People with serious mental illness frequently contend with general medical conditions and have significantly shorter life expectancies than the general population (
1,
2). Yet the geographic, organizational, financial, and cultural separation of traditional behavioral health and general medical care systems is widely assumed to limit behavioral health consumers’ access to primary medical care (
3–
8). As such, improving the integration of behavioral and general medical services has become a focus of multiple policy initiatives (
5,
6,
9–
11).
In a recent review, the Agency for Healthcare Research and Quality defined integration as systematic communication and coordination across behavioral health and general medical providers (
12). Early models of integrated care, in which mental health providers were brought into medical settings to identify and treat mental health conditions, have been associated with improvements in symptom severity, treatment response, and disease remission (
12). Less is known about the opposite approach, in which primary care services are integrated into behavioral health settings (
12,
13). This approach is expected to show particular promise for persons who have frequent contact with behavioral health providers but not with general medical providers (
14,
15).
Recent reviews have also summarized the few studies of approaches for integrating primary care into behavioral health treatment for adults with serious mental illness, and they noted improvements in consumers’ general medical and behavioral health and in quality of care (
12,
13). However, the studies reviewed may not reflect all variants of this approach to integration that are occurring in the field. Specifically, most of the studies were conducted in large, integrated health systems, such as the Veterans Health Administration or other large health maintenance organizations, where primary care providers and behavioral health staff were already working together to provide care. Less systematic accounts of integration of primary care in smaller behavioral health systems described implementation barriers related to inadequate space for primary care activities and difficulty integrating primary care activities into the existing organizational work flow (
16). More research is needed on the benefits and challenges of integrating primary care into diverse behavioral health settings.
To address this gap, we provide a descriptive analysis of the Substance Abuse and Mental Health Services Administration (SAMHSA) Primary and Behavioral Health Care Integration (PBHCI) grants program, which supports the integration of primary care services into community behavioral health settings for adults with serious mental illness. Specifically, we describe the grantee organizations, their integrated care programs and implementation plans, and the challenges associated with this integrated approach at start-up and after one year of program implementation.
Methods
Participants
Participants were 56 publicly funded community behavioral health agencies or agency partners that provide mental health or substance abuse services and that received a PBHCI service grant in September 2009 (N=13) or September 2010 (N=43). The programs received up to $500,000 per year for four years.
PBHCI grantee characteristics
Each grantee’s model of service integration included required and optional program features. The required program features were screening, assessment, and referral for the prevention and treatment of general medical illnesses and risk factors, including hypertension, obesity, smoking, and substance abuse; a registry or tracking system to house data about consumer-level primary care needs and outcomes; care management, defined as individualized, person-centered planning and coordination to increase consumer participation and follow-up with primary care services; and illness prevention and wellness support services.
Examples of optional program features included colocation of primary care providers in behavioral health settings, supervision by a primary care physician, and embedding of nurse care managers.
Up to 25% of the grant could be used to fund infrastructure development for the program, and up to 20% could be used to support data collection, performance measurement, and assessment.
Data extraction from grant proposals
Project team members used qualitative and quantitative templates to systematically extract information from grantees’ funded proposals, including general characteristics of participating behavioral health agencies and their PBHCI programs, such as number of locations and type of primary care partner; information about current and targeted patient populations, such as volume, gender, age, and insurance status; and implementation plans. Data extraction was completed between October and December 2010.
During November and December 2010, project team members conducted hour-long, semistructured telephone interviews with core PBHCI grantee agency staff, such as program managers, to verify the accuracy of the extracted information. Sixteen grantees (37%) did not respond to repeated telephone and e-mail requests to meet.
Coding of quarterly reports
SAMHSA requires grantees to submit quarterly reports describing program implementation progress. We coded an item from these reports describing “barriers [grantees] have experienced in implementing [their] programs.” Reports submitted at baseline (after the first quarter of implementation) by 56 programs and at one-year follow-up by 55 programs were used for coding.
Categories of barriers were developed by consensus: together two raters examined a subset of responses from the baseline reports and came to agreements about thematic content. The raters then independently evaluated the remaining baseline reports. To assess interrater reliability, ten baseline reports were randomly selected to be evaluated by both raters. The raters were in agreement about the categorization of 13 of the 16 barriers identified, yielding an interrater reliability of 81%. The same approach was used with a randomly selected group of five follow-up reports to identify any new categories of barriers, and the two raters agreed on 16 of 19 categorizations (84%).
Results
Characteristics of PBHCI grantees
The 56 grantees collectively represented 65 behavioral health agencies at 86 sites or locations. Most grantees consisted of a single treatment site (N=34, 61%), and most were located in a suburban or urban setting (N=47, 84%). The characteristics of the sites before the grant varied tremendously. For example, annual patient volume ranged from 100 to 13,000 (median=1,585), and the number of consumers with serious mental illness served ranged from 14 to 9,800 (median=1,000).
A majority of grantees included only one participating behavioral health agency (N=50, 89%). Grantees with multiple agencies generally offered the same array of services at each participating agency, so results are reported at the grantee level. Prior to receiving a PBHCI grant, all behavioral health agencies offered outpatient mental health services, nearly all offered substance use disorder services (N=51, 91%), and most offered crisis or emergency (N=38, 68%) and residential services (N=31, 55%). Less than half of the behavioral health agencies offered any primary care services prior to receiving the grant (N=25, 45%).
Organizational partnerships
PBHCI grantees tended to include at least one primary care provider organization partner (N=43, 77%), of which a majority were Federally Qualified Health Centers (FQHCs) (N=31, 67%). Most partnerships between primary and behavioral health care organizations were relatively new (median=two years) when grant applications were submitted, although some partnerships were longstanding.
Organizational partnerships were operationalized through formal contracts (N=9, 21%), memoranda of understanding or agreement (N=28; 65%), letters of commitment or agreement (N=5, 12%), and unspecified arrangements (N=1).
Primary care partner organizations varied widely in annual volume of patients (median=15,000; range 2,518–150,000) and physical proximity to partner behavioral health organizations. Some primary care partners were located in the same building as behavioral health providers, whereas others were up to 23 miles away (median=1.5 miles).
Anticipated clientele
Most programs (N=48, 86%) planned to prioritize integrated care for a subset of consumers with serious mental illness. Criteria used to identify priority groups included demonstrating inadequate connections to primary care, such as an inability to name a primary care provider (N=31 programs, 55%); having high-risk clinical factors, such as obesity (N=14, 25%); and having been diagnosed with a chronic general medical condition, such as diabetes (N=14, 25%). Six (11%) programs planned to prioritize care of consumers using psychotropic medication, and four (7%) targeted consumers lacking health insurance.
Table 1 describes characteristics of consumers that PBHCI programs expected to serve. The number of consumers expected and the racial-ethnic composition and insurance status of the target populations varied widely by program.
Required program features
Screening and referral.
Grantees reported a wide range of capacities for providing screenings and referral to primary care. On average, grantees planned to screen 300 consumers (range 65–3,000) and expected that 250 consumers would receive primary care services (range 40–2,000) during the first year. No grantees quantified the referral services they planned to provide.
Registry or tracking system.
Grantees generally reported the ability to collect consumer-level data (N=54, 96%), often as part of clinical care. Most grantees (N=34, 61%) reported that they had or were developing electronic health records (EHRs) for behavioral health information, and 46% (N=26) reported using or planning to use EHRs for general medical data. Plans for shared general medical and behavioral health EHRs were less common (N=16, 29%).
Care management.
Grantees planned to use PBHCI funds to support a median of 5.3 full-time employees, and most expected that at least one would be a care manager (
Table 2). Care managers included nurse care managers, non–nurse care managers, and mental health case managers.
Illness prevention and wellness support.
Grantees unanimously reported plans to provide illness prevention and wellness services in group and individual formats. Smoking cessation programs, which grantees are now required to provide but which were optional for the first cohort of grantees funded in 2009 (N=13), were proposed by most grantees (N=34, 77%). Smoking cessation approaches included counseling (N=34, 61%), nicotine replacement (N=18, 32%), other medications (N=12, 21%), and incentives (N=7, 13%).
Optional program features
Grantees planned to implement optional program features, including colocation of nurse practitioners and primary care providers (N=53, 95%), supervision by primary care providers (N=53, 95%), and embedding of nurse care managers (N=49, 88%). Nearly all grantees (N=53, 95%) planned to deliver evidence-based behavioral health interventions, such as Screening, Brief Intervention, and Referral to Treatment (
17); Illness Management and Recovery Program plans (
18); Wellness Recovery Action Program plans (
19); and U.S. Preventive Services Task Force Recommendations (
20) (
Table 3).
Implementation challenges and barriers
Figure 1 shows barriers to program implementation reported by grantees at baseline, one-year follow-up, and both time points.
Start-up.
At baseline, difficulty recruiting and retaining qualified staff was the most commonly identified barrier (N=18, 32%). Grantees also reported challenges related to data management (N=13, 23%), including using EHRs and clinical registries and meeting the data collection requirements of the grant.
Grantees also reported challenges related to the availability and readiness of space in which to implement the PBHCI program (N=11, 20%); difficulty getting licensing or approvals from the grantee’s own agency administration (for example, for clinics that were part of larger health care networks), the city, the state, or the federal Health Resources and Services Administration (N=11, 20%); and problems sharing consumer data across behavioral health and primary care partners for the purpose of patient care (N=8, 14%).
Both implementation challenges related to merging primary care and behavioral health protocols and barriers related to billing and administrative issues were reported by five grantees (9%). Four grantees (7%) described staffing issues other than hiring, for example, staff conflict related to the program and lack of staff buy-in or low morale. Four grantees (7%) reported no barriers.
One-year follow-up.
Problems related to space and problems getting licensing or approvals were reported by a smaller proportion of grantees after one year (N=3, 5%, and N=5, 9%, respectively) than at baseline (N=11, 20%) (
Figure 1). Two (18%) of the 11 grantees that reported problems with space or with licensing or approvals at baseline reported them again one year later.
Some program barriers were reported at similar or identical rates at baseline and follow-up, including recruiting and retaining qualified staff (N=18, 32%, and N=17, 31%, respectively); other staff issues, such as conflict or low morale (N=4, 7%); and increased or unexpected costs (N=2, 4%, and N=3, 5%, respectively). Fewer than half of the 18 sites with hiring or staffing problems at baseline reported such problems again one year later (N=8, 44%). However, three of four (75%) grantees that reported other staff issues at baseline continued to report them at follow-up. None of the grantees that reported barriers related to increased costs, billing, or other administrative issues at baseline reported them one year later.
The following barriers to implementation were reported more often at follow-up than at baseline: data management, including issues related to EHRs, clinical registries, and data collection (N=17, 31%, and N=13, 23%, respectively); sharing consumer data for the purposes of care (N=11, 20%, and N=8, 14%, respectively); and merging primary care and behavioral health protocols (N=9, 16%, and N=5, 9%, respectively). Among grantees that reported these barriers at baseline, four (31%) reported data management issues, two (25%) reported difficulties sharing consumer data, and three (60%) reported challenges merging primary and behavioral health protocols at follow-up.
The most salient barrier at follow-up was difficulty recruiting consumers to participate in PBHCI (N=19, 35%). Grantees commonly reported difficulty in getting staff to refer appropriate consumers to the program, and some reported difficulty enrolling consumers who were referred. Several reported difficulties caused by potential program participants not showing up for appointments. As one grantee said, “Many clients [are] referred to the program, but then they will either avoid our case manager’s attempts to contact them, or they will not show up for their intake.” Thirteen grantees (24%) reported difficulties maintaining participation and following up with consumers who had already enrolled in PBHCI.
Nine programs (16%) reported inadequate capacity to serve interested consumers because of higher than expected program participation and scheduling difficulties.
Fewer than four grantees (<8%) reported challenges related to consumer access to specialists, transportation, and payment and insurance. Three programs (5%) reported no implementation barriers at follow-up, and one grantee reported no barriers at either time.
Discussion
SAMHSA’s PBHCI grants support an array of programs integrating primary care services into community behavioral health settings. Because of its size and breadth, the PBHCI initiative provides a unique opportunity to learn about the range of benefits and challenges of this approach to integrated care. Currently, data describing the impact of PBHCI on consumers’ behavioral and general medical health are still being collected. This article characterizes the first 56 PBHCI grantees and their programs and describes the grantees’ early experiences implementing this nontraditional approach to integrated care.
Although the PBHCI grantee programs are constrained by the grant requirements, programs differed widely in terms of size, staffing, capacity for providing behavioral and primary care services, use of EHRs, ability to share data across provider groups, targeted clientele, and demographic characteristics and insurance status of their consumers. This diversity of programs suggests that an approach that integrates primary health care into behavioral health care may appeal to a wide variety of community behavioral health centers. In addition, the PBHCI grantees’ experiences could represent an important resource for programs unlike the large, well-integrated care systems that have been described previously in the literature (
12,
13).
PBHCI grantees reported diverse challenges to program implementation at start-up and one year later, some of which resembled the challenges previously reported by studies of programs that integrate behavioral health into primary care settings, such as staffing issues, organizational resistance to change, and data sharing challenges (
12). Below, we discuss each of these barriers and other concerns reported by PBHCI grantees.
Recruiting and retaining staff were major challenges at both baseline and follow-up, and approximately one in three grantees who reported these issues at baseline continued to report them one year later. Other staff issues, such as conflict and low morale, were less common but more persistent, given that three out of four programs that reported these issues at baseline continued to report them one year later.
Consistent with other research (
16), grantees that brought primary care providers on site reported problems with physical space; however, follow-up data suggested that most grantees resolved space-related issues within the first year of the program.
PBHCI grantees also reported challenges related to EHRs, registries, and data sharing at both baseline and follow-up. Grantees had funds available to purchase EHRs or registry software but reported concerns with the quality and capacity of available packages. Until electronic systems that efficiently and securely integrate behavioral and primary health information are widely available, progress in effective utilization of integrated data to enhance clinical care is expected to be slow.
Licensing issues, such as delays in approval for FQHCs to provide and bill for behavioral health services, were frequently mentioned by grantees, and often the issues were specific to the state in which the program was located. SAMHSA’s recently funded Center for Integrative Health Solutions, a technical assistance center intended to support integrated care programs, now provides information about licensing and regulation requirements in each state as well as strategies to expedite the licensing processes.
Finally, approximately one in four grantees reported difficulty keeping consumers engaged in the program one year into the grant, which suggests a need to further develop care management services, many of which were new as a result of PBHCI. Care managers can help consumers to attend appointments and wellness programming and should be a part of integrated care programs. In addition, some grantees have had success enlisting the help of peer specialists, who may have the time, inclination, and unique skills to effectively reach out to consumers to encourage participation.
This study had several limitations. Our descriptions of staffing, service capacity, and programming more accurately reflect the grantees that participated in structured interviews than the grantees that did not. The description of program features may be a conservative reflection of actual practice because the information was obtained by a review of grantee proposals and reports and through telephone conversations and did not include features that programs had implemented but failed to mention. Conversely, we may have overestimated the availability of program features that were proposed but had not yet been put in practice.
Because of insufficient statistical power, we are unable to determine whether characteristics of PBHCI programs or grantees, such as urban versus rural location, predicted grantee reports of the presence or absence of barriers to implementation or of the resolution of barriers from start-up to one-year follow-up.
Also, the scope of the barriers to program implementation identified in this report was limited. For example, because the PBHCI grantees were awarded four years of funding, they were unlikely to report issues related to funding and program sustainability during the first year of program implementation.
Conclusions
Few studies have described the process of implementing primary care in community behavioral health settings. The experiences of SAMHSA PBHCI grantees may inform organizations considering integrating care in diverse community-based settings. Information about the process of integrating primary care into behavioral health settings has important policy and practice implications, given that several states are pursuing integrated care through statewide initiatives, such as the Health Homes and specialty Health Homes programs (
14,
15). Future research should continue to monitor the benefits and challenges of integrated care and investigate potential relationships among the program characteristics and the implementation challenges identified in this article. Research on program costs may also facilitate program sustainability as funding for behavioral health programs becomes increasingly scarce.
Acknowledgments and disclosures
This research was supported by the Substance Abuse and Mental Health Services Administration and by the Office of the Assistant Secretary for Planning and Evaluation, U.S. Department of Health and Human Services (contract OS 42345). The authors acknowledge the efforts of current and past project officers Hakan Aykan, Ph.D., Kirsten Beronio, Ph.D., Lisa Patton, Ph.D., and David DeVoursney, M.P.P. They also thank Susie Lovejoy, M.P.P., and Sarah Hauer, M.A., for outstanding research and other support for this project, respectively.
The authors report no competing interests.