Sharing a Playbook: Integrated Care in Community Health Centers in the United States

Abstract

Objectives. We investigated basic measures used to assess collaboration between colocated providers and to gauge the extent to which health centers practice integrated care.

Methods. We used the Assessment of Behavioral Health Services survey and the 2010 Uniform Data System to explore the elements of integrated care for behavioral health conditions. We used multivariable regression models to examine the correlates of integrated care.

Results. More than 85% of health centers provided mental health services in 2010, and almost half offered substance use treatment. Health centers commonly reported shared access to information among behavioral health and medical providers and joint care planning. A higher degree of integrated care involving joint case conferences was less common. Health centers without electronic health records and those with lower percentages of total staff composed of behavioral health workers were less likely to provide integrated care.

Conclusions. A 2-pronged strategy involving financial incentives and technical assistance to spread best practices might increase integrated care, particularly among health centers that are not maximizing the potential of electronic health records and health centers with low behavioral health staffing levels.

The treatment of behavioral health conditions is a key component of quality care.¹ Behavioral health encompasses mental health and substance use disorders as well as health behaviors.² Improving access to screening and treatment services for mental health and substance use disorders is critical to the success of wider efforts to improve the health care system to pursue the triple aim³ of better health, better care, and lower per-person costs.^4,5 However, medical and behavioral health care providers have historically practiced in isolation, with little communication or coordination. The need to better integrate behavioral and medical care is especially pronounced for underserved patients; according to the Institute of Medicine, “[t]he single greatest flaw of the mental health safety net is its nearly total disconnection from the core [general medical] safety net.”^6(p189)

Mental health and substance use disorder services are frequently provided in primary care settings; in fact, many patients with behavioral health disorders never receive care in a specialty behavioral health setting.^7,8 Community health centers are key portals of access to medical and behavioral health services in underserved communities.⁹ Community health centers are also called “federally qualified health centers” or “health centers.” We used data from federally qualified health centers that received grant funding in 2010 under Section 330 of the Public Health Services Act through the Bureau of Primary Care at the Health Resources and Services Administration of the US Department of Health and Human Services. Because many health center patients face additional access barriers—40% of health center patients were uninsured in 2010—treatment initiation and engagement might be improved if on-site behavioral health services are available where patients access medical care and links to social services.¹⁰ The “warm handoff” to a behavioral health provider can create trust, because colocation with medical services can destigmatize behavioral health treatment. Patients already visit health centers for medical and other types of services, so accessing behavioral health services on-site at the health center is likely to be convenient.¹¹ In addition, colocating primary care and behavioral health services is a strategy to mitigate barriers to accessing care related to cultural beliefs among patients.¹²

Health centers are required to provide mental health and substance use disorder services on-site or by referral. Most health centers have on-site behavioral health specialists, particularly larger health centers, those located in urban areas, in the Northeast and West, in local areas with greater availability of behavioral health specialists, and in states that allow Medicaid same-day billing for medical and behavioral health services.^13,14 Health center capacity is expanding under the Affordable Care Act (Patient Protection and Affordable Care Act, Pub L No. 111-148, 124 Stat. 855 [March 2010]) to increase access to care for underserved patients and communities.¹⁵ Improving access to behavioral health services at health centers is currently a priority; more than 1 in 3 health centers received funding to expand behavioral health capacity in 2014 of more than $105 million.¹⁶

Building on the foundation of colocated behavioral health specialists and primary care providers, health centers are exploring how to integrate behavioral health services into primary care.^17,18 A commonly used continuum specifies 3 basic levels of orchestration between behavioral health and medical care: coordinated from 2 separate locations, colocated in a shared space, or integrated.^10,19 The definition is still evolving, but integrated care is distinguished by colocated, team-based care and, optimally, a shared care plan with both behavioral health and medical elements.^10,20–23

Integrated care typically refers to providing behavioral health services in the primary care setting, whereas the closely related terms “coordination” and “collaboration” are used to describe shared access to information, communication, and consultation between medical and behavioral health providers, regardless of whether the services are colocated.^24,25 We examined the processes used by primary care and behavioral health clinicians in health centers to conduct evidence-based activities to improve integration: colocating medical and behavioral health services, shared access to information in patient records, joint case conferences, and joint care planning.²⁶

It is important to note that colocating medical and behavioral health services does not necessarily lead to communication and collaboration; sustained technical assistance might be needed to support providers as they make the necessary changes to cultures, structures, and processes to allow more interdisciplinary communication and collaboration.^27,28 Barriers to integrated care include a lack of consensus regarding team members’ roles^29,30 and interprofessional conflict stemming from differing cultural norms and mental models of practice.³¹ The siloed and fragmented reimbursement landscape is another factor, particularly because reimbursement is often fee for service on the basis of the volume of patient encounters; funding streams that cover provider-to-provider communication might be necessary to support integrated care.^32,33

Prohibitions on same-day billing for medical and behavioral health services are another roadblock.^13,34 Additional financial barriers include staffing costs and health information technology (IT) implementation costs.³⁵ There are many other issues related to health IT, including usability issues of care coordination and registry functions, limited interoperability hindering health information exchange, and additional privacy protections for information on substance use disorders.^20,36–39

We explored some basic measures that can be used to assess collaboration between colocated providers and to gauge the extent to which a health center is practicing integrated care. We asked 2 main questions. First, to what extent is integrated care occurring for health center patients with behavioral health conditions? Second, which health center characteristics are associated with practicing integrated care? We hypothesized that larger health centers, those with electronic health records (EHRs), and those with higher percentages of total staffing composed of behavioral health specialists might be more likely to provide integrated care.

Our study makes a unique contribution to the literature by presenting nationally representative data on the elements of integrated care for patients with behavioral health conditions in health centers. The findings on contextual and health center characteristics associated with practicing integrated care in health centers might guide policies designed to reduce unmet needs for behavioral health treatment services among underserved patients.

METHODS

We used the 56-item Assessment of Behavioral Health Services in Federally Qualified Health Centers survey. The National Association of Community Health Centers distributed this survey on the SurveyMonkey platform via e-mail in February 2010 to 1080 health center grantee organizations reported in the 2008 Uniform Data System. Each grantee may administer numerous care delivery sites. We sent 2 reminder e-mails between February and May and conducted telephone outreach in June and July 2010. We sent the survey to the chief executive officer of each health center and instructed them to answer the survey themselves or delegate the task to “the member of your staff responsible for behavioral health services.” We did not give respondents incentives to complete the survey. There were 390 nonduplicate responses to the survey, yielding a 36.1% response rate.

Survey Questions

The survey asked whether mental health and substance use disorder services were provided on-site. Elements of integrated care captured by the survey include shared access to information, joint case conferences, and joint care planning. Shared access to information was assessed by asking, “Do behavioral health and medical staff have access to lab results and medication lists in the article chart or electronic health record?” The same question was asked for “problem lists” and “behavioral health treatment plans.” The frequency of joint case conferences was explored by asking, “Do your behavioral health and medical staff routinely work together on a specific day and time to discuss mental health cases?” The question was also asked separately about joint case conferences to discuss substance abuse cases. For both, respondents indicated whether the meetings were daily, weekly, monthly, informally as needed, or some other frequency.

Joint care planning was gauged by asking, “Do behavioral health and medical staff make joint decisions on the patient’s plan of care?” The primary care provider’s level of involvement with the formulation and approval of each patient’s behavioral health treatment plan was measured in the survey by asking, “Do medical and behavioral health staff both sign the treatment plan?” Finally, the self-reported quality of communication between medical and behavioral health staff was elicited by asking, “How would you rate the communication/coordination between medical and behavioral health staff at your center?” Respondents had 5 options: excellent, very good, good, fair, and poor.

We linked the survey data with the 2010 Uniform Data System (UDS) using the respondent’s UDS number. All health center grantee organizations report UDS data annually under Section 330 of the Public Health Service Act to the Health Resources and Services Administration, and the data are validated and audited for outliers. We conducted our analysis using the 363 survey responses that could be matched with the UDS. We constructed a variable from the UDS staffing data to capture the percentage of total staffing composed of behavioral health specialists. We considered other potential variables, including the percentage of total costs, patients, and encounters allocated to behavioral health services. Because we examined interdisciplinary collaboration, we used the staffing variable.

Analysis

Weighting for selection bias was not necessary because we surveyed the universe of health centers. We reviewed the data for potential nonresponse bias (descriptive statistics of whether each health center responded to the survey are available as a supplement to the online version of this article at http://www.ajph.org). We performed the differences of means t test for health center characteristics and stratified them by survey response. Rural health centers and those in the South and Midwest were less likely to answer the survey, as were health centers that had not adopted EHRs. We developed sample weights to adjust for nonresponse on the basis of the probability of survey response in each cell of a 2-by-2 table, stratified by rurality and whether the health center was located in the South. To create the weight we took the inverse of the probability of response to yield the weight for each cell. Weighted results are nationally representative.

Using the 7 survey questions on shared access to information, joint case conferences, and joint care planning, we constructed an ordinal composite variable to use as a dependent variable in multivariate regression analysis, representing varying degrees of integrated care. Guided by the theoretical models and evolving operational definition of integrated care, we tested the magnitude and statistical significance of the associations between the variables using bivariate logistic regression. We then computed the Cronbach α to measure the reliability of the assertion that the variables measure the same theoretical construct.

The results indicate an average interitem α of 0.667, which is within the acceptable range. These results support the use of a composite variable constructed of these 7 variables, so we constructed an ordinal variable ranging from 0 to 7 on the basis of a count of the number of elements of integrated care reported by each health center. If a health center did not provide behavioral health services on-site, we assigned the variable a value of 0. This variable specification is not meant to imply that the 7 elements of integration are equally important. We conducted sensitivity analysis using other specifications for the dependent variable.

We calculated weighted descriptive statistics on staffing and capacity for behavioral health screening and treatment services, both among all health centers and among the health centers that provided behavioral health services. We explored correlates of integrated care using bivariate methods and then constructed multivariate ordinary least squares regression models to examine how the measures of integrated care were correlated with health center characteristics. On the basis of the bivariate results, we added independent variables to the multivariate models.

The final model included region, rural location, number of full-time equivalent staff members, number of clinic sites, whether an EHR system was used by all providers in all sites, the percentage of total staff composed of behavioral health clinicians, whether the health center received funding to serve migrant and seasonal farmworkers and homeless individuals, the percentage of patients who were uninsured, the percentage of patients covered by Medicaid, the percentage of patients below 100% of the federal poverty level (FPL) as specified by the US Census Bureau for 2010, and the percentage female. We weighted and conducted all analyses using Stata version 11.0 (StataCorp LP, College Station, TX).

RESULTS

In 2010, each health center organization served an average of 17 729 patients, 39.60% of whom were uninsured and 1 in 3 of whom was covered by Medicaid (33.90%). Table 1 provides descriptive statistics on health centers. Almost half of the health centers (46.25%) were located in rural areas, 7 in 10 (72.60%) used an EHR system, and more than half reported that all providers in all clinic sites used an EHR system (57.30%). One in 6 health centers (17.70%) received special populations funding to serve homeless patients. The majority of health centers (84.70%) provided mental health treatment services in 2010, whereas just fewer than half (47.70%) provided substance use disorder services.

TABLE 1—

US Health Centers and Patient Characteristics: 2010

Characteristic	Mean %
Region
South	34.91
Northeast	16.59
West	25.94
Midwest	20.59
Rural	46.25
Received funding to target special populations
Homeless	17.66
Migrant and seasonal farmworkers	15.81
EHR installed	72.63
EHR used by all providers in all sites	57.33
Provide on-site behavioral health services colocated with medical care
Mental health	84.69
Substance use disorders	47.71
Specialty behavioral health providers as a percentage of total staffingb	8.41
Mental health	7.48
Substance use disorder	1.27
No. of patients served annually	17 729.45
No. FTE staff	118.23
% of patients below 100% FPLa	68.33
Insurance status and type
Uninsured	39.62
Medicaid	33.86
Other public	2.13
Private	16.28
Medicare	8.10

Note. EHR = electronic health record; FPL = federal poverty level; FTE = full-time equivalent. The sample size was n = 363.

Source. 2010 Uniform Data System data and weighted data from the 2010 Assessment of Behavioral Health Services in Federally Qualified Health Centers.

^aDefined by the 2010 US Census.

^bPercentage of total staffing comprised of behavioral health staffing was calculated among health centers that employ behavioral health staffing.

Medical and behavioral health providers at most health centers shared access to information, which is a precursor to higher levels of collaboration (Table 2). Both medical and behavioral health staff had access to medical information in patient charts in most health centers, including laboratory results and medication lists (73.5% of health centers) and problem lists (74.0%). Two in 3 health centers (66.1%) allowed both medical and behavioral staff to access behavioral health treatment plans.

TABLE 2—

Behavioral Health Service Provision and Integrated Care in US Health Centers: 2010

Variable	Mean %
Shared access to information in patient charts
Laboratory results and medication lists	73.51
Problem lists	74.00
Behavioral health treatment plans	66.11
Issue discussed in joint case conferences at least weekly
Mental health cases	19.99
Substance use disorder cases	17.80
Joint care planning of behavioral health and medical staff
Decide on patient care plans	64.75
Sign behavioral health treatment plans	14.40
Self-rated communication and coordination between medical and behavioral health staff
Excellent	19.99
Very good	33.94
Good	30.10
Fair	13.33
Poor	2.65

Note. The sample size was n = 363.

Source. 2010 Uniform Data System data and weighted data from the 2010 Assessment of Behavioral Health Services in Federally Qualified Health Centers.

Joint case conferences to discuss and comanage behavioral health conditions regularly occurred in about 1 in 5 health centers for mental health cases (20.0%) and substance use disorder cases (17.8%). Two thirds of health centers (64.8%) reported that behavioral and medical staff made joint decisions on patient care plans. We found expanded interdisciplinary decision-making authority also encompassing behavioral health treatment plans was less common; both types of staff signed behavioral health treatment plans in about 1 in 7 health centers (14.4%). Communication and coordination between medical and behavioral health staff was self-rated as good, very good, or excellent by 84.0% of health centers.

In adjusted analysis, EHR adoption, the share of total staffing composed of behavioral health specialists, and region were associated with the degree of integrated care practiced by each health center. Health centers with EHRs that were used by all providers in all sites scored higher on the composite integration variable than did health centers without EHRs in all sites (P = .004; Table 3; descriptive statistics of the composite variable available as a supplement to the online version of this article at http://www.ajph.org). As the share of total staffing composed of behavioral health specialists increased, the score on the composite integration variable also increased (P = .02). Finally, health centers in the Midwest scored lower on the integration variable than did health centers located in the South (P = .003).

TABLE 3—

Factors Associated With Integrated Care in US Health Centers: 2010

Characteristic	b (95% CI)	P
Region
South (Ref)	1.00
Northeast	–0.01 (–0.50, 0.49)	.982
Midwest	–0.72 (–1.20, 0.25)	.003
West	–0.34 (–0.75, 0.08)	.108
Rural	–0.26 (–0.63, 0.11)	.168
Number of FTE staff members	0.00 (0.00, 0.00)	.767
Number of clinic sites	0.02 (0.00, 0.03)	.055
EHR in all sites	0.49 (0.16, 0.83)	.004
Behavioral health staff as percentage of total staffing	0.82 (0.13, 1.50)	.02
Received funding to target special populations
Homeless	–0.09 (–0.52, 0.33)	.666
Migrant and seasonal farmworkers	0.26 (–0.23, 0.74)	.3
Insurance, %
Private, Medicare, or other	1.00
Uninsured	0.45 (–0.54, 1.45)	.37
Medicaid	0.49 (–0.70, 1.67)	.422
Patients below 100% FPL,a %	0.08 (–0.94, 1.11)	.874
Patients who were female, %	–1.06 (–3.68, 1.56)	.426

Note. CI = confidence interval; EHR = electronic health record; FPL = federal poverty level; FTE = full-time equivalent. Estimates are from multivariate ordinary least squares regression. The sample size was n = 340.

Source. 2010 Uniform Data System data and the 2010 Assessment of Behavioral Health Services in Federally Qualified Health Centers.

^aDefined by the 2010 US Census.

DISCUSSION

In addition to medical services, 85% of health centers offered mental health services in 2010, but less than half offered substance use treatment. Approaches integrating care are still evolving. Health centers commonly reported shared access to information among behavioral health and medical providers and joint care planning, but a higher degree of integrated care involving joint case conferences was less common. Health centers with EHRs in all sites and health centers with higher percentages of their staffing composed of behavioral health specialists were more likely to provide a higher degree of integrated care for patients with mental health and substance use disorder conditions.

These results are consistent with previous research that shows widespread mental health treatment capacity in health centers but more limited capacity to address the needs of patients with substance use disorders.¹³ This difference might arise because of the more limited availability of staff members with expertise and training in treating substance use disorders compared with the availability of the mental health workforce. The finding on EHR use affirms previous research indicating that health IT can support communication and collaboration by capturing and exchanging data in a structured and actionable way, enabling simultaneous chart access, asynchronous provider-to-provider communication, and joint care planning.^37,40–43 In addition, clinical decision support promotes the use of evidence-based care, and registry functionalities assist with longitudinal care management.⁴⁴

Previous research highlights 2 main strategies to increase integrated care within health centers that have behavioral health staffing: financial incentives and training and technical assistance to spread and sustain the adoption of best practices for integrated care.^44–46 The financial incentives and reimbursement methods being tested and implemented as part of new care models are likely to speed the adoption of integrated care. Many elements of integrated care do not occur during patient encounters and thus are not reimbursed under fee-for-service arrangements.

Initiatives from the Centers for Medicare and Medicaid Innovation such as Health Homes for Medicaid Enrollees with Chronic Conditions alter reimbursement rules to allow activities such as provider-to-provider communication—whether synchronous or asynchronous—and shared decision-making when patients have cooccurring medical and behavioral health conditions.⁴⁷ The Medicare Federally Qualified Health Center Advanced Primary Care Practice Demonstration Program includes a per-member per-month payment that can be used to cover the costs of the elements of integrated care that occur outside the strict definitions of a patient encounter.⁴⁸ In addition to changing reimbursement methodologies, providing assistance with fixed costs such as staffing costs and the costs of EHR implementation will enable health centers to increase the level of integration.

Beyond financial incentives, training and technical assistance are critical for helping with the workflow process and cultural changes needed for integrated care. Competencies like communication and collaboration should be included in initial clinical training^49,50 and in continuing education curricula.⁵¹ This will require not only including core competencies in curricula but also providing interdisciplinary training and continuing education opportunities.⁵¹ Training and technical assistance should focus on supporting health centers as they assess current provider roles and identify opportunities to allow provider roles to evolve, informed by evidence-based best practices.²⁰ Creating infrastructure to support learning collaboratives is an effective way to spread best and promising practices in addressing issues on provider roles.^52–54

Training and technical assistance are also needed to assist health centers with maximizing the potential of health IT.^55,56 The use of EHRs was associated with practicing integrated care, but previous research suggests that providers need extensive technical assistance to optimize the use of health IT.^57,58 Continued investment in technical assistance for health centers through the Health Center Controlled Network program will enable health centers to adopt and optimize advanced health IT functionalities, data repositories, and health information exchanges, particularly in light of the phasing out of the Regional Extension Center program.^59–61 In addition, assistance should be targeted to health centers that have not yet adopted EHRs.

Finally, consistent with previous research,¹³ these findings indicate that because many patients experience barriers to accessing behavioral health treatment off-site, there is still a need to build behavioral health capacity in some health centers, particularly for substance use disorder screening and treatment. The costs associated with recruiting behavioral health specialty staffing can be offset with expanded service funding targeted at behavioral health services, like the funding opportunity that distributed more than $105 million to 1 in 3 health centers in 2014 to expand on-site behavioral health capacity.¹⁶ Another factor that may spur utilization of substance use treatment and mental health services is expansions in insurance coverage under the Affordable Care Act.^62–64 Mental health and substance use disorders are included among the “essential health benefits,” so they should become more broadly covered under private health insurance as well as Medicaid plans. In addition, the Affordable Care Act also expanded the parity requirements of the Mental Health Parity and Addiction Equity Act to the individual and small group markets.⁶⁵

Limitations

The limitations of our analysis include the implications of using self-reported survey data. Although we contacted some grantees to verify the UDS data, we did not verify the assessment survey data. There may be errors because of misunderstanding questions or social desirability bias. In addition, the survey instrument did not allow full examination of how EHRs were used or elements of integrated care such as patient engagement, coordination during care transitions, and whether the providers were working from a unified care plan. The survey question about interdisciplinary provider-to-provider meetings did not capture asynchronous communication, so the frequency of communication might be understated. The survey does not capture which types of services were provided or by which staff types, and there are likely other missing factors that might lead to omitted variable bias, such as more detail on state Medicaid reimbursement policy.

Another weakness is the low response rate for the survey (36%); we weighted the data to yield nationally representative results. The dependent variable used in the multivariate model might not capture every facet of integration and might not accurately reflect the level of integration because of the data constraints. Finally, the survey data were from 2010, so these results do not reflect any changes resulting from increased sophistication of EHR functionalities that have occurred since then because of the certification requirements tied to payments through the Medicaid and Medicare EHR Incentive Programs.

As treatment of behavioral health disorders increasingly shifts to the primary care setting, further research is needed to elucidate, test, and build the evidence base around best practices for integrated care and inform policies that aim to build the necessary infrastructure to support shared access to information, joint case conferences, and joint care planning. Although we focused on behavioral health, it is important to include case managers on the care team. The foundational elements of integrated care can be used to better integrate all types of specialty care with primary care as the patient-centered medical home model evolves into the medical neighborhood.^66–68 Finally, further research is needed about how best to engage patients and families as additional members of the care team; patient education is needed to improve health literacy, and tools such as health risk assessments can help to activate patients.^69–72

Conclusions

Behavioral health is a key component of overall health, and we examined how integrated care is used to meet the behavioral health needs of patients in community health centers. These findings indicate that although some health centers are using integrated care to address the treatment needs of patients with behavioral health conditions, there is room for improvement in joint case conferences and joint care planning. These findings highlight the importance of monitoring for the possibility of disparities and targeting support to health centers that might be falling behind in integrated care provision, such as health centers without EHRs and health centers with low behavioral health staffing levels. Financial incentives and targeted technical assistance and support for communication and collaboration can boost the level of integrated care in health centers.