability mechanisms, metrics, and receptivity to such changes.

Theory/Conceptual Framework Many gaps or failings in the quality of care can be attributed to poor coordination across payers and providers of care. Organizational structures and fee-for-service payment models reinforce the fragmentation of the system, whereby performance measurement and accountability are focused at an individual level and ‘‘poor performance is seen as a consequence of individual failure, rather than flawed systems’’ (Fisher & Shortell, 2010, p. 1715). The ACO model attempts to address the failings of current payment systems that continue to encourage more independent provision of care and potentially discourage cost containment (Fisher, Staiger, Bynum, & Gottlieb, 2007). As such, the Institute of Medicine suggests that providers share accountability for the cost and quality of care (as promoted through ACO developments) as a potential means for promoting greater coordination and integration. Fisher and Shortell (2010) further suggest that the challenge for policy makers will be to ensure this collaborative approachV alongside other payment reformsVoffers a sufficient governance and accountability framework for medical professionals and the system. The question remains, however, whether physicians and others involved in health care payment and delivery are ready and willing to collaborate and be held to account in this manner. Each of these ‘‘parties’’ in ACO collaborationsVphysicians, hospitals, and health plans (among others)Vis traditionally


272

Health Care Management Review

autonomous, relatively unaccustomed to being held to account in a more collaborative arrangement. Physicians are traditionally self-regulating and largely determine their own domains of accountability while being responsive to payment programs and incentives from employers, health plans, and others (Mechanic, 1991). ACOs may represent a model where physicians are simultaneously accountable to their patients, to each other, and also to ACO executives and members from other professional groups and organizations. Little is known about the factors that will impact on the long-term success of ACOs (Fisher, Shortell, Kreindler, Van Citters, & Larson, 2012). However, we suggest that the model is built on an implicit assumption that the interplay between governance structures and accountability mechanisms will have a significant role in achieving their ambitions. We expect to see governance structures that provide a forum for feedback on performance metrics, allowing for decisions to be made regarding appropriate actions and sanctions (i.e., the accountability processes). We explore how these structures and mechanisms operate and work together in practice. In these emergent governance structures and accountability relationships, there are a number of structural and processual features of ACOs that are unclear. We used the following questions as a framework for identifying and exploring these governance structures and accountability mechanisms.

& What

(if any) collaborative governance structures and accountability mechanisms are being established to guide the operation and performance of emergent ACOs? & How are ACO members held to account, individually and collaboratively? & What are ACO members accountable for? And what is the standard (or threshold) that must be achieved? & How are measures and thresholds established? What is the level of agreement and ‘‘buy in’’ regarding these measures? & What sanctions or consequences are employed in instances where thresholds are not met? It is possibleVand even likelyVthat there could be multiple and overlapping accountability relationships operating simultaneously within an ACO, which may also differ across these collaborations. Some of these overlapping structures and processes may be compatible, whereas others might operate in conflict. Alongside formal accountability mechanisms, informal (and historical) relationships may also persevere. Shortell, Waters, Clarke, and Budetti (1998) have previously questioned whether managed care arrangements in the 1990s represented such a conflict, where physicians had multiple accountabilitiesVto patients, payers, and other organizational structures. These accountabilities may differ (e.g., clinical and financial) and may often be incompatible.


All of these featuresVand how they are played out in practiceVare likely to factor in ACO success or failure. Formal processes and structures of accountability will interact with informal relationships and the receptiveness of different member groups to more collaborative governance models. We examined how ACOs are managing these issues and what lessons might be learned.

Methods Four case studies of ACOs across the United States were undertaken between February and May 2012, triangulating data from semistructured interviews, nonparticipant observation, and documentary analysis. These cases were purposively chosen to represent a variety of types and stages of developmentVall had (two-sided) at-risk contracts in the commercial setting, as well as existing or planned Medicare ACO agreements through the Shared Savings or Pioneer Programs. Because of resource constraints, the analysis is limited to the four sites, but following Eisenhardt (1989), we believe this is sufficient to provide an empirical and conceptual basis upon which others can build. Eisenhardt (1989) suggests that random selection of cases is not necessary for building theory or even preferable. There are only a limited number of cases that can reasonably be considered within a research study. This exploratory study may provide the foundations for future theory-building, and therefore, it is potentially more useful to choose situations where processes are transparent and observable. These four case studies were selected from an identified list of 221 ACOs with commercial contracts (Muhlestein, Croshaw, Merrill, & Pen˜a, 2012) and cross-referenced with those in the initial cohort of Pioneer ACOs and those known to be applying to the Shared Savings Program. Two were geographically located in the Midwest, one on the West Coast and one on the East Coast. The four ACO case studies involved 34 semistructured interviews with ACO ‘‘members’’V19 physicians and 15 administrative executives (from the medical group, hospital, or health plan). All interviewees (classified for simplicity as ACO ‘‘members’’) were board members and/or held an executive role in the ACO (e.g., contracting or executive support). Through the interviews, we examined governance structures, accountability relationships and processes, and perceptions of the operation of the model in practice. Interviews were audiotaped, and key themes and verbatim quotes were drawn from the recordings. The interview questions provided a basis for the thematic coding structure. The data were then scanned for specific cases that illustrated and provided evidence for the themes. We observed nine meetings (totaling 22 hours), including ACO board meetings, strategic planning meetings, and operational committees. These observations further sought to consider representativeness, collaboration in practice, the


273


nature and content of decision-making, and any ongoing implementation issues. Third, we undertook a review of documentary materials from each ACOVMedicare ACO applications, annual reports, meeting minutes, committee charters, and strategic plans. This review provided background information on the history of the ACO and decision-making and the relationships between its members.

Findings Table 1 provides a summary of the key characteristics of each of the participating ACOs. As shown, the sites were similar in that all had 1. commercial at risk contracts (upside and downside risk); 2. existing/proposed Medicare ACO contracts (since the time of data collection, the two sites who applied to the Shared Savings Program have had their applications approved); and 3. established around the hub of a medical group, with varying degrees of involvement from hospitals, large employer groups, and health plansVdepending on the nature of the ACO-like contract. However, the ACOs under examination in this project also varied significantly in their level of integration, corresponding governance structures, and mechanisms of accountability. The cross-case variation in the thematic questions are highlighted and discussed below. 1. What structures and processes are being established to demonstrate collaborative governance?

The degree and history of integration (particularly showed through formal structures) were closely reflected in governance arrangements. More formally integrated ACOs (through joint ownership, for example), like ACO 3, had a wider and more senior representation of member groups within the governance structures. In all cases, the board was primarily comprised of executive physicians with attendance by financial and legal representatives. It was common across ACOs that the board contained members from the medical group(s). All but one of the ACOs (ACO 1) included hospital partners on the board. ACO 1 did not formally engage local hospitals in their ACO contracts. The medical group had made the decision that they could make faster progress on their own, rather than working through the shared governance of involving hospital partners. I don’t trust that [the hospital] wants to manage them [patients] in the same way that I want to manage them. So there is a lot of tension there right now. (Medical Director, ACO 1) Furthermore, there was only one ACO (ACO 3) that included representation from a health plan on their board or in other governance arrangements. In this instance, the health plan and medical group were formal partners in an integrated health system. This ACO was applying to be part of the Shared Savings Program and acknowledged that they would need to establish a separate governance structure for this contract, as the health plan had no involvement in the application or management of this patient population (where

Table 1

ACO site comparison Key dimension

1

2

3

4

At-risk payer contracts

Commercial insurers Pioneer

Multiple commercial insurer contracts Pioneer Medical group Hospital

Commercial insurers Possible SSP

Commercial insurers Possible SSP Medical group Multiple hospitals

Composition

Medical group Hospitals? Dominant employer group Level of integration Contractual relationship with hospitals Shorter history Payment types FFS

Other issues

IPA supported by a medical management group Mix of capitation and FFS

Contentious relationship Complex payer mix Large Medicaid between providers population

Medical groups Hospital Health plan Fully co-owned

Medical groups paid on capitation Physicians paid FFS Potential acquisition FTC investigation

Informal integration Long history Mix of capitation and FFS FTC investigation

Note: SSP = Shared Savings Plan; FFS = fee for service; FTC = Federal Trade Commission.


274


CMS represents the payer). In ACOs where there were multiple payers, these were not represented in formal governance structures. Payers and, on occasion, the employer groups that they represented were more likely to be consulted through regular managerial interaction with ACO executives. These managerial relationships were less formal. ACO executives met with payer representatives to negotiate contracts, discuss performance metrics, and monitor progress against quality and cost targets. It was often the case that organizational decisions were made through these managerial interactions, rather than through formal governance mechanisms. In one ACO observed, the Chief Executive Officer had several one-to-one meetings with board members in the days preceding the formal board meeting, where negotiations and decisions were made. These less formal meetings (outside the structured board meeting) were reportedly where much of the collaboration was taking place. We don’t even today have shared governance. A lot of this is driven by shared management. We developed a common strategic plan. Leadership from all parts of the system meets regularly to talk about strategy, better care, and lower costs. Frankly, that’s where a lot of the integration has happened. (Chief Executive Officer, ACO 3) Patients and consumers were largely unrepresented in governance structures; however, there was some progress being made. It is a requirement of Medicare ACOs that patients are represented on the board. As such, the four ACOs in this research were developing alternative arrangements to accommodate patients into their governance structure. This was achieved in different waysVby including patient representatives on their existing board, establishing a separate Medicare ACO board, or creating a new patient advisory group. At the time of our research during 2012, the specific role that patients would play in ACO governance was unclear. 2. How are ACO members held to account? Alongside structures of governance, there were also changes in processes of accountabilityVthat is, what ACO members are responsible for and how this process is managed. Through the ACO model, the number of accountability relationships was increasing, with physician members being held to account by immediate clinical colleagues, associated medical groups, and payers. Similar to governance structures, patients were largely unrepresented in these accountability relationships. The primary mechanism for holding physicians in the ACO to account was through a pay-for-performance program. Physicians were monitored and financially rewarded on established measures of performance. In some cases, these financial rewards were tied to the amount of savings achieved


by the ACO as a wholeVthat is, the pool of potential performance reward was determined by how much money the entire ACO saved. Only one of the ACOs did not have an overarching pay-for-performance arrangement (ACO 4), but they did issue bonuses to physicians and were beginning to use financial incentives in specific clinical areas, such as diabetes. ACO-member physicians continued to be predominantly held to account as individuals through performance measures. Less progress or success was reported in establishing mechanisms of shared accountability where a group (e.g., clinic or specialty group) might be collectively held to account. In some cases, there were some group level incentive payments (usually at the clinic level) alongside individual physician pay-for-performance goals. However, the ACO model had little impact on establishing mechanisms of shared accountability. Similarly, physicians in particular were being increasingly incentivized to achieve goals that would be more readily achieved through increased collaborationVwhereas rewards and related risks largely remained at the individual level. Patient satisfaction measures (discussed in the following section) were an example of this approach. Although most ACOs that we observed used grouplevel bonuses sparingly, one ACO (ACO 4) allocated 30% of its total incentive pool to how the local member organization as a whole performsVin this case, the Physician Hospital Organization (PHO). To measure this group-level effect, scores on performance indicators were aggregated at the group level. Part of the reward that the physicians get is for their individual work, but it is also for their group workV the performance of the local PHO plays a big part in how big your check is. It doesn’t matter if you achieved 100% of your measures, if the PHO failed or didn’t do well, then you’re not going to see 100% of your reward. (PHO Chair, ACO 4) This proportion of the payment is given to the group to distribute as it chooses. Principally, the distribution is allocated between the physicians in the groupVbut there have been occasions (however limited) where the money was shared with nursing and support staff. 3. What are ACO members accountable for? And what is the standard (or threshold) that must be achieved? Although the ACO model places greater emphasis on both reducing cost and increasing quality of care, most of the focus was on financial and procedural accountabilityV with limited direct consideration of quality or outcomes. Cost was largely monitored through process measures, such as prescribing rates for generic drugs. Quality was also largely measured through proxy process measures, such as hospital readmissions and other utilization or referral rates. Over time,


275


measures were introduced or adapted to reflect current highcost areas or quality concerns. For instance, over time, ACO 3 had increasingly focused on measuring the number of inappropriate emergency department visits, which the associated health plan had identified as a high-cost activity. Alternatively, ACO 4 originally had measures relating to the use of high-speed internet. Over time, this metric has been sufficiently met to enable it to be retired from the performance program and had instead become a membership criterion. At the time of data collection, there were two metrics in particular that were the focus of attention for all four of the ACOs in this studyVin-network care and patient satisfaction: In-network care: Increasingly, providers were being held to account on the number of referrals they made to providers outside of the ACO network. The patient populations that were of concern to ACOs were those in preferred provider organizations (PPOs) where patients were entitled to choose care from any provider. Contracts typically stipulated that the ACO was accountable for the quality and cost of all the care that these patients receivedVeven that provided by non-ACO providers. If the ACO received a bundled payment for these PPO patients, they were still responsible for the care costs from non-ACO providers. As such, ACO providers were under pressure to keep patients within the ACO network in order to have greater control over care management, prevent unnecessary utilization, and contain costs. In the forthcoming year, ACO 4 will measure in-network care as the percentage of days that PPO patients remain within the ACO networkVpresently, this percentage ranges between 93% and 100% across their ACO providers. However, as patients are part of a PPO plan, there are no formal mechanisms for ensuring they remain within the ACO network of providers. The ACO must use more informal and indirect mechanisms of influence, and many providers found this to be insufficient for meeting the stipulated in-network provision thresholds.

patient satisfaction. This is particularly relevant for the patient populations participating in these ACOs who are part of a PPO and not restricted to a provider or network. Physicians were anxious that there are many factors involved in how patients perceive their satisfaction, some of which were considered to be outside of their personal control. These anxieties resulted in considerable opposition to patient satisfaction measures. Who set up these measuresVare they accountants, are they physicians, are they politicians? (PHO Executive, ACO 4) ‘‘This isn’t statistically valid, you have no idea what you’re doing, and why are they rating me, I provide great care, my patients love me.’’ I’ve heard all of it. My answer to them is that unfortunately patient satisfaction ratings are not going away. You can’t blame the messenger. (Medical Director, ACO 1) Both the in-network care and patient satisfaction metrics are examples of measures that have caused considerable anxiety. The actual metric, how it is measured, and the threshold to be achieved were all contentious, as noted below. 4. How are measures and thresholds established? What is the level of agreement and ‘‘buy in’’ regarding these measures?

We wanted to have transparency to try and encourage each other to do better. (PHO Executive, ACO 4)

The process for establishing the measures by which ACO members are held to account differed slightly across the case study sites, largely depending on the member groups represented. For instance, if the ACO represented a single, dominant health plan or employer group (such as ACO 1), then they tended to be more involved in the negotiations. Alternatively, others had much more control over devising the performance metrics. ACO 4 developed a set of quality measures internally for their commercial population. They consult with their contracted health plans but were adamant they will only operate to a single set of measures, developed and agreed internally. This ACO is also participating in the Medicare Shared Savings Program, and fortunately, the quality measures are similar.

However, ACOs still had some stronger measures to encourage behavior change.

It needs to be consolidated across all payers, otherwise it’s unmanageable. (Vice President, ACO 4)

The organization has demonstrated enough value to its members that there are some sticks we can useV through membership criteria, etc. (ACO Executive, ACO 4)

ACO 2, however, has multiple commercial at-risk contracts, and the terms and performance metrics within each of these have been individually negotiated. This ACO and its providers are currently operating to multiple at-risk contracts, with multiple sets of quality measures and thresholds. Translating these into a coherent set of performance measures was largely the concern of ACO executives, but inevitably there were inconsistencies and conflicts that were felt by participating physicians. Prioritization decisions regarding where providers would most sensibly focus their efforts have

Patient satisfaction: Partly because of the requirements of the Medicare programs, ACOs are becoming more focused on measures of patient satisfaction. There is a realization that a more sophisticated, competitive, and consumer-oriented network will need to monitor and maintain a high standard of


276


been made. This ACO sees one of their main priorities to be alignment across all of their at-risk contracts. The current way of monitoring performance is not sustainable. Ibecause you can’t really manage 219 [different measures]. It’s one thing to measure it, and it’s another thing to do something about itVand that costs resources, time and expertise. (President, ACO 2) Each ACO had a Quality Committee and a Physician Compensation Committee where decisions on metrics and incentive payments were made, then approved by the governing board, and communicated across members. The metrics primarily reflected the overarching ACO system goals, relating to cost savings and quality improvement. In more formally integrated systems, these system goals were more transparent and shared across the different member groups. As such, they were then more specifically reflected in the established performance measures. For instance, a corporate goal relating to service in ACO 3 was reflected in a performance measure for the percentage of orthopedic patients satisfied with discharge. Or the quality corporate goal was measured through readmission rates. The process for establishing criteria was reportedly consensual, yet the process for establishing the thresholds or standards to be met was more contentiousVwith arguments that they should be individually (rather than collectively) set. Some physicians argued that their geographical location or the characteristics of their patient population meant that they would have more difficulty in achieving the threshold. For instance, physicians geographically on the edge of an ACO network argued that they had greater challenges in keeping all of their patient referrals within the networkV compared to those more centrally located. Yet, all physicians within the ACO are required to meet the same percentage threshold of in-network referrals. This pressure and perceived inequity was causing some anxiety for physicians who are not centrally located within an ACO network. As mentioned previously, the use of group or clinic level measures was limited; however, there were occasions where achieving a particular standard required several providers to cooperate. In these instances, however, it was ultimately a single physician who was accountable for achieving the metric. There were ongoing concerns regarding being accountable for measures that involved others in their achievement. For instance, some physicians argued that, when patients make judgments on their satisfaction, they would base their scores on overall impressions of the episode of careVparking, greeting from the receptionist, waiting times, etc.Vrather than specific satisfaction with the physician. However, ultimately, it is the physician who is held to account through the metrics. Some ACOs were considering more creative ways to amalgamate some measures at a group or clinic level to reflect the interdependencies around achieving the score, but such efforts were still being formed.


So far, there had been less performance dollars invested in group measures, and physicians were reportedly less focused on achievements in these areas. Some of those interviewed considered this to be a result of individual performance being masked within an amalgamated score. Given the limited experience of establishing group-level measures of performance, there was however only limited empirical evidence to date regarding their impact on behavior change. 5. What sanctions or consequences are employed in instances where thresholds are not met? There were different sanctions and consequences available to ACOs to incentive adherence to goals and performance thresholds. At the most severe end, individual physicians could be removed from the ACO network if they did not meet performance thresholds or comply with other membership criteria. The repercussions of this were that the physician loses their affiliation with the medical group and the clinical and administrative infrastructure that it provides. In some situations, physicians were removed from the ACO network for noncompliance with established guidelines or requirementsVparticularly around use of electronic health records. At this organization we have the luxury of telling people that if we’ve tried all the levers and none of them workVthat they might not be a good fit in the organization. (Chief Executive Officer, ACO 3) These more punitive sanctions were used rarely and only in extreme circumstances in response to consistent and particularly poor performance. More commonly, a physician would be visited by the ACO’s medical director or a ‘‘coach,’’ typically offering developmental opportunities rather than disciplinary actions. Generally, ACO executives relied on informal peer influence and the apparent competitiveness of physicians in order to influence behavior. Transparency of credible dataV with the possibility of peer comparisonVwas considered to be a significant motivator for behavior change. Physicians tend to be high performers, so when they see these comparisons or report cards, they get very engaged about it. Nobody wants to get a score lower than any of their peers. (Vice President, ACO 1) ACO 1 uses its predictive risk modeling software to calculate likely utilization rates for the patient population of each of its physicians. On the basis of these forecasts, ACO executives monitor and report variances in utilization (number of visits, referrals, and emergency department visits). So far, there have been no formal consequences for physicians, and simply seeing their own data has not had the expected impact on behavior change. This ACO is planning to provide these data in a more frequent,


277


comparative, and transparent formVand link results to more explicit consequences. The intention being that this approach will have a greater impact on improving performance.

Practice Implications The ACO model appears to have been established on the foundation that governance and accountability operate in tandem to support the shared aims of improving quality and containing costs. The ACO model suggests a governance structure and mechanisms of accountability where providers operate collaboratively within an overarching ACO framework. ACOs around the United States are at various stages of development in their governance models and accountability processes and measures. Through this exploratory study, we found very little experience of formal shared accountability across providers within an ACO, with few examples of measures that reflect the performance of a collaborative group of individuals. As such, experience with collaborative measures was limited, and there was therefore little evidence that shared measures and incentives had been influential on behavior change. It is likely, however, that as performance measures, incentives, and sanctions more specifically function to encourage collaboration, more attention will be paid to improving performance in these areas. It was evident that attempts to encourage collaboration across physicians needed to be linked to tangible incentives and/or consequences. ACOs have many levers available to change physician behaviorVfrom transparency of data and playing on competitive instincts, peer influence, through to removal from an ACO network. However, measures and incentives are only infrequently tied to group performance or collaborative working. On the basis of what we have learned from the four case studies, we suggest that accountability structures and processes will need to be tailored to each ACO’s membership composition, historical evolution, and current stage of development. But there are also some common lessons that can be drawn. It is important that shared goals and incentives are reflected through performance criteria. The aspirations of the collaborative or interorganizational context should be reinforced clearly through establishing measures that are directly tied to their achievement. Incentives for reaching these goals were instrumental in encouraging greater collaboration across providers. The findings demonstrate the importance of governance models that reflect current operation and stage of development. Those ACOs with a longer history of collaboration were more confident in establishing formal and integrated governance structures. ACOs with shorter histories relied more on managerial interaction outside of a formal governance setting. Rigid or formal structures should not be

ascribed to emergent collaborations, but rather formality should evolve over time to reflect the changing nature of the collaborative relationships and degree of interdependency. It is important to align measures and thresholds across payers to ensure a more manageable approach to meeting these thresholds. Those ACOs with multiple accountabilities to different payers reported continuing confusion and potential compromise in reaching certain thresholds. There was great interest among all the ACOs studied to develop aligned measures and thresholds on cost and quality criteria across all of the at-risk contracts within their ACO in order to simplify the establishment, communication, and achievement of their overarching goals. ACOs preferred to work to a single set of operational measures and thresholds, across all payer contracts. We emphasize the importance of credible, transparent data. The case study respondents highlighted the value of providing credible and transparent data to providers as a motivator of performance improvement. Those holding their colleagues to account relied on methods of informal influence and openness of performance data, as opposed to the use of formal sanctions and punitive consequences, in order to improve and align processes and quality. Access to transparent and comparable data is a valuable step in the direction of improving performance across the range of providers within an ACO network. We also saw little evidence of patient involvement in governance forums and accountability mechanisms, beyond, in the latter case, traditional patient satisfaction surveys. Recognition appears to be growing that to control costs while maintaining or improving quality will require greater emphasis on patient activation and engagement in their care and in the design of the delivery system itself (Audet, Davis, & Schoenbaum, 2006; Hibbard, 2007). The information provided by the four exploratory case studies provides a baseline of knowledge. There are opportunities to expand the focus of research in this area. The findings presented here are limited to a small number of case studies, which examined Medicare ACOs at an early stage of development and focused on board members within these organizations. Nevertheless, this article provides some exploratory findings, on which to base future research and theory-building. This article deliberately focused on boardlevel representatives; however, future research could usefully focus on a wider range of ACO ‘‘members.’’ We also used existing literature to establish a framework by which to consider governance and accountability. Future research could consider how ACO members themselves understand and interpret these mechanisms and their role in the success of the ACO model. Longitudinal studies that track the developmental trajectory of governance and accountability arrangements over time linked to cost and quality performance data will be particularly useful. All of the features of accountability presented in this articleVand how they are played out in practiceVare


278


likely to factor in the success or otherwise of ACO arrangements. Formal processes and structures of accountability will interact with informal relationships and the receptiveness of different member groups to shared governance models. As the requirements for improved performance in constraining costs and improving care coordination increase, there is likely to be even greater demand for datadriven and transparent accountability with the ultimate sanction being the financial viability of the ACO itself. This analysis of the four case study sites provides important information to CMS and those in the commercial sector for their ongoing monitoring efforts and to ACOs themselves in their quest to meet the ambitious goal of providing cost/ effective coordinated care. ACOs have a role to promote collaborative accountability, guided by a framework of quality measurement and incentivization that ensures supporting measures are available and effective. Acknowledgments

We thank the four participating ACOs for their willingness to be interviewed and be involved in this project. We also wish to acknowledge the financial support that the Commonwealth Fund provided for Rachael Addicott through their Harkness Fellowship in Health Care Policy and Practice. This project received institutional review board approval from the University of California, Berkeley.

References Audet, A. N., Davis, K., & Schoenbaum, S. C. (2006). Adoption of patient-centered practices by physicians: Results from a national survey. Archives of Internal Medicine, 166, 754Y759. Berwick, D. (2011). Launching accountable care organizationsV The proposed rule for the Medicare shared savings program. New England Journal of Medicine, 364(16), e32. Blackmore, C., Mecklenburg, R., & Kaplan, G. (2011). At Virginia Mason, collaboration among providers, employers, and health plans to transform care cut costs and improved quality. Health Affairs, 30(9), 1680Y1687. Burns, L. R., & Pauly, M. V. (2012). Accountable care organizations may have difficulty avoiding the failures of integrated delivery networks of the 1990s. Health Affairs, 30(11), 2407Y2416. Day, P., & Klein, R. (1987). Accountabilities. Five public services. London, UK: Tavistock. Delbanco, S., Anderson, K., Major, C., Kiser, M., Toner, B., & Hamilton, B. (2011). Promising payment reform: Risk-sharing with accountable care organizations. New York, NY: Commonwealth Fund. Eisenhardt, K. (1989). Building theories from case study research. Academy of Management Review, 14(4), 532Y550. Emanueal, E. (2012). Difficulties in making accountability practical. In B. Rosen, A. Israel, & S. M. Shortell (Eds.), Accountability and responsibility in health care (pp. 207Y218). New Jersey: World Scientific Press. Fisher, E. S., & Shortell, S. M. (2010). Accountable care orga-


nizations. Accountable for what, to whom and how. Journal of the American Medical Association, 304(15), 1715Y1716. Fisher, E. S., Shortell, S. M., Kreindler, S. A., Van Citters, A. D., & Larson, B. K. (2012). A framework for evaluating the formation, implementation, and performance of accountable care organizations. Health Affairs, 31(11), 2368Y2378. Fisher, E. S., Staiger, D. O., Bynum, J. P., & Gottlieb, D. J. (2007). Creating accountable care organizations: The extended hospital medical staff. Health Affairs, 26(1), 44Y57. Goldsmith, J. (2012). Accountable care organizations: The case for flexible partnerships between health plans and providers. Health Affairs, 30(1), 32Y40. Guterman, S., Schoenbaum, S., Davis, K., Schoen, C., Audet, A., Stremikis, K., & Zezza, M. (2011). High performance accountable care: Building on success and learning from experience. New York, NY: Commonwealth Fund. Hibbard, J. N., Mahoney, E. R., Stock, R., & Tusler, M. (2007). Do increases in patient activation result in improved self-management behaviors? Health Services Research, 42(4), 1443Y1463. Higgins, A., Stewart, K., Dawson, K., & Bocchino, C. (2011). Early lessons from accountable care models in the private sector: Partnerships between health plans and providers. Health Affairs, 30(9), 1718Y1727. Larson, B., Van Citters, A., Kreindler, S., Carluzzo, K., Gbemudu, J., Wu, F., I Fisher, E. (2012). Insights from transformations under way at four Brookings-Dartmouth accountable care organization pilot sites. Health Affairs, 31(11), 2395Y2406. Lewis, V., Colla, C., Carluzzo, K., Kler, S., & Fisher, E. (2013). Accountable care organizations in the United States: Market and demographic factors associated with formation, Health Services Research, Research Special Issue Best of Academy Health. doi 10.1111/475-6773.12102 Lieberman, S., & Bertko, J. (2011). Building regulatory and operational flexibility into accountable care organizations and ‘‘shared savings.’’ Health Affairs, 30(1), 23Y31. Lynn, L., Heinrich, C., & Hill, C. (1999). Studying governance and public management: Challenges and prospects. Chicago, IL: University of Chicago. Mechanic, D. (1991). Sources of countervailing power in medicine. Journal of Health Politics, Policy and Law, 16(3), 485Y498. Mechanic, R., Santos, P., Landon, B., & Chernew, M. (2011). Medical group responses to global payment: Early lessons from the ‘‘alternative quality contract’’ in Massachusetts. Health Affairs, 30(9), 1734Y1742. Muhlestein, D., Croshaw, A., Merrill, T., & Pen˜a, C. (2012). Growth and dispersion of accountable care organizations: June 2012 update. Salt Lake City, UT: Leavitt Partners. Reed, M. (2003). New managerialism and changing forms of organizational governance: Governance without government? In H. Dumez (Ed.), Gouverner les organisations (pp. 287Y327). Paris, France: L’Harmattan. Shortell, S. M., Gillies, R. R., Anderson, D. A., Erickson, K. M., & Mitchell, J. B. (2000). Remaking health care in America: The evolution of organized delivery systems (2nd ed.). San Francisco, CA: Josey-Bass. Shortell, S. M., Waters, T. M., Clarke, K. W., & Budetti, P. P. (1998). Physicians as double agents. Maintaining trust in an era of multiple accountabilities. Journal of the American Medical Association, 280(12), 1102Y1108. Tollen, L. (2008). Physician organization in relation to quality and efficiency of care: A synthesis of recent literature. New York, NY: Commonwealth Fund.


Accountable Care Organizations and Otolaryngology.

Accountable care organizations 2.0: linking beneficiaries.

The Economics of Medicare Accountable Care Organizations.

Accountable Care Organizations and Oral Health Accountability.

Accountable care organizations: what radiologists should know.

Accountable Care Organizations and Professional Liability Risk.

Pioneer accountable care organizations: traversing rough country.

Accountable care organizations: an opportunity for gastroenterologists.

Accountable care organizations: impact on pharmacy.

Accountable care organizations: financial advantages of larger hospital organizations.

Accountable Care Organizations in the U.S. Health Care System.

Pharmacists belong in accountable care organizations and integrated care teams.

Can accountable care organizations "disrupt" our fragmented child health system?

Moving Forward With Accountable Care Organizations: Some Answers, More Questions.

CMS Invites Feedback on the Proposed Accountable Care Organizations Rules.

Breast Imaging: A Paradigm for Accountable Care Organizations.

A new opportunity for pharmacists: accountable care organizations.

Radiology in pioneer accountable care organizations: much ado about nothing?

Accountable care organizations: what you need to know.

Accountable Care Organizations: Ensuring Focus on Cardiovascular Health.

Surgery and Medicare Shared Savings Program Accountable Care Organizations.

Early Performance of Accountable Care Organizations in Medicare.

Accountable care organizations in the USA: types, developments and challenges.

Reality bites: The establishment of accountable care organizations in otolaryngology.