Behavioral Health and the Comprehensive Primary Care (CPC) Initiative: findings from the 2014 CPC behavioral health survey

Primary care plays a critical role in health care and payment reform redesign. One of the most promising new care models is the Patient-Centered Medical Home (PCMH), which aims to treat the whole person, not only particular conditions or body parts, in a timely, coordinated, and comprehensive manner [1–5]. There has been increasing awareness that despite historically based separation in clinical training, treatment locations, insurance coverage, and billing practices in the United States, it is virtually impossible to separate behavioral health care needs from primary care [6]. Behavioral health conditions, which encompass mental health and substance abuse disorders, are usually identified and treated in primary care settings, [7, 8] a setting in which patients often prefer to receive behavioral health care [9]. Many consider it essential to have behavioral health care providers in primary care settings to promote health behavior change, treatment adherence, develop comprehensive treatment plans, and improve outcomes [7, 10, 11]. Given the high prevalence of both behavioral health conditions, and psychiatric comorbidities among individuals with chronic medical conditions, PCMHs that address behavioral health care needs could improve patient health and care quality and reduce costs [11–19].

Evidence-based approaches to addressing behavioral health needs in primary care often focus on integrating behavioral health providers with the primary care team [20, 21]. Behavioral health integration (BHI) involves care conducted with primary care and behavioral health providers working together with patients and families to provide systematic, patient-centered care [22, 23]. Examples include assisting a patient with health behavior change or performing an evidence-based intervention for depression. Multiple studies have demonstrated effectiveness of team-based approaches to managing behavioral health conditions in primary care [24] and addressing behavioral health and physical health needs simultaneously [25, 26].

In 2007, the American Academy of Family Physicians, American Academy of Pediatrics, American College of Physicians, American Osteopathic Association developed initial PCMH principles, and in 2014, they developed joint principles for BHI in PCMHs [13]. Although numerous randomized controlled trials have found that BHI improves care processes and outcomes, [27] and in spite of recommendations to incorporate BHI into PCMHs, widespread adoption of BHI has not occurred, [7] particularly outside of large integrated health systems (e.g., Department of Veterans Affairs, which mandated integrated care in 2008) [28].

Increasingly, primary care practices are conducting BHI [29]. However, several barriers exist to integrating behavioral health providers into primary care [30], including payment [10, 31, 32]. Primary care practices interested in BHI typically have fragmented payment models, making sustainable funding difficult. However, despite historical differences leading to separate payments for behavioral health and medical services, novel payment models exist aiming to better support primary care BHI.

A key challenge to measuring BHI implementation in primary care practices is that it is difficult to ascertain whether, how, and to what extent practices have adopted BHI. Most studies that evaluate any type of health care use employ administrative data to document the presence or absence of a particular diagnosis or treatment. However, being able to discern whether practices have co-located providers or whether primary care physicians and behavioral health providers communicate with one another is virtually unknowable from these data sources. Therefore, using surveys to learn more about whether and how BHI is operationalized within a given practice provides valuable and typically unavailable information.

This survey provides a benchmark to compare future assessments of BHI, both with and outside of CPC, including those that will become feasible with the implementation of Medicare payment for BHI (i.e., four new codes created for the 2017 Medicare Physician Fee Schedule to allow payment to health care providers for furnishing BHI services) in 2017 [33]. Further, our study identifies key associations between practice characteristics and the extent of BHI implementation, which will allow us to illuminate possible factors that facilitate success of implementing BHI.

We found expected differences between survey respondents and survey non-respondents. There were regional differences across the seven regions in practices that did and did not respond to the survey (χ² = 17.47, p<0.01) and BHI respondents were more likely than non-respondents to be from a multiple specialty practice (χ² =161.00, p<0.0001).

Answers to individual questions within a given domain illustrate variation possible within that domain (see Additional file 2). For example, the mean score for the integrated space domain question was 2.39 (sd=1.24; Table 3). Within that domain, the individual item question found that 41% of practices had entirely separate space, with the remaining 59% of practices having some level of integrated space, including 24% with fully integrated space (i.e., clinic space shared between behavioral health providers and PCPs). In the training domain, most practices reported that PCPs had been trained in principles of behavioral health care, with a mixture of those who reported feeling comfortable handling these issues and others not. In contrast, other clinical professionals and non-clinic staff in most practices did not address behavioral health issues or play a very limited role. In the access domain focused on scheduling, the preponderance of practices reported PCPs’ ability to accommodate addressing behavioral health issues, but a much more limited number of practices had the ability for behavioral health providers to accommodate addressing behavioral health issues. In the communication and coordination domain, respondents indicated that PCPs and behavioral health providers engaged in limited communication and did not regularly meet to review cases. In the treatment planning domain, shared care plans between PCP and behavioral health providers were rarely recorded and shared electronic health records were not routinely available. Many practices were able to link patients to community resources and follow up with patients regarding their behavioral health needs. In the resource domain, most practices reported not having readily available resources for patients’ behavioral health needs, such as staff and time.

Findings from disease-specific domains revealed high levels of disease screening (depression, anxiety, pain, alcohol use disorder, and cognitive function; Additional file 2). However, practices had treatment targets and consistent outcome monitoring for depression, anxiety, and pain; practices did not regularly provide care for alcohol use disorders or address cognitive disorders, nor did they typically have treatment targets for these disorders.

This paper provides a unique assessment of the real-world baseline status of BHI implementation in a multi-state primary care redesign initiative. We found that being part of a multispecialty practice, having a PCMH designation, having any behavioral health provider (especially a psychologist or social worker), and any communication between behavioral health providers and PCPs were each associated with significantly higher BHI domain and disease area scores. Conversely, practice ownership, metropolitan location, proportion of African American patients, and number of patients per FTE provider had almost no discernible impact on scores. We are unaware of any other survey that assesses BHI across practices, let alone one that includes this many primary care practices and patients, and that documents a broad range of experiences at the initiation or expansion BHI, particularly within the context of a primary care redesign effort. BHI is typically inferred from administrative data, and this survey provides key findings not available from those data sources.

Our study had several strengths including its breadth, depth, and response rate. We achieved a high response rate to this survey because practices were accustomed to regular reporting requirements as part of CPC participation, in addition to the multiple reminders they received to complete the survey. Findings also identified strengths and opportunities for improvement in BHI. Bringing together two historically disparate clinical treatment disciplines and cultures can be challenging and rewarding, and consistent with other findings, behavioral health providers were co-located in practices only roughly half of the time, with a very limited number of practices having any psychiatrists available [29]. Alternatively, most practices (83%) indicated that they had some form of communication with behavioral health providers, and nearly half (46%) had a behavioral health provider who had been involved with the practice for at least one year. Although regional differences were not a focus of this paper, regions that had initiated BHI prior to CPC were both more likely to be represented in the survey and were more likely to have higher domain scores (e.g., Oregon; data not shown).

Large national model tests programs such as CPC are investing substantial financial and educational resources in assisting practices to move from encounter-based, fee-for-service patient care towards a team-based, patient-centered approach with population-based payments. In the future, an increasing proportion of Medicare payments will be tied to quality or value, [42] which will likely continue to support BHI goals. As an essential element of primary care, inclusion of behavioral health holds great promise. As the survey indicates, however, primary practices have multiple opportunities to deepen their BHI work.

Though no causality can be implied from survey results, primary practices with experience in expanding the medical neighborhood, working with other medical specialties (as in multispecialty practices) and developing strong communication between a provider and a specialist, had a higher degree of BHI implementation. For policymakers and practices integrating behavioral health, these insights provide new and useful data on opportunities and challenges of BHI integration linked to primary care transformation.

Survey limitations include that it was cross-sectional given at the start of BHI implementation among CPC practices; it does not address how BHI implementation changes over time. We did not formally evaluate survey psychometric properties (such as inter-rater reliability) nor conduct a validation study prior to fielding this quality improvement tool. We did not link survey responses to staff or patient satisfaction or cost data, although the overall independent CPC evaluation could do that work. Domain scores were likely substantially higher than typical PC practices, as included practices chose to participate in BHI and CPC, therefore, may not be representative of typical primary care practices. It is also unknown whether findings could be replicated in other practices with less financial and technical support, though increasing numbers of practice are being paid differently for primary care services [43]. Finally, we are not certain who within each practice completed each survey, and whether administrators included front line clinicians when completing the survey.

A final concern is the lack of a consistent definition of BHI, which makes assessment of progress towards BHI challenging. As others have noted, the field needs a “lexicon” that can consistently explain what BHI is and is not [23]. Tools such as the Agency for Healthcare Research and Quality (AHRQ) lexicon have been created to help further elucidate what is and what is not BHI [44]. Implementing such tools would allow for a clearer distinction that helps with better identification and implementation of BHI. Since many practices do not have onsite behavioral health providers, future research using clearer criteria could better delineate between practices with fully integrated clinical workflows and those that provide high degrees of coordination and co-management between primary care and behavioral health external to the practice.

This study found a few key factors that influenced BHI, some of which may be more easily attainable for practices than others. It may be easier to develop collaboration with a psychologist or social worker than a psychiatrist. Focusing specifically on communication and/or securing access to any type of behavioral health provider may assist practices towards successful BHI implementation, even in the context of significant variation in practice size, location, and patient demographic characteristics.

Innovative multi-payer initiatives such as CPC hold great promise for catalyzing primary care to more effectively address needs of patients with behavioral health needs. To continue to promulgate BHI, there must be adequate practice transformation supports that allow for variability in BHI implementation while maintaining consistency with an overall definitional framework [22, 23]. Multi-payer alignment regarding BHI should be encouraged and continue. Future studies should examine the impact of new payment models and billing practices on BHI to help achieve the value added proposition of BHI, and should assess BHI-related outcomes.