Hub and Spoke Model

The system-based Hub and Spoke model was developed in Vermont.^29–32 The model consists of two levels of care, with the patient’s needs determining the appropriate level. In this model, “hubs” are OTPs that serve as regional specialty treatment centers (currently numbering 6) that provide traditional treatment for OUD and also have the capacity to either directly provide or to organize comprehensive care and continuity of services in a home health model. “Spokes” are clinics in the community that provide MAT and comprehensive care for less clinically complex patients. Patients are screened to determine whether they are appropriate for initial stabilization and management in a hub or spoke. The hubs provide care for clinically complex patients, support tapering off MAT, dispense methadone if needed, and provide consultative services to the spokes. Following stabilization, patients initially managed at a hub who do not require ongoing management at the hub may have their management transferred to a spoke; conversely, patients managed in a “spoke” who require a higher level of care may be transferred to a hub. Buprenorphine/naloxone has been the primary pharmacological component in the spokes within the Hub and Spoke model. The model is financed through a Medicaid health home model waiver state block grant. Its effect on outcomes has not been published. Vermont incentivized implementation of buprenorphine/naloxone prescribing by funding online training for physicians to obtain buprenorphine waivers and providing other technical assistance to physicians prescribing buprenorphine. The Hub and Spoke model includes some educational outreach in the community to increase the number of buprenorphine waivered physicians. Coordination and integration occurs between the hub and spoke as well as within each spoke site, and is typically carried out by a registered nurse, clinician case manager, or other “care connector” (e.g., via peer-to-peer support or behavioral health workers). Psychosocial services are embedded within spoke sites, including social workers, counseling, and community health teams.

An important advantage of the Hub and Spoke model is the availability of tiered care and the availability of regional expertise in the management of OUD. The established relationships between the hub and spokes promote ongoing coordination and integration, including efficient consultation with the hubs and transfer of care to the hub as needed. Within the spoke sites in this model of care, the use of designated nonphysician “care connectors” at the spoke sites and availability of embedded psychosocial services are important advantages over models in which the coordination/integration roles are less well defined or in which psychosocial services are not available on-site. A potential disadvantage of the Hub and Spoke model is that a hub with the appropriate expertise and resources may not be available in all settings that wish to implement a MAT model of care. Also, the spokes in the Hub and Spoke model are likely to vary in the degree of expertise and types of services provided.

Collaborative Opioid Prescribing Model

The system-based Collaborative Opioid Prescribing (Co-OP) model was developed in Baltimore.^33,34 Similar to the Hub and Spoke model, initial intake, induction with buprenorphine/naloxone, and stabilization is performed at a center (in the Co-OP model, this is an OTP). Patients are shifted to primary care clinics for ongoing MAT after stabilization on medication. Unlike the Hub and Spoke model, in the Co-OP model psychosocial services are generally provided concurrently on an ongoing basis by the OTP, rather than at the primary care site. Some outreach and education is performed by counselors involved in Co-OP to community physicians. Financing is through Medicaid and private insurance.

Like the Hub and Spoke model, an advantage of the Co-OP model is that initial evaluation and management occurs in a specialty center; in addition, the specialty center continues to provide psychosocial services following the handoff to the primary care site. Therefore, this model takes advantage of the expertise and resources available at the OTP on an ongoing basis. A potential disadvantage of the Co-OP model is that because ongoing psychosocial services are provided by the OTP, it may require relatively close proximity between the primary care sites and the OTP, which may not be available in all settings that wish to implement a MAT model of care. Also, because the OTP in the Co-OP model provides ongoing services, this could limit the number of patients that could be managed compared with the Hub and Spoke model, in which ongoing care for most patients is more dispersed and provided more independently within the spoke centers.

Office-Based Opioid Treatment

An early model for Office-Based Opioid Treatment (OBOT), a practice-based model, has been widely disseminated throughout the United States. In OBOT, physicians who complete 8 hours of training and receive a DEA waiver number may prescribe buprenorphine/naloxone in the context of primary care While many providers offer OBOT without staff assistance, some practices designate a clinic staff member, or “glue person” (often a nurse or social worker) who works in collaboration with a primary care clinician to coordinate services.^35–37 The glue person is instrumental for coordinating and integrating care, including primary care and mental health. Psychosocial services include regular brief counseling provided by the physician and glue person or other staff; other psychosocial services vary but can include integrated cognitive behavioral therapy or motivational enhancement therapy. Psychosocial services may be located on-site or off-site. Early OBOT trials provided education and training of new buprenorphine prescribers, which led to the development of the PCSS-Buprenorphine (now PCSS-MAT) model nationally, including mentoring by more experienced prescribers. OBOT is financed through provider reimbursement of billable visits. Medicare and many state Medicaid programs cover buprenorphine, though prior authorization is frequently required.

A key advantage of the OBOT model is its use of a glue person to coordinate ongoing care. This provides an efficient way for the prescribing physician to manage more patients. The model also takes advantage of a training and mentoring resource available via the Web. Although regular brief counseling is a core aspect of this model, a potential disadvantage is that the availability of additional psychosocial services is highly variable, which could make management more difficult for more complex patients. In addition, coordination and ongoing relationships with OTPs appear relatively informal or undefined in this model.

Massachusetts Nurse Care Manager Model

This system-based model was developed in Massachusetts, where Medicaid reimburses Federally Qualified Health Center nurses for OUD care management.^38–40 This model is similar to the OBOT model in that a key aspect is the use of a nonphysician to coordinate and manage much of the care. Unlike the OBOT model, the Massachusetts model specifically uses nurse care managers who team with primary care physicians to provide MAT (primarily buprenorphine/naloxone, with integration of extended-release naltrexone over the last 2 years). The nurse care manager performs initial screening, intake, and education, often with assistance from a medical assistant. The nurse care manager also provides ongoing management of OUD and other medical issues, including drop-in or same day visits, management of acute issues, coordination of prior authorization requests, communication with pharmacists, and perioperative care coordination. The diagnosis of OUD and appropriateness of MAT are confirmed by the prescribing physician, who comanages the patient with the nurse care manager. One Key Informant described an adaptation of this model at a community-based health care system in Massachusetts in which a “care partner” (usually a master’s level individual who is not a nurse care manager) performs this role. This model uses a training program to get more primary care physicians involved in prescribing buprenorphine and education is provided on best MAT practices; the nurse care manager receives training in MAT and addiction. Psychological services are integrated on-site or nearby, though the specific services that are available vary from site to site. Patients who require a higher level of care can be expedited into treatment in an OTP. The model is financed through direct Medicaid reimbursement to FQHCs for nurse care manager time as a billable service, in addition to usual Medicaid coverage for pharmacotherapy and physician visits.

A key advantage of this model is that it uses a nonphysician to offload some of the burden from prescribing physicians, which in turn enables the prescribing physicians to manage more patients. This model also emphasizes training and education to engage more primary care physicians in prescribing buprenorphine. Another advantage of this model is that it may be more sustainable financially, because Medicaid reimburses federally qualified health center (FQHC) nurses in Massachusetts for OUD care management and the state supports additional coordination services using Block Grant resources. However, this reimbursement mechanism is not available in all states. A disadvantage is that the availability of psychosocial services and whether they are present on-site vary. In addition, the model is highly dependent on the availability of a skilled person who can assume the nurse care manager or analogous role effectively.

Buprenorphine HIV Evaluation and Support Collaborative Model

The practice-based Buprenorphine HIV Evaluation and Support (BHIVES) Collaborative model uses the OBOT framework to provide a chronic care model for providing buprenorphine in HIV primary care settings.^41–51 Like the OBOT Model, a clinic coordinator glue person (typically a counselor or social worker) is essential for coordinating care, working in conjunction with the primary care provider. HIV care can be provided by the primary care provider or by another on-site provider in coordination with the primary care provider. BHIVES sites generally have on-site psychological services, including individual counseling, though the types of services vary. HIV clinics coordinate with affiliated OTPs for patients switching to or from methadone. A HRSA⁵² monograph promotes adoption of BHIVES in United States HIV clinics and BHIVES is considered the standard of care for engaging HIV-infected patients with OUD in treatment.^53,54 Buprenorphine and HIV care are typically covered by patient insurance. Ryan White Care Act funding supplements medication coverage, care coordination and counseling services in some states.

An advantage of the BHIVES model is that it is specifically designed to address MAT, HIV care, and primary care within a single setting. It also has the same advantages as other models that use a glue person for chronic care management and coordination. A potential disadvantage is that the availability of on-site psychological services and the types of available services vary and are not well specified. In addition, it requires clinicians with expertise and knowledge in both MAT as well as HIV care, which may not be available in all settings. PCSS now includes physician mentors with expertise in HIV care, an educational model that could potentially be expanded for other chronic comorbid conditions.

Project Extension for Community Healthcare Outcomes

Project Extension for Community Healthcare Outcomes (ECHO), a system-based model of care first developed in New Mexico, links primary care clinics in rural areas with a university health system utilizing an Internet-based audiovisual network for mentoring and education^55–57 regarding an array of medical conditions. The University of New Mexico developed a module for supporting rural primary care providers in MAT management. It emphasizes nurse practitioner- or physician assistant-based screening with referral to a collaborating physician prior to initiation of MAT and for ongoing treatment, typically with buprenorphine/naloxone. Counseling and behavioral therapies are offered from all ECHO team members. Complex patients can be referred for further assessment and/or evaluation at an OTP. There is also an emphasis on recruitment of physicians for buprenorphine waiver training and provision of continuing medical education in OUD. It is financed through various federal grants and Medicaid.

An important advantage of the ECHO model is that it enhances the ability of rural primary care clinics to provide MAT though its Internet-based mentoring and educational network. The ECHO model may be considered a rural adaptation of the Hub and Spoke or Co-OP models, in that it engages the expertise of a “hub” center to assist in provision of MAT. A potential disadvantage of the ECHO model over traditional tiered care models is that due to the geographic distance between the primary care sites and the hub, initial intake and assessment does not occur at the centralized hub, due to the dispersed and rural settings in which care is provided. Rather, all care, including initial intake and assessment, occur at the primary care sites. The limited availability of on-site or face-to-face expertise in MAT could pose challenges for the management of complicated or high-risk patients. The ECHO model may have had some impact in New Mexico placing among the top states in buprenorphine-waivered physicians per capita; New Mexico has also had more rapid growth in the number of waivered physicians practicing in rural areas than in other areas of the United States since its initiation in 2005.⁵⁵ In addition, the ECHO model focuses on utilizing mid-level care providers for performing initial screening, which may be critical for expanding access to MAT in many rural settings. There is also a strong emphasis on provision of psychosocial services in the ECHO model. The ECHO model is a tele-education/tele-consulting approach considered distinct from telemedicine, as there is no direct doctor-patient relationship between off-site experts and patients, who are de-identified. A potential advantage of this approach is that it only requires basic, widely-available teleconferencing technology and does not require the high startup costs required for Health Insurance Portability and Accountability Act (HIPAA)-compliant telemedicine expansion or the sustainable funding necessary to purchase and maintain telemedicine technology and services. A potential disadvantage is the lack of direct contact between off-site experts and patients, which could make it more difficult to manage complicated patients and obtain reimbursement for providing consultative expertise.

Medicaid Health Home Model for Those With Opioid Use Disorder

The Medicaid Health Home Model is a flexible, system-based model through Centers for Medicare and Medicaid Services that allows states that apply for a Medicaid waiver to integrate MAT and behavioral health therapies with primary care for patients with OUD.^58,59 Provider and community education is emphasized to increase uptake (by clinicians and patients) and to decrease stigma. A core aspect of this model is that core psychosocial services are required (i.e., comprehensive care management, care coordination, health promotion, comprehensive transitional care/followup, individual and family support, and referral to community and social support services). Some telehealth services are also offered, though their availability and use vary. Implementation of Medicaid Health Home Models differs from state to state with differences in how the models are structured and overlap with other models of care (e.g., Hub and Spoke) described in this section. In several states (e.g., Rhode Island and Maryland), implementation of the Medicaid Health Home Model has been in OTPs or psychiatric clinics, rather than in primary care clinic settings,⁵⁹ although as described above, the Hub and Spoke model involves a tiered model of care that includes community-based “spokes.” Buprenorphine/naloxone has been the primary pharmacological component of treatment, with integration of injectable naltrexone over the last 2 years. States determine the structure of health care delivery, for example with Hub and Spoke models in Vermont, and approach to payment, which may include per member per month payments (Maryland) and weekly bundled payments (Rhode Island) that fund care coordinators in addition to other billable health care services.

An advantage of the Medicaid Health Home Model is that it requires care coordination and a set of core psychosocial services. In addition, provider and community education are emphasized as key aspects of this model. The flexibility of this model is an advantage in enabling service delivery and provision to vary according to the needs and resources of the particular setting. At the same time, the flexibility of the model may be viewed as a disadvantage in that some aspects (e.g., who provides coordination/integration, who performs initial screening and assessment) are not standardized or well-defined.

Southern Oregon Model

The Southern Oregon Model is an example of a local and informal system-based model for delivery of MAT in a rural primary care network.⁶⁰ It focuses almost exclusively on buprenorphine/naloxone. A notable characteristic of the Southern Oregon Model is that it has used regular meetings of stakeholders (including regional Medicaid-accountable care organizations) for education, training, and development of practice standards around the prescription of opioids for chronic pain and addiction treatment. Coordination or integration of care is variable and often limited, though an on-site clinical social worker is available. A leader of this model is also medical director of a local federal oversight OTP clinic, providing a source of referral and consultation to providers in the region. However, access to OTPs for complex patients is not formally integrated. The model is financed through direct support from Accountable Care Organizations and usual fee for service billing.

An advantage of this model is that it is a grass-root, community-based effort, which may promote buy-in from clinicians and those in the community. This could serve as a model for implementation of MAT in rural settings where there may be increased stigma associated with MAT and resistance to its use. However, a number of key components of this model are not yet well-defined, and a Key Informant noted that psychosocial services and coordination/integration of care is often limited. The Key Informant also noted that the relationship with the local OTP is suboptimal and at times office-based MAT is viewed as a competitor rather than a partner by the OTP.

Emergency Department Initiation of Office-Based Opioid Treatment

This system-based model focuses on the emergency department (ED) identification of OUD, with buprenorphine/naloxone induction initiated in the ED.⁶¹ Patients are connected to ongoing OBOT, then transferred to ongoing, office-based maintenance treatment or detoxification. Brief “medical management” counseling is performed by physicians; other psychosocial services vary. Medications, ED visits, and OBOT are funded through patient Medicaid and other insurance plans.

An advantage of this model is that it identifies patients who might benefit from MAT and may not have access to primary care, or only sporadic access. Initiation of buprenorphine/naloxone in the ED also appears to increase retention in care rates versus a simple referral. A potential disadvantage of this model is added congestion in the ED as a means to access treatment. In the randomized trial that evaluated this model, ongoing management in primary care settings was provided through the OBOT model, which may not be the model available in all settings. However, the ED initiation model could be used to “feed” into various office-based models of care, depending on what is available in the community.

Inpatient Initiation of Medication-Assisted Treatment

This system-based model involves the identification of OUD in the hospital, with initiation of MAT (methadone, buprenorphine/naloxone, or naltrexone) during the hospitalization by a multidisciplinary addiction consult service.⁶² Patients are connected with primary care or specialty addictions care (patients initiated on methadone must be followed in an OTP), where treatment continues following hospital discharge. In some programs, when relevant, there is a buprenorphine “bridge” clinic for stabilization prior to transitioning to primary care. Ongoing psychosocial services are provided at primary care sites. A variation of this model involves identification of OUD in the hospital and brief counseling, with facilitated referral to a community-based clinic for induction of MAT and ongoing care following hospital discharge.⁶³ Another variation uses a program nurse to identify inpatients with OUD, a bridge clinic for initiation of methadone following discharge with provision of psychosocial services (case management, group health education, counseling), and transition to another OTP for long-term management; such a program could be adapted for office-based prescribing of buprenorphine/naloxone.⁶⁴ This model requires hospital support for initial development of inpatient consult services.

Like the model involving ED initiation, an important advantage of inpatient screening and initiation is that it identifies patients with complex morbidity and high risk of mortality who otherwise may have had limited or no access to MAT. Likewise, inpatient initiation appears to enhance retention in care rates versus simple referral for outpatient initiation of MAT after hospitalization. Like the ED initiation model, this model of care focuses on the inpatient aspect, but could be linked to one of the office-based models of care described above for ongoing management. Patients initiated on methadone would not be eligible for referral to office-based care.

Integrated Prenatal Care and Medication-Assisted Treatment

This practice-based model involves the provision of prenatal care to pregnant women who are treated with buprenorphine in primary care. Women receive prenatal and postpartum care, with care continued in an office-based setting after birth. Psychosocial services are provided on-site as well as through affiliated OTPs.

Like the models of ED and inpatient MAT initiation, this model can identify women with limited or no access to care who come into contact with the medical system for prenatal care and might benefit from MAT. In addition, women may be more amenable to MAT in the prenatal setting due to concerns about the fetus and the desire to integrate care in one location. An additional advantage of this model is that it provides ongoing care in the postpartum period, providing additional continuity. Outcome studies conducted in OTP settings suggest that there is a reduction in Neonatal Abstinence Syndrome when pregnant women with OUD are maintained with buprenorphine rather than methadone.^65,66 This model is typically financed through existing Medicaid and other insurance reimbursement. A potential disadvantage is the need to transition at some point to a setting that can provide ongoing, long-term care, unless the office-based setting is equipped to do so. In one model (Southern Oregon), ongoing care is provided through transition to a primary care clinic that can provide MAT.

One Stop Shop Model

The One Stop Shop model was developed in response to an outbreak of HIV infection in rural Indiana due to sharing of infected syringes.⁶⁷ Based in an existing mental health clinic, it provides integrated care including management of HIV/HCV infection, MAT, mental health, and primary care needs, as well as other services including syringe exchange.⁶⁸ This practice-based model focuses on use of extended-release naltrexone as the pharmacological component. Peer navigators and social workers provide coordination with primary care providers. Because it is based in an existing mental health clinic, this model provides comprehensive on-site psychological services, including a visiting psychiatrist who is available on a weekly basis for consultation. Financing is from a combination of existing Medicaid and federal funding.

An advantage of this model is that it makes use of an existing mental health clinic to provide comprehensive integrated care, including extensive psychosocial services under a single roof. However, Key Informants noted that this model represents a unique response to the HIV outbreak and may not be reproducible in other settings due to the resources and unique clinical setting (i.e., an existing mental health clinic prepared to provide MAT) required. In addition, this model was implemented recently, with more data needed to understand how successfully it can be implemented.

Facilitators and Barriers for Implementing Medication-Assisted Treatment in Primary Care

Our Key Informants and literature review identified a number of important considerations for implementing MAT in primary care. Insufficient institutional support is frequently cited as a barrier to implementation.^87,88 Institutional support may include sponsored training, resources and staffing for coordination and integration of care, and provision of nonphysician staff with expertise in OUD in order to implement a team-based approach, utilizing the skills appropriate to each profession, as well as offloading some of the burden from prescribing physicians. Primary care physicians also report important knowledge gaps in the area of addiction. These gaps reduce the likelihood that they will prescribe MAT unless they have ready access to addiction expertise (e.g., for complex patients). Addiction expertise could be accessed through telehealth initiatives (e.g., Project ECHO), mentored prescribing (e.g., PCSS-MAT), coordination with local OTPs or experts in addiction (e.g., Hub and Spoke model or Co-OP model), or other methods. Barriers to telehealth include substantial start-up costs to be HIPAA-compliant, the need for ongoing resources for staffing and maintenance, and variable reimbursement. Implementing MAT also requires the integration of enhanced psychosocial services that may not be readily available in all primary care settings. Because provision of MAT involves multiple practitioners with varying types of expertise, improvement in communication and exchange of health information could greatly facilitate implementation.

Another consideration is whether there are enough patients and sufficient reimbursement to justify the resources and time required to implement MAT in primary care settings. Key Informants noted that there needs to be a minimum number of waivered physicians available to provide cross-coverage to avoid burn-out among prescribing physicians. In rural settings, Key Informants observed that travel time can be a significant barrier, with some patients facing a 2-hour commute to clinic; this can result in high travel costs and jeopardize the ability of patients to maintain employment.⁸⁹

Key Informants and the literature describe other barriers to implementation of MAT in primary care settings.^87,88,90 A key barrier is the relative lack of physicians with an FDA waiver to prescribe buprenorphine for treatment of OUD. In December 2013, the average state had only eight waivered physicians per 100,000 residents.⁹¹ Increasing the limit on the number of patients that a physician can prescribe buprenorphine for OUD (currently 30 or 100) could be more effective at increasing buprenorphine use and access than increasing the number of addiction treatment facilities or increasing the number of waivered physicians.⁹¹ One study found that the greatest impact on the amount of buprenorphine prescribed was the number of waivered physicians able to treat up to 100 patients with buprenorphine.⁸⁵ Although some Key Informants felt that the current patient limits could be a barrier to implementation, most primary care clinicians are not close to the prescribing limit and there are concerns that increasing the limits could result in suboptimal care. Most (70% to 95%) physicians prescribing buprenorphine never turned away any patient because of patient prescribing limits.⁹² As noted above, there seems to be an unwillingness on the part of some physicians to prescribe, even though they have a waiver.⁹⁰ The same survey found that about two-thirds of physicians with a buprenorphine waiver elected to not be included on the public Centers for Substance Abuse Treatment Locator List in 2008; among these, about two-thirds reported no prescribing of buprenorphine in the last 90 days. Among physicians on the Locator List, 86 percent reported prescribing in the last 90 days. A related barrier is that DATA 2000 only permits “qualifying physicians” to prescribe schedule III, IV, or V medications for treatment of OUD. The inability of physician assistants and nurse practitioners to prescribe buprenorphine is especially important in rural areas and low income clinics, where these providers often outnumber physicians. One Key Informant noted that in Oregon, such providers can prescribe any amount of schedule II opioid for pain, but cannot prescribe buprenorphine for OUD. Pharmacists also play an important role in providing MAT and could assist with dispensing, monitoring for adherence and diversion, and patient education.

Key Informants consistently noted that stigma towards MAT remains an important barrier to implementation. Surveys of physicians⁹⁰ describe stigma as pervasive and present among physicians, clinic staff, patients, law enforcement, policymakers, insurers, and the community. Key Informants noted that some patients do not even want to be in the same waiting room as patients who are receiving MAT. This could result in significant barriers due to the need to create separate clinic areas. In some states and other settings, abstinence is still viewed as a “better” treatment than MAT, despite evidence to the contrary. The perception persists that using an opioid agonist is replacing one addicting drug with another and promotes a preference for detoxification and abstinence rather than agonist or antagonist therapy. In rural settings in particular, Key Informants noted that MAT is often discouraged due to these beliefs. The Key Informants noted a general lack of training and understanding⁹⁰ regarding MAT even among physicians, and emphasized the need for education of physicians as well as the community regarding the evidence on effectiveness of MAT in order to increase the number of buprenorphine waivered physicians, increase uptake of MAT by patients, and increase buy-in among the community.

Other barriers to prescribing buprenorphine for OUD frequently cited in a survey of family physicians in Vermont and New Hampshire includes inadequately trained staff, insufficient time, inadequate office space, and cumbersome regulations.⁹¹ Several Key Informants noted that a fear of potential Drug Enforcement Agency site visits,⁹³ as per DATA 2000, was a deterrent to obtaining a buprenorphine waiver.

Key Informants also noted barriers to use of extended-release naltrexone in primary care settings. These include unfamiliarity with its use (this medication was approved by the FDA for treatment of OUD in 2010), perception of low patient demand (due in part to its mechanism of action as a pure opioid antagonist), the need to taper patients off opioids prior to starting naltrexone, high cost, and potential for overdose in patients who relapse, since they are no longer opioid-tolerant.

Reimbursement remains an important barrier.⁸⁷ For example, although nurse care managers in the Massachusetts model are reimbursed for their services, people serving similar functions in other models are not necessarily reimbursed in the same way. Several Key Informants noted that lack of reimbursement is a barrier to use of extended-release naltrexone. In the Project ECHO model, off-site experts provide consultative expertise to primary care providers. There is no doctor-patient relationship, and therefore these services are not reimbursable. Key Informants also noted variability in policies related to reimbursement of provision of telemedicine services in which there is an established, direct doctor-patient relationship. Without adequate reimbursement, implementation of MAT models of care in many primary care settings is unsustainable financially. Key Informants also noted onerous prior authorization requirements as a barrier to prescribing buprenorphine, as well as arbitrary limits on the treatment duration and doses. A survey of 45 states found that in 2013, only 11 percent of states had Medicaid policies that excluded coverage for methadone and buprenorphine, whereas nearly three-quarters (71%) had policies to cover both buprenorphine and methadone in Medicaid enrollees.⁹⁴ However, there was also an increase in adoption of policies that could hinder access to buprenorphine or methadone, such as prior authorization requirements.

Training, Certification, and Staffing Needs

DATA 2000 allows physicians to provide MAT using buprenorphine outside of licensed OTPs if they complete 8 hours of training and submit an application to receive a waiver. Physicians who obtain a waiver may be subject to periodic DEA audits of patient records (a potential barrier to obtaining a waiver). DATA 2000 further specifies that brief counseling be offered in conjunction with buprenorphine; this can be provided by the physician or nonphysician staff. Models that integrate treatment of OUD with management of other chronic conditions require expertise in management of those conditions; this can be provided by the same physician that is managing the OUD or by other clinicians (not necessarily a physician).

Additional staffing and training requirements vary depending on the model of care. Several models use a designated staff person to support the prescribing physician and serve as a main point of clinical contact. In the Massachusetts model, an RN case manager performs screening, supports the prescribing physician, and coordinates care and in Project ECHO, nurse practitioners and physician assistants assume similar roles. There are no formal certifications or trainings required to fulfill these roles, though DATA 2000 buprenorphine waiver trainings are open to and attended by nonphysicians. The success of such models is likely to depend to a large degree on the knowledge and skill that such people have in the area of addiction. Additional staffing largely depends on the types of psychosocial services that are offered and may include psychologists, social workers, peer counselors or mentors, psychiatrists, addiction specialists, and others.

Medication-Assisted Treatment Models of Care

We identified six trials on the effectiveness of MAT models of care in primary care/office-based settings^{36,37,46,48,61,62} (Table 4). Two trials compared buprenorphine/naloxone with more intensive versus less intensive counseling in the OBOT (Yale) model.^36,37 One trial compared buprenorphine/naloxone with more intensive versus less intensive counseling among HIV-infected patients in the BHIVES model⁴⁸ and another trial of HIV-infected patients compared clinic-based buprenorphine/naloxone in the BHIVES model versus case management and referral to an OTP.⁴⁶ One trial compared the Emergency Department Initiation of OBOT model with buprenorphine/naloxone versus referral for treatment (with or without a brief intervention)⁶¹ and one trial compared the Inpatient Initiation of MAT model with buprenorphine/naloxone versus linkage to care.⁶² No trial compared the effectiveness of one MAT primary care model versus another.

Detailed tables of included trials for Guiding Question 3 are available in Appendix F.

Psychosocial Interventions

A number of trials have evaluated the comparative effectiveness of different psychosocial interventions given as a component of MAT. However, relatively few trials on psychosocial interventions have been conducted in office-based settings. A Cochrane review included 35 trials on the effectiveness of psychological therapies plus any agonist maintenance treatment as a component of MAT for OUD (Table 5).⁹⁵ Thirty-one trials were conducted in the United States. In six trials the pharmacological component was buprenorphine/naloxone; the remainder evaluated methadone (no study evaluated naltrexone). Of the trials, only one was conducted in a primary care/community-based setting.³⁶ It compared standard medical management with brief (20 minutes/session) medically-focused counseling versus extended medical management with more in-depth counseling (45 minutes/session) in patients prescribed buprenorphine/naloxone and found no clear differences in effectiveness. We identified nine additional trials that evaluated the effectiveness of more intensive psychosocial interventions or compared one psychosocial intervention versus another in office-based settings (Table 4). The comparisons evaluated were internet-based community reinforcement approach plus contingency management versus contingency management alone,⁹⁶ cognitive behavioral therapy versus standard counseling,^97,98 network therapy versus standard medication management,⁹⁹ cognitive behavioral therapy plus directly observed, thrice-weekly buprenorphine versus physician management with weekly buprenorphine, brief versus extended counseling,^100–102 guided drug counseling plus standard medical management versus medical management alone,³⁷ and brief physician management versus brief physician management plus nurse-administered drug counseling and adherence management.⁴⁸ The evaluation of different comparisons makes it difficult to assess overall findings of the trials, but in most studies there were no clear differences in outcomes between different psychosocial interventions.

Detailed tables of included systematic reviews for Guiding Question 3 are available in Appendix G.

Pharmacological Therapies

A number of trials evaluated the pharmacological component of MAT. In all trials, psychosocial interventions were also provided, though the psychological component was often not well-described. Relatively few trials were conducted in office-based settings. Some trials evaluated methadone and sustained-release morphine, which are not approved by the FDA for this indication. We included those medications in this section as they could inform future MAT strategies if they become available in the United States.

Special Populations

One Cochrane review evaluated the effectiveness of MAT in pregnant women, but evidence on effectiveness of FDA-approved office-based treatments for MAT was extremely limited (Table 5).¹²¹ In addition, although three trials (sample sizes 18, 30, and 175) evaluated buprenorphine versus methadone maintenance treatment, none were conducted in primary care or community-based settings. One trial evaluated buprenorphine/naloxone in community settings for treatment of OUD in young people (15 to 21 years of age), but did not meet inclusion criteria because it compared treatment for 12 weeks versus a 2-week taper.¹²² A Cochrane review evaluated effectiveness of oral agonist treatment for OUD in injecting drug users on risk behaviors and rates of HIV,¹²³ but did not focus on medications approved for use in office-based settings and only included two trials in which patients were managed in primary care settings (Table 5).^124,125 A trial of HIV-infected patients with OUD found no difference between office-based treatment with buprenorphine/naloxone versus referral to an OTP in HIV RNA levels and CD4 counts.⁴⁶ Trials of MAT in office-based settings primarily enrolled patients with OUD due to heroin; we identified no systematic review or randomized trial on effectiveness of MAT in primary care settings, specifically patients with OUD related to prescription opioids. Another Cochrane review of MAT for OUD related to prescribed opioids included six trials that found that methadone or buprenorphine appeared equally effective for outcomes related to opioid use and treatment retention (Table 5).¹²⁶ Five of the trials were conducted in the United States, but none of the studies were conducted in primary care settings.

New and Innovative Strategies

Key Informants uniformly noted that the most promising models of care are those that emphasize the integration of management of OUD with primary care and other medical and psychological needs. The chronic disease management paradigm is particularly suitable for populations with OUD who also have other conditions that require ongoing care, such as HIV or HCV infection.¹²⁹ The BHIVES model was specifically designed to integrate office-based treatment with buprenorphine/naloxone with HIV management. Some important innovations in implementation of MAT models of care include the use of a nonphysician glue person (e.g., OBOT [Yale], Massachusetts Nurse Care Manager model, ECHO Project), integration of more comprehensive psychosocial services (e.g., One Stop Shop, Medicaid Health Home Model), coordination and integration of office-based management with centralized centers of excellence (e.g., Hub and Spoke, Co-OP), and identification and initial treatment in ED, inpatient, or prenatal settings. Peer-delivered recovery support services are promising and could be integrated into primary care settings;¹³⁰ as of 2007, such services are Medicaid reimbursable. Several Key Informants noted that models of care that also integrate education, training, and outreach, such as the Massachusetts Nurse Care Manager model, are important for increasing the pool of buprenorphine-waivered physicians, decreasing stigma, and increasing uptake of MAT, while also promoting higher-quality care. Existing resources such as PCSS-MAT, which provides physician training and access to a national network of experts in MAT who can provide mentoring to those less experienced in prescribing buprenorphine, could be leveraged by models of care that lack resources for their own educational and training component; such resources were used successfully in the initial dissemination and expansion of office-based buprenorphine in the United States. Utilization of existing training and educational resources would also be more efficient than developing new resources in each implementation setting.

Recent MAT models focus on the identification of patients with OUD and initiation of treatment in the ED, inpatient, and prenatal settings. These strategies can help identify patients with OUD who otherwise might not have access to primary care, have a higher prevalence of OUD (e.g., in the ED and inpatient settings), or facilitate initiation and engagement in treatment. Ideally, such models of care would be linked to an integrated, office-based model that can provide ongoing management.

In rural settings, major barriers to MAT include the lack of addiction and psychiatric expertise, distances that patients must travel to access care, lack of buprenorphine-waivered physicians, and negative attitudes and beliefs regarding MAT. Strategies to overcome these barriers include Web-based learning networks (e.g., Project ECHO), use of telemedicine for consultation with experts, utilization of nonphysician providers in key roles (e.g., screening, counseling, coordination of care, provision of primary care), and educational and outreach efforts. In the Southern Oregon Model, for example, local stakeholders meet regularly and discuss issues in management of OUD and develop practice standards using a collaborative model. One Key Informant has developed and evaluated computer-assisted delivery of cognitive behavioral therapy for addiction.^131,132 Resources such as these could supplement face-to-face psychosocial services and would not be constrained by geographical barriers. In rural settings, the availability of extended-release formulations (e.g., currently approved extended-release naltrexone and emerging products such as implantable and injectable buprenorphine preparations) could potentially reduce the need for frequent visits, particularly in less complex patients who have long distances to travel, and if coupled with psychosocial services conducted over the phone or via the Web.

MAT models of care in primary care settings could also integrate pharmacist-based management strategies. A recent small (n=12 patients) pilot project evaluated a physician-pharmacist collaborative model in which patients were managed using a drug therapy management model.¹³³ The pharmacist conducted intake assessments and followup appointments and documented each interaction after debriefing with a physician, who appended additional notes as needed and cosigned records. The pharmacist was responsible for gathering data from outside providers and pharmacies regarding prescribed medications and results of urine drug testing. Prescriptions were written by the physician or called in by the pharmacist. In addition, the pilot study projected that the model would be cost savings for the health system. Another 2-year pilot study in San Francisco evaluated a tiered model with centralized induction and stabilization followed by management in a community-based center, with buprenorphine dosing and dispensing provided through a designated pharmacy.¹³⁴ The pharmacist at the dispensing pharmacy worked in collaboration with the clinicians at the community center, with a secure database specifically designed to facilitate communication. However, for both models, details regarding the provision of psychosocial services and coordination of care within this model are limited.

Implications for Diffusion of Medication-Assisted Treatment

Key Informants consistently noted that MAT is effective in office-based settings, but access remains limited, particularly in rural settings. Increasing the number of buprenorphine waivered physicians as well as the number of buprenorphine waivered physicians who actually prescribe are critical for increasing the diffusion of MAT. Enhanced use of extended-release naltrexone could also increase diffusion of MAT since it does not require a waiver to prescribe and provides patients with additional options. As an opioid antagonist, naltrexone may be preferred by patients who do not wish to use opioid agonist or partial agonist therapy.

This report describes a number of MAT models of care viewed as effective or promising by Key Informants. Although evidence is lacking with regard to how one model of care performs compared with another, comparative effectiveness research may not be the most important determinant for informing further diffusion of MAT. Rather, the most effective model of care is likely to depend in part on the specific implementation setting, including unique characteristics of the target patient population (e.g., HIV infection, pregnant, or adolescent), what resources are available locally, and financing options. Implementation of the Hub and Spoke or Co-OP models, for example, requires a relatively local center of expertise in addiction that is willing to partner with community centers in an integrated model. A model developed for patients with HIV infection requires expertise in both OUD and HIV care. In rural settings, models of care that integrate Web-based training, consultation, and mentorship may be needed to overcome the lack of local expertise. One support model, for example, is the Oregon Addiction Education and Prevention Initiative, in which academic medical center addiction medicine specialists partner with accountable care organizations to conduct DATA 2000 waiver training for rural primary care providers, who are then linked to PCSS and offered personal ongoing phone consultation support in MAT management. In some cases, effective diffusion of MAT may involve adaptation of an established model of care to the needs of the particular setting. For example, the Massachusetts Nurse Care Manager model represents an adaptation of the OBOT model developed at Yale and the BHIVES model represents an adaptation of the OBOT model for patients with OUD and HIV infection. Models of care could also integrate models that target different parts of the treatment process. For example, models that involve ED or inpatient screening for OUD and initiation of treatment could be integrated with models that provide ongoing care based on the Massachusetts Nurse Care Manager or Hub and Spoke models.

Given the barriers to implementing MAT in primary care settings, effective strategies for implementation are likely to require multifactorial interventions that involve partnerships between payers and clinics that use financing, contracting, policy change, process improvement to improve workflow, and customer input to facilitate organizational change. Although one such intervention (Advancing Recovery) has been shown to increase access to MAT in addiction treatment settings,¹³⁵ studies on the effects of Advancing Recovery in primary care settings are not yet available. Several Key Informants also commented that with increased diffusion of MAT comes the possibility for suboptimal provision of care. They noted the need for clear standards to measure the quality of care and ensure that care is adequate. Key Informants also noted that there is a general lack of knowledge regarding treatment of addiction in primary care, and that dissemination of addiction education into primary care could help with diffusion of MAT in primary care.

Ethical, Equity, and Cost Issues

Key Informants noted equity issues with regard to access to MAT in rural areas due to lack of prescribing physicians, ongoing stigma, and lack of policy and funding support. Efforts to expand MAT in Medicaid programs and Federally Qualified Health Centers represent an opportunity to increase equity. Although evidence indicates that OUDs often begin during adolescence, no models of care have been developed to address adolescent populations.¹³⁶ A multi-site clinical trial documented improved short-term outcomes for adolescents and young adults supported on buprenorphine/naloxone compared with those who completed a brief taper.¹²²

Key Informants consistently noted that MAT is effective when, and it is important from an ethical standpoint that, patients have access to these treatments and be provided with accurate information about the risks and benefits of MAT and alternative treatments. Although substance use disorder benefits are included as Essential Health Benefits in the Affordable Care Act, insurers may try to avoid paying for MAT medications through onerous prior authorization requirements or arbitrarily limit the duration or dose of therapy.¹³⁷ Key Informants noted that prevention of buprenorphine diversion has been a major concern of some payers and providers and in some cases has impacted the ability to provide MAT, due to the effects of efforts to prevent diversion.

Financing remains a major issue in many settings. They noted that some models have been run largely by volunteers or are unable to remain financially viable due to inadequate reimbursement and a lack of state or other financial support. One Key Informant noted that some private clinics have gone bankrupt trying to work with Medicaid. Some Key Informants noted that the 100-patient limit for prescribing buprenorphine may make provision of MAT noneconomically viable for some physicians. Other Key Informants noted that some for-profit clinics involve several physicians banding together to increase the number of patients treated and increase economic viability, but this could result in provision of MAT which may not meet quality of care standards. Key Informants noted that showing that MAT is cost-effective or even cost-savings in the long run would be very helpful for convincing policymakers and clinicians to support and use MAT.

Areas of Uncertainty and Future Research Needs

Based on our review of the literature and Key Informant input, we identified a number of important areas of uncertainty regarding MAT that warrant additional research. These include:

• Research to identify factors associated with high-quality care and how to measure it. With improved access to MAT, it is also critical to insure that the quality of care that is delivered is high. This will require development of new quality of care indicators for use of MAT in primary care settings.
• Research on management of patients with OUD and concomitant chronic noncancer or cancer pain,^138,139 benzodiazepine use, and/or alcohol use disorder (e.g., use of buprenorphine/naloxone for transitioning off high doses of opioids in patients with chronic pain). Treatment of OUD in patients who also have pain is a major challenge given the high prevalence of opioid prescribing. A systematic review of 10 studies of limited quality evaluated the role of buprenorphine for management of chronic pain, but only one study was conducted in primary care.¹⁴⁰
• Research on effectiveness of MAT in patients with prescription OUD. Most research on MAT has focused on patients with heroin use disorder. Research would be helpful for determining the degree to which evidence on MAT for heroin use disorder can be extrapolated to those with prescription OUD.
• Research on effectiveness and safety of mid-level prescribing of buprenorphine, such as by nurse practitioners and physician assistants. Currently, DATA 2000 only permits physicians to prescribe buprenorphine for OUD. Allowing mid-level providers to prescribe buprenorphine could help improve access in rural areas with few or no physicians.
• Research to identify patients more likely to benefit from more intensive psychosocial services, and methods for effectively targeting specific types of psychosocial services. The need for more intensive psychosocial services is likely to vary. Understanding which patients require which services would be very helpful for designing and implementing effective models of care.
• Research on effectiveness of peer-delivered support services as part of MAT in primary care settings.¹³⁰
• Research to understand optimal methods for coordination and integration of care. Although Key Informants consistently noted that this is a critical component of successful MAT models of care, methods for coordination and integration of care varied among models and no study evaluated the effectiveness of different coordination and integration methods.
• Research to better understand the costs and cost-effectiveness of implementing MAT models of care. Although long-term treatment with buprenorphine/naloxone in office-based settings appears to be cost-effective¹⁴¹ and provision of MAT using the Hub and Spoke model in Vermont is associated with decreased health care utilization and costs than treatment of OUD without medication,¹⁴² there are relatively few cost- and cost-effectiveness studies and analyses have not compared different MAT models of care or evaluated the use of newer pharmacological therapies. Such research would be of particular importance for policymakers, and that such research should address societal outcomes impacted by OUD (e.g., ability to work, criminal activity) in addition to impacts on drug use.
• Research on effective methods implementation of MAT models of care in primary care settings and increasing uptake of MAT. Although some multicomponent implementation strategies appear to be effective for enhancing access, they have not yet been studies in primary care settings.¹³⁵
• Research to better understand optimal duration and doses of treatment. This is particularly important because otherwise payers may (and sometimes do) impose arbitrary duration limits for MAT.
• Research on effectiveness of telehealth and Web-based training, mentoring, and educational resources. These would be particularly useful in rural and other settings where addiction and other expertise are not available locally. As noted elsewhere in this report, one Key Informant described a Web-based cognitive-behavioral resource that has been developed^131,132,143 and another described psychiatric consultation using computer tablets.
• Research on effectiveness of alternative medications or formulations (e.g., implantable and injectable buprenorphine preparations). Such formulations could reduce the frequency of followup, increase uptake and compliance, and mitigate barriers related to long travel distance. However, there is almost no evidence on injectable buprenorphine used in primary care settings.
• Research on effectiveness of methods for reducing diversion (e.g., use of extended-release medications, thrice weekly observed dispensing, or pharmacy-based dispensing). Pharmacy-based dispensing is done in Canada and Europe for buprenorphine and methadone prescribed in primary care and has been piloted in small studies in the United States.^133,134 Key Informants noted that preventing diversion has been a major concern of some payers and policymakers.
• Research to understand why buprenorphine waivered physicians don’t prescribe, factors associated with prescribing, and methods to increase prescribing. The gap between the number of waivered physicians and the number prescribing indicates that that there is substantial untapped capacity to prescribe buprenorphine.⁹²
• Research to better understand patients who are appropriate for office-based treatment versus those who require treatment in an OTP. Key Informants noted that current methods to determine who is appropriate for office-based treatment are largely based on anecdotal experience.
• Research on patients who are more likely to benefit from extended-release naltrexone, comparative effectiveness of buprenorphine/naloxone versus extended-release naltrexone, and optimal models of care for provision of extended-release naltrexone. Most models of care have focused on provision of buprenorphine/naloxone, and there is very little evidence on use of extended-release naltrexone in primary care settings. Although there is evidence supporting the efficacy of extended-release naltrexone, Key Informants reported the perception that this treatment was not in high demand by patients and that some patients might not do well with opioid antagonist therapy. In addition, a recent study found a low rate of linkage to ongoing treatment with extended-release naltrexone following an initial injection during inpatient opioid detoxification.¹⁴⁴ On the other hand, expanding the medication choices for patients could increase uptake and that extended-release naltrexone may be associated with less stigma by some patients and providers.
• Research on effectiveness of methadone for office-based treatment. Methadone is not authorized under DATA 2000 but has been evaluated in office-based settings in some clinical trials^118,119 and observational studies in the United States,^145–147 and is used in primary care settings in other countries. Primary care providers in Canada, parts of Europe, and some other countries prescribe methadone for directly observed daily dispensing in local pharmacies. This model has not been tested in the United States, but could expand access to OUD treatment while limiting diversion.
• Research to understand optimal MAT models of care in adolescents and children,^122,136 who often differ from adults in their treatment needs.¹⁴⁸ In 2014, an estimated 18,000 adolescents had heroin use disorder and 168,000 had OUD related to prescription opioids,³ but data indicate that treatment for OUD is markedly underused in this population.¹⁴⁹

Ongoing Studies

We identified several ongoing randomized trials of MAT models of care in primary care settings that may address some of the research gaps described above (Table 6). One ongoing trial compared effects of an organizational readiness intervention (including implementation tools and activities) plus an integrated collaborative care service delivery intervention (based on a chronic care model) versus usual care for implementing substance use disorder treatment in primary care.¹⁵⁰ Two ongoing trials focused on MAT models of care that involve screening and initiation of MAT in emergency department¹⁵¹ or inpatient¹⁵² settings. One other trial compared effects of group visits (5 to 10 patients with primary care provider and behavioral specialists) versus usual care (individual visits) in patients receiving buprenorphine/naloxone.¹⁵³ Another trial compared a strategy of an interim bridging buprenorphine treatment intervention for patients on a waitlist for MAT.¹⁵⁴ An AHRQ-funded demonstration project is focused on improving access to MAT in rural primary care practices.¹⁵⁵

Table 6

Ongoing studies of MAT for OUD.