Paula H. Song, PhD

 To describe the extent and implications of “churn” between different Medicaid eligibility classifications in a pediatric population: (1) aged, blind, and disabled (ABD) Medicaid eligibility, determined by disability status and family income; and (2) Healthy Start Medicaid eligibility, determined by family income alone.

 As a result of a 2013 policy change, children with ABD eligibility transitioned from fee-for-service to capitated care. We used Ohio Medicaid claims data from July 2013 through June 2015 to explore the relationships among instability in eligibility category, demographics, and utilization.

 To examine the potential financial effect of categorical churn, an effective capitation rate was created to capture the proportion of the maximum potential capitation rate that was realized.

 More than 20% of children exited ABD-based eligibility at least once. Switching was associated with younger age and rural residence and was not associated with healthcare use.

 Switching between eligibility categories is common and affects average capitation but not health service use.

Analyses of Ohio Medicaid claims data for children aged 3 to 16 years from 2013 to 2015 reveal that instability among eligibility categories is common and affects average capitation but not health service use.

The potential financial impact of churn should be considered when designing alternative payment models (APMs) to target high-cost populations. 

In addition, the impact of churn should be considered when setting rates because the risk associated with revenue uncertainty may deter APM organizations from taking on risk-bearing contracts.

Policy changes that enhance categorical stability (eg, reduced frequency of redetermination) could reduce churn without adding payment system complexity.

In most states, children who qualify for Supplemental Security Income (SSI) under the Social Security Administration’s definition of disabled children also automatically qualify for Medicaid under the aged, blind, and disabled (ABD) eligibility category. Because of the complex needs of these children, state Medicaid programs have traditionally retained them in the fee-for-service (FFS) system rather than mandating them into managed care, accountable care organizations (ACOs), or other alternative payment models (APMs).

 Increasingly, states are rethinking this choice.

 There is a growing belief that the benefits of APMs may outweigh the perceived risks regarding inherent limitations on service access and provider networks.

 Rather than exclude Medicaid-eligible disabled children from APMs, states are instead experimenting with modified payment approaches, including differential experience-based capitation rates, that allow APMs to provide this patient population with healthcare services, care coordination, and social supports that meet their higher needs. Changes in Medicaid enrollment categories due to changes in children’s health conditions and their families’ financial conditions can, however, complicate these efforts. For example, data from Ohio—a state that transitioned children from FFS to an APM&mdash;show that in a 2-year period, more than 20% of children moved between higher and lower capitation rates, potentially blunting the impact of differential capitation (

In this case study, we explore the extent and impact of “churn” between different Medicaid eligibility classifications for enrollment reasons on the financial exposure of organizations that use APM models, drawing on the experience of Partners for Kids (PFK), an Ohio Medicaid ACO that serves approximately 330,000 Medicaid-covered children in 34 of Ohio’s 88 counties. PFK began enrolling children in Medicaid’s ABD eligibility classification as part of a policy change that ended the ABD managed care exclusion, effective July 2013.

In the time frame studied, July 2013 through June 2015, Ohio operated as a 209(b) state, which allowed the eligibility criteria for Medicaid ABD to be more restrictive than for SSI.

 Specifically, ABD eligibility required both an SSI determination and a family income level of less than 64% of the federal poverty level (FPL). This linking of disability status and income selects a very high-risk population, given research documenting that family poverty is associated with worse health status and higher health-related financial burden, even after controlling for insurance status.

On a monthly basis, changes in household income could change categorical eligibility from ABD to Healthy Start, Ohio’s implementation of the State Children’s Health Insurance Program, which expands Medicaid to cover children 18 years or younger with family incomes up to 206% of the FPL. Changes in eligibility category generated changes in capitation, with the per-month experience-based capitation rate for ABD children being nearly 5 times the experience-based capitation rate for children who qualify for Healthy Start. The changes in capitation rates did not, however, change the services for which enrollees were eligible. Whether covered under ABD or Healthy Start, children had access to the full range of Medicaid-covered services and were eligible for care coordination appropriate for their clinical needs.

Analyses used 2 years of Medicaid administrative claims data (July 1, 2013, to June 30, 2015). The study cohort included 7262 children aged 3 to 16 years who were enrolled in PFK as ABD in July 2013. Children younger than 3 years were excluded because they did not have sufficient claims history to calculate their medical complexity status. Those older than 16 years were excluded so that the full sample would be age eligible for Medicaid throughout the follow-up period. Enrollees whose initial entry to ABD occurred after July 2013 were also excluded from the analysis in order to focus on transitions over 2 years.

Categorical churn was defined as switching between ABD eligibility classification (SSI disabled 

 <64% FPL) and Healthy Start classification (≤206% FPL). Demographic correlates of churn were tested with a bivariate χ

 test. Medical complexity was calculated using the Pediatric Medical Complexity Algorithm, which classified enrollees’ conditions as complex chronic, noncomplex chronic, or noncomplex/nonchronic.

 In order to determine whether children affected by categorical churn had differential use of costly services, we tested the association between churn and the probabilities of emergency department (ED) use and inpatient use.

To examine the potential financial effect of categorical churn, we created an effective capitation rate (ECR) that represents the proportion of the maximum potential capitation rate that was realized. The ECR was calculated as:

We used a capitation adjustment of 0.21 to approximate PFK’s 4.79-fold differential between the capitation rates during the study period. Using this formula, a child covered for a full year as ABD would have an ECR of 1.0, meaning that PFK would receive 100% of the potential ABD capitation rate for that child. Alternatively, if that child experienced a 2-month switch to Healthy Start eligibility, the ECR would be 0.868 ([10 + (0.21 × 2)] / 12) and PFK would realize only 86.8% of the potential ABD capitation rate.

The mean (SD) age of the study cohort at the start of the observation period was 10.1 (3.8) years. Demographics are presented in 

By June 2015, 28% (n = 2052) of the cohort (N = 7262) were no longer in ABD coverage. Of those, 67% (n = 1377) remained in PFK but had switched to Healthy Start Medicaid and 33% (n = 675) were no longer in PFK. The latter category includes children who were still in Medicaid but no longer in the PFK coverage region and children who were no longer covered by Medicaid.

Overall, 22.3% of the cohort switched between ABD and Healthy Start at least once (mean, 1.17 switches; range, 1-5). Rural residence and younger age were both significantly associated with churn. Churn was consistent across all medical complexity levels (Table 1).

Churn was not associated with “any ED visit” or “any inpatient admission” in the 2-year follow-up period (

); however, in children whose medical conditions were classified as noncomplex chronic, there was a marginally higher probability of ED and inpatient use among those with at least 1 eligibility change.

The mean ECR was 0.925, reflecting that the realized financial capitation was 7.5 percentage points lower than it would have been without categorical churn. Examining the distribution, we see that, overall, 8.2% of children had an ECR between 80% and 99%, 5.2% between 60% and 79%, and 8.7% below 60%. This distribution was similar across levels of medical complexity (

). Lower ECR was not associated with lower probability or lower count of ED visits or inpatient admissions (results not presented).

This case study demonstrates that categorical eligibility churn is extensive. Movement out of (or back into) ABD status was not associated with ED or inpatient use in the full sample, indicating that children who spend time in a non-ABD eligibility status are no less likely to use intensive services than their continuously ABD-enrolled peers. Utilization rates were at least as high and, in some cases, marginally higher in those who experienced churn compared with those with uninterrupted ABD eligibility status. This was most noticeable in children with noncomplex chronic diseases, including conditions such as asthma and attention-deficit/hyperactivity disorder, for which disease management and care coordination strategies could yield significant improvement in care.

This relationship may reflect a heterogeneous relationship with the type of conditions included in the noncomplex chronic category. Noncomplex chronic conditions involving single-system diseases like asthma and diabetes may have health impacts that are more episodic in nature. Exacerbations or periods of better disease control for these conditions may be associated with family income reductions or improvements, respectively, and could thus spur eligibility changes, linking churn and healthcare use. In the other categories, the health status may be more stable (consistently higher in the complex category and consistently lower in the noncomplex nonchronic category) and, thus, changes in income would be less likely to be associated with changes in healthcare use.

Month-to-month eligibility category changes create, in effect, a blended capitation level and an environment in which the funding realized for this high-need population is somewhat less than the capitation rate set based on the experience. Findings, therefore, suggest that APMs taking on new high-cost populations should consider the potential financial impact of churn when designing targeted care models for these populations.

These results represent a single state and a unique period in which the state’s Medicaid program was implementing multiple process and system changes simultaneously with the studied policy change. As such, the results are not generalizable, but they do inform the conversation on the extension of managed care into traditionally carved-out pediatric populations. They suggest that rate setting should include consideration of the impact of categorical churn because the risk associated with revenue uncertainty may deter APM organizations from taking on risk-bearing contracts or, worse, encourage “cherry-picking” in enrollment, as was seen in the early days of Medicaid managed care programs.

 Beyond capitation, innovative payment models like bundled payments for noncapitated settings or complexity bonuses can also be implemented to limit the effect of eligibility instability. Alternatively, policy changes that enhance categorical stability (eg, reduced frequency of redetermination) could reduce churn without adding payment system complexity.

In the case of the population discussed in this study, the state Medicaid agency incorporated program design elements from the start to mitigate the potential dangers of enrolling the disabled population in Medicaid managed care. The state began risk adjusting capitation rates among the managed care entities 1 year after the children enrolled in ABD were mandated to enroll into PFK, allowing sufficient time for differences in the membership complexity to surface. The state’s 3 geographic contracting regions each contain 1 of the 3 largest cities in addition to rural counties, ensuring that APMs do not avoid rural populations.

Not only did the state Medicaid program attempt to address churn at the start, it transitioned from the 209(b) waiver program to section 1634 regulations in 2016. Under section 1634, ABD Medicaid eligibility is directly tied to disability and income level, without the need to calculate spend-down impacts. This administrative simplification has stabilized and expanded the ABD membership, thereby decreasing churn. Subsequent revisions of the program’s qualification hierarchy have also been enacted to reduce churn among membership.

Several limitations should be mentioned. As noted previously, these results are representative of a single organization in a single state, so they cannot be generalized. The period in which the ABD population policy change occurred overlapped with changes in the state’s implementation of a new Medicaid eligibility system and associated processes. These changes may have contributed to instability. We also note that detailed cost data for children in managed care are not currently available in the state claims data available for our analysis, so our results are limited to utilization. Analysis using cost data would allow for a more detailed consideration of the cost implications of instability. Finally, our analysis examined only the cohort of ABD-eligible children who rolled into the ACO at the onset of the policy change. We did not study the complementary population who started in the Healthy Start Medicaid program and migrated to the ABD program or ABD enrollees who were new to Medicaid after the definition of the cohort. As such, these results are only representative of the experiences of the defined cohort. The full utilization and financial implications may differ. This larger-scale research is underway.

As states continue to migrate traditionally carved-out pediatric populations, including children in foster care or juvenile detention, into APMs, it is valuable to understand potentially unexpected challenges faced by those that have blazed the trail. This case study gives one example regarding the importance of considering categorical instability and offers insight into the potential planning challenges in building fixed clinical support infrastructure for an “unfixed” population.

 Department of Pediatrics, The Ohio State University College of Medicine (DJC, KJK), Columbus, OH; Partners for Kids, Nationwide Children’s Hospital (SPG), Columbus, OH; Department of Health Policy and Management, The Gillings School of Global Public Health (MED, PHS), and The Cecil G. Sheps Center for Health Services Research (MED, PHS), University of North Carolina at Chapel Hill, Chapel Hill, NC; The Research Institute at Nationwide Children’s Hospital (EA), Columbus, OH; Division of Health Services Management and Policy, College of Public Health, The Ohio State University (WYX), Columbus, OH.

 This research was funded through Patient-Centered Outcomes Research Institute (PCORI) award #IHS-1310-07863. The views presented are solely the responsibility of the authors and do not necessarily represent the views of PCORI, its Board of Governors, or its Methodology Committee.

 Drs Chisolm and Kelleher are employed by an academic medical center that is a partner in the accountable care organization (ACO) whose data are described, and Dr Gleeson is employed by that ACO. They receive no direct benefits from the findings. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (DJC, SPG, KJK, WYX, PHS); acquisition of data (DJC, SPG, KJK); analysis and interpretation of data (DJC, SPG, EA, PHS); drafting of the manuscript (DJC, KJK, MED, PHS); critical revision of the manuscript for important intellectual content (DJC, SPG, KJK, MED, EA, WYX, PHS); statistical analysis (EA); obtaining funding (DJC, KJK, WYX, PHS); administrative, technical, or logistic support (KJK, MED); and supervision (DJC, KJK).

 Deena J. Chisolm, PhD, Department of Pediatrics, The Ohio State University College of Medicine, 700 Children’s Drive, RM FB3322, Columbus, OH 43205. Email: Deena.Chisolm@NationwideChildrens.org.

1. Lloyd PC, Simon AE, Parker JD. Characteristics of children in Medicaid managed care and Medicaid fee-for-service, 2003-2005. 

2. Kleinman R, Kehn M, Wishon Siegwarth A, Brown J. State strategies for coordinating Medicaid and housing services. 

. 2017;40(2):225-232. doi: 10.1037/prj0000238.

3. Hall JP, Kurth NK, Chapman SL, Shireman TI. Medicaid managed care: issues for beneficiaries with disabilities. 

. 2015;8(1):130-135. doi: 10.1016/j.dhjo.2014.08.010.

4. Mitchell JM, Gaskin DJ, Kozma C. Health supervision visits among SSI-eligible children in the D.C. Medicaid program: a comparison of enrollees in fee-for-service and partially capitated managed care. 

. 2008;45(2):198-214. doi: 10.5034/inquiryjrnl_45.02.198.

5. SI 01715.010 Medicaid and the Supplemental Security Income (SSI) program: policy for states and state choices. Social Security website. policy.ssa.gov/poms.nsf/lnx/0501715010. Updated October 2, 2017. Accessed July 10, 2017.

6. Sommers BD, McMurtry CL, Blendon RJ, Benson JM, Sayde JM. Beyond health insurance: remaining disparities in US health care in the post-ACA era. 

. 2017;95(1):43-69. doi: 10.1111/1468-0009.12245.

7. Okoro CA, Zhao G, Fox JB, Eke PI, Greenlund KJ, Town M. Surveillance for health care access and health services use, adults aged 18-64 years — Behavioral Risk Factor Surveillance System, United States, 2014. 

. 2017;66(7):1-42. doi: 10.15585/mmwr.ss6607a1.

8. Parish SL, Rose RA, Dababnah S, Yoo J, Cassiman SA. State-level income inequality and family burden of U.S. families raising children with special health care needs. 

. 2012;74(3):399-407. doi: 10.1016/j.socscimed.2011.10.035.

9. Simon TD, Cawthon ML, Stanford S, et al; Center of Excellence on Quality of Care Measures for Children With Complex Needs (COE4CCN) Medical Complexity Working Group. Pediatric medical complexity algorithm: a new method to stratify children by medical complexity. 

. 2014;133(6):e1647-e1654. doi: 10.1542/peds.2013-3875.

Articles

Analyses of Ohio Medicaid claims data from 2013 to 2015 reveal that instability among eligibility categories is common and affects average capitation but not health service use.

Implications of Eligibility Category Churn for Pediatric Payment in Medicaid

To explore accountable care organizations (ACOs) as they develop in the private sector, including their motivation for development, perspectives from consumers regarding these emerging ACOs, and the critical success factors associated with ACO development.

Comprehensive organizational case studies of 4 full-risk private sector ACOs that included in-person interviews with providers and administrators and focus groups with local consumers.

Sixty-eight key informant interviews conducted during site visits, supplemented by document collection and telephone interviews, and 5 focus groups were held with 52 consumers associated with the study ACOs.  

We found 3 main motivators for private sector ACO development: 1) opportunity to improve quality and efficiency, 2) potential to improve population health, and 3) belief that payment reform is inevitable. With respect to consumer perspectives, consumers were unaware they received care from an ACO. From the perspectives of ACO stakeholders, these ACOs noted that they prefer to focus on patients’ relationships with providers and typically do not emphasize the ACO name or entity. Critical success factors for private sector ACO development included provider engagement, strategic buy-in, prior experience managing risk, IT infrastructure, and leadership, all meant to shift the culture to a focus on value instead of volume. 

These organizations perceived that pursuing an accountable care strategy allowed them to respond to policy changes anticipated to impact the way healthcare is delivered and reimbursed. Increased understanding of factors that have been important for more mature private sector ACOs may help other healthcare organizations as they strive to enhance value and advance in their ACO journeys.

Improving the understanding of factors that have been important for more mature accountable care organizations (ACOs) can help organizations that are developing and continuing to evolve as ACOs. Our study is among the first to explore private sector ACO development, and we found the following: 

		An adequate information technology infrastructure enabling timely access to meaningful data was essential for ACOs to manage and monitor quality and outcomes. 

		Consumers generally lacked knowledge of ACOs and were unaware they were part of an ACO. 

		Physician engagement and strong organizational leadership to shift culture from a volume focus to a value focus were crucial to ACO development.

The introduction of the accountable care organization (ACO) as a form of healthcare delivery and payment system reform supported by the Affordable Care Act is hoped to improve quality of care and reduce healthcare costs by aligning the incentives of physicians, hospitals, and other clinicians and healthcare organizations.

 the ACO label has been defined as “groups of providers who are willing and able to take responsibility for improving the overall health status, care efficiency, and healthcare experience for a defined population.”

Although early adopters of the ACO model commonly operated under the CMS Medicare Shared Savings Program (MSSP) or the Pioneer ACO Program,

 recent market entrants have included ACOs in the private market.

 These private sector ACOs have goals similar to those of the MSSP and Pioneer models in that the contracts with payers include both cost and quality targets; however, they have more flexibility with respect to contract terms relative to Medicare ACOs.

 For example, private sector ACOs are able to develop multiple contracts with multiple payers to extend coverage across broader, commercially insured populations and those including children. In addition, private sector ACOs can have very different quality reporting requirements and financial relationships with, and incentives for, providers—within the boundaries of antitrust considerations&mdash;relative to Medicare ACOs.

Early reports have suggested that provider groups decide to become ACOs for a variety of reasons, including the desire to save money and improve care,

 and due to a sense of inevitability about the direction of health reform.

 However, few studies to date have focused on private sector ACOs

—particularly those that bear downside risk&mdash;and none have asked both ACO stakeholders and consumers affiliated with these organizations about their perspectives. Because private sector ACOs are not bound by the federal regulations and reporting requirements of Medicare ACOs, they have the potential to develop quickly and for different reasons and in different environments than Medicare ACOs. We undertook this study to improve our understanding of ACO development and implementation in the private sector, which is one of the most rapidly growing sectors of the ACO market.

 Our research objective was to explore the rationale for private sector ACO development, the perspectives consumers had about these emerging ACOs, and the critical success factors interviewees associated with private sector ACO development.

We conducted 4 comprehensive case studies of private sector ACOs that assumed full risk from different geographic regions of the United States. Unlike most studies that focus only on the perspectives of ACO providers and leaders, we also focused on the consumer perspective. We elected to use a multiple case study design

 because of the exploratory nature of our study and our desire to learn from multiple stakeholders to improve our understanding of a relatively new organizational form.

We selected ACOs for study that both operated in the private sector and assumed financial risk for substantial portions of the populations they served. Organizations had to self-identify as an ACO and, consistent with common definitions of ACOs,

 they had to involve groups or networks of providers that assume responsibility for the cost and quality of care for defined populations, including some level of downside financial risk. We also maximized the variability of our sample to include ACOs that differed along several important dimensions, including structure (ie, physician-owned vs physician-hospital organizations), size, geography, and population (ie, pediatric vs general). Summary information about our 4 ACOs is included in 

We collected data through site visits during the spring and summer of 2013. Across the 4 sites, we held a total of 68 in-person or telephone key informant interviews. In addition, we held 5 focus groups comprising 52 consumers (

). Our data collection process included a concomitant assessment of interview and focus group transcripts and discussion of preliminary findings to permit probing for new concepts and to ensure that we reached saturation in data collection consistent with standards for rigorous qualitative research.

Site visits and key informant interviews.

 During 2- to 3-day visits, we conducted semi-structured interviews with ACO stakeholders from 4 main areas: 1) organizational leaders (eg, chief executive officer [CEO], chief operations officer, chief marketing officer); 2) administrators and staff members who could provide insight about the ACO decision making and development process (eg, informants from business development, strategic planning, finance, contracting, marketing, external affairs, information systems, quality and performance improvement); 3) administrators, staff members, and clinicians involved in ACO operations (eg, ACO administrator, contracts manager, practice manager, medical director); and 4) physicians affiliated with the ACO.

 To ensure consistency in our data collection, we used 2 standard interview guides: 1 tailored for providers and 1 for administrators and staff. The interview guides consisted of multiple-question domains, including: 1) History and Context; 2) ACO Implementation; 3) Consumer Involvement; and 4) Critical Success Factors. These interview guides were tested in a pilot study of a different ACO to improve question clarity and refine the guides prior to our study site visits. Most interviews were conducted in person and lasted 30 to 60 minutes. All interviews were recorded and transcribed verbatim. 

Each focus group was conducted using a standard guide that included open-ended questions exploring consumers’ perspectives about ACOs and their own healthcare. A study investigator moderated each of the focus groups, with an additional investigator present as a co-moderator. One focus group was conducted at 3 of the ACO sites and 2 were conducted at the fourth site to capture consumer perspectives from both the urban and rural markets this ACO served. Each focus group comprised 7 to 16 participants who lived in the ACO market and interacted with the ACO as patients or parents of patients of ACO providers. Sessions lasted 90 minutes and participants were given a $50 gift card as a token of appreciation for their time. All sessions were recorded and transcribed verbatim.

 For each participating ACO, we also requested key documents related to ACO development. Examples of these documents included the ACO strategic plan, communication vehicles, ACO progress reports, board presentations, and annual reports. 

Our analytic approach included using both inductive and deductive methods to analyze the 1158 pages of transcripts from our interviews and focus groups in an iterative process. After each site visit, the research team first compiled a “Site Visit Summary” that included summary impressions. This initial analysis was helpful for conceptualizing findings and themes and provided insight into more focused inquiry.

Our second analytic step involved coding the interview transcripts to break this qualitative data into smaller, meaningful segments for analysis.

 We developed an initial set of codes based on our interview guides and emergent themes from the site visit summaries; we then summarized these codes in a “coding dictionary” that included detailed definitions to ensure consistent application of the codes.

 This preliminary coding dictionary was further refined as themes and patterns emerged in the data. Three members of the research team then coded the data and, throughout the coding process, conducted cross-checks of the coded data to ensure that codes were consistently applied.

 Data were coded and managed using a qualitative software program, Atlas.ti (Scientific Software Development GmbH; Berlin, Germany).

The third step of analysis we conducted for this paper involved examining the codes of “motivations” and “critical success factors” to inform our understanding of private sector ACO development. We used an inductive approach to identify themes and patterns in the coded data, first focusing on patterns within cases (ie, study sites) and then evaluating whether these were replicated across cases, paying particular attention to any differences in the contrasting cases (eg, in pediatric ACOs vs others).

 At each stage of this coding and analysis process, we also sought and considered disconfirming evidence

 to ensure the robustness of our findings. 

When asked about their motivation for developing a private sector ACO, interviewees across the study sites noted 3 reasons, explained here and with supporting evidence presented in 

The first was to lower costs while improving care quality. As one ACO executive summarized, “We came together with a fundamental mission of improving cost of care and improving quality of care.” Second was to improve coordination of care and population health management. This motivation was acknowledged even in the face of understanding that all incentives were not aligned. As one interviewee explained, “You can’t see either the hospital or the health plan or the physician group or whoever the players are as adversaries, even though your incentives might not always be completely aligned. You’ve got to find a way, a common ground to do what’s right for the patient.” The third common motivator had to do with a sense of inevitability given the direction of health reform. One interviewee reflected, “I think if you polled hospital CEOs, 90% of them are going to say, ‘We are on the path to an accountable care organization.’ …there is now a general belief and acceptance in that this is the right thing to do in that whether you’re an insurance company, a provider, a hospital, you are moving in this direction.” Although individual interviewees at the sites noted other reasons, such as the desire to consolidate market power so that the organization could “leverage the best payments” or to be “a market leader” in this new model of care, these comments did not reflect the collective sentiment across interviewees and institutions. 

Consumers’ Perspectives about ACO Development

Across the focus groups, none of the consumer participants was familiar with the terms “accountable care organization” or “ACO,” nor were any individuals aware they were part of an ACO. This became apparent during each session when no participants provided answers to the questions, “Prior to coming here [to this focus group session], did you know what an Accountable Care Organization was?” and “Did you know that [this organization] was an ACO?” As a result, participants’ answers to open-ended questions instead reflected their experience with their own physicians or a named hospital and not the ACO entity. Consumers’ lack of familiarity with the ACO model had been predicted by our sites’ key informants, who consistently indicated that their ACO outreach efforts purposely did not include branding the ACO to consumers, nor promoting the ACO as something new and different. 

Critical Success Factors for Private Sector ACO Development

When asked what interviewees believed to be critical success factors for development and implementation of an ACO model in the private sector, responses could be classified into 5 main categories: 1) physician involvement, 2) strategic buy-in to the ACO vision, 3) information technology (IT) infrastructure and data, 4) experience and understanding, and 5) leadership. These factors are explained below, with additional supporting evidence provided in 

Provider buy-in and engagement were mentioned often and across all 4 sites, and were perceived to be the key factors limiting whether the ACO model could be implemented. As one senior executive explained, “The providers actually have to buy into it…Because if they don’t, you can provide reports and have meetings and do everything that you want to, but if the providers haven’t actually bought into it, it’s an uphill battle.” Engagement and involvement of physicians in ACO governance were also important, with one interviewee summarizing this factor as “absolutely bringing physicians to every point in the table.”

 Strategic vision and buy-in at the organizational level (ie, moving beyond provider-level buy-in) involved commitment from and across the organization rather than only individuals (providers) engaging with the ACO concept. One senior executive reflected, “I think you really need that strategic buy-in … the fact that the healthcare system needs to take some responsibility to figuring out how to provide more effective, affordable, and efficient care that still meets the needs of the patients.” Interviewees at other sites made similar comments about the importance of buy-in to help ensure support for required investments and the effort required to support cultural changes for the organization. 

 IT was commonly recognized as critical. As summarized by one executive, “Well, an IT infrastructure is key. You can’t control what you can’t measure. You need to define your quality and utilization parameters. … You’ve got to build the structure right. You’ve got to have adequate data from your health plans so that you can figure out what you need to figure out.” Beyond infrastructure, interviewees also recognized the need for the data that the systems supported. As one ACO director noted, “You can have physicians who are bought into it and are engaged in it, but they need to know how they’re doing, and they want to know how they’re doing, right?” 

 At all 4 sites, interviewees explicitly noted that past experience with risk was contributing to their successful implementation of the ACO model. One director commented, “What we find is that with the experience that we’ve had previously, we’re sort of anticipating what’s coming down the road. And I think that’s positioning as well, as we move forward. Otherwise, you’re spending a lot of time scrambling.” This understanding was also valued at another site, as an interviewee explained, “We felt that we could take risk for a population that we knew very well and that we could manage the health of that population within the fiscal constraints.”

 Comments emphasizing leadership as a success factor reflected the importance of the leadership function overall, as well as the need for leadership of the cultural changes required to implement the ACO model. One manager commented, “The cultural shift is pretty huge to move from fee-for-service. And that whole mindset and getting people to really be able to embrace that.” Leadership was noted to be especially important as the ACO model evolved. As one executive explained, “And even when everybody’s on board it’s tough, because there are winners and losers, and people’s roles are going to change, how they’re perceived to perform, their incentives will change. It’s a tough cultural change even when everybody’s on board. So if there’s any hesitation by an important part of the organization, it’s going to be tough.” 

The ACO concept has spread beyond the Medicare ACO initiatives, as provider and payer organizations in the Medicaid and private markets are developing ACOs to serve these populations. Lacking clearly defined models for private sector ACOs, however, payers and providers in these spaces are left to develop their own models. Furthermore, because most research to date has focused on ACOs developed in the MSSP and Pioneer ACO programs, there is little information about how and why ACOs are developing to serve other populations. In this paper, we have begun addressing this gap by providing early insights from the development and implementation of 4 ACOs serving privately insured and Medicaid populations. 

Although private markets and populations may vary from those served by the MSSP and Pioneer programs, our findings suggest that many of the motivators for developing private sector ACOs are similar. The need to control the rising costs of care while simultaneously improving the quality of care, the need to develop capabilities to manage population health, and a sense of the inevitability of healthcare reform were the primary motivations for ACO development espoused by our interviewees. In a sense, providers want to “do the right thing,” and they perceive the ACO model as providing an opportunity to do that. This mix of motivators suggests both push and pull forces at work. The sites we studied are moving away from legacy approaches to healthcare financing and delivery, including fee-for-service and fragmented, siloed care, which they feel are no longer sustainable nor desirable. At the same time, these organizations are moving toward new contractual and organizational arrangements that they perceive both as inevitable and as allowing them to improve the health of populations.

 In short, healthcare reform efforts to better align incentives with quality and population health goals for traditional Medicare populations appear to be affecting transformations more broadly. 

Our findings that consumers lacked knowledge of ACOs and were unaware that they were part of an ACO are notable, especially given growing emphasis on the engagement of patients in their own care.

 Providers and payers involved in developing ACOs are investing considerable time, money, and effort to transform the healthcare system, and yet the parties with arguably the greatest stake in that transformation are neither informed of nor empowered to shape these efforts. Given the linkage between patient behaviors and many health outcomes, we expect there are numerous opportunities to involve consumers in ways that will improve the abilities of ACOs to accomplish their goals. For example, there may be opportunities to provide incentives, including financial ones, to engage patients or to provide physicians training in patient engagement.

 Other opportunities may be to include patient representatives in governance, as required in the MSSP and Pioneer ACO programs, but not required in any private sector ACO model of which we are aware. 

Our findings suggest the importance of having strong leadership and physician engagement as part of ACO development, thus highlighting the importance of physician engagement reported in prior studies.

 Moving beyond physician engagement, however, all of the ACOs in our study had physicians in formal leadership positions and 3 of the ACOs had strong physician representation on their governing boards. The ACOs in our study included employed and community physicians in various aspects of decision making, ranging from compensation and incentive committees to quality improvement initiatives. Our findings about the importance of physician leadership are consistent with a recent national survey of public and private ACOs, which found that more than half of all ACOs are physician-led and that physicians make up the majority of governing boards in over 75% of the ACOs surveyed.

 Strong physician leadership could also help facilitate high levels of communication across providers, thereby improving patient care.

These findings also underscore the need for adequate IT infrastructure for ACO development. Policies, such as Meaningful Use, helped motivate providers to implement electronic health records, which is a necessary first step. However, ACOs need the ability to access timely and meaningful data to monitor and improve outcomes and efficiency. The ACOs we studied are making significant capital investments in data and reporting capabilities; meanwhile, the ACOs that are unable to make the necessary capital investments in IT infrastructure will likely lag in terms of development. The Medicare Advance Payment Model attempts to address this capital need for smaller or rural Medicare ACOs; however, there do not appear to be similar arrangements in the private sector ACO markets.

 Rather, private sector ACOs either rely on capital infusions from physician groups or health system partners or turn to the credit markets. The limited options for capital support may hinder private sector ACO development, particularly for newer and smaller private sector ACOs. 

The ACOs in this study had all been in risk arrangements well ahead of the emergence of the term “ACO” and had considerable experience (and success) in managing risk. Thus, they were able to reorganize relatively quickly as ACOs and had the confidence that they could take on the associated risk and responsibility successfully. New ACOs may face more challenges ramping up and may need to enter the private sector ACO market with lower-risk arrangements, such as shared savings, as they work to gain experience with contracts and population health management.

One important limitation of this study was the small number of organizations involved. Given the time and energy constraints of qualitative studies, there are significant barriers to large-scale studies. Future work can include the development of surveys based on this research to explore and validate our findings in large samples. However, the ACOs included in our study are relatively mature, and their experience and perspectives provide important insights for newer ACOs that are considering entering the private market and assuming downside risk. Another possible limitation is that the ACOs in our study are likely to have experienced substantial changes from the time of our initial visits, given the rapid development of the ACO market. 

The private sector ACO market is developing rapidly. Recent estimates suggest that approximately half of the 600 ACOs identified in the United States are private sector ACOs.

 Given this rapid growth, future research should include larger-scale studies to explore the structural and financial arrangements in private sector ACOs that facilitate their development. Future studies should also focus on the performance and sustainability of the ACO model in this market. To date, approximately one-third of the ACOs participating in the Medicare Pioneer ACO program have withdrawn

; however, far less is known about exits within the private sector. 

Private sector accountable care organization development has been motivated by perceived opportunities to improve quality, efficiency, and population health, and the belief that payment reform is inevitable.

Paula H. Song, PhD

Articles

Implications of Eligibility Category Churn for Pediatric Payment in Medicaid

Private Sector Accountable Care Organization Development: A Qualitative Study