Skip to main content

Abstract

Objectives: The integration of behavioral health and primary care services may be a way to reduce racial and ethnic inequities in medical and behavioral health care services and outcomes for people of color. We reviewed the evidence pertaining to integrating behavioral health and primary care for people of color and assessed social work’s role and impact on their health outcomes. Method: We systematically searched PubMed, PsycINFO, Social Science Citation Index, and Social Services Abstracts to identify studies of integrated primary and behavioral health care published between 2000 and 2015. English-language publications of randomized controlled trials (RCTs) conducted in the United States focusing on people of color were included in our systematic review. Results: Of the 37 RCTs we reviewed, 7 focused on people of color and examined behavioral health integration in primary care and related health and/or health care utilization outcomes. Six of the studies focused on people of color reported social work involvement, and 5 found that integrated care was associated with improved outcomes. Conclusions: Although benefits are associated with integration of behavioral health and primary care, evidence to suggest that integration per se improves health outcomes for people of color remains limited.

The Patient Protection and Affordable Care Act boasts a number of provisions promoting the integration of behavioral health and primary care. Integration exists along a continuum—ranging in scope from chronic disease management to medical complexity—and is implemented via coordinated delivery systems, such as colocating the provision of care through patient/person-centered medical homes (PCMH) and other settings that feature systems-level integration of primary and behavioral health care (Nardone, Snyder, & Paradise, 2014; U.S. Department of Health & Human Services, 2010). Integrated care delivery fostered by the Affordable Care Act has the potential to improve quality, increase access, and reduce fragmentation for people with mental illness and substance use disorders (Croft & Parish, 2013; McLellan & Woodworth, 2014), particularly for people of color seeking services in community settings (Sprague Martinez, Walter, Acevedo, Lopez, & Lundgren, 2018). As such, integration efforts could play a role in reducing racial and ethnic inequities in medical and behavioral health care services (Chin et al., 2012; Institute of Medicine, 2003; Purnell et al., 2016). Despite the promise of integration, the benefits for people of color (i.e., racial minorities, including African Americans, Hispanics/Latinos, Native Americans/Alaskan Natives, Asians/Pacific Islanders; E. Williams et al., 2010) have not been examined. Therefore, we set out to systematically review the literature to answer the following questions:

1.  What is the impact of integration on health outcomes and health care delivery for people of color?

2.  What integration characteristics are associated with improved care?

3.  What is the role of social work in integration efforts designed to improve care for people of color?

Research has demonstrated significant racial and ethnic inequities in mental health (Alegría et al., 2015; Ault-Brutus, 2012; Hahm, Lê Cook, Ault-Brutus, & Alegría, 2014; Miranda, McGuire, Williams, & Wang, 2008) and substance use disorder treatment (Lundgren, Amodeo, Ferguson, & Davis, 2001; Mulvaney-Day, DeAngelo, Chen, Cook, & Alegría, 2012). Specifically, people of color (POC) are less likely to receive treatment for mental health and substance use disorders (Alegría et al., 2015; Wells, Klap, Koike, & Sherbourne, 2001), and they often receive poorer quality of care (Schmidt, Greenfield, & Mulia, 2006). Additionally, there are significant and well-documented racial inequities in physical health status among people in the United States; these inequities are observable from birth to childhood and are persistent throughout the life course (Fortuna, Alegria, & Gao, 2010; D. R. Williams, Mohammed, Leavell, & Collins, 2010). For example, diabetes and hypertension are 2–3 times more prevalent among Black Americans than White Americans (Cooper et al., 2005; Geiss et al., 2014; Gillespie & Hurvitz, 2013), and American Indian populations have increased lifetime risk of post-traumatic stress disorder (PTSD) and alcohol dependence (Beals et al., 2005) as well as growing disparities in suicide rates (Sondik, Huang, Klein, & Satcher, 2010). Given that POC comprise the majority population in about half of the 100 largest cities in the United States (Kaufmann, 2007) and are projected to represent the numerical majority of the population by midcentury (Roberts, 2008), identifying promising practices (e.g., integration) to improve their health is essential.

Care integration aims to enhance the overall patient experience, improving health outcomes and the quality of care while simultaneously containing costs (Peek, 2009). To that end, integrated health care delivery models have focused broadly on chronic disease management while enhancing clinical decision-making and quality (Peek, 2009). Meanwhile, research has indicated that integration increases access to care and reduces fragmentation for individuals with co-occurring conditions (Coleman, Austin, Brach, & Wagner, 2009). Moreover, integration studies have demonstrated improvements in both physical and behavioral health (Bogner, Morales, de Vries, & Cappola, 2012; Coleman et al., 2009; Thota et al., 2012). As health care delivery systems shift toward integrated systems of care, it is imperative to critically examine the potential benefits of this model of care for POC.

Method

This review is reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines (Moher, Liberati, Tetzlaff, & Altman, 2009). Prior to starting the review, we registered it with PROSPERO under the title “Is behavioral health and primary care integration a promising strategy to improve health and health care for people of color? Evidence and the role of social work” (Registration No. CRD42016042532, https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=42532).

Eligibility Criteria

We examined studies of integrated primary and behavioral health care, inclusive of collaborative care teams, medical health homes, fully integrated systems, and colocated primary care/behavioral health providers using integrated care treatment models. With consultation from a research librarian and feedback from the behavioral health learning community, a search protocol was developed. The initial search included peer-reviewed English-language publications, published in 2000–2015, that reported on studies conducted in the United States. These years are consistent with the movement toward PCMH recognition (National Committee for Quality Assurance, 2014), development of joint principles of the PCMH (American Academy of Family Physicians, American Academy of Pediatrics, American College of Physicians, & American Osteopathic Association, 2008), and evaluation of PCMH demonstration projects aiming to transform patient care in the U.S. The search results were then restricted to POC-focused studies that used a randomized controlled trial design. Dissertations, newspaper articles, conference abstracts, and books or manuals were excluded. Information sources included PubMed, PsycINFO, Social Science Citation Index, and Social Services Abstracts.

Search and Study Selection

Two research assistants entered search terms (see Table 1) into each database in February and March 2016 and recorded the number of hits by term and database. Search results were systematically reviewed first by title and then by abstract.

Table 1. 

Search Terms

Primary Care and Behavioral Health Integration
Integrated Primary and Behavioral Health Care
Integrated Behavioral Health (Care)
Integrated Care and African Americans
Integrated Care and Latinos
Integrated Care and Native Americans
Mental Health and Primary Care Integration
Pharmacological Treatment for Substance Abuse in Medical Care
Substance Use Disorder and Mental Health Integration

Papers with the following related subject matter were excluded during the initial search: integrative health care technology; integrated medical/psychological education; chronic disease self-management (e.g., diabetes, HIV); integrated approaches to public health prevention, outreach and screening; integration of nutrition services in health settings; integrated chiropractic care; and integrated psychotherapeutic treatment modalities. Papers meeting the inclusion criteria were sorted alphabetically, stored in an Excel spreadsheet, and then downloaded in an Endnote database. The spreadsheet included the authors, title, article URL, and abstract, as well as blank fields for data to be extracted (Table 2): study design, population, health outcomes, delivery outcomes, social work role, and integration type.

Table 2. 

Code Book

Level 1 AnalysisLevel 2 Analysis
POPOUTCOME_H (Health Outcomes)OUTCOME_D (Delivery Outcomes)SWRINT_TYPE (Integration Type)
(Population)(Social Work Role)
0 None (majority White)0 None0 None0 None1 Behavioral health in primary care
1 Described as African American/Black1 Substance use disorder (SUD)1 Increased utilization of behavioral health services1 Team member2 Primary care in behavioral health
2 Described as Latino/Hispanic2 Severe mental illness (SMI)2 Increased utilization of primary health care2 Author affiliation with a school of social work and no SW credential3 Collaborative care
3 Described as Asian/ Pacific Islander3 Co-occurring SMI and SUD3 Reduced number of missed appointments3 Study PI4 Other definition
4 Described as Native American/Hawaiian/Alaska Native4 Depression–anxiety4 Improved patient satisfaction4 Social work author 
5 Described as diverse immigrant (no other details)5 Diabetes5 Improved patient provider relationship  
6 Not described6 Hypertension6 Cost/cost-effectiveness  
7 Other7 Other7 Other  
8 Majority POC (but not described or addressed in paper)    

Note.  POC = people of color; SW = social work; PI = principal investigator.

View Table Image

Two researchers met to reconcile the database, eliminating duplicate files. Database files were then examined to determine study type. Papers using randomized controlled trial (RCT) designs were retained, and those with other study designs (i.e., quasi-experimental, longitudinal, case report, and other) were eliminated from the data set. We used a conservative approach to ensure that all included RCT publications met the inclusion criteria. The lead author and two research assistants each read all RCT papers in full and met in person to determine if any of the remaining RCT papers were outside the scope of investigation. For example, studies focusing exclusively on integrated therapies (e.g., mind–body treatment for sleep and PTSD [Nakamura, Lipschitz, Landward, Kuhn, & West, 2011], integrated PTSD and smoking treatment [Feldner, Smith, Monson, & Zvolensky, 2013]) without explicit mention of primary care and behavioral health integration were removed from the data set.

Similarly, we excluded studies exploring the integration of substance use and mental health treatment if primary care was not also addressed. RCTs were also excluded if integration was restricted to any of the following definitions: community integration (e.g., provision of mental health crisis services in public housing; McHugo et al., 2004); community-integrated home-based care (Gitlin et al., 2012); telemedicine (e.g., phone-only therapy for depression in primary care; Ross, TenHave, Eakin, Difilippo, & Oslin, 2008; Simon, Ludman, & Rutter, 2009); Web-based care coordination with advanced cancer patients in a palliative care setting (Steel et al., 2016); brief intervention (e.g., brief treatment for problem drug use in primary care settings; Roy-Byrne et al., 2014) and/or referral (e.g., referral to primary care treatment for adults in a detoxification unit with alcohol and substance use disorders; Samet et al., 2003); and colocated care.

Once the data set was complete, the team reviewed RCT publications to determine study population. Publications with majority (50% or greater) African American/Black, Latino/Hispanic, Asian/Pacific Islander, and/or Native American/Hawaiian/Alaska Native study populations were retained for inclusion in the final data set as long as some mention of the study population was made within the body of the paper. Papers with majority POC populations that were not mentioned or addressed at any point in the body of the paper were coded separately and retained for inclusion in the final data set. Papers were excluded if their study populations were majority White and/or if the racial/ethnic composition of the study population was not described. The final sample included integration RCTs focused on POC.

Data Collection and Analysis

Three members of the research team extracted data using a standardized form. Extracted information included study design and setting; study population, participant demographics, and baseline characteristics; details of the intervention and control conditions; study methodology; recruitment and study completion rates; outcomes and times of measurement; measures; and social work involvement. Social work involvement was conceptualized as social workers involved in the integration effort, social work coauthorship, social work lead author, and social work principal investigator and coinvestigator. Social work author and/or investigator was defined as an individual with a social work credential (and/or degree) or a faculty member at a school of social work (with or without social work credentials), and not necessarily individuals self-identifying as social workers.

Researchers extracted data independently and then met to review and reconcile their extraction forms and resolve any inconsistencies. In instances where items were coded differently, each individual explained their reasoning, and final decisions were made using consensus. Of note, there were three reviewers, which facilitated the reconciliation process. To resolve inconsistencies in coding for social work involvement in the form of authorship, a database of all listed authors was constructed in Excel. The institutional website for each author was searched to determine affiliation. When available, the curriculum vitae for authors were downloaded from institutional websites and LinkedIn pages to determine if their academic training included graduate training in social work.

Summary Measures and Methods for Synthesis of Results

Indicators used to determine the extent to which integration improves health status and health care delivery among POC included utilization of outpatient behavioral health services (e.g., counseling/therapy, psychiatry appointments); utilization of outpatient primary care health services; number of missed appointments; patient satisfaction; patient–provider relationship; cost and cost effectiveness; and disease outcomes for substance use disorder (SUD), severe mental illness (SMI), co-occurring SUD and SMI, diabetes, depression and/or anxiety, and hypertension. These indicators were extracted from all RCT papers. All health and delivery outcomes that did not meet our established categories were coded as “other” (e.g., emergency room visits, hospitalizations, etc.).

To determine risk of bias in individual studies, researchers independently assessed the risk of bias in each study. The team then discussed the findings. A narrative synthesis of findings from the included studies structured around the conceptualization of integration, target population, and outcomes of interest was constructed based on themes that emerged through data extraction. In addition, intervention effects for each of the seven studies by outcome were summarized.

Results

The PRISMA Study Flow Chart (see Figure 1) illustrates the results of each phase of the search and data-extraction processes. The initial search yielded 67,550 records, of which 3,995 were relevant according to initial search criteria. After screening for duplicate records and removing any publications that did not meet initial search criteria (i.e., dissertations and conference abstracts, studies with a non-U.S. focus, etc.), a total of 1,371 studies underwent further analysis to determine study type. Of these, 86 RCTs were retained in Level 1 analysis, and the remaining 1,297 were eliminated for non-RCT study design. Upon review during Level 2 analysis, 37 RCTs meeting the outcomes criteria were identified; 49 studies were removed for not meeting the operational definition of integrated care. Of the remaining 37 relevant RCTs, 7 papers (18.9%) focused on POC. These papers included studies that explicitly focused on racial and ethnic minority populations as well as those that described a study sample of at least 50% POC but did not focus on race/ethnicity in the framing of the paper.

Figure 1. 
Figure 1. 

PRISMA study flow (Moher et al., 2009). RCT = randomized controlled trial; POC = people of color.

Targeted health and/or health care outcomes explored in each study are summarized in Table 3. In the 7 POC-focused studies, integration was characterized as behavioral health interventions delivered in primary care settings or as collaborative care involving a coordinated interdisciplinary team. Three studies examined the benefits of integrated care for patients receiving pharmacological treatment, 2 addressed the integrated delivery of care, and 2 included behavioral health interventions delivered in primary care. Four of the 7 studies were small RCT pilots: 1 was a site-specific analysis, and 2 were full-scale RCTs. Of note, 3 of the pilot studies, although distinct RCTs, had the same lead author.

Table 3. 

Coding Outcomes for Integration RCTs Focused on People of Color

Study CharacteristicsAyalon et al. (2007)Bedoya et al. (2014)Bogner & de Vries (2008)Bogner & de Vries (2010)Bogner et al. (2012)Kwong et al. (2013)Saitz et al. (2013)
Design       
 RCTYesYesYesYesYesYesYes
Population
 African American/BlackYesYesYesYes
 Latino/HispanicYes
 Asian/Pacific IslanderYes
 Majority people of color (not discussed/addressed in paper)Yes
Health outcomes       
 NoneYes
 Substance use disorderYes
 Depression–anxietyYesYesYesYesYes
 DiabetesYesYes
 HypertensionYes
Delivery outcomes       
 NoneYes
 Increased utilization of behavioral health servicesYesYesYes
 Increased utilization of primary health careYes
 Reduced number of missed appointmentsYes
 Improved patient satisfactionYes
 OtherYes*Yes*
Social work role       
 NoneYes
 Team memberYes
 Author affiliation with a school of social workYesYes
 Study PIYes
 Social work authorYesYesYesYes
Integration type       
 Behavioral health in primary careYesYesYesYesYesYes

Note.  RCT = randomized controlled trial; PI = principal investigator.

* Medication adherence.

View Table Image

Populations of focus included African Americans (4 studies), Latinos (1 study), Chinese immigrants (1 study), and a mixed population that was 53% individuals of color (1 study). Of the 4 studies that included African Americans, 2 explicitly focused on this population. The 2 remaining studies had samples comprised of at least 50% African American individuals but did not discuss the population beyond sample characteristics or implications for the broader population.

The most common behavioral health problems assessed in the 7 POC-focused studies were depression (5 studies), followed by diabetes and SUD (2 studies; see Figure 2). One study assessed hypertension. None reported on outcomes for SMI or co-occurring SMI/SUD. Measures used by studies examining depression outcomes were inconsistent and included the Center for Epidemiological Studies-Depression (CES-D) symptoms index and the Patient Health Questionnaire (PHQ-9), whereas studies of patients with diabetes consistently used hemoglobin A1c as an outcome indicator. Health care outcomes were measured by examining service utilization, including behavioral health visits (4 studies), primary care visits (2 studies), and number of missed appointments (1 study). In addition, multiple studies measured adherence using the Medication Event Monitoring System. None of the studies included cost/cost-effectiveness or patient–provider relationship in the assessment of health care delivery.

Figure 2. 
Figure 2. 

Summary of outcomes. Appts = appointments; BH = behavioral health; HTN = hypertension; PC = primary care; POC = people of color; RCT = randomized controlled trial; SUD = substance use disorder.

We set out to explore the extent to which integration improves care for POC. As outlined in Table S1, we found only 1 study that examined the impact of integration on a POC population in comparison to a White population. In a site-specific analysis of a multisite RCT, Ayalon, Arean, Linkins, Lynch, and Estes (2007) found that for African American elders, integrated mental health and substance use services—set in the context of primary care as opposed to a specialty mental health clinic—was associated with increased utilization of outpatient mental health and substance use services (Ayalon et al., 2007). Of the 3 POC population-specific studies in our sample, 2 demonstrated statistically significant improvements for POC who received integrated care. Bedoya et al. (2014) found that Latino adults with depressive symptoms receiving integrated, culturally focused psychiatric consultation delivered in primary care experienced greater symptom reduction than Latino adults with depressive symptoms receiving care in usual-treatment settings. Meanwhile, Bogner and deVries (2010) reported that, when compared to usual care, an integrated primary care intervention for older African American patients receiving pharmacological treatment was associated with reduced depressive symptoms, and both systolic and diastolic blood pressure medication adherence, as well as increased medication adherence for depression and hypertension at 6 weeks. However, Kwong, Chung, Cheal, Chou, and Chen (2013) did not see statistically significant differences for integrated depression care management in primary care setting for Chinese immigrants.

Of the 7 studies, 3 did not report outcomes specific to POC, despite having samples with more than 50% POC. Accordingly, we do not have evidence from these studies as to whether or not study outcomes apply to specific POC populations. Only 3 of the studies we examined demonstrated improved outcomes for POC, limiting our ability to identify the specific characteristics of integration on health status and care quality for POC.

Finally, we documented social work involvement in 6 of the 7 (85.7%) RCTs specific to POC populations. With respect to the RCTs in non-POC populations (n = 30), 10 (33.3%) had social work involvement. This included 2 studies with lead authors who were also principal investigators. In addition, we identified social work coauthors of 4 studies. Overall, we found that RCTs focused on POC appeared to be more likely to have social work involvement compared to the total population of RCTs.

Discussion

We found limited research on populations of color participating in studies of integrated care. The data reported in the few RCTs and pilot RCTs we identified indicated that integration can improve health outcomes and care over usual-care strategies. This was particularly evident for African American patients with regard to increased mental health and/or substance abuse visits (Alegría et al., 2015), as well as for improvements in depression and diabetes indicators over periods of 6 and 12 weeks (Bogner & de Vries, 2008, 2010; Bogner et al., 2012). Not surprisingly, better medication adherence was associated with better outcomes. Other groups (e.g., Latinos) also showed mental health benefits (Bedoya et al., 2014), but no benefits were evident for Chinese immigrants with depression and anxiety (Kwong et al., 2013) or for a mixed sample of patients with SUDs (Saitz et al., 2013). Overall, 5 of 7 studies pointed to better outcomes with integrated care. Unfortunately, evidence of benefit was completely lacking for those with serious mental illness as no studies were identified in this area.

Although there is a substantial amount of research on care integration in the form of case reports and theoretical studies, as well as cross sectional and longitudinal studies, there is little definitive research in the form of RCTs. Accordingly, the effectiveness of integration in improving health care and outcomes for POC remains unclear. Furthermore, the small number of studies available for review prevented our assessment of the characteristics of integration studies associated with improved care for POC.

We also found studies that included sizable portions of POC but did not report population specific outcomes. This is not surprising given the limits of subgroup analyses in RCTs, including their impact on study power (Lagakos, 2006; Wang, Lagakos, Ware, Hunter, & Drazen, 2007). Nonetheless, there is a need for evidence of the impact of integrated delivery models on POC populations, as well as the characteristics of integration that are associated with improved health outcomes and care.

Our data indicate that when it comes to integration RCTs targeting POC, social workers are likely to be involved in the leadership and dissemination of such research, specifically in the roles of principal investigator/lead author and coauthor (5 of the 7 studies). Social workers also served as members of integrated care teams, but this was reported for only 1 study. This finding suggests that social work professionals are already involved in the development of integration studies and the delivery of integration interventions. Educational and training opportunities for social workers need to align with this role at both the master’s and doctoral levels. Finally, it is clear from our review that social work plays a key role in research on integration of behavioral and physical health specifically on populations affected by health inequities. This research indicates that social work researchers adhere to their professional commitment to promote racial justice.

Limitations

Based on our initial search, we anticipated limitations to our analysis given the number of outcomes we were examining. Integrated care as envisioned by the Affordable Care Act ranges in scope from chronic disease self-management, care navigation, and colocated care to the provision of care through PCMH and other systems-level physical/behavioral health care entities (Nardone et al., 2014; U.S. Department of Health and Human Services, 2010). No single definition of integrated care is used in the literature; rather, integrated care is described as a continuum of different services and systems. By operationalizing integrated care to include collaborative care teams, medical health homes, fully integrated systems, and colocated primary care/behavioral health facilities working with integrated care treatment models, we may have eliminated a sizeable portion of the literature. The small number of RCTs we identified and the large number of outcomes together limited our ability to assess the effectiveness of integration as a strategy to improve care for communities of color or to identify the characteristics of integration associated with improved care. Moreover, there is a risk of bias in the study results. There is a risk of publication bias, given our focus on peer-reviewed English-language papers and U.S.-based studies. Including government reports might have expanded our sample size and reduced the risk of publication bias. In addition, the majority of included studies had small sample sizes, which presented an additional risk of bias.

Conclusion

Although there may be benefits associated with integration for POC populations, the evidence that integration significantly improves health outcomes or care is limited by the small number of RCTs on such populations. Unfortunately, we were unable to identify characteristics of integration that may be associated with improved care for POC. Evidence was particularly lacking in the areas of SMI and SUD. We did, however, find social work involvement in the majority of POC-focused RCTs and social work leadership in study authorship.

We would like to acknowledge the support of the Center for Innovation in Social Work and Health, which convened the Behavioral Health Learning Community (BHLC) to collaboratively design this review and provided administrative support and guidance for the BHLC. We would also like to thank the Center for Addictions Research and Services, which supported the time of research assistants dedicated to this study. In addition, we would like to thank Ms. Eliza Campbell, who assisted with the search database.

Notes

Linda Sprague Martinez, PhD, is an assistant professor in the Macro Department, Boston University School of Social Work.

Lena Lundgren, PhD, is a professor at the University of Denver Graduate School of Social Work.

Angela Wangari Walter, PhD, is an assistant professor in the Department of Public Health, Zuckerberg College of Health Sciences, University of Massachusetts Lowell.

Jessica Sousa, MSW, MPH, is a research assistant at the Center for Addictions Research, Boston University School of Social Work.

Nermeen Tahoun, MSW, MPH, is a research assistant at the Center for Addictions Research, Boston University School of Social Work.

Gail Steketee, PhD, is a professor emeritus at the Boston University School of Social Work.

Hyeouk Hahm, PhD, is an associate professor and chair of the Social Research Department, Boston University School of Social Work.

Kim T. Mueser, PhD, is a professor in the Boston University Center for Psychiatric Rehabilitation.

Ivy Krull, PhD, is an assistant professor in the Sociology Department, Emmanuel College, School of Humanities and Social Sciences.

Daniel Lien Do, MSW, MPH, is an adjunct faculty member in the Boston University School of Social Work.

Richard Saitz, MD, is a professor and chair of Community Health Sciences, Boston University School of Public Health.

Correspondence regarding this article should be directed to Linda Sprague Martinez, 264 Baystate Road, Boston MA 02215 or via e-mail to

References