Financial sustainability of novel delivery models in behavioral health treatment

Abstract

Background:

In the US, much of the research into new intervention and delivery models for behavioral health care is funded by research institutes and foundations, typically through grants to develop and test the new interventions. The original grant funding is typically time-limited. This implies that eventually communities, clinicians, and others must find resources to replace the grant funding – otherwise the innovation will not be adopted. Diffusion is challenged by the continued dominance in the US of fee-for-service reimbursement, especially for behavioral health care.

Aims:

To understand the financial challenges to disseminating innovative behavioral health delivery models posed by fee-for-service reimbursement, and to explore alternative payment models that promise to accelerate adoption by better addressing need for flexibility and sustainability.

Methods:

We review US experience with three specific novel delivery models that emerged in recent years. The models are: collaborative care model for depression (CoCM), outpatient based opioid treatment (OBOT), and the certified community behavioral health clinic (CCBHC) model. These examples were selected as illustrating some common themes and some different issues affecting diffusion. For each model, we discuss its core components; evidence on its effectiveness and cost-effectiveness; how its dissemination was funded; how providers are paid; and what has been the uptake so far.

Results:

The collaborative care model has existed for longest, but has been slow to disseminate, due in part to a lack of billing codes for key components until recently. The OBOT model faced that problem, and also (until recently) a regulatory requirement requiring physicians to obtain federal waivers in order to prescribe buprenorphine. Similarly, the CCBHC model includes previously nonbillable services, but it appears to be diffusing more successfully than some other innovations, due in part to the approach taken by funders.

Discussion:

A common challenge for all three models has been their inclusion of services that were not (initially) reimbursable in a fee-for-service system. However, even establishing new procedure codes may not be enough to give providers the flexibility needed to implement these models, unless payers also implement alternative payment models.

Implications for Health Care Provision and Use:

For providers who receive time-limited grant funding to implement these novel delivery models, one key lesson is the need to start early on planning how services will be sustained after the grant ends.

Implications for Health Policy:

For research funders (e.g., federal agencies), it is clearly important to speed up the process of obtaining coverage for each novel delivery model, including the development of new billable service codes, and to plan for this as early as possible. Funders also need to collaborate with providers early in the grant period on sustainability planning for the post-grant environment. For payers, a key lesson is the need to fold novel models into stable existing funding streams such as Medicaid and commercial insurance coverage, rather than leaving them at the mercy of revolving time-limited grants, and to provide pathways for contracting for innovations under new payment models.

Implications for further research:

For researchers, a key recommendation would be to pay greater attention to the payment environment when designing new delivery models and interventions.

1. Introduction

A notable feature of behavioral health care is the pace of innovation in delivery models. Researchers and others continually develop new interventions and delivery models and seek to encourage their dissemination across the treatment system. Examples in recent decades have included the collaborative care model (CoCM) for depression; Screening, Brief Intervention and Referral to Treatment (SBIRT); and assertive community treatment (ACT).

In the US, a lot of this research into new intervention and delivery models is funded by the National Institutes of Health (NIH) and by foundations, and the funding typically takes the form of grants to develop and test the new interventions. Another key player is the Substance Abuse and Mental Health Services Administration (SAMHSA) which funds states and providers to demonstrate novel models. The original grant funding from these sources is typically time-limited. This implies that eventually communities, clinicians, and others must find resources to replace the grant funding. If providers are not able to find long-term funding for the new services, the innovation will not be disseminated across the health system – even if the initial research showed that the innovation would be effective, and cost-effective to society.¹ This problem is not rare. In 2008, two prevention researchers reported that among evidence-based grant-funded interventions aimed at reducing drug use among at-risk youth, a substantial proportion do not survive beyond their initial funding period, due to lack of financial sustainability.²

One could ask why the US health system does not automatically fund innovations that are found to be effective and cost-effective. One answer might consider the fragmented nature of health care payment, where services are paid for by a variety of public and private payers. In many areas there are multiple competing private insurers making coverage decisions, which reduces incentives to be the first plan to innovate. Innovation may result in cost savings that are not captured by the innovating plan, or may result in attracting more costly enrollees (adverse selection).³ However, delays in adoption are also seen at large public insurers such as Medicare and Medicaid, which do not face the same adverse selection issue.

A more serious challenge is the continued dominance in the US of fee-for-service (FFS) reimbursement as the way individual health care providers are paid, especially for behavioral health care. This means that if a new delivery model includes new types of service, they will initially lack a standard procedure code, so providers cannot bill insurers for those services until a procedure code is established. And new procedure codes can take a long time to be designed, approved, and implemented by the many payers. Establishing new Current Procedural Terminology (CPT) procedure codes requires applying to the CPT Editorial Panel, which is authorized by the American Medical Association to revise, update, or modify the codes. Criteria for approval include the extent to which the new procedure differs from existing procedures, whether it is already in use, and clinical evidence of effectiveness. One study found that it takes it takes 11.2 years on average to go from the initial research study to approval of a provisional (CPT III) code.⁴

Some novel delivery models face additional challenges to their dissemination because they include components that fit poorly with fee-for-service payment. This includes services that do not involve face-to-face patient contact, such as consultation among primary care physicians, specialists, and care managers (including by phone or email). Similarly, health insurance more generally has had trouble adapting to pay for services of non-medical providers (e.g., peer support) or non-medical services for health-related social needs (e.g., transportation, housing, or employment services). Finally, some delivery innovations are designed without considering the payment issues present in a fragmented health care system with many insurers, separate specialty plans for behavioral health and pharmacy, and the interactions with other siloed systems such as criminal justice and child welfare.

In this paper, we seek to better understand the financial challenges to disseminating innovative behavioral health delivery models in a fee-for-service environment. While recognizing that dissemination also faces nonfinancial challenges such as stigma and organizational culture, we focus on financial challenges, in the hope that these might be more addressable by policy. We proceed by reviewing US experience with three specific novel delivery models that emerged in recent years. The models are: the collaborative care model (CoCM) for depression, outpatient based opioid treatment (OBOT), and the certified community health clinic (CCBHC) model. These examples were selected as illustrating some common themes and some different issues resulting from the characteristics of each model. For each model, we discuss its core components; evidence on its effectiveness and cost-effectiveness; how its dissemination was funded; how providers are paid; and what has been the uptake so far. The final section draws out lessons for the diffusion of future models, in the US and potentially also in other countries to the extent that they too use fee-for-service reimbursement approaches.

2. Example 1: Collaborative care for depression

2.1. Delivery model

The collaborative care model for depression (CoCM) is a specific type of team-based integrated care that involves systematic follow-up and coordination between the behavioral health providers and the primary care medical providers. Behavioral health care managers are deployed in medical practices in order to provide assessments, brief psychosocial interventions, and medication management support. The care managers receive back-up and regular consultation from a designated psychiatric consultant, and the primary care practice and care manager maintain a CoCM registry to track patients’ progress toward treatment goals.⁵ The model originated in research conducted at the University of Washington from the 1990s onward by Wayne Katon and colleagues, who were looking for new ways to assist primary care providers in the diagnosis and treatment of mental disorders using psychiatric consultation.⁶ The model was developed and tested using a series of grants from the National Institute of Mental Health (NIMH) and the Agency for Healthcare Research and Quality (AHRQ), and from private foundations.

2.2. Evidence on effectiveness

A 2012 Cochrane review of 79 randomized controlled trials concluded that collaborative care was associated with significant improvement in depression, anxiety, and select primary care outcomes compared with usual care.^7,8 The CoCM also tends to increase cost, so the evidence on its cost-effectiveness has been somewhat mixed, according to one review. However, only one-third of the studies in that review had used follow-up periods longer than one year, as would be desirable given the typical duration of depressive disorders.⁹

2.3. How dissemination has been funded

Unlike some other innovations we discuss, the incubation and dissemination of CoCM into practice has not been specifically targeted for grants by the Substance Abuse and Mental Health Services Administration (SAMHSA). Early adoption was championed by Kaiser Permanente of Southern California¹⁰ and by the US Department of Veterans Affairs (VA), which required most of its community-based outpatient clinics nationwide to implement CoCM.¹¹ More recently, payers including Group Health (now Kaiser Permanente of Washington State) and Intermountain Healthcare launched CoCM initiatives.⁵ Perhaps not coincidentally, several of these are integrated delivery systems that pay most of their providers on a salaried basis, so they are less subject to the challenges that fee-for-service billing poses for the CoCM (see below). In 2023 SAMHSA’s Primary Care and Behavioral Health Care Integration grants are forecast to provide support to States collaborating with either Federally Qualified Community Health Centers or community behavioral health providers to advance integration of services in both settings, but promotion of CoCM per se is not expected to be called out.

2.4. How providers are paid

In a traditional fee-for-service setting, psychiatrists and other providers can bill for their face-to-face patient visits as usual, but several CoCM components are not billable. These components include meetings between the psychiatrist and the primary care physician or care manager (e.g., for consultation, or to review registry updates); therapeutic interventions that are too short to meet the requirements to bill for a diagnostic assessment or therapy session; and time spent entering data into the CoCM registry and analyzing it.¹²

Recognizing the billing challenge, various payers have developed payment approaches that are intended to help fund some of the nonbillable costs. For example, in Minnesota starting in 2008, nine health plans and 25 medical groups joined to implement an evidence-based collaborative care program for depression in over 85 clinics (the DIAMOND program). Clinics were paid by health plans using a case rate that covered evidence-based collaborative care (care management and psychiatric caseload supervision and consultation) in addition to traditional fee-for-service billing by the patients’ regular primary care providers.¹³ The initiative was coordinated by a regional healthcare improvement collaborative, the Institute for Clinical Systems Improvement. Similarly, in 2008, the Washington state government joined with a local health plan to fund an integrated care initiative in a diverse network of community health clinics across the state, to provide CoCM for various disadvantaged populations. Separately from the usual fee-for-service payments to primary care providers, participating clinics received lump sum payments (adjusted by caseload size) to fund on-site care managers.⁵ From 2009 onward, the health plan withheld 25 percent of payments to each clinic until it had met a number of agreed upon quality indicators. The initiative drew on strong support from researchers at the University of Washington who had been involved in developing the model.¹⁴

A much larger-scale intervention occurred in 2017, when the Centers for Medicare and Medicaid Services (CMS) adopted dedicated billing codes for services to support and reimburse enhanced care teams (including collaborative care) in primary care settings, through a monthly case rate. (In 2012, CMS had announced willingness in principle to consider such codes, so a workgroup of the American Medical Association proceeded to design them.)¹⁵ In 2018 these codes were formally integrated into the CPT code set, making them available to all payers to implement. Two of the new codes reimburse the treating clinician for 60–70 service minutes spent in a month on patient assessment, management, and treatment directed by the treating clinician, in consultation with a psychiatric consultant, and delivered by a behavioral health care manager as a supplement to usual care. Another code is an add-on for 30 additional service minutes, and a fourth code is for 20 minutes of monthly services and coordination for more general behavioral health case management with a designated care team member. These were followed by special codes for Federally Qualified Health Centers and Rural Health Centers. Medicare implemented the codes. Subsequently, various state Medicaid programs adopted the codes too (24 states as of 2022), although some have set very low reimbursement rates which are likely to discourage billing.¹⁶ Some commercial insurers have also adopted the new codes.

One study using simulation methods in 2017 concluded that primary care practices adopting the CoCM approach for their Medicare patients would likely maintain net positive revenues, even at relatively low rates of behavioral health treatment attendance and follow-up among those screening positive for behavioral health disorders. The results were not particularly sensitive to variations in patient demographics and practice type.¹⁷

2.5. Uptake of the model

The CoCM appears to have spread extensively in more integrated health systems such as the VA and the Kaiser Permanente health system. Dissemination has been slower in other health systems where fee-for-service payment predominates, although some recent growth has been reported.¹⁸ A survey of accountable care organizations in 2017 found that only 17% reported implementing all collaborative care components.¹⁹ The new billing codes could be helpful in those settings, but as noted above, they have only been activated by some payers. Even in settings where providers can bill using those codes, there is still little evidence of how often they are doing so. Early studies found extremely low uptake in the first years after code activation in Medicare nationally²⁰ and in North Carolina Medicaid.²¹

Slow uptake of CoCM may reflect in part the remaining challenges even in settings where the new billing codes are available. These include low payment rates in many states, lack of financial support for the initial investment costs,²² and even lack of provider awareness about the codes. In addition, Carlo et al. noted that providers face challenges in using the billing codes. First, it is potentially cumbersome to track all CoCM-specific minutes for each enrolled patient throughout the calendar month. Second, CMS requires that monthly CoCM minute totals be exclusively based on accrued time for the behavioral health care manager, separately from other team members’ collective efforts. Clinic data systems would not typically collect such data, so Carlo et al. report on the work-arounds that their health system developed, such as adapting its electronic medical record to track minutes and alert clinicians when billing time thresholds are reached.²³

2.6. Comment

Even if more commercial payers and states activate the available billing codes for CoCM, the model still faces substantial financial challenges to its dissemination. Some have suggested the need to go beyond the current billing codes and create a more aggregated payment unit, such as per-member per-month payments based on number of patients potentially eligible for CoCM services.²⁰ This would give providers greater flexibility in arranging the services needed to implement CoCM. More generally, it is worth noting that as long ago as 2003, there was already published discussion of the problems with funding key components of CoCM under fee-for-service.¹ However, it took until 2017 to establish national billing codes to at least partially address the issue.

3. Example 2: Outpatient based opioid treatment (OBOT)

3.1. Delivery model

Opioid misuse and overdose have rapidly increased in the last two decades in the US and been classified as an epidemic. Over the same period, office-based opioid treatment (OBOT) with buprenorphine has emerged as a newer delivery model which seeks to make medication for opioid use disorder (MOUD) widely available in general medical settings. In the US, OBOT first became legal in 2002 under the 2000 Drug Addiction Treatment Act. It allowed physicians to prescribe buprenorphine/naloxone in the context of primary care, provided they completed 8 hours of training and received a Drug Enforcement Administration (DEA) waiver number - and, subsequently in 2016, 24 hours of training for select non-physician advanced practice providers such as nurse practitioners and physician assistants. (Those training requirements were eliminated in April 2021, and the need for a waiver was eliminated in January 2023). In the US, this approach is often referred to as Office-Based Opioid Treatment (OBOT), although more recently some proponents have expanded the term to “Office-Based Addiction Treatment”, applying similar approaches to other substance use disorders. The label ‘office-based’ highlights the key difference from methadone, which in the US can only be dispensed in specialized ‘opioid treatment programs’, not in physician offices. (These regulations too may be relaxed soon.)

While buprenorphine prescriptions are the defining feature of OBOT, the model also typically includes psychosocial counseling, and some version of the following additional features: (1) coordinated care management, where a clinic staff member (e.g., nurse or social worker) coordinates patient follow-up, buprenorphine prescription refills, drug testing, and prescription drug monitoring program queries; (2) technical assistance to the clinic, including both training sessions and provider-to-provider consultations around opioid use disorder (OUD) treatment; and (3) (in some versions) a link to a ‘hub’ where the initial induction of MOUD occurs, as part of a ‘hub-and-spoke’ model. Several different models have emerged for delivery of OBOT, which differ in how they organize the above-listed components, as described in a report to the Agency for Healthcare Quality and Research.²⁴

3.2. Evidence on effectiveness

A Cochrane review of 31 trials concluded that buprenorphine was superior to placebo medication in retention of participants in treatment at all doses examined.²⁵ However, this review did not analyze results by treatment setting or delivery model, implying that it does not provide evidence specifically for the OBOT model. Another systematic review examined 35 interventions (10 randomized control trials and 25 quasi-experimental interventions) that all tested buprenorphine or methadone in primary care settings across eight countries. Based on programs’ retention rates and risk assessment scores, the review classified seven of the interventions as successful (six of which used buprenorphine as the medication). It concluded that all of those seven studies had used coordinated care models with multidisciplinary teams.²⁶ In terms of economic evaluation, other reviews have concluded that buprenorphine is also cost-effective compared to no pharmacotherapy, although again the finding is not specific to the OBOT model.^27,28

3.3. How dissemination has been funded

SAMHSA did not choose to offer states grants destined solely for OBOT, in the way that it has done with some other novel delivery models (e.g., CCBHC and SBIRT). Rather, it included OBOT among the list of potential ways that states could spend money from more general-purpose SAMHSA grants. One such grant mechanism was SAMHSA’s “State Targeted Response to the Opioid Crisis” (STR) grant program launched in 2017, which provided $1 billion over a 2-year period to 57 states and U.S. territories. States had flexibility in allocating the funds, but their strategic plan had to include implementation or expansion of access to clinically appropriate evidence-based treatment—particularly the use of MOUD. In 2018, 35 states reported using at least some of their SAMHSA STR funds to expand the number of patients served with OBOT.²⁹ Examples include California and Washington, which both launched hub-and-spoke versions of OBOT, to increase access to MOUD.^30,31 Of note, the STR funds may not be used to fund services, but states have still found them useful in funding other costs that insurance would not cover, such as startup costs and nonbillable activities.

Even before the STR grant program started in 2017, SAMHSA had introduced funding opportunities to expand access to MOUD through both OBOT and OTP models, including through the Medication Assisted Treatment – Prescription Drug and Opioid Addiction (MAT-PDOA) program which has awarded a cumulative $104 million to states, local health authorities, and individual providers since 2015. While this funding has supported OBOT expansion projects across the country, it is not focused on dissemination of OBOT per se. Meanwhile, some states have taken advantage of the SAMHSA State Substance Use Disorder Block Grant (SUDBG) to increase the uptake of OBOT specifically, including Massachusetts which has provided grants to community health centers for OBOT since 2007 using SUDBG funds.³²

3.4. How providers are paid

Some OBOT services are naturally billable in a fee-for-service environment, such as the face-to-face visits for counseling and medication management that are delivered by physicians and various other providers. However, until 2020 (and to this day in many plans), several key components of the OBOT model were not billable to insurers, even if the payer was a managed care plan. This has been an issue for services delivered by the nurse case manager (except where nurses can bill), and for services like administrative support, and specialist training or consultation to primary care providers, which are typically not billable under FFS because there is no doctor-patient interaction.²⁴ Clinics have had to find creative ways to cover these nonbillable costs, including taking advantage of state grants where available.³³ Two studies identified the breakeven patient caseloads which would allow clinics that provided OBOT to cover the salaries of the non-billing staff, and the sensitivity of those breakeven levels to a variety of assumptions.^{34 35}

Some states have attempted to address this non-billability issue. The Massachusetts grants to community health centers were intended to cover nonbillable services, including the salary of a full-time registered nurse, staff training, and technical support on implementation of protocols for each program. More recently the state switched to funding those nonbillable services with a ‘unit rate’.³² In Vermont, for every 100 Medicaid patients receiving MOUD at a ‘spoke’ (OBOT) practice, the state Department of Health Access pays the cost of one nurse and one licensed mental health/addiction counselor to support the prescribing providers.³⁶ Maine’s Medicaid program approves practices to be “opioid health homes,” and pays them capitation amounts per member per month which depend on the level of care a patient needs, and on whether the health home also provides coordinated case management.³⁷

In 2020 Medicare took a major step toward addressing the non-billability problem by establishing a new monthly bundled payment option for OBOT, supported by three new procedure billing codes. Rather than paying per visit in the usual way, instead the codes cover a specified number of minutes per month of “office-based treatment for OUD, including development of a treatment plan [first month only], care coordination, individual therapy and group therapy and counseling.” The monthly payment bundle allows the therapy or substance use disorder (SUD) counseling services to be provided by any qualified practitioner, and covers care coordination or peer services under the bundle without subjecting these services to additional time documentation requirements. Notably, these codes do not cover payment for the OUD medications or for related toxicology testing, which continue to be billed separately, as do additional mental health treatment services for individuals with co-occurring disorders. The codes are specific to office-based treatment, as CMS established a separate bundled payment rate for Opioid Treatment Programs, and they can be used for telemedicine services. Various states are reportedly considering applying a Medicare-type OBOT bundled payment approach in their Medicaid programs, but it is still too soon to tell if the CMS codes will take off in Medicaid.

Some similar examples of bundled payment for OBOT have been advanced by providers and successfully negotiated with payers. One such model is a prospective bundled payment that fronts the provider an anticipated monthly expenditure for specific CPT codes involved in OBOT and then requires the provider to submit FFS claims for these services when they are delivered. These claims are then “zero paid” by the insurer. Operationally this model departs very little from fee-for-service and some contracts even allow payers to recoup the difference between zero-paid claims and the bundled payment rate, discouraging provider investment in components of the service for which FFS claims do not exist. To the positive, such contracts may include upside performance payments for meeting quality measures such as initiation and engagement in treatment, adherence to MOUD, or reductions in emergency department (ED) utilization or admission to 24-hour levels of care. While there are no quality payments associated with the new CMS OBOT bundled payment, the new HCPCS codes eliminate the need to “zero-pay” FFS claims for individual components of the service, creating greater flexibility to provide care management and recovery supports.

3.5. Uptake of the model

Buprenorphine prescribing in the US has increased in recent years, even during the Covid-19 pandemic, with the number of patients reaching around 850,000 by end-2020.³⁸ Until 2021, OBOT’s dissemination was constrained by the willingness of prescribers to obtain a federal waiver training to prescribe buprenorphine, as required under the 2000 DATA law. The elimination of that requirement in 2023 has removed one obstacle, although even among the prescribers who obtained waivers before that change, rates of prescribing buprenorphine were low on average.³⁹ The creation of Medicare’s bundled payment could increase uptake by flexibly funding the casework and therapeutic complements to prescribing MOUD that make clinics and physician practices more willing and able to offer OBOT. However, in the first year the proportion of OUD claims billed using the new codes was extremely low.⁴⁰ Even with the bundled payment, other obstacles remain such as payer restrictions on which patients may receive buprenorphine,⁴¹ lack of institutional and clinician peer support, poor care coordination, provider stigma, and even provider awareness of the new codes.⁴²

3.6. Comment

In addition to payment, the challenges to OBOT’s diffusion have included physicians’ comfort with prescribing and caring for the SUD population, their capacity to do it (having waivered prescribers), and their competency (as in, having sufficient experience/training/support resources to take on more than a few OBOT cases). Stigma likely also plays a role. Removal of the prescribing waiver requirement eases the capacity challenge, but the other remaining issues are likely to continue to impede wider adoption of OBOT. It remains to be seen whether other payers will adopt Medicare’s bundled payment approach, easing the payment constraints on providers. Some payers may be concerned that in the behavioral health area, they do not have adequate performance measures to monitor quality under bundled or other value-based payment approaches. There is considerable uncertainty about the effects of bundled payment for OBOT, which could be addressed by studies of the Medicare implementation or by new randomized trials of this payment approach.

4. Example 3: The CCBHC model

4.1. Delivery model

The Certified Community Behavioral Health Clinic (CCBHC) model is a new delivery approach developed by the National Council for Mental Wellbeing and has been promoted and funded by SAMHSA since 2014. CCBHCs are intended to provide ready access to integrated care to adults and children with a variety of mental and substance use disorders. To be certified, a CCBHC must offer a specified list of comprehensive services either directly or (for some services) through a designated collaborating organization. The services must be accessible to all individuals who seek care, regardless of their ability to pay, including but not limited to those with serious mental illness, serious emotional disturbance, and substance use disorders. The services specified in the authorizing legislation are: (1) 24-hour crisis services, (2) screening, assessment, and diagnosis, (3) patient-centered treatment planning, (4) outpatient mental health and substance use treatment, (5) screening and monitoring of key health indicators, (6) targeted case management, (7) psychiatric rehabilitation services, (8) peer and family support and counselor services, and (9) intensive, community-based mental health care for members of the armed forces and veterans. The model is also intended to provide coordinated care that addresses both behavioral and physical health conditions,^43,44 although the form of this coordination is less specified than in some other novel delivery models. The model is inspired by studies of co-occurring mental health and substance use disorders which recommended integrating specialty behavioral health treatment and coordinating with primary care to address both disorders.⁴⁵

4.2. Evidence on effectiveness

The CCBHC model was not developed and tested through grant-funded research, although various of its components derive from such research. As a result, the main evidence to date on this model’s effectiveness comes from a national evaluation conducted by Mathematica Policy Research.⁴⁴ The evaluators concluded that the participating clinics had expanded services, hired and trained staff, developed partnerships with external providers, enhanced their data systems, and changed many of their care processes to become CCBHCs. They also concluded that the quality of care provided to clients of CCBHC services was comparable to available benchmarks, and, for some measures, improved substantially over time. In some states there was evidence of decreasing hospitalizations and ED visits associated with CCBHC activity.⁴⁴ According to one report, “in areas with CCBHCs, from patient baseline to 6-month reassessments, hospitalizations for mental health care decreased by approximately 73 percent, visits to emergency departments for a behavioral health issue decreased by approximately 69 percent, and stable employment increased by 14 percent”.⁴⁶ However, that report also noted that these findings were limited by a lack of baseline data, a lack of comparison groups, and a lack of standardization in coding across states. No studies so far have examined the cost-effectiveness of CCBHC, to our knowledge.

4.3. How dissemination has been funded

The CCBHC program was first authorized by the Protecting Access to Medicare Act (PAMA) of 2014. In October 2015, the U.S. Department of Health and Human Services awarded planning grants to 24 states to begin certifying clinics to become CCBHCs as part of a Medicaid demonstration project. In 2016, eight of those states were selected for the demonstration project and started receiving demonstration grants from SAMHSA, which they then used to fund treatment organizations that met criteria to be considered a CCBHC. Subsequently two more states were awarded demonstration grants in 2020 as part of the CARES Act (COVID-19 relief). Demonstration states also receive enhanced federal matching funds for services provided to Medicaid beneficiaries at CCBHCs during the demonstration. More recently, SAMHSA awarded planning grants (each lasting 4 years) to 15 more states, to result next year in 10 more full grantees in 2024, and with plans to expand to all 50 states by 2030.

In addition, SAMHSA provides direct capacity-building grants to providers to create CCBHCs in states that are not participating in the Medicaid demonstration (‘capacity building and expansion grants’), using federal appropriations from PAMA, and additional funding that SAMHSA received under the 2020 CARES Act. Recipient providers are obliged to develop plans for sustainable payment similar to the models adopted in Demonstration states.

The Medicaid Demonstration grants are time-limited, but none have expired yet, as Congress recently renewed the current demonstration states through 2025. Nonetheless, the current grantees have been planning ways to eventually replace the grant funding. Some states have passed “bridge grants” to support the continuation of the model beyond the end of federal grants.⁴⁷ Five states have modified their Medicaid state plans to add CCBHC permanently as a covered service,⁴⁸ which could give providers more confidence to invest in this model.

4.4. How providers are paid

The CCBHC Medicaid demonstration has departed from standard FFS payment by requiring states to reimburse CCBHC sites for care of Medicaid patients with a cost-based prospective payment. The CCBHCs in these 10 states receive a clinic-specific daily or monthly prospective Medicaid payment that is based on the anticipated costs of providing the demonstration services to a complex population (similar to how federally qualified health centers are reimbursed). The services include recovery supports and rehab services which some state Medicaid programs might not otherwise fund. The model also requires access to urgent care and extended hours of operation. In some states, CCBHCs also receive a quality bonus payment based on their performance on required quality measures.⁴⁵ The prospective payment approach does not apply to other insurers, whose enrollees’ care at CCBHCs is reimbursed by those insurers in their usual way.

However, in the 32 states that have CCBHCs but are not participating in the CCBHC Medicaid demonstration, this prospective payment approach does not apply. In the non-demonstration states, providers who have received CCBHC capacity-building or expansion grants directly from SAMHSA must bill on a traditional Medicaid FFS basis for state plan covered services delivered by the CCBHC. For grant-funded CCBHCs, SAMHSA funding essentially covers both non-billable costs⁴⁹ and funding shortfalls from existing FFS rates.

4.5. Uptake of the model

As of late 2022, there are over 500 CCBHCs nationwide, with a presence in 47 states, and they serve an estimated 2.1 million people.⁴⁸ Growth of the model has been rapid, but uptake is constrained by various factors. One constraint is the number of states that participate in the Medicaid demonstration, although this is scheduled to increase to cover all states in the next few years. In non- demonstration states, clinics may be deterred from seeking CCBHC certification by the less favorable coverage of nonbillable costs there, as noted above. Second, the demonstration states have varied in the geographical extent of their CCBHC implementation, ranging from 10% of counties in Pennsylvania to 78% of counties in Missouri.⁴⁴ Third, in both demonstration and non- demonstration states, uptake depends on how many clinics apply and/or are accepted for CCBHC status, although in February 2023 CMS issued guidance to allow the group of ten states to add new CCBHC sites to their demonstrations. No research has compared diffusion rates between demonstration and non- demonstration states, although this could be informative for policymakers.

Finally, CCBHC officials interviewed in one report listed a variety of financial challenges. These included behavioral health workforce shortages, difficulty recruiting staff for positions funded by grant dollars, and barriers to collecting and sharing data. The report authors concluded that “The greatest challenge to the long-term success of CCBHCs is the lack of sustainable funding”.⁴³

4.6. Comment

The CCBHC model appears to be diffusing more successfully than some other innovations. This is probably in part due to multi-year SAMHSA grant allocations offering coverage of both nonbillable costs and subsidization of underfunded services, and to the CCBHC Medicaid demonstration requiring more flexible approaches to provider payment (cost-based prospective payment) as a condition for receiving enhanced federal matching funds. Challenges remain to folding CCBHC into mainstream health care financing, as described in the report by Foney et al.,⁴³ but perhaps also lessons for plans to disseminate other grant-funded delivery models. One issue is that states can limit the number of CCBHCs through their procurement activities, and benefit financially from doing so because non-CCBHC programs continue to be paid at the lower rates of traditional fee-for-service, even when providing comparable services or certified by SAMHSA under the expansion grant program. Until this year when CMS removed the limitation on adding sites to the CCBHC Medicaid demonstrations, federal policy reinforced the notion of CCBHCs as a limited network. This contrasts with Medicare’s bundled payment for OBOT, which from the outset was established under the Physician Fee Schedule as billable by any provider able to deliver the required standard of care.

5. Discussion

From these three case studies, several common themes emerge. First, all these novel delivery models involved care-management type components that were not face-to-face patient contacts, and therefore not automatically billable in a fee-for-service environment. One challenge becomes whether payers can establish procedure codes for those novel services, and how quickly that happens. For OBOT and CCBHC, this process went quickly because the delivery models were embodied in legislation and soon afterward CMS established billing codes. For collaborative care, there was no similar push, and proponents had to apply to the American Medical Association to establish CPT procedure codes, which were only recently established, decades after the model’s emergence. At the same time, creation of a procedure code is necessary but not sufficient for dissemination of the delivery model. Payers must then activate the code and providers must find it convenient to use the code for billing – and all three delivery models that we considered have faced challenges in those areas.

A second theme is that establishing procedure codes may not be enough to give providers the flexibility needed to implement these models, in which case payers should consider creating more ‘bundled’ procedure codes, as was done for all three of the models we considered. Availability of a bundled procedure code paves the way for payers to use more flexible payment models such as bundled payment or capitation (and potentially reduce the paperwork burden on providers). Another example is Assertive Community Treatment, an evidence-based model of integrated, team-based care for serious mental illness that was developed in response to deinstitutionalization, with a proven track record of reducing hospitalization. This model has long been paid for under a global per diem rate. Efforts by the Alliance for Addiction Payment Reform to develop an episodic payment model for early and maintenance phases of recovery from addiction are garnering increased attention among providers and health plans, with payer-provider demonstrations underway in six states, although no studies of the model’s implementation have yet been released to our knowledge.⁵⁰ Similar considerations may apply to other emerging delivery models such as those aimed at linking released prisoners to community-based behavioral health treatment.⁵¹

Third, sustainable bundled payment rates will need to cover the actual costs of care – which for new delivery models may include some startup costs. Bundling the existing fee-for-service rates may be inadequate, as they are often set below operating cost and ignore indirect cost – unlike grant funding which is often based on explicit budgets that include allocation to indirect cost.

We have focused on time-limited funding, but research on novel delivery models faces other challenges worth mentioning. Such research is often initiated by academic investigators rather than by delivery systems or payers, and consequently often disconnected from routine health care reimbursement systems, and harder to scale up. The problems described here are not unique to behavioral health care, as other areas of health care also have novel delivery models involving nonbillable components, such as Wagner’s Chronic Care Model (which was an important influence on the CoCM).⁵² However, the challenges may be more severe for behavioral health care given its greater reliance on psychosocial interventions, and the fact that diffusion of those interventions - unlike novel medications - is rarely encouraged by commercial marketing.⁵³

More generally, a lot of resources go into developing new interventions and new delivery models. If those new models will not eventually be disseminated, then the value of those initial investments becomes questionable. To reap the rewards from its investment in innovation, society needs to give the innovations the best chance to disseminate, by removing unnecessary financial obstacles. To make progress on this, below we propose actions for different kinds of stakeholders.

Implications for Health Care Provision and Use

For providers who receive time-limited grant funding to implement these novel delivery models, one key lesson is the need to start early on planning how services will be sustained after the grant ends. This is most obvious for providers in cases where the provider organization itself is the grantee, as with SAMHSA’s direct grants to CCBHCs. But it also applies to situations where providers are billing a state agency which received a time-limited grant, meaning that providers must consider whether the state will sustain the new services beyond the end of the grant period.

Implications for Health Policy

For policymakers, a key lesson is the need to fold novel models into stable, existing funding streams such as Medicaid and commercial insurance coverage, rather than leaving them at the mercy of revolving time-limited grants. Without that security, it may be harder to recruit skilled staff to work on these new models, especially given the current workforce shortages in behavioral health care. In discussing other lessons, it is important to distinguish payers such as Medicaid, Medicare, VA, and commercial insurers from the research funding agencies like NIH, SAMHSA, and private foundations, which have provided many of the grants aimed at disseminating novel delivery models. Different lessons are relevant to each of these stakeholder types.

For funders of research, it is clearly important to speed up the process of obtaining billable procedure codes for the non-billable components of each novel delivery model, and to plan for this as early as possible. Without those codes, dissemination will be delayed. Second, funders should identify and remove factors that discourage providers from billing the codes once they are available. Third, funders should issue Requests for Applications aimed at generating evidence on funding and payment models, with priority specified for researchers who are working with payers and providers during the design phase, so that their models are more likely to be scalable. Finally, funders should require provider grantees to initiate sustainability planning for the post-grant environment, ideally early in the period of grant funding. SAMHSA already typically requires this of its grantees. But in addition, funders may need to help less experienced grantees (e.g., smaller clinics) by providing more technical assistance with sustainability planning, including assistance with advocating service coverage to payers using aggregate data related to the outcomes and cost impact of various grant-funded programs.

For payers, a key lesson is the need to cover novel delivery models once they are found to be effective, and to pay for them in ways that facilitate uptake by providers. For example, more payers need to consider use of alternative payment models like bundled payment, and to implement them in ways that give providers more flexibility than fee-for-service payment. Payers could also reduce workforce shortages by increasing provider payment rates, perhaps making use of funds from the upcoming legal settlements from opioid manufacturers, or by moving services paid for under SAMHSA block grants into the Medicaid benefit to maximize federal financial participation and reinvest matching funds in service expansion. Of course, such changes to payment should be evaluated to make sure they are having the intended impacts. This requires greater collaboration between payers and providers on the collection and sharing of data to measure the impact of investments on treatment adherence, health care utilization, patient-centered outcomes, and disparities. To this end, payers should get involved earlier in the grant-stage evaluation of medical expense offsets as well as quality and outcomes measures, since only they typically have the data to observe multiple dimensions of health care treatment and utilization, and to assess financial impact from the payer perspective.

Finally, the federal Department of Health and Human Services should take steps to better align SAMHSA and CMS priorities, policies, and long-term healthcare financing strategy. CMS has the potential to exert much greater influence over state Medicaid policy than it has historically, in order to drive the adoption and sustainability of promising approaches to solving profound behavioral healthcare challenges, including those approaches funded by SAMHSA through both state block grants and direct-to-provider programs. It can specify payment rate adequacy requirements in its approvals of state 1115 waivers (as it recently did for Oregon and Massachusetts). By tying the availability of federal matching funds in Medicaid demonstrations to sustainable payment, CMS could use its federal matching authority to catalyze adoption and expansion of both promising service delivery and alternative payment models. This would accelerate the movement of models into Medicaid as standard, adequately funded services, while preserving the important role of SAMHSA, other federal agencies, and private funders as incubators for innovation.

Implications for further research

For researchers, a key recommendation would be to pay greater attention to the payment environment at the time of designing new delivery models and interventions. What would make your new delivery model more sustainable beyond the end of grant funding? This could include investigating what evidence might persuade payers or states to cover the intervention, and what performance measures payers could use to reward providers who adopt the intervention and implement it with fidelity. Goldstein et al. recently suggested some ways that prevention research could be designed from the outset to inform the evidentiary requirements of potential payers.⁵⁴ Researchers on the Healing Communities Study recently applied a similar approach to ask Massachusetts payers about their willingness to cover interventions to reduce overdose, which the parent study had initiated in four states with time-limited NIH grant funding.⁵⁵ However, interviewing payers and providers is costly, and is not likely to occur much unless research funders announce that they value such work and will reward it in the grant review process.

It is worth noting that our focus on three specific innovations is a limitation of this paper. This focus has the value of allowing a richer understanding of those specific cases, but may lead us to miss information that would come from a broader survey. See for example Drake et al.’s discussion of the diffusion of naltrexone, multisystemic therapy and the IMPACT model for geriatric depression.⁵³ Future work could build on these and other case studies.

Financing is clearly an important aspect to be addressed. However, in order for new delivery models to be disseminated, other barriers need to be addressed as well. In particular, various forms of stigma may affect providers’ willingness to adopt new delivery approaches,⁵⁶ and thus weaken the effect of payment interventions. In the real world, there is not usually a single silver bullet to deal with the complex issues around sustainability.

6. Conclusion

The experiences reviewed in this paper suggest that financial issues play an important role in limiting the dissemination of novel delivery models in behavioral health care. But these experiences also suggest that many of the financial issues are addressable with greater attention to financial sustainability by all stakeholders, at an earlier stage in the dissemination process. Focusing on financial sustainability could ensure more successful dissemination, and therefore also a higher return on the substantial resources invested in developing novel delivery models.