Brief Behavioral Therapy for Pediatric Anxiety and Depression in Primary Care: A Follow-Up RH = Brief Behavioral Therapy in Primary Care

Abstract

Objective:

To report on the 32-week outcome of the Brief Behavioral Therapy (BBT) for Pediatric Anxiety and Depression in Primary Care clinical trial.

Method:

185 youth aged 8–17 with anxiety and/or depression identified through 9 pediatric primary care (PPC) settings in San Diego and Pittsburgh were randomized to receive Assisted Referral to Care (ARC) or up to 12 sessions of BBT over 16 weeks. The primary outcome was clinical response across anxiety and depression, defined as a Clinical Global Impressions-Improvement Score ≤ 2. Secondary outcomes included interview-rated functioning, depression, and anxiety. Here, we report on outcomes at 32 weeks after randomization. All analyses with primary outcomes are corrected for multiple comparisons utilizing the False Discovery Rate technique.

Results:

At 32 weeks, BBT was superior to ARC with respect to response (67.5% vs. 43.1%, q=0.03, NNT=5) and functioning (d=0.49, q=0.04). BBT was superior to ARC with respect to its impact on anxiety (f=0.21) but not depressive symptoms (f=0.05). These findings persisted after controlling for the number of sessions received. Ethnicity moderated the impact of BBT on outcome, (NNT for Hispanic youth=2), due to a much lower response rate to ARC in Hispanics than in non-Hispanic youth (16.7% vs. 49.2%, p=0.04).

Conclusion:

BBT is a promising intervention that can be effectively delivered in PPC and may be particularly effective for Hispanic patients. Further work is indicated to improve its impact on depressive symptoms and to test BBT against other treatments delivered in pediatric primary care.

Clinical Trial Registration:

Brief Cognitive Behavioral Therapy (CBT) for Pediatric Anxiety and Depression in Primary Care; http://clinicaltrials.gov; NCT01147614.

INTRODUCTION

Anxiety and depression in youth are common, impairing, often comorbid, under recognized, and undertreated¹. Moreover, there are considerable ethnic disparities in treatment rates, with Hispanic youth being less likely than non-Hispanic youth to be identified and treated for these disorders.^2,3 Untreated, these conditions lead to sustained functional impairment in interpersonal relationships, education, and work, as well as an increased risk for suicidal behavior.^4–6 Therefore, successful treatment of internalizing problems in youth during the developmental periods of childhood and adolescence is critical.

There are several barriers that contribute to under-treatment of anxiety, depression, and the accompanying ethnic disparities. First, even when recognized, many young patients and their families have difficulty gaining access to care.^3,7–9 Second, ethnic minority youth experience even greater barriers to obtaining care due to health care beliefs, perceived stigma, and discrimination.¹⁰ Third, even when patients are able to access care, treatment is often neither diagnostically-specific nor evidence-based.¹ Finally, the majority of evidence-based treatments do not address the common co-morbidity between anxiety and depression, potentially resulting in inefficient and incomplete treatments with sub-optimal results.¹¹

To address these barriers to optimal access and outcomes, Weersing and colleagues developed a trans-diagnostic Brief Behavioral Treatment (BBT) for anxiety and depression, designed to be delivered in a pediatric primary care setting. BBT is an intervention that combines exposure for anxiety and behavioral activation for depression to encourage graded engagement and approach toward avoided life tasks.¹² Of note, BBT does not rely on cognitive restructuring, an important ingredient in many evidence-based interventions for depression and anxiety that has been shown to be challenging for community clinicians to learn and deploy and for youth to assimilate.¹³ BBT relies on problem-solving, one of most frequently used techniques among community clinicians learning evidence-based treatments, and emphasizes behavioral activation and exposure, both identified as key effective elements for treatment response in depression and anxiety, respectively.^14–16 Delivery of treatment in the primary care setting overcomes many barriers to care, is widely believed to have less accompanying stigma than accessing care in a specialty mental health care setting, and thus may be effective in overcoming disparities in access and utilization.^17–19 Moreover, a recent meta-analysis shows conclusively that collaborative care is an effective treatment approach for pediatric mental health disorders.²⁰

We have previously reported on the acute effectiveness of BBT in the treatment of depression and anxiety in 185 children and adolescents treated in PPC in San Diego and Pittsburgh.²¹ Participants were randomized to either BBT or assisted referral to care (ARC). Sixteen weeks after randomization, youth receiving BBT were more likely to show significant clinical improvement (56.8% vs. 28.2%, Number Needed to Treat [NNT]= 3.5), greater reduction in symptoms (Cohen’s f=0.28), and better functioning (Cohen’s d=0.58) than those in the ARC condition. Moreover, treatment effects were moderated by ethnicity, with Hispanic youth showing a more robust response to BBT vs. ARC than non-Hispanic youth (NNT=2).

While these results are promising, it is important to determine if the benefits from BBT were enduring. Therefore, in this paper, we report outcomes of BBT vs. ARC at 32 weeks post randomization, or a mean of 4.9 (standard deviation [SD]=1.2) months after the end of treatment. Specifically, we report on the differential impact of BBT vs. ARC on overall improvement, functioning, symptomatology, and service use, and the extent to which ethnicity and baseline symptom severity moderated clinical and functional status at follow-up. Finally, within those treated with BBT, we examine predictors of treatment response, including “dose” of treatment, and test whether the differences between BBT and ARC persisted after adjustment for difference in treatment “dose.”

METHOD

Sample

Youth ages 8 to 17 years were recruited from 9 PPC practices in San Diego (n=4) and Pittsburgh (n=5) from October 6, 2010 through December 5, 2014 (see Figure 1 for a Consort Diagram). Youth were eligible for the study if they met full or probable criteria for DSM-IV²² anxiety (separation, generalized, social phobia) and/or depression (major, minor, dysthymic disorder), lived with a consenting legal guardian for at least 6 months, and both the participant and one participant caretaker spoke English. Participants were excluded if they had current suicidality with a plan, bipolar disorder, psychosis, post-traumatic stress disorder, substance dependence, current maltreatment, report of intellectual disability, school placement below the second grade, or reported unstable chronic physical illness. Participants were randomized using a modification of the biased coin-toss,²³ balancing on sex, race/ethnicity, and presence of a diagnosis and/or elevated symptoms of depression. Written assent and consent from youth and parents was obtained in accordance with procedures approved by the Institutional Review Boards of the University of Pittsburgh, Kaiser Permanente Southern California, and the San Diego State University Human Research Protection Program.

Figure 1: Study Flow of Participants: Screening to Analysis.

Note: ARC = assisted referral to care; BBT = brief behavioral therapy.

Interventions

Brief Behavioral Therapy (BBT).

As noted above, BBT is a transdiagnostic intervention designed to address depression, anxiety, and their co-occurrence.^12,21 BBT consists of psychoeducation about depression and anxiety for parents and patients, “graded engagement” through exposure and behavioral activation to avoided (and consequential) activities, along with relaxation to manage somatic symptoms and problem-solving skills for stress management. BBT was delivered individually in 8 to 12 weekly 45-minute sessions over a 16-week period by Master’s level therapists. Training of therapists consisted of a half-day workshop, review of recordings of 2 training cases, and completion of a session-by-session role play. Audio recordings of the first 4 cases were reviewed in detail with therapists, and in subsequent cases, therapists received one hour of weekly supervision across their BBT caseload.

Assisted Referral to Care (ARC).

ARC was based on evidence-based approaches to improving adherence to referrals to community mental health care.²¹ Our referral list included clinicians and programs that had expertise in pediatric depression and anxiety, had a sliding scale, and were willing to see referrals in a timely manner. ARC was delivered over the phone to parents by the study research coordinators, who were Master’s level. Coordinators provided feedback to the participant and parent about the youth’s symptoms and the benefits of services, referrals and education about how to obtain services, and facilitated problem-solving about how to overcome initial barriers to accessing treatment. Parents were called at least every 2 weeks to encourage adherence and the clinician problem-solved with the parent about how to cope with continuing obstacles to obtaining care. More than 80% of parents assigned to ARC (81.1%) answered at least one call; the median number of calls was 4 (Interquartile range [IQR]=9, range 1 to 12). The median total duration across all calls was 20 minutes (IQR=82, range 5–150). ARC was successfully delivered, with 82.2% of families having connected to specialty mental health services for a mean of 6.5 outpatient sessions.²¹

Assessments

Assessments were performed by Master’s level independent evaluators (IEs), blind to condition. Baseline assessments were conducted in the primary care office. Subsequent follow up interviews at 8, 16, 24, and 32 weeks were conducted by phone.

Demographic data were collected by youth and parent report; race/ethnicity was based on parental report using US Census categories.²⁴ The Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL) was used to assess diagnostic eligibility.²⁵ The Clinical Global Impressions scale²⁶ was used to assess global severity (CGI-S) and improvement (CGI-I) across anxiety and depression. A CGI-I of 1 (very much improved) or 2 (improved) indicated clinical response. Functional status was rated by IEs using the Children’s Global Assessment Scale (CGAS),²⁷ anxiety severity using the Pediatric Anxiety Rating Scale (PARS),²⁸ and depression using the Children’s Depression Rating Scale-Revised (CDRS-R).²⁹ Clinically significant depression at baseline was defined as CDRS-R> 40 and/or diagnosis of depression (i.e., DSM-IV criteria for minor or major depression or dysthymic disorder). Inter-rater reliability was good across all outcome measures (ICC’s=0.70–0.95). The Child and Adolescent Services Assessment was administered independently to parents and youths to document health service use for both conditions.³⁰

Statistical Methods and Data Analytic Plan

The primary registered outcome of the trial was clinically significant improvement across anxiety and depression (CGI-I ≤ 2). Secondary outcomes included the PARS, CDRS-R, and functioning (CGAS). For Week 32, we further probed maintenance of response over follow-up and rates of remission, defined as CGI-I=1 as per the definition used in the largest clinical treatment trial for pediatric anxiety.³¹ In addition, we analyzed the impact of treatment dose on outcome to determine if differences between BBT and ARC persisted after controlling for the number of sessions attended. Chi-square tests and logistic regression were used for categorical measures, and t-tests or linear regression for continuous measures. Specifically, independent sample t-tests were used to compare treatment groups on functioning, and linear regression to test whether ethnicity moderated functioning, both at 32 weeks. Given the association between change in anxiety and depression in the literature, we adopted a doubly-multivariate repeated measures approach to jointly analyze changes in the PARS and CDRS-R. These models took into account anxiety and depression symptoms at baseline. As per one of the aims of this study, we conducted pre-planned moderator analyses to test the impact of ethnicity and presence of clinically significant depression at baseline on outcome by following the standard Barron and Kenny and Kraemer frameworks.^32,33

All significance tests were 2-tailed, and analyses were designed as intent-to-treat. To adjust for multiple comparisons in the main analyses, we applied the false discovery procedure with the Yekutieli method³⁴ using the qqvalue package in Stata.³⁵ However, we used α=0.05 for the secondary analyses (i.e. moderation, stratification) given their exploratory nature. To estimate the impact of missing data, we conducted sensitivity analyses for each Week-32 outcome by first performing the planned analyses in the unadjusted intent-to-treat sample, then repeating these models in an imputed dataset in which missing data were estimated by multiple imputation with chained equations, and finally including as covariates participant characteristics significantly associated with attrition. Across these methods, interpretation of results remained unchanged; thus, for simplicity, we present the unadjusted intent-to-treat analyses. In addition, we probed for site effects on outcomes and included site as a covariate in all logistic regression and doubly multivariate repeated measures models. Finally, as 9 sibling pairs were enrolled, we accounted for the presence of siblings by including a term for clustering in the multivariate analyses except for the doubly multivariate approach, for which this option was not available. In those models, we repeated the analyses randomly excluding one sibling; results did not differ from the model with both siblings included.

RESULTS

Sample Characteristics and Retention

As reported previously, the sample (N=185) did not significantly differ by treatment arm on baseline demographic or clinical characteristics, including having private health insurance coverage.²¹ At baseline, 57.8% were female and 32.4% self-identified as members of a racial and/or ethnic minority group, with 21.7% of the sample identifying as Hispanic. Participants most frequently presented as primarily anxious with 62% of youth meeting criteria for one or more anxiety disorders without clinically significant depression, 32% (n=60) displaying anxiety and clinically significant depression, and a small number (n=11, 6%) who were depressed without comorbid anxiety.

Of those randomized in the clinical trial, those who completed the week 32-week assessment (vs. non-completers) did not differ by treatment assignment (BBT vs. ARC; 84.2% vs. 80.0%, χ²₁=0.56, p=0.46; see Figure 1), site (83.2% vs. 81.3%, χ ²₁=0.11, p=0.74) or by response status at week 16 (responders vs. non-responders, 89.9% vs. 88.6%, χ ²₁=0.07, p=0.79), nor were there differences in age (p =0.70), gender (p =0.97), or income (p =0.14). Compared to non-completers, completers were less likely to be Hispanic (p=0.003) and to have parents with less than a college level education (p=0.02), and more likely to live with both biological parents (p=0.03). Non-completers were more likely to be diagnosed as depressed (p=0.01), and to have higher self-reported (p=0.02) and interview-rated (p=0.01) depressive symptoms at baseline (Table 1).

Table 1:

Baseline Characteristics by Completion of Week-32 Assessment

	Total	Did not complete	Completed	test	df	P
	N=185	n=33	n=152
Age, M(SD)	11.3 (2.6)	11.4 (2.7)	11.2 (2.5)	t=0.38	183	0.70
Female sex, n(%)	107 (57.8)	19 (57.6)	88 (57.9)	χ2=0.001	1	0.97
Caucasian, n(%)	144 (77.8)	22 (66.7)	122 (80.3)	χ2=2.91	1	0.09
Ethnic Hispanic, n(%)	38 (20.7)	13 (39.4)	25 (16.6)	χ2=8.62	1	0.003
Living with both biological parents, n(%)	125 (67.8)	17 (51.5)	108 (71.1)	χ2=4.72	1	0.03
Parent at least college graduate, n(%)	116 (63.7)	15 (45.5)	101 (67.8)	χ2=5.83	1	0.02
San Diego Site, n(%)	96 (51.9)	18 (54.6)	78 (51.3)	χ2=0.11	1	0.74
Monthly income (thousands of dollars), Median (range)	4.4 (0–21)	3.6 (0.6–15.0)	4.5 (0–21.0)	z=−1.49		0.14
Depression n(%)	71 (38.4)	19 (57.6)	52 (34.2)	χ2=6.26	1	0.01
CDRS-R, M(SD)	32.9 (12.6)	38.2 (13.3)	31.8 (12.1)	t=2.65	182	0.01
CGI-S, M(SD)	4.2 (0.8)	4.4 (0.9)	4.1 (0.8)	t=1.67	183	0.10
CGAS, M(SD)	56.1 (6.8)	54.7 (6.7)	56.5 (6.8)	t=−1.38	183	0.17
PARS 6-item score, M(SD)	14.9 (5.2)	15.4 (5.6)	14.8 (5.1)	t=0.61	183	0.54

Note: 10 of the completers are participants who had missed the week-16 follow-up.

CDRS-R = Children’s Depression Rating Scale-Revised; CGAS = Children’s Global Assessment Scale; CGI-S = Clinical Global Impressions – Severity; M = mean; n = number of cases; PARS= Pediatric Anxiety Rating Scale; SD = standard deviation

Clinical Outcomes

Clinical response.

At week 32, participants assigned to BBT had a higher treatment response rate compared to those assigned to ARC (67.5% vs. 43.1%, χ²₁=9.19, p=0.002, q=0.03; number needed to treat [NNT]=5, 95% confidence interval [CI]: 2.5, 11). In the BBT group, 95.6% (43/45) of those who responded at week 16 maintained response at 32 weeks, and 32.4% (11/34) of those who were non-responders at 16 weeks became responders at 32 weeks (χ²₁=6.23, p=0.01, q=0.06). All participants (n=17) in the ARC group who responded at week 16 maintained response at week 32, and 26.1% (n=12/46) of non-responders became responders.

Rates of remission.

The rates of remission were not significantly higher in BBT than ARC (36.3% vs. 22.2% [χ²₁=3.58, p=0.06, q=0.32; NNT=8, 95% CI: 3.5, 441.5]). There was no difference by treatment in the proportion of the week-16 responders who became remitters (BBT: 64.4% [n=29/45] vs. ARC: 64.7% [n=11/17], χ²₁=0.0004, p=0.99).

Effect on global functioning.

The BBT group showed better functioning (CGAS) at week 32 than the ARC group (70.9±11.4 vs. 65.0±13.1, t=2.95, df=148, p=0.004, q=0.04, Cohen’s d=0.49).

Depression and anxiety symptoms.

In the doubly multivariate analyses, BBT showed a non-significant trend with regard to the combined outcome of interview-rated depression and anxiety compared to ARC (treatment by time interaction: F_4,133=2.61, p=0.04, q=0.23, Cohen’s f=0.28). Post-hoc, BBT showed a significant effect on anxiety symptoms (F_2,274=6.03, p=0.003, q=0.04, Cohen’s f=0.21), but not on depressive symptoms (F_2,272=0.38, p=0.64, q>0.99, Cohen’s f=0.05) (see Figure 2).

Figure 2: Pediatric Anxiety Rating Scale (PARS) and Children’s Depression Rating Scale-Revised (CDRS-R) Over Time.

Note: Solid lines represent the Brief Behavioral Therapy (BBT) group, dashed lines the Assisted Referral to Care (ARC) group. Error bars represent 95% confidence interval.

Allocation concealment.

At week 32, IEs guessed treatment assignment at a rate greater than chance (n=96, 63.6%, χ²₁=11.13, p<.001). The rate of response between those for whom the IE guessed correctly vs. those when the IE did not guess correctly was not statistically significant (60.4% vs. 47.3%, χ²₁=2.45, p=0.12, q=0.54, Cohen’s f=0.13).

Moderator analyses.

Ethnicity moderated treatment response (odds ratio [OR]=8.17, 95% CI: 0.97, 68.95, z=1.93, p=0.05), with Hispanic youths showing a robust response to BBT (76.9%) and little response to ARC (16.7%), χ²₁=9.07, p=0.003, NNT=2, 95% CI: 1.1, 3.4). The response rate to BBT did not differ between Hispanic and non-Hispanic participants (76.9% vs. 65.7%, Fisher’s exact test [FET], p=0.53), but the response to ARC was much lower in Hispanic compared to non-Hispanic youths (16.7% vs. 49.2%, χ²₁=4.28, p=0.04). Although Hispanic ethnicity did moderate functioning at week 16, it did not moderate treatment effects on functioning at week 32 (β=6.67, 95% CI: −2.41, 15.75, t=1.45, p=0.15). Ethnicity also did not moderate treatment outcome in the doubly multivariate model of CDRS-R and PARS (F_4,130=1.44, p=0.23, Cohen’s f=0.21).

The presence of clinically significant depression at baseline did not moderate overall treatment response (OR=0.44, 95% CI: 0.11, 1.80, z=−1.14, p=0.26, Hosmer and Lemeshow χ²₈=2.08, p=0.91) or functioning at week 32 (β=−2.07, 95% CI: −10.95, 6.81, t=−0.46, p=0.65). However, depression did moderate symptom outcomes in the doubly multivariate model (depression by treatment by time, F_4,131=3.88, p=0.01, Cohen’s f=0.34). In the presence of depression at baseline, there was no significant treatment by time interaction (F_4,39=0.49, p=0.75, Cohen’s f=0.22), but in its absence, there was a strong effect of BBT (F_4,88=5.99, p<0.001, Cohen’s f=0.52) driven by the superior effect of BBT on anxiety symptoms (F2,182=12.19, p<0.001, Cohens f=0.37. see Figure 3). In fact, BBT showed a substantial reduction in depression symptoms, but it was not superior to ARC in this respect.

Figure 3: Pediatric Anxiety Rating Scale (PARS) and Children’s Depression Rating Scale-Revised (CDRS-R) Over Time by Depression.

Note: Solid lines represent the Brief Behavioral Therapy (BBT) group, dashed lines the Assisted Referral to Care (ARC) group. Error bars represent 95% confidence interval

Site effects.

As was found in the week 16 outcomes, in a model that included treatment, site, ethnicity, and two-way interactions, site was significant (OR=0.40, 95% CI: 0.19, 0.84, z=−2.40, p=0.02, Cohen’s d=−0.40), but there was no site by treatment interaction (OR=2.76, 95% CI, 0.70–10.80, p=0.14). As was the case at week 16, there were no site differences in BBT response (Pittsburgh: 71.8% vs. San Diego: 63.4%, χ²₁=0.64, p=0.42). However, a site difference emerged at week 32, with a higher rate of response to ARC in Pittsburgh than in San Diego (60% vs. 27%, χ²₁=7.98, p=0.01). At baseline, participants at the San Diego site were more symptomatic with respect to anxiety (16.0 (SD=5.0) vs. 13.7 (SD=5.2), t=3.04, df=183, p=0.003) and functional impairment (54.1 (SD=6.9) vs. 58.3 (SD=6.0), t=−4.40, df=183, p<0.001) and were less symptomatic with respect to depression (30.4 (SD=11.8) vs. 35.7 (SD=12.9), t=−2.87, df=182, p=0.01). Adjusting for these variables attenuated the effects of site on ARC response (OR=0.45, 95% CI: 0.14, 1.41, z=−1.37, p=0.17).

Treatment implementation, treatment effects, and service use.

Those who were assigned to BBT began their treatment within a shorter period of time from referral than did those assigned to ARC (7.7 days (SD=7.0) vs. 29.1 days (SD=24.4), t=5.95, df=51.1, p<.001). Participants assigned to BBT received a mean of 11.2 sessions (SD=2.2), and more than 90% received at least 8 sessions, which is considered an adequate dose.³⁶ Within BBT, receiving at least 8 sessions was not related to week-32 response (OR=4.43, 95% CI: 0.42, 46.33, z=1.24, p=0.21, , q=0.88, Hosmer and Lemeshow χ²₁=0.10, p=0.75), but showed a non-significant trend towards being positively related to week-32 functioning (β=14.18, 95% CI: 3.33, 25.03, t=2.61, p=0.01, q=0.06), and was significantly related to better symptomatic outcome on the doubly multivariate model (F_4,71=5.44, p=0.001, q=0.02, Cohen’s f=0.55). Logistic regression identified lower baseline CDRS-R as the only predictor of receipt of an adequate number of BBT sessions (OR=0.88, 95% CI: 0.82, 0.94, z=−3.89, p<0.001, q<0.001, Hosmer and Lemeshow χ²₈=3.82, p=0.87). After controlling for baseline CDRS-R, receipt of an adequate number of BBT sessions was no longer related to week-32 functioning (β=10.35, 95% CI: −1.57, 22.27, t=1.73, p=0.09), but was still associated with outcome in the doubly multivariate model (F_4,70=4.71, p=0.02, Cohen’s f=0.52).

Those in the ARC condition received a mean of 4.2 calls (SD=2.6) and a mean of 6.5 (SD=5.7) sessions of specialty mental health care over the first 16 weeks. We divided the ARC group according to the number of sessions (<8 visits versus ≥8 visits). There were no baseline differences between those attending fewer than 8 visits vs. those who attended 8 or more. We found no effect of “dose” of ARC sessions on response (OR=0.37, 95% CI: 0.12, 1.15, z=−1.71, p=0.09, q=0.44) or symptoms (doubly multivariate analysis: F_4,55=0.47, p=0.76, q>0.99, Cohen’s f=0.18), but found a trend towards an inverse relation between the receipt of at least 8 sessions and week-32 functioning (β=−7.62, 95% CI: −13.53, −1.71, t=−2.58, p=0.01, q=0.06).

Service Use.

(See Table 2). The BBT and ARC group were similar except in the use of non-study outpatient services where, by design, ARC participants were referred to external outpatient care and, thus, had higher rates and frequency of use of non-study outpatient services (p’s<0.001). However, when we include the number of BBT sessions received in the calculation of outpatient services, BBT had significantly higher number of total sessions compared to ARC (12.9 (SD=4.7) vs. 9.3 (SD=8.7), z=4.36, p<0.001, Cohen’s d=0.52).

Table 2:

Service Use by Condition^a

		Total (N=185)	ARC (n=90)	BBT (n=95)	Test	P
Use of Services/medications
Prior to Study	Outpatient, n(% Yes)	119 (64.3)	58 (64.4)	61 (64.2)	χ21=0.001	0.97
	School, n(% Yes)	71 (38.4)	34 (37.8)	37 (38.9)	χ21=0.03	0.87
	Emergency Room, n(% Yes)	14 (7.6)	8 (8.9)	6 (6.3)	χ21=0.44	0.51
	Services in Justice System, n(% Yes)	2 (1.1)	1 (1.1)	1 (1.1)	FET	0.97
	Inpatient, n(% Yes)	1 (0.5)	1 (1.1)	0 (0.0)	FET	0.49
	Antidepressant, n(% Yes)	2 (1.1)	1 (1.1)	1 (1.1)	FET	>0.99
	Stimulant, n(% Yes)	20 (10.8)	7 (7.8)	13 (13.7)	χ21=1.67	0.20
Through week 16 (excl. baseline)	Outpatient, n(% Yes)	100 (57.8)	74 (91.4)	26 (28.3)	χ21=70.31	<.0 01
	School, n(%)	58 (33.5)	30 (37.0)	28 (30.4)	χ21=0.84	0.36
	Emergency Room, n(% Yes)	16 (9.3)	6 (7.4)	10 (10.9)	χ21=0.62	0.43
	Antidepressant, n(% Yes)	7 (4.1)	4 (4.9)	3 (3.3)	FET	0.71
	Stimulant, n(% Yes)	8 (4.6)	4 (4.9)	4 (4.4)	FET	>0.99
Between weeks16 and 32 (incl. week 32)	Outpatient, n(% Yes)	94 (57.0)	58 (74.4)	36 (41.4)	χ21=12.25	<0.001
	School, n(%)	48 (29.1)	25 (32.1)	23 (26.4)	χ21=0.63	0.43
	Emergency Room, n(% Yes)	21 (12.7)	13 (16.7)	8 (9.2)	χ21=2.07	0.15
	Services in Justice System, n(%Yes)	1 (0.6)	1 (1.3)	0 (0.0)	FET	0.47
	Antidepressant, n(% Yes)	12 (7.3)	7 (9.0)	5 (5.8)	FET	0.43
	Stimulant, n(% Yes)	6 (3.6)	3 (3.9)	3 (3.5)	FET	>0.99
Through week 32 (excl. baseline)	Outpatient, n(% Yes)	123 (69.1)	81 (94.2)	42 (45.7)	χ21=49.04	<.0 01
	Outpatient/BBT sessions, n(% Yes)	173 (95.6)	92 (96.8)	81 (94.2)	FET	0.48
	School, n(%)	82 (42.1)	41 (47.7)	41 (44.6)	χ21=0.17	0.68
	Emergency Room, n(% Yes)	33 (18.5)	18 (20.9)	15 (16.3)	χ21=0.63	0.43
	Services in Justice System, n(%Yes)	1 (0.6)	1 (1.2)	0 (0.0)	FET	0.47
	Antidepressant, n(% Yes)	15 (8.4)	9 (10.5)	6 (6.5)	χ21=0.90	0.34
	Stimulant, n(% Yes)	13 (7.3)	7 (8.1)	6 (6.5)	χ21=0.17	0.68
Number of encounters
Baseline	Outpatient, M(SD)	1.3 (2.5)	1.5 (2.8)	1.2 (2.2)	z=0.32	0.75
	School, M(SD)	2.0 (6.0)	1.7 (3.7)	2.3 (7.6)	z=0.17	0.86
	Emergency Room, M(SD)	0.1 (0.5)	0.2 (0.7)	0.1 (0.2)	z=0.71	0.48
	Services in Justice System, M(SD)	0.01 (0.1)	0.01 (0.1)	0.01 (0.1)	z=0.04	0.97
	Inpatient, M(SD)	0.04 (0.5)	0.1 (0.7)	0 (0.0)	-	-
Through week 16 (excl. baseline)	Outpatient, M(SD)	3.2 (4.8)	6.1 (5.3)	0.8 (2.4)	z=9.22	<0.001
	School, M(SD)	2.3 (7.8)	3.4 (10.1)	1.4 (4.8)	z=0.99	0.32
	Emergency Room, M(SD)	0.1 (0.4)	0.1 (0.4)	0.1 (0.4)	z=−0.73	0.47
	Study BBT sessions, M(SD)		-	11.2 (2.2)	-	-
Between weeks16 and 32 (incl. week 32)	Outpatient, M(SD)	2.6 (4.6)	3.9 (5.7)	1.3 (2.8)	z=4 .97	<0.001
	School, M(SD)	2.0 (8.7)	2.9 (11.9)	1.2 (4.3)	z=0.81	0.42
	Emergency Room, M(SD)	0.2 (0.7)	0.3 (0.9)	0.1 (0.3)	z=1.45	0.15
	Inpatient, M(SD)	0.2 (1.9)	0.3 (2.7)	0.1 (0.8)	z=0.09	0.93
Through week 32 (excl. baseline)	Outpatient, M(SD)	5.5 (7.6)	9.3 (8.7)	2 (3.9)	z=7.93	<0.001
	Outpatient/BBT sessions, M(SD)	11.2 (7.1)	9.3 (8.7)	12.9 (4.7)	z=−4.36	<0.001
	School, M(SD)	4.1 (12.0)	5.9 (15.4)	2.5 (7.2)	z=0.90	0.37
	Emergency Room, M(SD)	0.3 (0.8)	0.3 (0.9)	0.2 (0.6)	z=0.78	0.44
	Inpatient, M(SD)	0.2 (1.9)	0.3 (2.6)	0.1 (0.7)	z=0.06	0.96

Note: BBT = Brief Behavioral Therapy; Excl. = excluding; FET = Fisher’s exact test; Incl. = including; M = mean; n = number of cases.

^aAll services are part of usual care except for BBT.

Controlling for number of sessions attended throughout the 32 weeks and for site, BBT continued to be superior to ARC regardless of whether the outcome was response (OR=3.22, 95% CI: 1.59, 6.52, z=3.26, p=0.001), functioning at week 32 (β=7.28, 95% CI: 3.50, 11.06, t=3.81, p<0.001) or symptoms in the doubly multivariate model (F4,132=2.62, p=0.04, Cohen’s f=0.28).

DISCUSSION

In this study, we found that, 5 months after the end of treatment, the effects of BBT continued to be superior to ARC with respect to response, functioning, and symptomatology. Ethnicity moderated response, with Hispanic youth showing a more favorable response to BBT relative to ARC, driven by a poorer response to ARC than non-Hispanic youth. In some analyses, but not others, it appeared that BBT had a more robust effect on anxiety than on depressive symptoms, and, in those with clinically significant depression, BBT was less effective on overall symptom reduction. However, there was also a dose effect, insofar as those who received at least 8 sessions of BBT had a better outcome than did those who received less than 8 sessions, and those with depressive symptoms were less likely to receive the full dose of BBT. The clinical and research implications of these findings are discussed, putting them in the context of the strengths and limitations of this study.

One of the greatest strengths of this study is that it tested BBT in the locations and under the circumstances under which it was intended to be deployed, namely, for delivery in pediatric primary care. The study was multi-ethnic and multi-site, and the treatment effects were robust both with respect to ethnicity and site. Another strength of the study was the nature of BBT itself, which was trans-diagnostic, targeting both anxiety and depression. Moreover, BBT utilized a streamlined set of therapeutic strategies that focused on active behavioral ingredients of treatment and eschewed cognitive restructuring. One important limitation of the study was that there was an undeniable asymmetry to the two interventions: BBT was cost-free and offered on-site in pediatric practices; ARC was off-site and required a co-pay, depending on the youth’s insurance. Those who delivered BBT received a fair degree of training and 1 hour of supervision per therapist per week, which may exceed that which occurs in routine clinical care. Those assigned to BBT received treatment more quickly and received more treatment than those who were assigned to ARC. However, BBT’s superior performance persisted even after controlling for the number of sessions attended. BBT’s efficacy in this study cannot be generalized to those Hispanic youth who are not fluent in English, although it may be generalized to the majority of Hispanic individuals in San Diego, 79% of whom were estimated to be fluent in English at the time this study began recruitment.³⁷ A final limitation is that the follow-up after treatment is only 5 months, which is shorter than the follow-up duration of some other studies, a trade-off we made in order to increase time for recruitment. There was no evidence of a treatment by site interaction, and the response rates for BBT were similar across sites. However, while the response rates to ARC were similar across sites at 16 weeks, the response rates to ARC were higher in Pittsburgh by week 32. These findings were not attributable to differential attrition but might be explained by greater baseline symptomatology and impairment for participants in San Diego. After adjusting for these differences, the site difference in response rate to ARC was attenuated.

Comparison to other studies.

BBT continued to show superiority to ARC 5 months after treatment, with 67.5% showing a clinically significant response, and nearly 1/3 of non-responders at 16 weeks converting to responders at 32 weeks, against only 2/62 responders who became non-responders. The enduring effect of BBT is comparable to the results of Coping Cat cognitive behavior therapy (CBT) and sertraline monotherapy in the Child Anxiety Multimodal Study (CAMS), with response rates of 69.4% and 71.5% at weeks 24 and 36, respectively, despite a higher dose of psychotherapy in the CAMS study.³¹ Similarly, the rate of remission of 36.3% in BBT compares favorably with a similar outcome in the CAMS study (rates of remission for sertraline monotherapy, 34%, for CBT, 20%).³⁸

Moderation by ethnicity.

As with the primary outcomes of our initial report of outcomes 16 weeks after intake,²¹ BBT showed a particularly strong rate of treatment response relative to ARC in Hispanic youth at 32 weeks after baseline. While BBT worked comparably well in both Hispanic and non-Hispanic youth, ARC had a much smaller effect for Hispanic youth. Caution is required in interpreting these findings, since nearly all the Hispanic youth were at one of the sites, making it difficult to definitively disentangle the effects of ethnicity from that of site. However, the effects of ethnicity did persist even controlling for site. While we cannot pinpoint what it was about ARC that was less effective for Hispanic youth, delivering BBT on site in pediatric settings appeared to have the potential to attenuate ethnic disparities in attendance and outcome. These results were not explained by higher attrition rates by Hispanic participants, since the rates of attrition in Hispanic youth were similar across interventions. These findings are consistent with other studies that have found that the provision of integrated behavioral health care on site in primary care can reduce mental health disparities in access and outcomes found in Hispanic patients.¹⁸

Moderation by and impact on depression.

While there was no moderating effect of baseline depression on the outcomes of response or functioning, baseline depression did moderate the impact of the intervention using the doubly multivariate model. In the presence of baseline depression, the gap between BBT and ARC narrowed, with youth in both treatment conditions showing sizeable reduction in symptoms over time without differential benefit associated with participation in BBT. This effect is somewhat difficult to interpret, given that BBT maintained superiority to ARC for depressed youths on more global outcome measures. This may be due in part to the predominance of anxiety symptoms in this sample. Most depressed youths in this trial also met criteria for a comorbid anxiety disorder and only six youths were depressed without anxiety. Furthermore, depression at baseline was associated with both attrition from the trial and receipt of fewer BBT sessions. We cannot determine whether depressed individuals dropped out because they were not benefitting or whether depression led individuals to drop out before they could benefit, because we do not have additional time points of symptom assessments prior to week 8. It may be that treatment of comorbid depression and anxiety requires additional efforts at engagement and higher doses of psychotherapy than might be required to treat patients with just one condition.

These findings also fit with the literature at large suggesting that depression may be more difficult to treat than anxiety. Meta-analyses and systematic reviews consistently have found larger effects for the treatment of anxiety than for treatment of depression.³⁹ Also, some, but not all of the recently developed transdiagnostic interventions for depression and anxiety show greater effects for anxiety.^40–42 Moreover, a network analysis found that the efficacy of problem-solving therapy and behavior therapy, two components of our intervention for depression, were less efficacious relative to treatment as usual (effect size [ES]= −0.31), compared to interpersonal therapy or cognitive behavior therapy (ES=−0.60 and 0.51, respectively).⁴³ In any case, it may be that pediatric patients with comorbid depression and anxiety may require a stepped care approach to intervention that tests the benefits of increased early engagement efforts, additional sessions of the same intervention, or augmentation of psychosocial intervention with selective serotonin reuptake inhibitors.⁴⁴

Summary.

In summary, BBT is a promising intervention that can be delivered in pediatric primary care to anxious and depressed youths and demonstrates significant impact on overall severity and functioning. BBT shows particular efficacy reduction of anxiety symptoms. Further work is indicated to attempt to augment the efficacy of BBT for those with comorbid depression, which may involve improved treatment engagement, since these youths attended fewer sessions. The BBT intervention was particularly efficacious in Hispanic youth, although whether this was attributable to the nature of BBT or the fact that it was delivered in primary care, we cannot determine from this study.¹⁸ Nevertheless, given the concerning level of mental health disparities in access for Hispanic youth,³ our findings about the acceptability and effectiveness of BBT for Hispanic youth and families delivered in primary care are promising. Further work is indicated to learn if primary care-based interventions like BBT, when widely implemented can contribute to the overall improvement of youth mental health and the imperative elimination of extant ethnic mental health care disparities for American youth.

Clinical vignette.

Luis is a 11-year-old Hispanic boy who presented at his pediatric well-child visit with significant shyness and social withdrawal, worries about his health and future happiness, low self-esteem, and recurrent issues with stomachaches and headaches that frequently resulted in missing school. The pediatrician had made referrals for Luis’s anxiety in the past, but the travel time required to go to the therapist and concerns about labeling Luis with a mental health problem had interfered with Luis’s parents’ ability and willingness to take him to treatment. The pediatrician referred Luis to this study, resulting in Luis being diagnosed with both generalized and social anxiety disorder and randomization to receive Brief Behavioral Therapy (BBT) on-site in pediatrics. In BBT, the patient and family were educated about the symptoms and manifestations of anxiety and internalizing symptoms, and the behavioral model of treatment. Luis was taught relaxation techniques and problem-solving skills to cope with stress and practiced these in session. Then, the therapist developed a graded approach hierarchy in collaboration with Luis and practiced approaching situations that provoked anxiety, negative affect, and somatic symptoms. The final steps in the plan involved in-vivo trips to a nearby mall. For example, during one visit to the mall, Luis asked a stranger for directions to a particular store. He successfully completed his graduation task, which was to go into a pharmacy, and ask the pharmacist where he could find over-the-counter medication for diarrhea (triggering both social and health related worries). Luis attended 11 sessions and by the end of treatment had achieved a significant reduction in his anxiety symptoms and improvement in his everyday functioning, including improved school attendance.