Increasing Access to and Utilization of Behavioral Health Care Through Integrated Primary Care

During our evaluation period (September 2015 to October 2018), our academic health center’s primary care network included 12 clinics. All clinics are located in underserved urban communities. The primary care network averaged 134 992 visits per year for a total of 404 977 patient visits over the course of the project. Most patients are Medicaid-enrolled and members of our accountable care organization (see Table 1 for demographic details). In October 2016, a psychologist was integrated into each of the 4 primary care clinics 4 days per week, with 1 day each week for administrative activities. At the start of integration, our institution employed 325.8 full-time equivalent (FTE) specialty psychologists and therapists who worked in the hospital, local schools, and community outpatient centers. Total effort increased by 61.5 by the end of the evaluation period (October 2018) to a total of 387.3 FTE. We ended the evaluation period in 2018 because of later changes that occurred across the primary care network, including the expansion of the number of clinics, total patient volumes, and processes. Integrated clinics were larger, serving an average of 11 047 unique patients per clinic site compared with 8001 unique patients over the evaluation period. All clinics were located in the same metropolitan area and were similarly distanced from specialty BH locations.

We integrated 4 of 12 primary care clinics to test the feasibility and impact of integration. The 8 nonintegrated clinics served as comparators. We chose clinics to integrate on the basis of several factors, including patient volume, community needs, and matching psychologists to the communities served (eg, race, ethnicity, and language when possible). When designing our integration model, we considered the needs of our community, the clinical and operational needs of PCCs, as well as gaps in our existing BH system. PCCs reported high rates of BH needs within the community, insufficient time to address BH concerns, stigma, long wait times, and family contexts that limited follow-through with referrals (eg, financial, transportation, and employment challenges).

We selected the Primary Care Behavioral Health (PCBH)²⁷  integration model because it prioritizes joint, team-based visits to maximize the care families get during their primary care visits with brief follow-ups on the BH clinician’s schedule (typically 3 to 5 30- to 45-minute sessions) when indicated. Integrated psychologists provided generalist BH care to children aged 0 to 21.³⁴  Patients who did not respond to IPC interventions, needed psychological testing, or had acute needs were referred to specialty BH services.

Compared with colocated models of care, the PCBH approach affords greater population reach, as well as requiring fewer transportation demands and fewer days off work and school because families do not have to return for separate BH and medical appointments.²⁷  PCBH was chosen over the Collaborative Care Model because our system already had a telephonic psychiatric consultation line to address psychotropic medication questions, PCCs wanted support with the full range of patients rather than a specific clinical population, and it was deemed more feasible to employ a psychologist who can bill for their services opposed to a care manager, consulting psychiatrist, and regularly scheduled consultation time for the PCC. Our model met the criteria for level 6, the highest level of integration, on the Integrated Practice Assessment Tool³⁵  and included all components of the PCBH model (see Supplemental Table 7 for PCBH components and Reiter & colleagues³⁰  for an overview and how it contrasts with other IPC models).

After selecting the clinical model, we mapped existing clinic processes, developed new workflows that prioritized joint, team-based visits, created a data dashboard to track implementation metrics, and conducted regular implementation meetings. Psychologists billed current procedural terminology (CPT) for mental health diagnoses, health and behavior billing codes for physical health diagnoses (eg, adherence to asthma treatment plan), and used a “non-billable” code to track clinically meaningful encounters that did not meet billing requirements (wellness promotion,³⁴  subclinical problems, and visits not meeting minimum time; see Table 3 for billing codes). Psychology charges were self-billed, not billed as “incident to.” IPC psychologists had the same institutional productivity expectation as specialty BH providers and delivered additional nonbillable services as need and capacity allowed.

Our project includes a 1-year baseline period from October 2015 to September 2016 and 2 years of evaluation data from October 2016 to October 2018. Changes in access to and utilization of BH services were compared between integrated and nonintegrated clinics to account for the growth in specialty BH services and other institutional interventions to increase access to and utilization of BH services.

Given documented inequities in access to BH care⁶  and our mission to increase the equity of BH care, we disaggregated data by race, ethnicity, preferred language, age, and insurance type. Demographic data were reported by families at the time of the appointment and entered into the medical record by registration staff as a standard part of clinical care.

We measured the utilization of BH treatment by calculating the percentage of primary care appointments each month that resulted in contact with a non-prescribing BH provider (masters- or doctoral-level clinician providing either specialty BH or IPC services) within 180 days from the primary care visit. BH contact was identified by a BH billing code associated with a visit within our health care system occurring within 180 days of an appointment in primary care. We selected 180 days to account for the delays in accessing specialty BH and to align with previous studies.⁶  The percentage of visits was calculated rather than using raw volume to account for the difference in visit volume between integrated (n = 193 541) and nonintegrated clinics (n = 211 436). We included all primary care appointments (for any presenting concern or age) in our calculations consistent with our generalist clinical model.

We used wait times as a proxy for access to care. Wait times were defined as the number of days from the date a referral was placed to attendance at the first appointment with a non-prescribing BH provider as defined above.

Data for analysis were acquired from the electronic health record data warehouse. Descriptive statistics were used to describe the sample and summarize variables.

Access to BH services was assessed by comparing the median wait time from identification of a BH concern to initial contact with a BH provider (specialty and IPC) between integrated and non-integrated clinics and the interquartile range (IQR; IQR = Q1–Q3).

To test for significant change in BH utilization, we used a comparative interrupted time-series analysis with ordinary least-squares regression-based approach estimate using Newey-West standard errors in Stata 15.0.³⁶  This allowed for both pre and post comparisons within integrated clinics as well as comparisons between integrated and control clinics. Autocorrelation was present at lag 1 but not at any higher lag orders (up to 6 tested); thus, our model specifying lag(1) correctly accounted for autocorrelation.³⁷ 

We assessed differences in the utilization rate of BH services across demographics at baseline using the χ-square test with regression-based planned contrasts.³³  Demographic changes in utilization postintegration were assessed by using a cross-sectional difference-in-difference comparison.³³  This was accomplished by using regressions with Bonferroni corrections to test for differences in utilization rates pre and post-integration within each demographic category (ie, age, race, ethnicity, language, and insurance) between integrated and nonintegrated clinics.

We compared cross-sectional differences in the percentage of specialty BH contact pre- and post-integration within and between integrated and nonintegrated clinics.

This project was considered exempt by the institutional review board.

Integrated psychologists were involved in ∼5.7% of all primary care visits in integrated clinics, and 74% percent of all patient contacts (initial and follow-up) occurred as a joint, team-based appointment; 89.1% of contacts were billed. Of the billable contacts, 89.2% were mental health CPT codes and 10.8% were health and behavior. The 3 most common diagnoses were adjustment, behavioral problems, and ADHD (see Table 4). IPC psychologists billed 12% less than specialty psychologists in the first year, and 6% less in the second year.

Within integrated clinics, 93% of initial IPC contacts occurred during their primary care appointment, resulting in no wait time. Of the 7% of patients whose first contact occurred on the psychologist’s schedule, the median wait time to see an integrated BH psychologist was 11.4 days (IQR = 5.3–17.7) compared with 48.3 days (IQR = 20.4–93.6) for specialty BH. Appointments scheduled from referrals to integrated psychologists were completed within 30 days 90.8% of the time compared with 36.2% for specialty BH.

Comparative interrupted time series analysis (see Fig 1 and Table 5 for details) indicated there were no statistically significant differences in the utilization of BH services between integrated and nonintegrated clinics during the baseline period. In the 2 years after integration, IPC clinics experienced an immediate and sustained increase in the utilization of BH services. The percentage of primary care visits that resulted in contact with a BH clinician in 180 days increased from an average of 4.6% in the year before integration to an average of 11.2% during the first 2 years of integration. During this same time period, there was no significant change in utilization for nonintegrated clinics (4.0% to 4.4%).

Table 1 reveals population-level demographic differences between integrated and nonintegrated clinics during the evaluation period, and Table 2 reveals BH utilization data across measured demographics. Of note, non-English-speaking parents and patients 0 to 2 years of age had lower rates of BH utilization, and non-Hispanic White patients had the highest levels of utilization within both integrated and nonintegrated clinics at baseline. Difference-of-difference analysis indicated that patients in integrated clinics had significantly more growth in utilization compared with nonintegrated clinics across all demographic categories, with the largest growth occurring for Nepali-speaking patients in integrated clinics. Somali-speaking parents did experience significantly more utilization in integrated clinics compared with nonintegrated clinics, but growth was notably smaller for these patients. Although integration significantly increased utilization for all demographics, non-Hispanic White and English-speaking youth were still more likely to receive contact with any type of BH provider (ie, specialty and IPC combined) within integrated clinics during the study period overall. However, there were significantly fewer and sometimes no demographic differences in the utilization of the IPC psychologist (see Supplemental Table 7).

Patients in integrated clinics were less likely to use specialty BH services (ie, non-prescribing BH services outside of the primary care clinic) compared with baseline and compared with nonintegrated clinics during the evaluation period. Specifically, contact with specialty BH reduced from 4.6% before integration to 3.5% after integration (25% reduction) and was statistically lower than the 4.4% of patients seen by specialty BH in nonintegrated clinics. Approximately 14% of patients in integrated clinics were seen by both IPC and specialty BH.

Integration resulted in a significant and immediate increase in the percentage of primary care visits that resulted in contact with a BH provider within 180 days. Given the long delays between symptom onset and treatment,¹¹  it is notable that ∼25% of patients seen by a BH provider within integrated clinics were ≤5 years of age and 10% were ≤3 years of age. Patients accessed IPC psychologists quickly, with 93% of initial contacts occurring during the primary care appointment when a need was identified. Patients in integrated clinics were also less likely to use specialty BH services after integration. Thus, the PCBH model improved access to integrated, team-based care at the population level and reduced the need for specialty services consistent with the mission of primary care.^38,39,40 

This study adds to the literature in several important ways. First, we used a highly integrated, team-based IPC model (PCBH). Second, our clinics serve a diverse and geographically urban population, an understudied population in IPC research.^27,40  Third, we used a single-component approach to assess the unique effects of integration. Finally, we included comparator clinics within the same network to help account for system-level and historical confounds. Notably, the use of comparator clinics led to the surprising finding that substantial growth of specialty BH clinicians over the 2-year integration period did not lead to a significant increase in the utilization of BH services for patients in nonintegrated clinics. Given the large improvement in BH utilization observed after investing a single integrated FTE per clinic, integration proved to be a highly efficient model for increasing the utilization of BH services for our urban primary care population.

Given existing racial and ethnic disparities in care, we designed our services to be culturally informed and connected to the community (eg, participating in community events, partnering with community organizations).³⁴  We prioritized team-based, same-day visits because of structural barriers limiting follow-up and matched psychologist demographics to the community when possible. These efforts likely contributed to the reduction of demographic disparities in BH care utilization within integrated clinics. However, disparities still remained in some areas, including in the use of specialty BH services, and we found only modest increases in utilization for Somali-speaking patients in integrated clinics. To address the modest improvements for Somali-speaking patients, we have engaged with Somali-oriented community organizations and hired a Somali-speaking therapist. Although integration typically improves linkage to specialty BH care,⁵  further investigation is needed to understand how integrated psychologists can address inequities in who accesses specialty BH services.

Although the use of comparator clinics is an advantage of this study, clinics were not randomly assigned, and some small baseline differences between integrated and nonintegrated clinics were observed. This quasi-experimental design has limits to its causal interpretations, although the observed effects were large. In addition, increased access to and utilization of BH care does not guarantee improved outcomes. Despite growing evidence supporting the effectiveness of single-session⁴¹  and brief co-located interventions,⁴²  further research on clinical outcomes in pediatric PCBH is needed. Although most patients receive care within our large BH system of care, we were unable to track the use of specialty care outside of our institution. Additionally, self-reported language data kept within the medical record do not allow us to account for the possible differences for families who speak English as a second language and those who do not, as well as families with generational differences in language barriers, both of which are areas for future research to explore. Although the evaluation of the financial impact is beyond the scope of this paper, it is worth noting, given the challenges of funding IPC models (including PCBH⁴³ ), that integrated psychology billing was similar to specialty psychologists in the second year of integration. Future studies should further explore the financial impacts of pediatric PCBH models.

This study adds to a growing body of literature that reveals the benefit of shifting some of the BH workforce “upstream” to support primary care in managing pediatric BH needs. Although the workforce is limited, integrating BH clinicians within primary care teams can result in patients being seen more quickly, equitably, and efficiently⁴⁴  in the setting they prefer²  while reducing the demand for specialty BH services and making them more accessible for patients with more acute needs.⁴⁵ 

BH

behavioral health

CPT

current procedural terminology

FTE

full-time equivalent

IPC

integrated primary care

IQR

interquartile range

PCBH

Primary Care Behavioral Health model

PCC

primary care clinician