Scott A. Berkowitz, MD, MBA

Little is known about how providers within accountable care organizations (ACOs) perceive ACO utility in early stages of operation. We explored initial perceptions of providers participating in the Johns Hopkins Medicine Alliance for Patients, LLC (JMAP), within 6 months of launching as an ACO.

 The survey was administered to primary care and cardiology providers in JMAP entities, including medical school faculty and physician practices, a regional ambulatory care network, and 3 additional community practices. We assessed their satisfaction with the care provided and perceptions of the ACO model, inter-provider coordination, and interventions likely to facilitate high-quality and cost-effective care within the organization. We compared perceptions by entity type and provider’s clinical area.

About half of invited providers responded, and a considerable proportion (29%-49%) was neutral in rating their comfort, understanding, or expectations about becoming an ACO. Community practice providers were the most positive about becoming an ACO, and faculty providers were the least satisfied with the chronic care they were providing. Cardiologists reported better coordination with hospital teams than did primary care physicians; cardiologists also reported a higher frequency of communication with primary care practices than vice versa. Interventions to facilitate care coordination and provide more comprehensive care, such as use of care coordinators, behavioral therapists, or support staff to facilitate gaps in care, were considered important for success.

Six months into JMAP operations, providers gave ambivalent responses about their understanding of the implications of being an ACO, and subgroups within the ACO held different views. Data pointed to room for improvement in care coordination and care delivery interventions moving forward.

To improve the quality and efficiency of healthcare, the US health system is increasingly shifting from fee-for-service to value-based payment models. An accountable care organization (ACO), through the Medicare Shared Savings Program (MSSP), is a model created under the Affordable Care Act to facilitate this transformation.

 In collaboration with a payer, provider groups voluntarily assume responsibility for quality and costs of care for a defined population of beneficiaries. If quality and spending targets are met, the provider organization may keep a portion of savings achieved. There are currently more than 740 ACOs in the public and private payer health sectors,

 and most of the US population lives in an ACO service area.

 Early results demonstrate potential flattening of the cost-curve, decreased utilization of emergency and postacute care, and improved clinical quality.

Engagement of healthcare providers facilitates success of ACOs.

 Perceptions of providers regarding ACOs in early stages of development have, however, been largely unexplored. Although previous research has identified effective inter-provider communication and shared objectives as essential to ACO performance,

 these dimensions of relational coordination have not been assessed among participants in ACOs at academic medical centers (AMCs). Compared with other settings, the challenges of provider engagement may be different for emerging ACOs at AMCs, which have to combine this new care delivery model with their clinical, research, and educational missions.

We surveyed providers within a new, large AMC ACO for their perceptions of the model, satisfaction with care provided, assessment of inter-provider coordination, and utility of interventions likely to facilitate high-quality and cost-effective care within the organization.

The Johns Hopkins Medicine Alliance for Patients, LLC (JMAP), is a non—risk-bearing MSSP ACO that was launched in January 2014 (

 Primary participating entities include: Johns Hopkins University Clinical Practice Association, comprising School of Medicine faculty and physician practices (hereafter referred to as SOM); Johns Hopkins Community Physicians (JHCP), including primary and specialty care outpatient sites; 2 primary care practices unaffiliated with Johns Hopkins Medicine (JHM); 1 cardiology community practice with a preexisting JHM relationship; and 5 JHM hospitals, which are all in Maryland and Washington, DC. There are approximately 2900 providers and 38,000 Medicare beneficiaries associated with JMAP.

JMAP undertook a multi-pronged communication effort at the program’s outset to inform providers and staff of JMAP goals and expected provider and beneficiary impact.

 Initiatives featured provider- and staff-focused webinars; practice-site “road show” visits; and clinical-, management-, and operations-related committee presentations. In addition, JMAP leadership established 3 regional advisory councils, which include medical and administrative leadership from each clinical site within the JMAP catchment area, to disseminate information and receive feedback on challenges or opportunities.

 To assess provider engagement and determine future interventions of focus, JMAP launched a provider survey in July 2014, approximately 6 months after ACO operations commenced. The survey was intentionally performed at outset of the organization to serve as a baseline of perceptions and also to assist in identifying priorities and strategies for management. At survey administration, the ACO outreach interventions previously described had been deployed. Medical and quality interventions were early in development and included implementation of care coordination in a high-risk subset of the Medicare beneficiary population in Baltimore through a related initiative. Within the ACO, providers or practices initially referred beneficiaries for care coordination support with JMAP. Early in 2015, the process transitioned to also employ a risk-prediction algorithm to identify beneficiaries most likely to benefit from available services. Other ACO interventions implemented in late 2014 included expanded population health functionality in the electronic health record (EHR) system and internal monitoring of ACO quality measures.

We invited all JMAP primary care (internal medicine practitioners, general practitioners, and geriatricians) and cardiology providers by e-mail to complete an online provider survey. The survey was launched in mid-July 2014 and closed in early August. The Johns Hopkins Institutional Review Board and the JMAP Board of Directors reviewed and approved the instrument, which was constructed using both previously validated

 and original items for the purpose of quality improvement (

). Response choices were yes/no or agreement on a 5-point Likert scale, or were open-ended to allow elaboration. The survey included questions in 8 domains, including provider demographics; satisfaction with the care of chronically ill beneficiaries; and coordination, staff support, and resource use in the provider’s group practice.

Another domain asked to which hospital the provider sent beneficiaries most often, how frequently the inpatient and outpatient teams communicated, whether communication was timely, and the extent of shared goals and problem-solving between the hospital and the responding provider—the same dimensions of relational coordination supported in a recent paper by Rundall and colleagues.

 Additional domains investigated coordination among providers, providers’ understanding of the ACO, and how likely they believed certain interventions were to facilitate high-quality and cost-effective care.

The final domain included 2 open-ended questions to elicit more general feedback: “Do you have any ideas of how to improve the quality or efficiency of care provided within the ACO?” and “What are things that are being asked of you day-to-day, that, if removed, would not negatively impact the quality of patient care?”

We compared respondents’ demographic and professional characteristics among ACO subgroups using Fisher’s exact tests for categorical variables and Kruskal-Wallis tests for continuous variables. To assess the differences among subgroups in rating coordination, communication, and perception of ACOs, we constructed ordinal logistic regression models to fit the ordered, discrete nature of the Likert-scaled dependent variables, and we adjusted for the respondent’s age, sex, years in practice, medical specialty, and clinical training. We accounted for clustering of responses from the same ACO subgroup using a fixed effect (dummy variable) for each group. To present how respondents rated interventions that may help achieve the ACO’s goals, we calculated the proportion of responses in the top 2 categories of the 5-point Likert scale—“very likely” and “likely”&mdash;by ACO subgroup. We examined differences among group preferences using binomial tests. All analyses were conducted using Stata software, version 14 (StataCorp LP, College Station, Texas). We report 2-sided 

 values and 95% CIs for statistical tests.

We distributed the survey electronically to 444 primary care and cardiology providers in more than 40 institutional and community practices. A total of 216 responses were collected, including 99 respondents from SOM, 96 from JHCP, and 21 from the 3 additional community practices. The aggregate response rate was 48.6%. Within SOM, JHCP, and the community practices, the response rates were 49%, 48%, and 50%, respectively.

Most providers were physicians (84%), although nurse practitioners and physician assistants represented 5% to 14% of survey respondents within ACO subgroups (

Cardiology providers formed 32% of respondents. Ninety-one percent of respondents were board certified. The average age of respondents was 48.1 years, and duration at current practice ranged from less than 1 year to 42 years (average = 10.6 years).

SOM respondents were more likely to assess frequency of communication with hospitals as high than JHCP providers were (odds ratio [OR], 2.3; 95% CI, 1.2-4.2) (

). Ratings of other dimensions of care coordination with hospitals were not statistically different among the ACO subgroups. Cardiology providers had higher odds than primary care providers of favorably assessing frequency of communication (OR, 2.6; 95% CI, 1.4-4.9), goals alignment (OR, 2.2; 95% CI, 1.2-4.1), and problem-solving communication (OR, 3.2; 95% CI, 1.7-6.2) with hospitals.

Cardiology and primary care providers rated alignment of goals between their respective practices similarly (Table 2). Cardiology providers, however, were more likely to report higher frequency of communication with primary care practices than vice versa (OR, 2.2; 95% CI, 1.1-4.4; 

 = .03). On the other hand, primary care practices were more likely to rate timeliness of communication from cardiology practices favorably than vice versa (OR, 7.9; 95% CI, 3.7-17.0; 

When asked to rate their comfort with, understanding of, or expectations about becoming an ACO, a considerable proportion of respondents (29%-49%) chose the neutral response option (

), indicating neither negative nor positive assessment. Generally, the 3 community practices had the most positive view of the ACO model, followed by JHCP respondents, and then SOM affiliates. SOM affiliates were less likely to be comfortable with ACO-induced changes (vs JHCP: OR, 0.32; 95% CI, 0.17-0.62; vs community practices: OR, 0.20; 95% CI, 0.08-0.52), less likely to indicate understanding of the implications of becoming an ACO (vs JHCP: OR, 0.40: 95% CI, 0.22-0.74; vs community practices: OR, 0.15; 95% CI, 0.06-0.38), less likely to express confidence that the ACO model can improve care (vs JHCP: OR, 0.33; 95% CI, 0.18-0.64; vs community practices: OR, 0.19; 95% CI, 0.07-0.48), and less likely to have informed patients that their practice had joined an ACO (vs JHCP: OR, 0.49; 95% CI, 0.27-0.91; vs community practices: OR, 0.10; 95% CI, 0.04-0.27). JHCP respondents were less likely than their counterparts in other community practices to report understanding the implications of becoming an ACO (OR, 0.37; 95% CI, 0.16-0.89) and less likely to tell their patients about the ACO (OR, 0.20; 95% CI, 0.08-0.52). All subgroups reported little desire for an organizational role in the ACO.

Respondents from all subgroups reported similar levels of satisfaction with communication with patients and families at their practices. SOM respondents, however, were more likely than their counterparts in other subgroups to assess teamwork and satisfaction with coordination of care at their practices as being significantly lower. We did not find significant differences in perceptions of the ACO and satisfaction with patient care by providers’ medical specialties.

Interventions Assessed by Providers to Facilitate ACO Effectiveness 

Eighty-four percent of respondents believed the provision of care coordinators was likely or very likely to facilitate high-quality and cost-effective care within the ACO (

). Other interventions favored by more than 80% of respondents were: improving communication and coordination of care throughout the care continuum (83%), using support staff to facilitate timely communication of test results (83%), and provision of behavioral therapists to address mental health issues (82%). Providing centralized scheduling for primary and specialty care was the only proposed initiative that garnered support from less than half (42%) of respondents.

There was some, but not complete, overlap among interventions supported by the respective ACO subgroups. Among SOM respondents, interventions with the greatest support were provision of behavioral therapists (79%), support staff for timely communication of test results (79%), and home-based monitoring and palliative services (78%). JHCP respondents expressed strong support for provision of care coordinators (92%), improved coordination throughout the continuum of care (87%), and support staff to identify service gaps (85%). Interventions most supported by providers in the 3 community practices were: quick follow-up with discharged patients (95%), support staff for timely communication of test results (95%), provision of behavioral therapists (90%), and improving communication and coordination throughout the continuum of care (90%). Compared with JHCP respondents, a significantly lower proportion of SOM affiliates recommended provision of care coordinators, support staff to identify service gaps, a 24-hour call center, evidence-based clinical practice guidelines within EHRs, or clinical decision support tools as interventions likely to foster ACO effectiveness.

This study provides a comprehensive initial assessment of providers’ perceptions within a new AMC ACO. The findings contribute to the sparse literature on a major “category of influence” on ACO impact: the readiness and perceptions of participating providers to adopt this new delivery and payment model.

 Whereas previous assessments of the early phases of ACO implementation have relied on the views and knowledge of ACO administrators only,

 the ACO model is typically provider-driven, and our study draws responses from front-line clinicians and staff for an early appraisal of perceptions.

Six months in, the providers surveyed were, on average, only slightly more positive than neutral on care provided to chronically ill beneficiaries, practice teamwork, and coordination of care with hospitals or other outpatient facilities. Furthermore, about half of respondents responded neutrally when asked to express their comfort with changes due to the new model. They were uncertain whether the ACO would improve care for their beneficiaries, did not understand the practice-level implications, and typically did not mention it to their patients. Considering this, it is not too surprising that few clinicians expressed interest in an ACO organizational role. Nevertheless, since others have suggested previously that involving providers in ACO leadership and governance is important for improved engagement

 this clearly represents an area of future opportunity. The community practices reported greatest contentment with joining the ACO and were most likely to discuss their ACO involvement with patients. This finding may reflect their interest in both value-based care models and the more intensive investigation process they conducted to determine whether to join JMAP.

Effective engagement of providers under the ACO model is critical to motivating the care transformations necessary to improve outcomes, and it also influences beneficiaries’ acceptance of this model of care.

 In addition to the awareness deficits discussed above, uncertainty among clinical staff may arise from possible alterations to organizational design or clinical systems that accompany transition to a more collaborative, team-based model. For AMCs in particular, the conversion implies that faculty practices may have more opportunity to participate in team-based care, consider standardization of clinical processes, and ensure that access to both primary and specialty care is emphasized.

 Furthermore, advancing population health may involve new collaborations with community practices, public health departments, and community social service organizations.

Innovative ways previously suggested to increase provider engagement include involvement in shared savings distribution schemes, incorporating quality of care into criteria for promotion, and rewarding clinical excellence.

 In response to the survey results, considerable emphasis has been placed on development of comprehensive communication tactics, as well as implementation of a more robust medical and quality strategy to support identified needs. This strategy includes broadening behavioral health support, access to specialty care, and use of population-based pharmacy support.

On average, respondents were slightly satisfied with the quality of care provided to chronically ill beneficiaries at their respective practices. In their open-ended comments, providers cited difficulty accessing EHR information on encounters outside of the ACO network as an impediment to providing better care. Other recurrent concerns were care coordination challenges within the ACO and inadequate time during office visits for proper attention to beneficiaries with complex conditions. When asked about day-to-day activities that detract from care of beneficiaries, providers similarly mentioned EHR inefficiencies, excessive requirements for documentation, and inadequacy of support staff. These represent overarching systems challenges typically not under direct control of the ACO. These challenges, however, must be addressed at the system level with the ACO as a key stakeholder.

To facilitate ACO implementation, providers favored initiatives to improve communication and coordination of care over enhancements to EHRs and scheduling systems. Coordination and adequate communication across the spectrum of providers in an ACO is integral to success in improving care and reducing spending.

 Cooperation among physicians and acute care facilities relies on alignment of incentives across settings.

 It is worth noting that emphasis on care coordination has been in place within Johns Hopkins Medicine’s Maryland hospitals for the last 5 years due to the state’s all-payer rate regulation,

 as well as other ongoing initiatives. Nevertheless, our results suggest that respondents are uncertain about whether the acute care facilities and outpatient providers share the same goals. Of note, SOM affiliates rated coordination with hospitals higher than did other subgroups. This may reflect the fact that SOM providers are more likely to practice in hospital-based clinics and to have concomitant inpatient responsibilities.

An important limitation of our study is that we conducted a cross-sectional assessment of perceptions at a single ACO without a comparison group or pre-intervention counterfactual. A recent taxonomy of ACO structure

 raises the question of whether our results are generalizable to other ACO types. We had a reasonable response rate of 49%, but we did not collect information on nonrespondents in our anonymous survey. It may be that providers who did not participate would have reported lower levels of satisfaction with care and engagement, or that those with greater concerns may have viewed the survey instrument as an opportunity to voice overarching concerns. It is also unclear whether providers’ comments in completing the survey reflect a true understanding of specific ACO goals or interventions, as these may be difficult to extricate from pre-existing programs. This general awareness may evolve over time, and it would be interesting to repeat the survey at a future date.

The early experience of JMAP, an AMC ACO, supports the notion that even for a relatively integrated system with experience in performance-based payment, engagement is an incremental process.

 Limited physician engagement can be a barrier to achieving goals as an ACO develops, and significant effort needs to be devoted to educating providers on ACO goals and providing opportunities for input. Case studies of emerging ACOs at varying stages of implementation are warranted to educate providers, administrators, and policy makers, and to advance dissemination of innovations successful in enhancing physician engagement and inter-provider coordination under this new, but promising, model of delivering care. 

Department of Health Policy and Management (JAM, OAF, YJH) and Department of Epidemiology (JHY), Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; Division of Cardiology, Department of Medicine (SAB) and Office of Physicians (Accountable Care) (SAB), Johns Hopkins University School of Medicine (JHY, MM, SAB), Baltimore, MD; Johns Hopkins Community Physicians (MM), Baltimore, MD; Welch Center for Prevention, Epidemiology, and Clinical Research, Johns Hopkins Medical Institutions (JAM, JHY), Baltimore, MD; Johns Hopkins Health Care, LLC (JAM, JHY), Baltimore, MD; Johns Hopkins Medicine Alliance for Patients, LLC (SAB), Baltimore, MD; Signature Partners, Inova Health System (MGP), Falls Church, VA.

 Dr Berkowitz is the executive director and a member of the Board for JMAP. Dr Poffenroth is the former medical director for JMAP. 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. JMAP is affiliated with Johns Hopkins Medicine. The Bloomberg School of Public Health is part of Johns Hopkins University.

Concept and design (SAB, JAM, MM, MGP, JHY); acquisition of data (SAB, JAM, MM, MGP, JHY); analysis and interpretation of data (SAB, OAF, YJH, JAM, MGP, JHY); drafting of the manuscript (SAB, OAF, YJH, JAM, MGP, JHY); critical revision of the manuscript for important intellectual content (SAB, OAF, JAM, MGP, JHY); statistical analysis (OAF, YJH); provision of study materials or patients (JAM, JHY); administrative, technical, or logistic support (OAF, JAM, MM, JHY); and supervision (SAB, JAM, JHY).

Jill A. Marsteller, PhD, MPP, Associate Professor, Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, 624 N Broadway, Baltimore, MD 21205. E-mail: jmarste2@jhu.edu.

1. Berwick DM. Making good on ACOs’ promise—the final rule for the Medicare shared savings program. 

. 2011;365(19):1753-1756. doi: 10.1056/NEJMp1111671.

2. Fisher ES, Shortell SM, Kreindler SA, Van Citters AD, Larson BK. A framework for evaluating the formation, implementation, and performance of accountable care organizations. 

. 2012;31(11):2368-2378. doi: 10.1377/hlthaff.2012.0544.

3. Muhlestein D. Growth and dispersion of accountable care organizations in 2015. 

 website. http://healthaffairs.org/blog/2015/03/31/growth-and-dispersion-of-accountable-care-organizations-in-2015-2/. Published March 31, 2015. Accessed July 6, 2016.

4. Lewis VA, Colla CH, Carluzzo KL, Kler SE, Fisher ES. Accountable care organizations in the United States: market and demographic factors associated with formation. 

. 2013;48(6, pt 1):1840-1858. doi: 10.1111/1475-6773.12102.

5. Nyweide DJ, Lee W, Cuerdon TT, et al. Association of Pioneer accountable care organizations vs traditional Medicare fee for service with spending, utilization, and patient experience. 

. 2015;313(21):2152-2161. doi: 10.1001/jama.2015.4930.

6. Pham HH, Cohen M, Conway PH. The Pioneer accountable care organization model: improving quality and lowering costs. 

. 2014;312(16):1635-1636. doi: 10.1001/jama.2014.13109.

7. Larson BK, Van Citters AD, Kreindler SA, et al. Insights from transformations under way at four Brookings-Dartmouth accountable care organization pilot sites. 

. 2012;31(11):2395-2406. doi: 10.1377/hlthaff.2011.1219.

Articles

The authors’ survey of providers in a new accountable care organization reports that initial perceptions of this care model are ambivalent and vary among participating practices.

Early Provider Perspectives Within an Accountable Care Organization

To determine if Medicare Shared Savings Program (MSSP) accountable care organizations (ACOs) are meeting public reporting requirements related to shared savings plans, to quantitate the composition of shared savings distribution plans, and to investigate whether early ACO success is associated with specific plan or ACO characteristics.

ACO descriptive characteristics and distribution plan details were abstracted from official ACO websites for all 338 active MSSP ACOs launched through January 2014. Publicly available MSSP results from 2012 and 2013 start date ACOs were used to investigate associations with successful shared savings generation.

 Of current MSSP ACOs, 313 of 338 (93%) maintain a website, 284 of 338 (84%) provided at least a general statement about shared savings distributions, and 176 of 338 (52%) reported detailed allocation percentages to ACO participants. On average, ACOs reporting detailed allocations planned to give 63% (range = 0%-100%; SD = 26.3) to their primary care providers (PCPs), specialists, and/or hospitals, and 33% (range = 0%-100%; SD = 25.6) to infrastructure. ACOs including a hospital planned to give a larger average percentage to participating entities than those without (69% vs 58%; 

 = .01). ACOs planning to give >50% to their PCPs and specialists were more likely to have generated savings (

 = .001), as were ACOs composed of >10 participating entities (

Just over one-half of MSSP ACOs report detailed shared savings distribution plans online, and these plans vary widely. There appears to be no single shared savings distribution plan determinate of ACO success. Continued investigation of predictors for generating savings is needed to inform future shared savings models.

Analysis of publicly reported organizational characteristics, shared savings distribution plans, and early financial success of accountable care organizations (ACOs) in the Medicare Shared Savings Program (MSSP) demonstrated that:

		Most ACOs are meeting their public reporting requirements related to shared savings distribution plans.

		ACOs vary widely in how they plan to distribute shared savings.

		In a preliminary investigation using early MSSP ACO results, ACOs that planned to allocate the majority of shared savings to their primary care providers and specialists or were composed of more individual participating entities (ie, clinicians, clinician group practices, hospital, and/or other participating providers) were more likely to have generated savings.

More than 440 Medicare accountable care organizations (ACOs) have been established nationwide since the Affordable Care Act was passed in 2010.

Medicare ACOs in the Medicare Shared Savings Program (MSSP) and Pioneer ACO programs voluntarily contract with Medicare to be responsible for the health outcomes and expenditures of a defined patient population. ACOs aim to promote higher-quality care and to reduce the growth of healthcare costs through improved care coordination, care management programs, information technology, and other interventions,

 and when ACOs succeed in these goals, they may share in the cost savings with Medicare. These potential “shared savings” are 1 important driver for ACO formation.

Of MSSP ACOs that initiated operations in 2012 or 2013, 118 (54%) lowered expenditures compared with benchmark projections. In total, these ACOs generated $383 million in net savings for Medicare during their first performance year, with 52 ACOs earning shared savings payments of more than $315 million.

 ACOs (and ACO-like models) continue to form, including in the commercial insurance sector and in state-based Medicaid programs.

 Nevertheless, significant questions remain about the financial viability of the ACO model.

Specific regulatory requirements governing the MSSP provide an opportunity to examine important questions about what ACOs plan to do with their shared savings (ie, their “shared savings distribution plans”). The MSSP requires ACOs to initially submit their shared savings plan in their application to the MSSP program, and later to publicly report their distribution plans on required ACO websites.

Guidance suggests that MSSP ACOs should also describe the percentage of total shared savings that will be allocated to infrastructure, primary care providers (PCPs), specialists, and hospitals in the ACO. Importantly, Medicare ACOs have flexibility in how they use shared savings, as long as use is consistent with the ACO program mission

; shared savings payments to beneficiaries, however, are not permitted.

Whether and how ACOs are meeting these policy requirements is unknown but of significant interest to all stakeholders,

 and patients may be interested in whether savings benefit them directly (eg, via new care programs and interventions) or indirectly (eg, via incentives to clinicians and hospitals to improve care quality).

Although savings distribution to ACO PCPs and specialists is considered a key component of ACO success, how much and on what basis ACOs will do so is unspecified and unknown.

Finally, ACO leaders and policy makers may be interested in what ACOs are doing regarding shared savings, because savings distribution plans likely reflect ACO strategic planning and what ACOs believe will help to incentivize desired care patterns.

The objective of this study was to begin describing the landscape of how MSSP ACOs publicly report and plan to use their shared savings. We also conducted a preliminary investigation on whether certain ACO organizational characteristics appear to be associated with certain intended uses of shared savings. Finally, using early results from the MSSP, we examined whether certain organizational characteristics or shared savings distribution plan design elements are associated with ACO financial success.

We examined the official ACO websites for all 338 active MSSP ACOs initiated in April and July 2012, January 2013, and January 2014. At the time of our study, 5 ACOs had been approved to participate, but were not included in the CMS database of active ACOs (these were not included in this analysis).

 Website addresses were found using information available on the CMS website,

 but because the CMS website did not include website addresses for the January 2014 start date, we located these on the Internet. Among 338 ACOs, 322 had an available official website for review; all 16 ACOs without a website locatable by routine search methods were in the January 2014 group. Pioneer ACOs were excluded from analysis.

We designed a data abstraction strategy based upon information expected from MSSP public disclosure requirements

and our study objectives, such as key ACO descriptive characteristics and composition, public reporting, and proposed shared savings distribution plan features (eg, percentage of distribution to infrastructure). A data abstraction form covering these domains is not publicly available to our knowledge; therefore, we developed one to capture desired data. We piloted a draft form on a subset (n = 30) of ACOs from the July 2012 group, then met and revised the form into a final version (

). This version was used to collect data on all MSSP ACOs from June 3 through July 23, 2014.

Some variables collected during data abstraction warrant explication. We hypothesized that organizational complexity might affect shared savings plan characteristics or ACO success, because larger, more complex organizations may complicate how an ACO forms and operates. As a proxy of complexity, we collected data on the number of individual entities (ie, clinicians, clinician group practices, hospital, and/or other participating providers) participating in each ACO. We refer to this as the number of “participating entities” in each ACO. To illustrate, by our definition, an integrated health system that forms an ACO within its existing organizational structure has 1 participating entity, whereas an ACO that forms from a private practitioner, a group practice, a multi-specialty group, and a hospital—all of whom were previously unaffiliated as de-fined by ACO rules&mdash;would have 4 participating entities. Our use of participating entity is generally consistent with the CMS regulatory definition of an ACO participant.

We also hypothesized that ACO composition might affect shared savings plan characteristics or success. Although ACO regulations refer to ACO “providers,”15 we identified whether an ACO included hospitals, PCPs, and specialists from information provided on official ACO websites. Whether an individual or group practice in an ACO is primary care or specialty may not be evident by name alone; therefore, we independently confirmed these practices in all cases by reviewing their websites. Moreover, because an ACO composed of a single participating entity (such as an integrated health system) might include multiple clinician groups (such as PCPs, specialists, and hospitals) relevant for shared savings plan characteristics, we elucidated the composition of these ACOs in similar fashion.

We also believed it was important to assess whether involvement of an external stakeholder might affect shared savings distribution plans. We defined “external stakeholder involvement” as present when an external stakeholder investor or company representative (not apparently otherwise affiliated with an ACO participating entity) was present on the governing board, or when the distribution plan included distributions to investors/shareholders.

Shared savings distribution plans were analyzed to determine if the composition of ACOs or the presence of external stakeholders were associated with differences in distribution plans. Once early MSSP financial performance results became available, we examined the 58 ACOs that generated savings to determine whether certain ACO characteristics or shared savings distribution plans were associated with success in generating shared savings.

Study data were collected and managed using REDCap version 5,

 and statistical analysis was conducted using STATA version 11 (StataCorp, College Station, Texas). For comparisons of categorical variables, a χ

 test was used, and for comparisons of means, an unpaired 

We found that 313 (93%) of the MSSP ACOs have information regarding their composition publicly available on their website. Of these ACOs, 131 (42%) have more than 20 participating entities, while 42 (13%) are composed of only 1 (

). More specifically, 140 (45%) included PCPs, specialists, and hospitals, 124 (40%) had PCPs and specialists, while 49 (16%) were composed entirely of primary care physicians. Those meeting our definition of involving external stakeholders numbered 67 (21%).

In rare instances, other types of clinicians were noted as being in the ACO. These included chiropractors, podiatrists, hospice, clinical laboratories, home care, pharmacies, dentists, physical therapists, optometrists, senior living centers, social services, and others. Among all ACOs, 49 (14%) included an academic medical center (AMC).

Public Reporting of Shared Savings Distribution Plans 

We found that 285 (84%) ACOs publicly reported their shared savings distribution plan in various levels of detail. Of these 285, 81 (28%) had “to be determined” (TBD) listed under their savings distribution plan, 28 (10%) had general statements regarding how distributions would be allocated (eg, “to incentivize physicians and build infra-structure”) but without specific percentages, and 176 (62%) had detailed plans with specific percentages of the savings allocated to distinct categories (

Among 176 MSSP ACOs with detailed plans available, 155 (88%) reported savings allotments for infrastructure (mean amount = 33%; range = 0%-100%; SD = 25.6; median = 25%); 166 (94%) reported savings allotments to their PCPs, specialists, and/or hospitals (mean amount = 63%; range = 0%-100%; SD = 26.3; median = 70%); and 29 (16%) reported savings apportionments for “other” reasons, such as investor allocations or payments to “strategic partners” or a “specialty advisory committee” (mean amount = 4%; range = 0%-50%; SD = 11.7; median = 0%) (

). Of note, 25 (14%) ACOs with detailed plans available reported planning to use shared savings to cover any expenses before apportioning savings to any participants (Figure 1).

In 166 cases, shared savings allocations were broken down into amounts for hospitals, PCPs, and specialists. Of the 77 ACOs including a hospital, 42 (55%) reported a specific percent designation to the hospital (mean amount = 9%; range = 0%-60%; SD = 15.0; median = 0%). Of the 166 ACOs reporting allocations to participating entities 137 (83%) designated a specific percent for clinicians, including PCPs and specialists (mean amount = 58%; range = 12.5%-100%; SD = 20.3; median = 59%). Of these 137, 95 (69%) reported specific allocations for PCPs (mean amount = 49%; range = 6.25%-100%; SD = 21.2; median = 50%) and specialists (mean amount = 11%; range = 0%-48%; SD = 12.5; median = 6%).

Finally, 46 plans (16% of plans with distribution plans online) explicitly referenced patients’ potential to benefit from shared savings. Furthermore, 16 (6%) referenced specific new programs that would be started to benefit patients and funded by shared savings. Examples included hiring case managers or launching educational sessions for patients.

Shared Savings Distribution Plans by ACO Type 

Comparing the ACOs composed solely of PCPs to the ACOs with PCPs and specialists showed little difference in the distribution to each category (Figure 2). However, ACOs that included hospitals gave a larger average percentage to PCPs, specialists, and/or hospitals within the ACO compared with those without (69% vs 58%; 

 = .01).Compared with all active ACOs, ACOs with external stakeholders had a higher percentage of plans that were TBD (55% vs 18%; 

 <.001) and a lower percentage of plans with specific percentage allocations listed (36% vs 62%; 

 <.001). In addition, compared with the 152 ACOs without external stakeholders, ACOs with external stakeholders planned to give a smaller average percentage to PCPs, specialists, and/ or hospitals within the ACO (53% vs 65%; 

Shared Savings Distribution Plans and Financial Performance

No differences in the ability to generate savings were found between ACOs based on the inclusion of a hospital, AMC, or external stakeholder in the ACO.

We also calculated the proportion of ACOs that generated savings based on numbers of participating entities within an ACO (

). ACOs with 6 to 10 participating entities were the least likely to have generated savings (13%), while ACOs with more than 20 participating entities were most likely to have generated savings (37%).

No association was found between the percentage of shared savings allocated to PCPs, specialists, and/ or hospitals within the ACO versus infrastructure and the ability to generate savings. However, we found that ACOs that planned to distribute greater than 50% of shared savings to PCPs and specialists were associated with a higher probability of generating savings compared with those that did not (39% vs 22%; 

 = .001). Similarly, we found that ACOs that planned to distribute greater than 60% of shared savings to PCPs were more likely to have generated savings compared with those that did not (53% vs 24%; 

Our study describes the landscape of MSSP ACO shared savings distribution plans with a focus on public reporting, plan details, and the association between ACO organizational or distribution plan characteristics and ACO early financial success in generating shared savings.

Analysis of publicly reported organizational characteristics, shared savings distribution plans, and early financial success of accountable care organizations in the Medicare Shared Savings Program.

Scott A. Berkowitz, MD, MBA

Articles

Early Provider Perspectives Within an Accountable Care Organization

Medicare Shared Savings Program: Public Reporting and Shared Savings Distributions