Five-Year Outcomes of Behavioral Health Integration in Pediatric Primary Care

The eligible BHIP population comprises 84 pediatric practices enrolled in a statewide, independent practice association affiliated with an academic medical center. The practices include nearly 400 PCPs serving >350 000 patients.

After notification of project launch and participation requirements through usual network communication channels, 71 practices (85%) agreed to participate in the BHIP. Program participation requires that practices (1) designate ≥1 PCPs (physicians or nurse practitioners) to attend the education component and serve as BH “champion” for the practice; (2) use the consultation component as needed; (3) engage in integrated practice transformation, including creating a BH team composed of PCPs, behavioral health clinicians (BHCs [psychologists, social workers, counselors]), and care coordinators; and (4) provide on-site clinical BH services. Participating practices phased into the BHIP at a rate of ∼10 to 15 per year. BHIP enrollment began in July 2013; once a practice is enrolled, participation is ongoing. Because the project is consistent with our institution’s definition of a quality improvement program, the need for individual informed consent was waived.

The sample for this article (BHIP, Phase-1) comprises the first 13 practices (with ∼105 PCPs serving ∼114 000 patients) reporting readiness to meet all BHIP participation requirements. The Phase-1 practices are located in the Greater Boston area and range in size from 3 to 17 PCPs. The practices’ communities range from 14 714 to 86 241 in population and $28 000 to $72 744 in per capita income (mean: $47 258, compared with $35 763 in Massachusetts⁵³) and are on average 78% white, 8% African American, 6% Hispanic, and 5% Asian American (compared with 74%, 9%, 11%, and 6%, respectively, in Massachusetts⁵³). Approximately 85% of practices’ patients are commercially insured; 15% have Medicaid. Structure, process, and outcomes were assessed in Phase-1 practices from July 2013 through June 2018.

The BHIP education component (Behavioral Health Learning Community [BHLC]) comprised 10 sessions (16 hours) delivered over the first 2 years of Phase-1; 12 hours were delivered in person in a geographically central location, and 4 hours were delivered by televideo. The adult-learning format⁵⁴ encompassed an interactive blend of didactic and case-based material. Formal sessions were supplemented by optional bimonthly televideo case discussions. BHLC faculty comprised child and adolescent psychiatrists (CAPs), developmental and behavioral pediatricians, adolescent medicine specialists, psychologists and neuropsychologists, and clinical social workers from the affiliated medical center.

The BHLC was designed to provide the knowledge, skills, and resources needed to safely and effectively identify, assess, and treat mild and moderate presentations of common child and adolescent psychiatric disorders, with an emphasis on anxiety, depression, ADHD, and related disorders (stress, behavior). The BHLC emphasized universal screening⁵⁵ to identify problems with anxiety, depression, and ADHD; use of symptom rating scales^56,–59 to assess problem severity and monitor response to treatment; provision of brief psychotherapy; prescription of guideline-congruent^60,–62 medications for anxiety, depression, and ADHD when indicated; and knowing which patients to refer to specialty care. Twenty category 1 continuing medical education credits were offered to physicians by the affiliated medical school; no other incentives for participation were offered.

The BHIP consultation component provided real-time, workday, 9 am to 5 pm telephone consultation by CAPs to PCPs. Consultation was designed to reinforce and extend the knowledge, skills, and resources acquired in the BHLC to the management of individual patients. Telephone consultations provided initial assessment and treatment suggestions and ongoing medication management support. For severe presentations, the CAP facilitated referral to specialty BH care for in-person consultation, interim treatment until stable, or ongoing treatment as indicated. To provide continuity of learning, the first 3 years of Phase-1 consultation was provided by CAP BHLC faculty; to ensure fiscal sustainability, in the subsequent 2 years, consultation was provided through a partnership between the BHIP and the Massachusetts Child Psychiatry Access Program (MCPAP).⁶³ 

The BHIP practice transformation component comprised 6 operational sessions (12 hours) focusing on clinical and business workflows; revenue-cycle management; BHC hiring, contracting, and credentialing; BH crisis protocols; linkages to specialty BH services; electronic health record (EHR) documentation; and support for practice-individualized quality improvement projects. Operational sessions were supplemented by 6 to 8 hours per month of practice-based support. Clinical support for BHCs by BHIP senior social workers comprised 1 to 2 hours per month of individual telephonic supervision, 1 hour per month of televideo case conferences, and twice-yearly 6-hour in-person trainings.

On-site clinical BH service used a stepped-care model in which each step has a clearly defined patient population, goal, and provider and ascends in treatment intensity in accordance with the severity of the clinical presentation (Fig 1). The clinical roles of PCPs were BH screening, guided self-management, and psychopharmacology; the roles of BHCs were focused assessment and brief psychotherapy. Practice-based care coordinators assisted practitioners with referrals to specialty care as indicated.

Progress of Phase-1 practices toward full BH integration⁶⁴ was assessed at baseline and 5-year follow-up by completion of the Behavioral Health Integration Readiness Assessment (BHIRA) by BH teams in each practice. This 35-item, 10-point Likert-scale instrument, adapted from the AAP Mental Health Readiness Inventory⁶⁵ and the Maine Self-Assessment Evaluation Tool,⁶⁶ is used to measure the degree of BH integration in 8 domains; examples of BHIRA items are presented in Table 1. A total mean score was calculated from the domain score means.

BHC use in Phase-1 practices was compiled from deidentified medical records maintained by the pediatric practice network; analysis was limited to 6 out of 13 practices using the same EHR. Use of consultation was compiled from deidentified records in an electronic database maintained by the consultation program through June 2016; after that, the consultation component merged with the MCPAP,⁶³ which maintained separate usage records.

BH service delivery in Phase-1 practices was defined as all psychotherapy visits and all ambulatory medical and hospital emergency department (ED) visits for a primary BH diagnosis (see Table 2 for billing codes) and all prescriptions for guideline-congruent (Table 3) medications for anxiety, depression, and ADHD. Total ambulatory BH costs were defined as any spending on outpatient and emergency BH visits and BH pharmacy. BH visits and costs, compiled from 2013 through the end of 2017 (the most recent complete project year), were derived from paid insurance claims from a single, large commercial insurance company that shares data with the practice network. All costs were adjusted to 2017 dollars by using the Consumer Price Index for Medical Care (Northeast region).

Provider satisfaction was assessed through anonymous electronic surveys completed by BHIP providers at 5-year follow-up. The BHC survey, addressing BH self-efficacy and professional satisfaction, was adapted from a previous survey administered to PCPs.⁶⁷ 

Baseline and follow-up BHIRA scores were compared by using paired-sample t tests. Interrupted time series analysis was used to assess changes in psychotherapy and medical BH visits and psychiatric prescriptions from 2013 (pre-BHLC) to 2015–2017 (post-BHLC), controlling for secular trends and postlaunch slope changes.^68,69 Autocorrelation among data points was assessed by using the Durbin-Watson statistic; when significant autocorrelation was found, autoregressive parameters were included in the model. All analyses were completed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC). Trends in ED use were assessed by using statistical process control methodology (QI Macros software; KnowWare International, Denver, CO) to determine if special cause variation coincident with BHIP implementation occurred. Ambulatory BH costs in 2013 (pre-BHLC) were compared with the average 2015–2017 costs (post-BHLC), adjusted to 2017 dollars.

Twelve Phase-1 practices (92%) completed the BHIRA at baseline and 5-year follow-up. The total mean BHIRA score increased from 4.9 (2013) (“some degree” of integration) to 7.0 (2018) (integration “most of the time”) (P < .001). Increases in 7 of the 8 domain mean scores also were statistically significant (Fig 2). At follow-up, all 8 integration readiness domains were demonstrated “most” or “all” of the time.

All Phase-1 practices designated ≥1 PCP BH champion to enroll in the BHLC. Twenty-nine PCPs enrolled (28% of 105 Phase-1 PCPs); of those, 23 (79%) completed continuing medical education requirements (attendance at >70% of sessions, completion of assigned quality improvement coursework).

All Phase-1 practices participated in telephone consultation. Over 30 months, consultations totaled 254 (mean: 8.5 per month; range: 0–62; mean: 19.5 per practice). The top 2 reasons for consultation pertained to medication management (52.2%) and diagnosis (13.1%); the remainder primarily pertained to level and type-of-care decisions. The top provisional postconsultation diagnoses were anxiety (26.6%), depression (22.7%), ADHD (14.8%), and behavior disorder (7.0%). Postconsultation, 67.3% of patients were mutually agreed to be appropriate for PCP management, and 32.7% were referred to specialty care. As rated by the CAP consultant, the mean Children’s Global Assessment Scale score (used to measure functional impairment⁷⁰) for referred patients was 48 (moderately severe). Of referred patients, 48.6% were referred locally, 35.7% were referred to the affiliated medical center, 10.0% were referred to the state BH system for Medicaid-insured youth, and 4.1% were referred to a hospital ED.

All Phase-1 practices received integration support. Eleven practices (85%) integrated a BHC, 80% by hiring and 20% by contracting with a community agency. Seventeen BHCs were hired, ranging from 1 to 3 per practice. The median age of BHCs’ patients was 11 years; 51.6% were boys. The median number of psychotherapy visits per patient was 3 (range: 1–40). Disorders most commonly diagnosed were stress related (40.3%), anxiety (32.4%), attention (13.0%), depressive (7.7%), and behavior (2.1%); comorbid disorders comprised 8.6%. Of all billable visits (8042), 46.6% were evaluation and 53.4% were psychotherapy; the nonbillable (telephone calls) to billable ratio was ∼2:1.

The integration of BHCs resulted in increased psychotherapy visits in Phase-1 practices from 11 out of 1000 patients per year (2013) to 230 (2015–2017) (P < .001; Fig 3). Psychotherapy visits in specialty settings remained unchanged (1508 [2013], 1499 [2015–2017]; P = .53). The proportion of total (BHIP plus specialty) psychotherapy visits delivered within BHIP practices increased from 0.7% (2013) to 13.3% (2015–2017). A marked seasonal pattern in psychotherapy visits (fewer visits in summer) observed in specialty care was not observed in BHIP practices.

Medical BH visits to Phase-1 PCPs increased from 137 out of 1000 patients per year (2013) to 174 (2015–2017) (P = .04; Fig 4) while remaining unchanged in specialty settings (246 [2013], 244 [2015–2017]; P = .99). The proportion of total medical BH visits delivered within BHIP practices increased from 35.8% (2013) to 41.6% (2015–2017).

Prescriptions for guideline-congruent anxiety and depression medications (selective serotonin reuptake inhibitors [SSRIs]) increased in BHIP practices from 55 30-day fills per 1000 patients per year (2013) to 186 (2015–2017) (P = .05; Fig 5). The proportion of total SSRIs prescribed within BHIP practices increased from 17% (2013) to 40% (2015–2017). Prescriptions for guideline-congruent ADHD medications (stimulants, α-agonists, atomoxetine) increased in Phase-1 practices from 299 30-day fills per 1000 patients per year (2013) to 397 (2015–2017) (P = .05; Fig 5). The proportion of total ADHD medications prescribed within BHIP practices increased from 48% (2013) to 53% (2015–2017).

Beginning in 2014, ED BH visits for patients in Phase-1 practices decreased by 33.9% (18.9 visits per 1000 patients per year to 12.5) (Fig 6). However in 2017, ED visits rose to 18.5 visits per 1000 patients per year, similar to the pre-BHIP rate.

Over 5 years, total ambulatory BH costs (BHIP plus specialty) increased by 8% in constant dollars, from $32.78 per patient per year (2013) to an average of $35.32 (2015–2017) (Fig 7). Care within BHIP practices accounted for nearly all of the increase, with BHIP outpatient costs increasing by 123%, from $2.06 per patient per year (2013) to an average of $4.60 (2015–2017) and BHIP pharmacy costs increasing by 40%, from $3.56 per patient per year (2013) to an average of $4.98 (2015–2017). Total BH-related emergency costs decreased by 19%, from $1.88 per patient per year (2013) to an average of $1.51 (2015–2017).

As reported from a separate study, survey responses from 66 PCPs in BHIP Phase-1 through Phase-3 (81% response rate) demonstrated high BH self-efficacy and satisfaction associated with BHIP participation.⁶⁷ Survey responses from 14 BHCs (93% response rate) in BHIP Phase-1 were similar (Fig 8). More than 90% of surveyed PCPs and BHCs believed that BHIP participation enables effective management of mild and moderate BH problems in pediatric primary care.

This study is the first to our knowledge in which researchers have examined, over an extended period, the structure, process, and outcomes of a multicomponent, transdiagnostic model of BH integration in real-world pediatric settings with diverse patient populations exercising treatment choice. In the findings, it is suggested that the BHIP was feasible to implement and successful in moving practices toward full BH integration, increased primary care BH access while maintaining quality and averting anticipated substantial increases in BH costs, and achieved high provider BH self-efficacy and professional satisfaction. Although favorable findings have also been demonstrated in several research-based studies of collaborative pediatric BH care,^71,–73 because this was a naturalistic study, the findings from the BHIP may have heightened applicability to other real-world pediatric settings and, as such, serve as a model for their development.

Over the 5-year observation period of BHIP Phase-1, clear evidence of increased BH integration⁶³ in participating practices emerged, encompassing leadership, BH resources, administrative, clinical management, family centeredness, care coordination, and quality improvement domains. In this enabling structural context, beneficial findings were observed in patient experience, cost, and provider satisfaction.

Primary care access to both psychotherapy and medical BH visits and to anxiety, depression, and ADHD medication increased significantly over time. By the end of the observation period, BHIP BHCs provided more than one-tenth of all psychotherapy visits, and BHIP PCPs provided more than two-fifths of all medical BH visits, two-fifths of all SSRI prescriptions, and one-half of all stimulant, α-agonist, and atomoxetine prescriptions (for comparison, in a recent national study, PCPs prescribed one-tenth of antidepressants to children and adolescents, compared with specialists⁷⁴). The access findings were congruent with ambulatory BH costs, in which shifts over time in both BH outpatient and pharmacy costs from specialty to primary care were demonstrated.

The shift in service venue from specialty to primary care (“task-shifting”⁷⁵) is a central tenet of integrated care that has substantial access advantages to patients and families, including proximity, continuity, familiarity, trust, reduced stigma, and lower cost.⁷⁶ Task-shifting did not appear to be associated with a decrement in the quality of care, in that increases in the prescribing of anxiety, depression, and ADHD medications by BHIP PCPs adhered to guideline-congruent medications. Moreover, all patients receiving psychotherapy from BHIP BHCs had diagnoses targeted for primary care management, and the average number of therapy visits was consistent with brief primary care intervention.⁷⁷ Other BHIRA-derived quality-of-care indicators were significant increases in the extent to which BH screening, evidence-based care, and quality improvement activities occurred in BHIP practices.

Total ambulatory BH spending increased by only 8% over 5 years, suggesting that BH task-shifting to primary care with the goal of increasing access did not substantially increase costs. In the only other collaborative pediatric BH study in which researchers examined spending,⁴⁸ ambulatory BH costs in primary care also increased during the intervention because of greater access to BH services in that setting.

Total emergency BH spending decreased by 19%, which may partially reflect decreased (albeit not sustained) ED BH usage by patients in BHIP practices. By comparison, the medical center serving as a major referral site for pediatric BH emergencies in the Greater Boston area reported an 86% increase in ED BH volume from 2013 (983 visits) to 2017 (1824 visits) (D.R. DeMaso, MD, personal communication, 2018). It may be that early identification and intervention in lower-cost primary care settings can decrease overuse of high-cost emergency BH services, although this remains to be definitively demonstrated.

Provider burnout⁷⁸ was of interest in the BHIP because of the expansion of PCPs’ and BHCs’ scope of practice to include BH and primary care, respectively. In follow-up surveys, it was demonstrated that both groups of providers developed high BH self-efficacy and professional satisfaction as a result of BHIP participation, and nearly all came to believe that mild and moderate BH problems can be effectively managed in primary care.

Because subsequent phases of the BHIP are scaling up to all 84 network pediatric practices with nearly 400 PCPs, sustained expansion was a major focus from the outset. With philanthropic support, the education component is being migrated to a free-access Web-based platform maintained by the affiliated medical center to enable on-demand PCP learning.⁷⁹ The consultation component merged with the MCPAP, which receives ongoing legislatively mandated state and commercial payer funding.^80,81 In preliminary data, it is suggested that integrated BHCs can be revenue-neutral, despite relatively high unbillable services (J. Bromberg, PsyD, personal communication, 2018).

Strengths of this study include the large study population, high program participation rates, standardized interventions and outcome measures across all practices, a broad array of quality indicators, availability of EHR and claims data, and lengthy follow-up. As a quality initiative, findings from BHIP Phase-1 are informing program content, delivery, and support in subsequent implementation phases.

Limitations include the nonexperimental design, possibility of selection bias creating an early adopter cohort, service use and cost data limited to a single (albeit largest) commercial payer source (and as such, potentially limited applicability to higher Medicaid populations), inability to disaggregate the effects of individual program components, lack of a comprehensive cost-revenue analysis, and absence of clinical BH patient outcomes (planned for assessment in later BHIP phases after migration of all practices to a single EHR). Additionally, the BHIP benefits from implementation in a quality-driven pediatric practice network affiliated with an academic medical center that provides grant-funded education and clinical and operational support and from participation in a state-funded CAP consultation program. As such, the BHIP model may need adaptation for other settings with different health care structures and resources.⁸² 

In the findings from this study, it is demonstrated that the BHIP model is feasible to implement, highly valued by providers, and useful in contributing to health care’s quadruple aim. If widely scaled, locally adapted, and sustainably funded, programs such as the BHIP can expand the BH workforce into pediatric primary care and in so doing, help alleviate the substantial gap between the millions of youth needing quality BH services and those receiving them.

We thank the following for their contributions to the BHIP: the practitioners and staff at Bridgewater Pediatrics, Children’s Medical Office of North Andover, Benjamin Spingarn Rottenberg Pediatric and Adolescent Medicine, Fellsway Pediatrics, Framingham Pediatrics, Hyde Park Pediatrics, Longwood Pediatrics, Needham Pediatrics, Pediatric Associates of Medford, Post Road Pediatrics, Revere-Winthrop Pediatrics, Wareham Pediatrics, and Westwood/Mansfield Pediatric Associates; the BHLC faculty from the Departments of Psychiatry and Pediatrics of Boston Children’s Hospital; and the BHIP staff, past and present.

AAP

American Academy of Pediatrics

ADHD

attention-deficit/hyperactivity disorder

BH

behavioral health

BHC

behavioral health clinician

BHIP

Behavioral Health Integration Program

BHIRA

Behavioral Health Integration Readiness Assessment

BHLC

Behavioral Health Learning Community

CAP

child and adolescent psychiatrist

ED

emergency department

EHR

electronic health record

MCPAP

Massachusetts Child Psychiatry Access Program

PCP

primary care practitioner

SSRI

selective serotonin reuptake inhibitor