Recognizing and Seizing the Opportunities That Value-Based Payment Models Offer Behavioral Health Care

Since the passage of the Affordable Care Act, investments in accountable care organizations (ACOs) and other value-based payment (VBP) models have accelerated markedly, seeking to address all three aspects of the “triple aim” for the U.S. health care system: bending the cost curve, improving health outcomes, and improving patient experience of care. Given the poor outcomes and the high cost of care for individuals with mental health and substance use disorders (collectively termed behavioral health disorders), it is imperative that health systems entering VBP contracts prioritize these populations. However, multiple barriers have become apparent. Many early ACOs did not adequately prioritize behavioral health, and those that did often replicated the existing fragmentation between general medical and behavioral health (1). ACOs that have prioritized behavioral health have identified several financial and logistical barriers, including difficulties sustaining funding for innovative models of care as well as challenges coordinating care because of behavioral health carve-outs in ACO contracts (2). Although some health systems have shown promising cost and quality improvements within an ACO model, largely through the expansion of integrated care and other innovative models of care delivery (3), the overall effect of early ACO programs on behavioral health costs and outcomes has been limited (4). Finally, whereas patients with mental health and substance use disorders are disproportionately disadvantaged in society, it has been noted that existing VBP or pay-for-performance programs disproportionately penalize health systems serving populations of low socioeconomic status, potentially worsening preexisting health disparities caused by unmet social needs (5, 6).

Despite these challenges, a shift from fee-for-service to a VBP model has potential to improve outcomes for those with behavioral health conditions. With greater use of VBP nationally, it is imperative that health systems find economically sustainable ways to operate in these models. This is an especially pressing issue for public and safety-net hospitals, because public payers are simultaneously trying to broaden access and contain escalating costs through VBP models (7). The experience of health systems that were early adopters of VBP models can thus be valuable for other health systems facing decreased reimbursement rates and increased costs and demand for care.

Characteristics of an Early Adopter Health System and Medicaid ACO Contract

Our organization is an integrated public safety-net community health system with a historic focus on primary care and behavioral health. Our operations are highly dependent on public payers, and roughly two-thirds of the patient population are either publicly insured or uninsured. The population served is disproportionately non-White, poor, and more likely to have a primary language other than English compared with the rest of the state’s population. The clinical population also has a high rate of mental health and substance use disorders, and efforts to improve care for these groups have been a focus of past payment reform efforts.

Our system was one of the first health systems in the state to adopt an accountable care model under Massachusetts’s 1115 Medicaid waiver and has adapted in a number of ways. After reviewing the options for VBP contracts under state regulations, our health system chose to partner with a local Medicaid managed care plan to carry out the contract and share governance of the ACO with the health plan. Financially, the ACO receives a risk-adjusted capitated payment from the state Medicaid program for all its members for the given year and shares risk with the health plan. The health system and health plan share savings and losses equally, and if gains or losses exceed a preset percentage of the global budget, the state Medicaid program will also share in them. Contractually, our ACO bears full risk of health care costs and does not carve out management of behavioral health to another entity, which makes it unique among area ACOs. This decision was made because of both the known disadvantages of carve-outs, including care fragmentation, conflicting goals, and increased administrative costs, and the health system’s strong commitment to behavioral health services.

An integral part of VBP models is quality measurement. Previous behavioral health quality metrics were extremely limited (early Medicare ACOs measured only “screening for depression”) and were therefore unlikely to produce meaningful changes in care (4). The system’s ACO contract includes a more robust set of quality metrics aimed at measurably improving behavioral health care. Some of these, such as follow-up plans after depression screening and depression remission or response, expand on earlier metrics by using a validated measurement instrument. Others, including hospital readmissions, follow-up within 7 days of discharge from a behavioral health hospitalization, and rates of emergency department (ED) visits for patients with behavioral health conditions, focus on utilization. The contract also includes process and health screening metrics, such as metabolic monitoring for patients on antipsychotics, initiation and engagement of substance use disorder treatment, and engagement with community partners who provide care management and coordination designed to promote community-based stabilization. Finally, the amount of funding tied to quality metrics will increase from 5% to 20% of total reimbursement over the life of the contract, thus increasing incentives for the system to meet quality benchmarks.

More recently, the state has modified its risk adjustment model used to better account for social determinants of health. From the beginning, the state Medicaid program attempted to adjust for broad indicators of social need as well as for preexisting health conditions. More recently, the state has increased capitated payments for individuals with substance use disorder or severe mental illness, for unstably housed individuals with behavioral health conditions, and for youths with severe medical or behavioral health conditions. The effects of these changes are not yet known but are welcome because of the high cost of providing high-quality care for individuals with significant social needs and severe behavioral health conditions.

Initial Approach to Value-Based Care

In response to the requirements of the ACO contract, our health system developed specialized initiatives to address the needs of high-need, high-cost populations as well as broader interventions to address more common problems. The system is investing in services for populations with three high-risk, high-cost conditions: serious mental illness, substance use disorders, and severe emotional disturbances of children and adolescents. To address the needs of these populations, the ACO has supported services not traditionally reimbursable by insurance, including intensive case management for those who use the ED frequently, peer recovery coaches for individuals with substance use disorders, and a “reverse integration” behavioral health home for those with psychosis (8). Rather than restricting access to inpatient care, the ACO works to coordinate care transitions and engage with individuals with a history of repeated readmissions in order to achieve longer periods of stability in the community. A small team of seasoned clinicians consults with treatment teams and facilitates access to a range of state-funded and community-based services. Preliminary internal analyses suggest that these interventions have reduced inpatient psychiatry readmissions and ED visits for high-risk patients with behavioral health conditions.

To improve the care of patients who have common conditions, the system introduced collaborative care and office-based substance use disorder treatment across its primary care clinics. Risk scores, based on past utilization and built into electronic health records (EHRs), inform determinations about level of care. Individuals deemed to be at lower risk are initially managed using less resource-intensive methods, such as group visits, indirect consultation in primary care, telepsychiatry, or brief time-limited interventions with integrated behavioral health staff. Individuals requiring more intensive levels of care are referred to specialty psychiatry outpatient treatment or community-based care, depending on the clinical assessment and the patient’s historical treatment use patterns. Shared care plans in the EHR serve to integrate treatment decisions across the continuum of care. Preferred provider agreements define how community-based providers will communicate with primary care teams and encourage the use of electronic communication where possible. In addition, claims data and new population-focused utilization tools enable us to track trends in cost and utilization and to provide timely feedback to providers.

Challenges to Implementation and Future Directions

Our health system faced several initial challenges to implementation. First, although our system has a large outpatient psychiatry department, it does not provide the full continuum of services necessary to manage the population at the appropriate level of care, to provide hospital diversion, or to offer timely access to less intensive care to step patients down from inpatient treatment. This gap has been especially evident in acute substance use treatment services (detoxification, step-down care) as well as in intermediate levels of care and has contributed to readmissions to acute care. This limitation also has created difficulty for patients to access care and follow up on referrals and has made it difficult for providers to communicate efficiently with external providers in the absence of a shared EHR. Additionally, although our ACO could offer a limited supply of inpatient psychiatric beds, demand has continued to exceed supply. Second, our system has continued to struggle with existing fragmentation among the broader health care delivery system, community-based services, and government agencies, creating challenges in coordinating care and containing costs. Third, as in many community mental health settings, workforce recruitment and retention has continued to be a challenge because of the high acuity and social needs of the patient population and the attendant stress on clinicians. Finally, the high degree of poverty, societal marginalization, linguistic diversity, and both acute and chronic trauma faced by the population we serve at times has made the effective engagement in and provision of behavioral health care challenging.

Other health systems may also experience the challenge of controlling costs for substance use disorders relative to other mental health conditions. Our health system placed significant initial emphasis on managing care of individuals with mental health conditions who did not have a primary substance abuse diagnosis and has had relative success in lowering costs and shifting avoidable care from inpatient to diversionary and step-down programs for this population. However, utilization management has proved more challenging for those with primary substance use disorders, for whom acute service utilization and costs have remained stubbornly high, particularly among homeless individuals. Additionally, although the opioid overdose crisis has received significant nationwide attention, acute service utilization in our system has been largely driven by those with alcohol use disorders. Whereas funding and treatment for opioid use disorders remains critical, there has been comparatively less attention to and investment in prevention and treatment of alcohol use disorders, which has made linkage to community supports more challenging. Health plan benefits have been expanded to include residential treatment and recovery coaching, but continued creative collaboration with local public and nonprofit agencies will be crucial, as will policy changes to better treat and ultimately prevent substance use disorders.

In response to these initial challenges, the health system is investing in new services, lowering thresholds to access existing services, and attempting to bridge links with community providers. For instance, the system has developed an EHR-based order set for substance use disorder to improve the consistency of treatment and patient safety. Our system is also developing closer relationships with external, preferred medically monitored detoxification and residential treatment services. Where needed resources do not exist, the system is evaluating business opportunities to create or contract for services to fill gaps. Finally, given the high degree of poverty and deprivation in our population and the key role these social determinants of health play in determining health outcomes, all patients are screened for social determinants of health, and patients are referred for case management support to address identified needs.

Conclusions

Implementation of VBP models will be tested by preexisting shortcomings of the behavioral health system and the impact of social determinants of health on our populations. Fragmentation in the behavioral health delivery system and the limited electronic capabilities of community-based providers remain challenges to better integration of care, and broader social pressures historically outside the health system’s control place financial strain on ACO models. Nonetheless, significant opportunities remain for capitalizing on the opportunities of VBP models, given the pivotal role that behavioral health plays in overall health outcomes. These experiments could be viable even in historically underfunded public and safety-net hospitals. Further progress will depend on collaboration between health systems, policy makers, community-based providers, and service users, both on issues directly relevant to health care payment and to the social support and policy changes that will affect the long-term viability of VBP models.