Mental Health Care in the Accountable Care Organization

Organized group medical practices are not new. However, recent interest in the ACO comes from ACA provisions that provide for:“[A] shared savings program . . . that promotes accountability for a patient population and coordinates items and services under [Medicare] Parts A and B, and encourages investment in infrastructure and redesigned care processes for high quality and efficient service delivery. . . . (A) groups of providers . . . may work together to manage and coordinate care for Medicare [fee-for-service] beneficiaries through an [ACO]. . . ; and (B) ACOs that meet quality performance standards . . . are eligible to receive payments for shared savings.”

A key point in these provisions is “promot[ing] accountability,” registering a shift from traditional fee-for-service care to models in which overall quality and cost influence revenue. Although a variety of provider-supplier combinations may participate, each must have a mechanism for shared governance and accept responsibility for at least 5,000 Medicare beneficiaries (3).

The first ACO component is the beneficiary population. To define this, CMS reviews all Part B claims to identify Medicare beneficiaries seen by ACO-participating providers. If at least 10% of these beneficiaries’ evaluation and management services are with ACO-participating primary care providers, they are aligned to the ACO. Using the aligned beneficiary pool, CMS then generates the ACO’s annual financial benchmark, an estimate of what Medicare Parts A and B fee-for-service expenditures would have been for the aligned population in the absence of the ACO. An aligned beneficiary has the freedom to visit any health care provider, but costs incurred will count toward the respective ACO’s annual benchmark. Finally, the ACO must report on and meet 33 quality measures spanning the following domains: patient and caregiver experience, care coordination and patient safety, preventive health, and at-risk population with specific chronic general medical conditions.

The two basic CMS ACO models are based on the components described above. In the Medicare Shared Savings Program (MSSP), Medicare continues to pay each ACO through a fee-for-service arrangement, but actual annual expenditures are compared with the projected benchmark. An MSSP organization that meets the 33 quality measures at a cost below its financial benchmark receives up to 50% of the savings from CMS as revenue. In providing care that is both high quality and cost-efficient, the ACO generates revenue by saving the payer (CMS) money.

The second model is the Pioneer ACO, designed for larger organizations (≥15,000 aligned beneficiaries) with more advanced infrastructure and a willingness to accept financial risk for a potentially higher proportion of savings. As with MSSP, a Pioneer ACO generates revenue by meeting quality and cost targets, but it can share up to 60% of the savings. In exchange, the Pioneer ACO also accepts financial risk: if costs exceed the annual benchmark, the organization must reimburse CMS up to 60% of the excess. This is in contrast to the traditional fee-for-service model, in which higher patient expenses simply generate more revenue for an organization. In year 3, Pioneer ACOs that meet the first two years’ cost and quality targets are eligible to move from fee for service to a prospective payment model, receiving 50% of anticipated expenditures per beneficiary, per month. This source of revenue is intended to allow ACOs the flexibility to provide services not covered under fee-for-service arrangements and invest in infrastructure to support care coordination.

Mental health conditions are among the most expensive as primary disorders (4) and, when comorbid with general medical disorders, are associated with increased costs for the primary general medical disorder (5). The cohort of older adults with mental illness is expected to increase from under eight million in 2010 to 15 million in 2030 for several reasons, including the aging of baby boomers, their higher rates of depression and anxiety, and onset of late-life psychiatric disorders in the expanding aged population (6). Despite this growing burden of mental illness and its cost implications, current ACO disease-specific quality and cost efforts are focused almost entirely on chronic general medical conditions. The one exception—depression screening with a documented follow-up plan—may have minimal impact on actual care.

Service delivery models such as collaborative care management consistently demonstrate the ability to improve care for both mental illness and comorbid general medical illness across a variety of conditions and treatment settings (7). However, these models are difficult to implement in traditional fee-for-service settings, where reducing acute care episodes decreases revenue. The ACO model, through its shared savings mechanism, provides the potential for revenue from cost savings that could justify investment in collaborative care management. However, without the addition of mental health quality measures that would be required to share savings, an ACO’s incentive to implement these evidence-based models is based solely on the potential to generate cost savings, for which evidence is limited (7).

Therefore, the current ACO rationale to engage patients with mental illness is to generate savings relative to the financial benchmark by proactively managing its most complicated and costly patients. This high-cost, high-risk cohort will likely overrepresent persons with mental illness, with potential savings realized only by providing preventive and coordinated care to limit unnecessary hospitalizations and overuse or duplication of services. To maximize savings, this would include efforts to engage these patients with ACO-affiliated providers, allowing the ACO more control over both the services and their cost.

One cohort of potentially high-cost patients includes those with serious mental illness, such as bipolar or psychotic disorders, at high risk for fragmented, inappropriate care. Hospitalization for these patients is expensive and associated with increased adverse events (8), and it increases across the lifespan (9). These patients’ mental health needs would likely be best managed in specialty mental health settings with consultant medical care managers proactively addressing general medical conditions (10). At risk for poor self-management of general medical and mental illnesses, these patients could benefit from ACO efforts promoting improved care transitions and provider communication.

Another potentially high-cost, high-risk cohort includes patients with general medical disorders and comorbid mental disorders, such as depression or anxiety. Their medical costs are significantly higher than those without mental disorders (5), but these Medicare beneficiaries seek care in primary care settings and are less likely to seek or be referred to specialty mental health care (11). There are effective models for mental health care management in primary care settings (12). However, because the ACO is not measured on the implementation of its depression or anxiety care, the only incentive to use these evidence-based models would be if they can have an impact on the ACO’s annual financial benchmark.

The first issue is whether ACO patient alignment includes patients with mental illness. It is unclear whether alignment based on primary care services will capture patients with mental illness, especially those with serious mental illness, or whether adverse selection will occur in these models (13). In addition, it is unclear to what extent mental health providers are participating, because the type and number of associated specialists are not specified by CMS.

Regardless of alignment, the nature and quality of mental health care in ACOs should be examined. Previous population-based demonstrations have shown improvement for chronic general medical conditions. The CMS Physician Group Practice Demonstration of population-based ACO precursors demonstrated success in improving quality and restraining cost, in particular for the dually eligible population, which was likely related to improved overall chronic care management (14). It could then be hypothesized that there also would have been improved care of chronic mental illness, but no mental health–specific analyses have been conducted. However, Massachusetts’ Alternative Quality Contract (AQC), wherein a commercial payer utilized global payments incorporating quality measures with several large provider organizations, demonstrated overall improvement in chronic care with the notable exception of depression (15). To at least begin the examination of mental health care delivery in ACOs, it will be important to understand whether or how organizations vary on the depression screening measure and whether this has an impact on care downstream. Based on the AQC experience, in which depression care was unchanged despite the presence of depression treatment quality measures, it is unlikely that the single ACO depression screening measure will have any impact on treatment.

Mental health conditions need to be examined for their impact not only as primary disorders but also for their impact on quality of care for comorbid general medical conditions. High-quality diabetes care, for example, is an explicit goal that has quality measures included for ACO beneficiaries; if the overall quality of diabetes care improves in an ACO, the improvements should include those with comorbid mental illness. Although improving mental health care is not an explicit ACO goal, part of the overall evaluation of medical care should focus on vulnerable populations, such as persons with mental illness.

The findings noted above should then be considered in light of how ACO organizational structure and setting influence care, because organizations exert significant influence on the manner in which clinicians practice and the processes and outcomes of patient care (16). As ACO implementation proceeds, the nature and quality of mental health services will be influenced by features of organization context, structure, and process. Although some of these may not be modifiable (geography and academic affiliation), aspects of structure (specialty and provider mix and use of incentives) or process (communication and collaboration) may be both important and potentially modifiable, with relevance to future policy.

It is essential to understand how the ongoing ACO experiment affects both the general medical and mental health care of Medicare beneficiaries with mental illness. This shift toward population-based care also gives mental health providers the opportunity to more broadly consider how population-based care should look for their patients. This new population orientation is a critical matter for the profession, because the workforce of geriatric mental health specialists is inadequate to meet the increasing demand. Optimal patient care as well as financial success will require that care delivery is optimized for patients with mental illness.

Dr. Marcus has received grant support from Ortho-McNeil Janssen and Forest Research Institute and has served as a consultant to AstraZeneca. The other authors report no competing interests.