Redesigning Behavioral Health

June 1st, 2019

Disclaimer: These are my views based on professional experience and do not necessarily represent the views of my current or former employers.


The Current State of Behavioral Health

As in primary care, behavioral health reimbursement is typically at or below breakeven, which leads to undersupply. Many psychologists, psychiatrists, and other licensed behavioral health professionals do not participate in insurance networks. Commercially insured populations in major cities struggle for months to find in-network behavioral health providers, and network adequacy issues are even more severe for Medicare/caid beneficiaries and those living in rural areas.


Behavioral health is an extremely broad clinical discipline that includes such diverse disorders as substance use/addiction, food/eating, sleep/wake (e.g., narcolepsy), sexual, trauma (e.g., PTSD), obsessive-compulsive, personality, anxiety, bipolar, dissociative, depressive, neurodevelopmental, and neurocognitive. Behavioral health treatments include such diverse therapies as individual psychotherapy, virtual reality, brain stimulation (e.g., ECT, TMS, and VNS), medication, acute residential programs, and inpatient hospitalizations. Behavioral health is intimately connected to social determinants of health (SDOH), such as food, housing, economic opportunity, and social/community issues. While one in four Americans face behavioral health challenges, most don’t receive any professional treatment.


Pockets of behavioral health delivery are somewhat organized (e.g., psychiatric hospitals and substance use centers), but most are not. It’s partially a cottage industry — many behavioral health providers who do not work in educational settings run their own private practice, which creates a highly fragmented ambulatory delivery system. Private providers practice a broad swath of clinical approaches and fewer than 20% measure outcomes. Most behavioral health occurs in isolation from primary care despite a mountain of research demonstrating the importance of primary care and behavioral health integration.


My early thinking suggests that three fundamental changes are required to make significant progress in this area:

  1. Evolve payment to incentivize the ideal behaviors and outcomes. This involves determining what behaviors and outcomes we want from our behavioral health system and developing new payment models that incentivize those results.
  2. Organize delivery with a segment-specific focus. We’ll likely need to organize providers into aggregative structures so they can succeed in new payment models. These groups will also want to focus on specific segments of the market so they can develop deep expertise.
  3. Redesign care delivery to be more patient-centered, high-leverage, and outcome-oriented.


Let’s dive into each of these elements in a bit more detail.

Part 1: Evolve Payment

Determine the desired behaviors and outcomes

We get what we pay for. Currently, we pay for most behavioral health based solely on 1) the service complexity or duration and 2) the clinical training of the individual rendering those services. We do not typically pay for outcomes. The result is a system that encourages many transactions but does not necessarily incentivize higher quality.


There are different approaches to evolving this incentive structure. We could incentivize provider capabilities that we think benefit patient care. For example, we could pay provider groups more if they offer extended visit availability on nights and weekends or offer a 24/7 behavioral health emergency phone line. We could also pay more based on process measures, such as medication adherence or the average time from referral to intake. We could also pay for patient outcomes, such as depression remission rates. Experts can create balanced approaches to measure outcomes and encourage the ideal provider behavior (to the extent it is generalizable) for each clinical condition.



Develop payment models that incentivize those behaviors and outcomes

If we want our behavioral health system to function differently, we need to evolve reimbursement away from transactional, volume-based payments and move toward value-based reimbursement models that facilitate new behaviors and models of care.


There are a variety of value-based structures that would shift care delivery, including episode-based or bundled payments, specialty capitation, and shared savings models. Value-based constructs should remove the constraints of reimbursable procedure codes and financially incentivize provider groups to generate superior outcomes. The first factor is necessary so that delivery groups have stable revenue and freedom to restructure care delivery, and the second factor allows the system to pay for higher quality.



Part 2: Organize Delivery with a Segment-Specific Focus

Aggregate delivery to succeed in new payment models

Our existing behavioral delivery system is ill-suited to execute on outcome-based models. It is difficult to attribute patients, measure outcomes, and evaluate performance across a set of well-intentioned but disaggregated and disorganized behavioral health providers. Since independent providers do not have the capabilities to succeed in these contracts, we will need new aggregative structures.


Different types of entities might help facilitate value-based execution, including employed behavioral groups, behavioral health IPAs, behavioral health marketplace aggregators, and tech-enabled behavioral health MSOs. These groups can enable consistent outcomes measurement using approved methodologies and aggregate enough patients to credibly facilitate outcome-based reimbursement, in both actuarial and practical terms. Ideally these organizations will be nationally scalable, though there are few national success stories in healthcare delivery so far.



Focus on specific market segments to develop deep expertise

In order to succeed in value-based behavioral health contracts, these new delivery groups will need to focus.


The idea of focusing on a specific segment sometimes feels foreign in healthcare, but it’s how most of our economy works. Just as Chipotle does not try to serve the high-end dining market, the Capital Grille is not interested in the fast casual lunch crowd. By focusing on a specific segment, organizations can get really good at serving their customers. This already happens at an individual level; while orthopedic surgeons are trained in the entire musculoskeletal system, most surgeons typically focus on one or two specific regions.


A delivery group can get focus in several ways. Payer type (i.e., commercial self-funded, commercial fully insured, managed Medicaid, Medicare Advantage, managed behavioral healthcare networks) will likely become a key segmentation factor since buying behavior will differ by payer. The clinical issue, or clusters of related clinical issues, will likely serve as another axis of segmentation. While transdiagnostic treatment approaches are increasingly common, delivery groups will need to develop programmatic care pathways. The boundaries will not always become obvious, but a patient with a substance use disorder needs a different pathway than a new mother with postpartum depression or a lonely senior who recently lost her spouse. Many patients will have comorbidities, but evidence-based pathways combined with the freedom to individuate care will help patients get the right type of treatment.


Under this approach, “focused factories” expert at managing specific segments within behavioral health would emerge. For example, a delivery organization might develop a highly effective program for teenagers with anxiety disorders. They might be paid a capitated rate for every patient they serve in this segment and earn bonuses for improvements in GAD-7 scores. They might have an entirely different program with a bundled payment for schizophrenia tied to outcomes bonuses for medication adherence and inpatient readmission rates.


Approaching behavioral health in this way could also lead to new health promotion and disease prevention models, such as a postpartum depression prevention module for high-risk populations. These programs could mirror other successful prevention models, such as the Diabetes Prevention Program (DPP), and create new treatment modalities to truly prevent illness rather than merely manage it more effectively.

Part 3: Redesign Care Delivery

In addition to developing new value-based payment models and organizing provider groups into aggregative structures with a focus on specific segments, we must also redesign the way we deliver care. Our new behavioral healthcare system will likely need to be more patient-centered, high-leverage, and outcome-oriented in order to succeed at scale.



Whole person: A robust understanding of a patient’s lifestyle and SDOH factors, such as community, education, and nutrition, will be fundamental to developing effective care plans.


High access: High-performing behavioral health groups will offer 24/7 access via chat, phone, and video to care for patients’ needs outside of traditional office hours. They may literally meet patients where they are — in the home, hospital, halfway house, etc.


Integration: Successful delivery organizations will figure out how to partner with the broader healthcare system, including primary care, specialty care, and inpatient facilities, as well as social and community-based services. These new behavioral health entities could even embed within existing primary care delivery organizations.


Interdisciplinary teams: Identifying the best resource for each patient is a key challenge and opportunity. For example, not every patient will need a clinical psychologist; some might need coaching instead. Delivery organizations will deploy a broad spectrum of licensed providers, including PhDs, MDs, MFTs, psychiatric NPs, LCSWs/LMSWs, LPCs, and health coaches. In addition, groups will need to design true team-based approaches to ensure that patients are cared for holistically.


One-to-many formats: While not all patients are suited for group or peer-based interventions, these models have proven effective in providing a community for normalization, support, and behavior change. These groups can act as a “third place” for behavioral health, especially for underrepresented populations. In addition to the clinical benefits, group-based models offer panel leverage for behavioral health practitioners by treating multiple patients at once, which reduces costs and increases access.


Self-care: Behavioral health providers will empower patients with tools for self-care where appropriate, such as app-based CBT-I for insomnia. Providers will compete on their ability to help patients form healthy habits that contribute to overall wellbeing (e.g., meditation, mindfulness-based stress reduction, nutrition, exercise). These approaches might be particularly valuable in prevention programs as well.


Data-driven: Providers and payers will collaborate on holistic outcomes measures, including patient-reported outcomes related to clinical symptoms and quality of life, utilization measures such as readmissions, and total cost of care metrics.


Programmatic individualization: Behavioral health providers will develop evidence-based approaches (e.g., step therapy prescribing) that respond to individual patient needs and clinical outcomes. They will adapt their programs in response to new evidence, including their own.

Challenges & Limitations

This conceptual approach to redesigning our behavioral health system will face several crucial challenges.


One hard problem is initiation. Who initiates: the payer or the provider? It would be difficult for a payer to create a new behavioral health reimbursement model if the delivery system is not yet organized to execute on it, but it’s also difficult to build a new delivery system without payment to support it. Codevelopment might work. Behavioral delivery groups will need to reach a critical mass of patients quickly enough to turn early demonstration pilots into durable relationships, as payers will not want to develop and administer custom contract structures with different metrics, incentives, and adjudication rules for a multitude of small partners.


Second, measurement will pose a significant challenge. Attributing patients to providers can be difficult, and attributing results to different interventions is even more challenging. For example, if a patient’s depression improves, how much of that was due to reduced knee pain after a surgical intervention vs. the therapy and medication regimen delivered by the behavioral health provider? Risk adjustment will be another critical but challenging problem for these models, particularly due to the role of non-medical factors (i.e., SDOH) in mental health. While there are new diagnosis codes for SDOH (e.g., inadequate housing), established risk adjustment models have little experience with these factors. The behavioral health market might benefit from a universal risk adjustment standard, just as the Medicare Advantage market uniformly relies on the CMS-HCC risk adjustment model. It’s imperfect, but at least everyone plays by the same rules. Bad actors will always try to game the system, but smart actuaries will work hard to get ahead of these issues.


Finally, regulatory changes might help better support new models of care. The elimination of state-based clinical licensure, or steps to reduce the burden of multistate licensure, might allow expertise to flow across borders and open the door to more efficient telebehavioral solutions. Unified credentialing approaches across payers and sites of care could reduce administrative overhead and add flexibility to care models. In addition, the system may benefit from further clarity on behavioral health parity questions. There are likely other policy and regulatory levers to consider as part of a broad approach to redesigning our behavioral health system.


I am encouraged by the growing interest in improving behavioral health. Visions for the future must be met with an equally realistic grasp of what it will take to develop sustainable care delivery models. As with most delivery problems, the real challenge is not the science of care, but instead exists at the nexus of incentives and capabilities. Progress will only happen through close partnerships between the organizations that own incentives and the organizations that own care delivery. With the right actors working on the problem, I’m confident the market could make meaningful progress within the next decade.