N EW R E S E A R C H

A Longitudinal Follow-up Study Examining Adolescent

Depressive Symptoms as a Function of Prior
Anxiety Treatment
Jennifer S. Silk, PhD, Rebecca B. Price, PhD, Dana Rosen, MS, Neal D. Ryan, MD,
Erika E. Forbes, PhD, Greg J. Siegle, PhD, Ronald E. Dahl, MD, Dana L. McMakin, PhD,
Philip C. Kendall, PhD, Cecile D. Ladouceur, PhD

Objective: Children who are fearful and anxious are at heightened risk for developing depression in adolescence. Treating anxiety disorders in pre-/
early adolescence may be one mechanism through which depressive symptoms later in adolescence can be prevented. We hypothesized that anxious
youth who responded positively to cognitive-behavioral therapy (CBT) for anxiety would show reduced onset of depressive symptoms 2 years later
compared to treatment nonresponders, and that this effect would be specific to youth treated with CBT compared to an active supportive comparison
treatment.
Method: Participants were 80 adolescents ages 11 to 17 years who had previously completed a randomized trial comparing predictors of treatment
response to CBT and child-centered therapy (CCT). Youth met DSM-IV criteria for generalized, separation, and/or social anxiety disorder at the time of
treatment. The present study was a prospective naturalistic 2-year follow-up examining trajectories toward depression, in which participants were
reassessed for depressive symptoms 2 years after anxiety treatment. Treatment response was defined as a 35% reduction in independent evaluator
rated
anxiety severity on the Pediatric Anxiety Rating Scale after treatment.
Results: As hypothesized, lower levels of depressive symptoms were observed in anxious youth who responded to CBT for anxiety (b ¼
0.807, p ¼
.004) but not CCT (b ¼ 0.254, p ¼ .505). Sensitivity analyses showed that the effects were driven by girls.
Conclusion: Findings suggest that CBT for anxiety is a promising approach to preventing adolescent depressive symptomatology, especially among
girls. The results highlight the need for better early screening for anxiety and better dissemination of CBT programs targeting anxiety in youth.
Key words: cognitive-behavioral therapy, anxiety, depression, prevention
J Am Acad Child Adolesc Psychiatry 2019;58(3):359–367.

epression is a leading cause of worldwide symptoms. The transitional period between preadolescence
D disability1,2 that increases markedly during
adolescence, with one in seven adolescents
experiencing an episode of depression prior to adulthood.3,4
and early adolescence, beginning with the onset of pubertal
changes around age 9 or 10 years,11 involves dramatic
multi-systemic changes that influence social and affective
Depression that onsets earlier in life is associated with functioning and learning.12 These include rapid physical
problems in physical health; emotional, social, and occu- growth, changes in endocrine function and neural develop-
pational functioning into adulthood; and a high likelihood ment, renegotiation of family and peer relationships, igniting
of recurrence.5-8 Yet, up to half of youth who meet criteria romantic interests, advanced hypothetical thinking skills, and
for depression do not receive treatment, and even among shifts in sleep and circadian regulation.13-15 These changes are
treated youth, many do not achieve full remission.9 For believed to converge in ways that amplify risk for depressive
these reasons, there is a critical need to identify early pre- symptoms during this developmental period.16,17 Because the
ventive and treatment approaches.10 preadolescent and early adolescent periods are marked by
neural plasticity, this may also be an opportune time to
Adolescence as a Period of Risk for Depression modify risk factors in ways that could potentially have a
The transition into adolescence may represent a sensitive positive impact on the life course trajectory. Providing effec-
period for the development and prevention of depressive tive interventions for vulnerable youth in pre- and early

Journal of the American Academy of Child & Adolescent Psychiatry www.jaacap.org 359
Volume 58 / Number 3 / March 2019
SILK et al.

adolescence could curtail a lifetime of disability.14 Research is treatment, compared to those who had not been responsive
needed to identify and test modifiable risk factors for to anxiety treatment, approximately 16 years prior.30
depression that can be targeted during this developmental However, Wolk et al.31 reported that despite a lack of
period. within-sample group differences, treated youth in this
sample experienced lower rates of depressive disorder di-
Anxiety as a Risk Factor for Depression agnoses over time than would have been expected based on
Rice et al.18 suggested that heightened fear and anxiety population prevalence. No prospective studies that we are
represent one pathway toward depression that could be aware of have examined the possible preventative role of
targeted in primary preventive interventions for children. CBT for anxiety on depressive symptoms during adoles-
Having elevated levels of fear and/or anxiety in childhood or cence, or compared CBT versus other therapies for pre-
early adolescence dramatically increases risk for developing vention of future depressive symptoms.
depression later in adolescence.18-20 This is especially true
for youth with social, separation, and generalized anxiety.21 Current Study
Although not all children with anxiety disorders go on to The present study examined depressive symptoms in adoles-
develop depression, about 75% of youth who do develop cents 2 years after receiving CBT or a comparison treatment
depression have a history of at least one anxiety disorder.22 for an anxiety disorder. In the original clinical trial (Silk
Genetic studies demonstrate that depressive symptoms et al.37), the majority of youth in both CBT and child-
during adolescence are correlated with earlier symptoms of centered therapy (CCT) were classified as treatment re-
anxiety, highlighting at least a partially heritable basis for sponders (CBT: 71%; CCT: 56%), but CBT showed supe-
this pathway.23 Psychological treatments for anxiety are riority in 1-year follow-up treatment response rates, as well as
efficacious for many youth, with cognitive-behavioral ther- full recovery rates (ie, no longer meeting diagnostic criteria) at
apy (CBT) treatment response rates of about 56%.24 both posttreatment and 1-year follow-up. In the present
Notably, reported treatment response rates are higher for study, we focus on depressive symptoms, given evidence that
anxiety disorders than for depressive disorders in childhood depression is a dimensional construct32 and that youth with
and adolescence, regardless of whether treatment was subclinical depressive symptoms show impairment and greater
delivered via CBT, medication, or combined CBT plus likelihood of subsequent diagnosis.33,34 This approach also
medication.24-28 Thus, treating anxiety prior to the onset of afforded more power to test hypotheses, as we did not expect
depressive symptoms may be more efficacious and cost- a large number of case-level diagnoses in our relatively small
effective than waiting to intervene until the emergence of sample. We hypothesized that anxious youth who were clas-
depressive symptoms. sified as treatment responders after anxiety treatment would
Silk et al.16 suggested that one mechanism through show lower depressive symptomatology at a 2-year follow-up
which elevated anxiety may contribute to the onset of compared to youth who were nonresponders, regardless of
depressive symptoms may be through reducing reward- initial depressive symptoms. We also hypothesized that the
seeking behaviors, which in turn reduces the experience of benefit of treatment on future depressive symptoms would be
pleasant events. Anxious youth often avoid potentially risky stronger for youth who were responders to CBT compared to
but also rewarding activities, such as going to a party, supportive therapy, given that CBT focuses on reducing
playing a sport, or making a new friend. This process could avoidance and increasing approach behaviors, whereas CCT
contribute to the social withdrawal and anhedonia that does not aim to directly change behavior. To determine
characterize—and often precede—depression.29 The goal of whether any effects observed could be accounted for by acute
CBT is to reduce avoidance by teaching youth skills to cope effects on depressive symptoms, we also explored whether the
with the feared scenarios and then directly exposing them to link between treatment response and 2-year depressive
these scenarios in a gradual and supportive fashion that symptoms was mediated by posttreatment depressive symp-
reduces the intensity of anxiety and promotes approach toms. Finally, we explored whether the link between treat-
behavior. Over time, successfully treated anxious youth ment response and 2-year depressive symptoms was
learn to approach feared situations even if they carry some moderated by sex, pubertal status, or age at the time of
risk, and to reap the resulting rewards. Our goal, therefore, treatment.
was to examine whether successful treatment of anxiety in
early adolescence could reduce risk for experiencing prob- METHOD
lems with depressive symptoms later in adolescence. One Participants
previous retrospective study did not find lower depressive Participants were 80 youth who met DSM-IV35 criteria for a
symptoms in adults who had been responsive to anxiety diagnosis of generalized anxiety disorder (75%), separation
360 www.jaacap.org Journal of the American Academy of Child & Adolescent Psychiatry
Volume 58 / Number 3 / March 2019
 CBT FOR ANXIETY AND SUBSEQUENT DEPRESSION

anxiety disorder (20%), and/or social anxiety disorder required current ongoing treatment with psychoactive medi-
(24%), who previously received CBT (n ¼ 54) or CCT cations including anxiolytics and antidepressants, were acutely
(n ¼ 26) as part of a randomized clinical trial37 and were suicidal or at risk for harm to self or others, failed to meet
subsequently enrolled in a follow-up study (Figure 1). magnetic resonance imaging (MRI) safety requirements, or
Participants were 9 to 14 years old (mean ¼ 11.21, SD ¼ had previously completed a course of CBT. Diagnostic
1.47) at the pretreatment assessment and 11 to 17 years of exclusionary criteria included current primary diagnosis of
age (mean ¼ 13.56, SD ¼ 1.52) at the 2-year follow-up. major depressive disorder, current diagnosis of obsessive-
Table 1 lists participant characteristics. Participants were compulsive disorder, posttraumatic stress disorder, conduct
excluded if they demonstrated an IQ below 70 as assessed by disorder, substance abuse or dependence, attention deficit/
the Wechsler Abbreviated Scale of Intelligence (WASI),36 hyperactivity disorder combined type or predominantly

FIGURE 1 CONSORT Diagram

Anxious Sample Randomized

Allocation

Allocated to CCT (n=43) Allocated to CBT (n=90)

Post Treatment Follow-up

Completed post-treatment Completed post-treatment

follow-up (n=39) follow-up (n=85)
 Developed medical condition  Developed medical condition
(n=1) (n=1)
 Referred out for severity of illness  Dropped out of study (n=4)
(n=1)
 Dropped out of study (n=2)

Enrolled in CATS-D (n=30) Enrolled in CATS-D (n=65)

 Dropped out of CATS (n=3)  Dropped out of CATS (n=9)
 Chose not to enroll in CATS-D  Chose not to enroll in CATS-D
(n=6) (n=10)
 Enrolled in CATS-D but did not
2-Year Follow-up

complete 2-year follow-up visit

(n=1)

Analyzed (n=26) Analyzed (n=54)

Excluded from analysis because of Excluded from analysis because of
missing data (n=4) missing data (n=11)

Note: CATS-D ¼ Child Anxiety Treatment Study
Depression Follow-up; CBT ¼ cognitive-behavioral therapy; CCT ¼ child-centered therapy.

Journal of the American Academy of Child & Adolescent Psychiatry www.jaacap.org 361
Volume 58 / Number 3 / March 2019
SILK et al.

TABLE 1 Descriptive Characteristics of Participants

Treatment Responders Treatment Nonresponders

(n ¼ 56) (n ¼ 24)
Age, y 10.99 (1.6) 11.30 (1.3)
Female participants, n (%) 25 (44.6%) 17 (70.8%)
Caucasian, n (%) 50 (89.3%) 21 (87.5%)
Baseline current diagnosisa,b, n (%)
Separation anxiety disorder 10 (17.8%) 5 (20.8%)
Social anxiety disorder 12 (21.4%) 6 (25.0%)
Generalized anxiety disorder 42 (75.0%) 18 (75.0%)
Specific phobia 10 (17.8%) 3 (12.5%)
Major depressive disorder 1 (1.8%) 0 (0.0%)
Baseline depressive symptomsc 18.1 (11.4) 17.9 (9.6)
Anxiety diagnosis at 2-year follow-upa,d, n (%)
Separation anxiety disorder 1 (1.8%) 0 (0.0%)
Social anxiety disorder 6 (10.7%) 4 (16.7%)
Generalized anxiety disorder 10 (17.9%) 5 (20.8%)
Specific phobia 3 (5.4%) 3 (12.5%)
Depressive disorder over 2-year follow-up, n (%)
Major depressive disorder 2 (3.5%) 1 (4.2%)
Depressive disorder not otherwise specified 2 (3.5%) 1 (4.2%)
2-Year depressive symptomsc 7.7 (8.1) 12.5 (11.6)

Note: Data are presented as mean (SD).

a
Diagnostic groups are partially overlapping because of inclusion of comorbid patients, meaning that percentages for the 3 diagnostic inclusion
groups will not sum to 100.
b
One participant in the treatment responder group and one participant in treatment nonresponder group met criteria for attention-deficit/
hyperactivity disorder (ADHD).
c
Depressive symptoms assessed using child-report on the Mood and Feelings Questionnaire.
d
Among treatment responders at 2-year follow-up, 4 participants received a current diagnosis of obsessive-compulsive disorder, Tourette’s disorder,
or posttraumatic stress disorder, and 2 participants received a diagnosis of ADHD. One participant in the treatment nonresponder group met criteria
for a current ADHD diagnosis.

hyperactive-impulsive type, or lifetime diagnosis of autism or nonspecific therapeutic ingredients such as active listening,
Asperger syndrome, bipolar disorder, psychotic depression, reflection, accurate empathy, and encouragement to talk
schizophrenia, or schizoaffective disorder. Participants were about feelings, but does not include directive problem solving,
originally recruited for the treatment study through com- psychoeducation about anxiety or coping skills, or exposure.
munity advertisements (84%) or referrals from pediatricians, Previous research on this sample has shown that both treat-
school counselors, or mental health professionals (16%) from ments resulted in a positive treatment response for a large
a metropolitan midwestern American city. number of youth, with acute treatment response rates at 69%
for CBT and 60% for CCT.37
Procedure Of the 133 youth who were originally randomized to
Full procedures for the randomized controlled trial are treatment (Figure 1), 124 completed a posttreatment
described in Silk et al.37 (see also Supplement 1, available assessment and 96 completed a 1-year follow-up. Of these,
online). Briefly, youth were randomized to 16 sessions of 95 were successfully recruited into the Child Anxiety
psychotherapy delivered in a University of Pittsburgh clinic, Treatment Study
Depression Follow-up (CATS-D) and
with a 2:1 ratio for assignment to CBT or CCT. The CBT were invited to return to the laboratory for assessments
was delivered using the Coping Cat manual38 and focused (conducted by interviewers blinded to treatment group)
on building coping and problem-solving skills during the approximately 2 years after their posttreatment assessment.
first half of treatment, followed by graduated exposure in The goal of the CATS-D Study was to examine the impact
the second half of treatment. CCT is a manualized nondi- of successful anxiety treatment on subsequent symptoms of
rective, supportive therapy based on humanistic principles depression and related neurobehavioral functioning in ado-
that was used as an active comparison treatment for the lescents. Complete data were available for a final sample of 80
original study.39,40 CCT places an emphasis on core participants (Figure 1, CONSORT Diagram). There were no
362 www.jaacap.org Journal of the American Academy of Child & Adolescent Psychiatry
Volume 58 / Number 3 / March 2019
 CBT FOR ANXIETY AND SUBSEQUENT DEPRESSION

baseline differences in age, sex, anxiety severity, depressive as the outcome. In Step 1, covariates were entered. In Step
severity, or treatment response between the subset of partic- 2, treatment response status (responder/nonresponder)
ipants who comprised the final sample (n ¼ 80) and those immediately after anxiety treatment was entered. In Step 3,
who participated in the larger trial that were not included in treatment type (CBT or CCT) and the interaction between
the final sample (n ¼ 53) (all p values >.2). Data on treatment response and treatment type were entered.
enrollment in follow-up intervention services in the com- Exploratory analyses included the same covariates at Step 1,
munity and enrollment in a behavioral sleep intervention and examined sex and development (chronological age and
offered through our clinic were also collected using the pubertal status at time of treatment) and their interaction with
Supplemental Services module from the Anxiety Disorders treatment response as predictors of 2-year MFQ-C, both in
Interview Schedule for Children.41 the full sample and in the subsample who received CBT.
Diagnoses were determined at pretreatment, posttreat- Analyses were not conducted within the CCT-only subsam-
ment, and follow-up by an independent evaluator who was ple because of the small sample size. For interpretability,
blinded to treatment assignment using the Schedule for continuous variables were standardized across the full dataset,
Affective Disorders and Schizophrenia for School-Age and dichotomous variables were coded as 0 and 1. We report
Children
Present and Lifetime Version (K-SADS-PL).42 standardized betas (with associated 95% CI), which denote
Treatment response was defined as a 35% reduction in the magnitude of the group difference (for dichotomous
independent evaluator
rated anxiety severity on the Pedi- variables) or the impact of a 1-SD change in a predictor
atric Anxiety Rating Scale (PARS)43 from pre- to post- variable (for continuous variables), expressed in SD units of
treatment.44 The present sample included 56 treatment the dependent variable.
responders (70%) and 24 nonresponders (30%). One Finally, mediation analyses were conducted to examine
participant met criteria for a depressive disorder at the time whether posttreatment depressive symptoms could help to
of enrollment in the original study, but the depressive dis- explain the link between anxiety treatment response and 2-
order was considered secondary to anxiety and did not recur year depressive symptoms. We used a standard three-
over the 2-year follow-up. Six participants (7.5%) developed variable path model and considered M to be a mediator of
a depressive disorder during the 2 years after anxiety treat- the relationship between X and Y if: (1) X is related to M
ment (12-month prevalence, 3.75%). Depressive symptoms (path “a”); (2) M is related to Y, controlling for X (path “b”);
were assessed at all time points using youth-report on the and (3) the effect of X on Y controlling for M is significantly
long version of the Mood and Feelings Questionnaire different from the direct effect of X on Y (mediation path
(MFQ-C).45 We focused on child-reported depressive “a*b”). We tested these path coefficients for a mediation
symptoms as the primary outcome variable, given evidence model in which X ¼ treatment response (as defined above,
that youth are typically better reporters of internalizing based on degree of anxiety reductions [PARS]), Y ¼ 2-year
symptoms than are parents,46 but we also report supple- MFQ-C scores, covarying baseline (pretreatment) MFQ-C,
mentary analyses using parent-report of depressive symp- and M ¼ acute post-treatment depression (MFQ-C) scores.
toms (MFQ-P; see Supplement 2, available online). These exploratory mediation analyses were conducted on the
Pubertal status was assessed at pretreatment using the Pu- subsample of participants (n ¼ 66; n ¼ 44 in the CBT
bertal Development Scale.47 subsample) for whom data were available for all measures in
the model. Reported p values for each path were obtained via
Analytic Plan bootstrapping (10,000 bootstrapped samples) using the
Analyses were conducted on all participants for whom Mediation Toolbox48 (http://wagerlab.colorado.edu/tools)
complete data were available. Consistent with an intent-to- implemented in Matlab (version R2016a).
treat approach, we included participants who withdrew
from treatment but continued to complete assessments (n ¼ RESULTS
4). Baseline demographic and clinical variables that could The covariates added in Step 1 (gender, baseline depressive
also contribute to risk for later depression were included as symptoms, and receipt of additional psychological services
covariates, including sex, baseline depressive symptoms, and during the 2-year follow-up enrollment) collectively pre-
receipt of additional psychological services during the 2-year dicted 2-year depressive symptom scores (F4,75 ¼ 5.01, p ¼
follow-up enrollment. Follow-up services were classified as .001, R2 ¼ 0.211, adjusted R2 ¼ 0.169). Specifically,
enrollment in community treatment, which included psy- gender (girls>boys: b ¼ 0.463, 95% CI ¼ 0.038
0.888],
chotherapy and/or medication (yes/no), and enrollment in a p ¼ .033) and self-reported enrollment in community treat-
supplementary sleep intervention (yes/no). Hierarchical ment during the follow-up interval (enrolled>unenrolled:
linear regression was performed with 2-year MFQ-C scores b ¼ 0.781, 95% CI ¼ 0.312
1.250, p ¼ .001) were
Journal of the American Academy of Child & Adolescent Psychiatry www.jaacap.org 363
Volume 58 / Number 3 / March 2019
SILK et al.

statistically significant predictors of 2-year depressive symp- Exploratory Analyses: Depressive Symptoms as a
toms, whereas enrollment in a sleep intervention (b ¼ 0.147, Mediator of the Effects of Treatment Response on
95% CI ¼
0.286 to 0.580, p ¼ .501) and baseline 2-Year Depressive Symptoms in Full Sample
depressive symptoms (b ¼ 0.163, 95% CI ¼
0.044, 0.370, and CBT Subsample
p ¼ .121) did not. At Step 2, treatment response immediately In the full sample with available data for the mediation
after anxiety treatment was added. This predictor did not analysis, mediation by posttreatment MFQ-C scores was
significantly predict 2-year depressive symptoms across the supported, suggesting that depression symptoms immedi-
entire sample, although it approached significance in the ex- ately after treatment helped to explain the link between
pected direction (b ¼
0.447, 95% CI ¼
0.901 to .006, acute anxiety treatment response and 2-year depressive
p ¼ .053, DR2 ¼ 0.039, DF1,74 ¼ 3.86). In the final step, symptoms. Specifically, while covarying the effect of base-
there was no significant main effect of treatment type line MFQ-C in all analyses, the “a” path (treatment
(b ¼
0.723, 95% CI ¼
1.476 to 0.030, p ¼ .060); response predicting posttreatment MFQ-C: b ¼
5.97;
however, the interaction between treatment type and t ¼
2.24; p ¼ .04), “b” path (posttreatment MFQ-C
treatment response was a significant predictor (b ¼1.061, predicting 2-year MFQ-C, controlling for treatment
95% CI 0.138
1.985, p ¼ .025, DR2 ¼ 0.051, DF1,72 ¼ response: b ¼ 0.42; t ¼ 2.81; p ¼ .01), and “a*b” mediation
5.25; final model: F7,72 ¼ 4.43, p < .001, R2 ¼ 0.301, path (b ¼
2.61; t ¼
1.65; p ¼ .02) were each significant.
adjusted R2 ¼ 0.233). As hypothesized, lower levels of In the CBT subsample (n ¼ 44), the mediation path was no
depressive symptoms were observed in treatment re- longer significant (a*b mediation path: b ¼
2.65;
sponders relative to nonresponders when they had t ¼
10.40; p ¼ .10).
received CBT (simple effect of responder status:
b ¼
0.807, 95% CI ¼
0.1.352 to
0.263, p ¼ .004) DISCUSSION
but not CCT (b ¼ 0.254, 95% CI ¼
0.502 to 1.010, The present results indicate that anxious youth who were
p ¼ .505) (Figure 2). It should be noted that CBT successfully treated with CBT for an anxiety disorder during
nonresponders did not differ from CBT responders in preadolescence or early adolescence showed lower levels of
level of pretreatment anxiety symptoms at baseline; depressive symptoms later in adolescence compared to
therefore this finding is not a reflection of differences in anxious youth who were not successfully treated. This as-
severity of psychopathology at study entry. Supplemen- sociation between early anxiety treatment and decreased
tary analyses revealed that parent-report of depressive development of depressive symptoms was specific to CBT
symptoms 2 years later was lower for treatment re- and was not found for those who responded acutely to a
sponders relative to nonresponders regardless of treatment supportive CCT. This is the first study to demonstrate that
type (see Supplement 2, available online). a specific therapeutic approach (ie, CBT) may disrupt the
commonly observed trajectory toward increasing depressive
Exploratory Analyses: Moderators of the Effects of symptoms among adolescents with a history of anxiety.
Treatment Response on 2-Year Depressive Symptoms in These data support recent suggestions that CBT in-
Full Sample and CBT Subsample terventions selectively targeting anxious and fearful youth
After controlling covariates in Step 1 (as above), there was a could represent an efficacious and cost-effective approach to
trend toward a main effect for age (b ¼ 0.195, 95% CI ¼
preventing depressive symptoms during this sensitive period
0.008 to 0.398], p ¼ .059), with older participants reporting for depression onset.10,18,21 Although we focused only on
higher depressive symptoms at follow-up. However, there depressive symptoms, these findings are important because
were no main effects for pubertal status at the time of treat- higher levels of depressive symptoms place youth at
ment (p > .359) or interacting effects of acute response to heightened risk for subsequent clinical diagnoses of
treatment by age (p > .552) or pubertal status (p > .223) on depression later in adolescence and adulthood.34
2-year depression scores, in either the full sample or the CBT Research is needed to determine the specific mecha-
subsample. In the CBT sample, there was a gender-by- nisms through which CBT for anxiety may reduce risk for
treatment response interaction (b ¼
1.133, 95% CI ¼
later depressive symptoms. We used a CBT protocol that
2.148 to
0.119, p ¼ .029) suggesting that female combines instruction in coping skills with graduated expo-
(b ¼
1.274, 95% CI ¼
0.1.935 to
0.613, p < .001) sure to provide anxious youth with repeated success expe-
but not male (b ¼
0.141, 95% CI ¼
0.908 to 0.627, riences approaching fearful situations that were previously
p ¼ .714) CBT responders had lower levels of depressive avoided. The protocol also allows youth to test their fearful
symptoms at 2-year follow-up compared to CBT expectations and to discover that feared outcomes are not
nonresponders. catastrophes. In contrast, the supportive therapy did not
364 www.jaacap.org Journal of the American Academy of Child & Adolescent Psychiatry
Volume 58 / Number 3 / March 2019
 CBT FOR ANXIETY AND SUBSEQUENT DEPRESSION

FIGURE 2 Treatment Type by Treatment Response Interaction Demonstrating Lower 2-Year Depressive Symptoms in
Cognitive-Behavioral Therapy (CBT) Responders but Not Child-Centered Therapy (CCT) Responders

20
p<.01
18
2-Year Depressive Symptoms

16
14
p<.01
12
10
8
6
4
2
0
n=15 n=39 n=9 n=17
CBT CCT
Non-responder Responder
Note: Error bars indicate standard error.

include coping skills, exposure, or direct therapist encour- protective effect of anxiety treatment on subsequent
agement to approach feared activities. Given that depression depressive symptoms.
during adolescence is characterized by high levels of with- In terms of moderators, we did not find that treatment
drawal and low approach, it is possible that CBT attenuates delivered earlier in adolescence/puberty was more or less
risk for depressive symptoms by reducing avoidance be- advantageous for preventing future depressive symptoms,
haviors and/or increasing approach behavior. Alternatively, although longer-term follow-up in larger samples is needed
protective effects may be mediated by changes in other core before ruling out this possibility, as the study was under-
characteristics shared by youth with depression and anxiety powered to detect small effects, and the youngest partici-
and targeted by CBT, such as levels of negative affect and pants in the study were only beginning to enter the period
associated cognitive biases. of highest risk for depression. We found that female, but
Consistent with these possible mechanisms, exploratory not male, CBT responders had lower levels of depressive
analyses indicate that, within the whole sample, the asso- symptoms at 2-year follow-up compared to CBT non-
ciation between anxiety treatment response and later responders. It is not clear what mechanisms might account
depressive symptoms is at least partially accounted for by for this sex difference, as previous studies have typically not
reductions in depressive symptoms that occur during the found sex differences in treatment response for anxiety or
course of anxiety treatment, which could have downstream depression in youth.49 It may be that anxiety and depression
effects on approach and avoidance behaviors as well as other are more tightly linked in girls, perhaps because girls are
cognitive and affective processes. It is also important to note more interpersonally sensitive, and there is a strong inter-
that mediation analyses did not hold within the CBT personal component to both anxiety and depression in
subsample, which is likely a result of limited power, and adolescence.21,50 Although future research is needed to
that our predictor (treatment response) and mediator clarify mechanisms, the finding that CBT for anxiety may
(posttreatment depressive symptoms) were both measured be especially helpful in preventing depressive symptoms for
at the same point of time. Future longitudinal research in girls is important because girls are at dramatically higher risk
larger samples with multiple posttreatment assessment for experiencing depression during adolescence.51,52
points is needed to clarify the role of acute changes in One limitation of the present investigation is the
depressive symptoms, as well as resulting changes in moti- absence of an untreated sample of anxious youth, which
vational, cognitive, and affective processes, in conveying the limits our ability to infer a causal effect. However, this
Journal of the American Academy of Child & Adolescent Psychiatry www.jaacap.org 365
Volume 58 / Number 3 / March 2019
SILK et al.

limitation is attenuated by our inclusion of an active com- functioning into adulthood, as well as high rates of recur-
parison treatment to control for nonspecific aspects of rence, morbidity, medical comorbidity, and lifetime
intervention, such as support and attention. A related lim- disability. The present results highlight the need for
itation inherent in the naturalistic follow-up design is that screening for anxiety early in adolescence and for better
we did not preclude participants from seeking additional availability of CBT programs targeting anxiety during this
services during the follow-up period; thus it is possible that developmental stage. It would be wise for future research to
additional intervention services provided after the CBT examine whether briefer53 or more dissemination-ready (eg,
treatment could have contributed to reduced depressive technology-based) interventions54,55 might be effective in
symptoms. However, we found that participants who attenuating risk for depression.
sought follow-up services in the community had higher, not
lower, rates of depressive symptoms at follow-up. A longer-
term follow-up will be informative to determine whether Accepted January 9, 2019.

benefits of early anxiety treatment on depressive symptoms Drs. Silk, Price, Ryan, Forbes, Siegle, Ladouceur, and Ms. Rosen are with the
University of Pittsburgh, PA. Dr. Dahl is with the School of Public Health,
persist through adolescence. In addition, the present study University of California at Berkeley, CA. Dr. McMakin is with Florida Interna-
focused on increases in symptoms of depression, and the tional University, Miami, and the Nicklaus Children’s Hospital, Miami, FL. Dr.
Kendall is with Temple University, Philadelphia, PA.
sample had low rates of depressive diagnoses. Future follow-
This work was supported by National Institute of Mental Health grant
up is needed to determine whether the identified benefits of MH091327. Support for research participant recruitment was also provided by
early CBT for anxiety translate into lower rates of clinical the Clinical and Translational Science Institute, University of Pittsburgh (NIH/
NCRR/CTSA grant UL1 RR024153).
diagnosis. Another limitation is that, as a result of the 2:1
The authors are grateful to the participants and their families. They are also
randomization for CBT to CCT used in the original trial, grateful to Anthony Mannarino, PhD, of Allegheny General Hospital, for clinical
power was much lower to detect effects within the CCT supervision, and to the following University of Pittsburgh research staff and
trainees for their assistance in data collection, analysis, and provision of clinical
group. In particular, the cell size for CCT nonresponders services: Laura Trubnick, Jennifer Jacubcak, Marcus Min, Jessica Wilson,
was small (n ¼ 9). For this reason, findings on the unique Marcie Walker, Kara Colaizzi, Melissa Milbert, Katie Burkhouse, Rebecca
Hartjen, Christine Larson, Abigail Martin, Kristin Pracht, Karen Garelik, Sherri
benefits of CBT relative to comparison need to be inter- Karas, Grace Chung, Suzanne Meller, Rosalind Elliott, Patricia Tan, Kristy Benoit
preted with caution until they can be replicated in a larger Allen, Caroline Oppenheimer, Kyung Hwa Lee, and Lindsey Stone.

sample. Disclosure: Drs. Silk, Price, Ryan, Forbes, Siegle, Dahl, McMakin, Kendall,
Ladouceur and Ms. Rosen report no biomedical financial interests or potential
In conclusion, we found that treating anxiety with CBT conflicts of interest.
during preadolescence or early adolescence appears to confer Correspondence to Jennifer S. Silk, Department of Psychology, University of
a preventive effect on adolescent depressive symptom- Pittsburgh, 210 S. Bouquet Street, Pittsburgh, PA, 15260; e-mail: jss4@pitt.edu

atology, especially for girls. This finding is important 0890-8567/$36.00/ª2019 American Academy of Child and Adolescent
Psychiatry
because adolescent depressive symptoms are associated with
https://doi.org/10.1016/j.jaac.2018.10.012
serious disruptions in emotional, social, and occupational

REFERENCES
1. Murray C, Lopez A. The Global Burden of Disease. Boston: Harvard School of Public 10. Glowinski AL, Rosen MS. Prevention targets for child and adolescent depression. JAMA
Health, WHO and World Bank; 1996. Psychiatry. 2017;74:160-161.
2. Ferrari AJ, Charlson FJ, Norman RE, et al. Burden of depressive disorders by country, 11. Herman-Giddens ME. Recent data on pubertal milestones in United States children: the
sex, age, and year: findings from the global burden of disease study 2010. PLoS Med. secular trend toward earlier development. Int J Androl. 2006;29:241-246.
2013;10:e1001547. 12. Dahl RE, Allen NB, Wilbrecht L, Suleiman AB. Importance of investing in adolescence
3. Kessler RC. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the from a developmental science perspective. Nature. 2018;554:441.
United States: results from the National Comorbidity Study. Arch Gen Psychiatry. 1994; 13. Steinberg L, Morris A. Adolescent development. Annu Rev Psychol. 2000;52:
51:8-19. 83-110.
4. Mojtabai R, Olfson M, Han B. National trends in the prevalence and treatment of 14. Dahl RE. Adolescent brain development: a period of vulnerabilities and opportunities.
depression in adolescents and young adults. Pediatrics. 2016;e20161878. Ann N Y Acad Sci. 2004;1021:1-22.
5. Weissman MM, Wolk S, Goldstein RB, et al. Depressed adolescents grown up. JAMA. 15. Spear LP. Neurobehavioral changes in adolescence. Curr Direct Psychol Sci. 2000;9:
1999;281:1707-1713. 111-114.
6. Bardone AM, Moffitt T, Caspi A, Dickson N. Adult mental health and social outcomes 16. Silk JS, Davis S, McMakin DL, Dahl RE, Forbes EE. Why do anxious children become
of adolescent girls with depression and conduct disorder. Dev Psychopathol. 1996;8: depressed teenagers? The role of social evaluative threat and reward processing. Psychol
811-829. Med. 2012;42:2095-2107.
7. Rohde P, Lewinsohn PM, Seeley JR. Are adolescents changed by an episode of major 17. Davey CG, Yucel M, Allen NB. The emergence of depression in adolescence: develop-
depression? J Am Academy Child Adolesc Psychiatry. 1994;33:1289-1298. ment of the prefrontal cortex and the representation of reward. Neurosci Biobehav Rev.
8. Horowitz JL, Garber J. The prevention of depressive symptoms in children and ado- 2008;32:1-19.
lescents: a meta-analytic review [see Comment]. J Consult Clin Psychol. 2006;74: 18. Rice F, Sellers R, Hammerton G, et al. Antecedents of new-onset major depressive
401-415. disorder in children and adolescents at high familial risk. JAMA Psychiatry. 2017;74:
9. Kennard B, Silva S, Vitiello B, et al. Remission and residual symptoms after short-term 153-160.
treatment in the Treatment of Adolescents with Depression Study (TADS). J Am Acad 19. Brady EU, Kendall PC. Comorbidity of anxiety and depression in children and ado-
Child Adolesc Psychiatry. 2006;45:1404-1411. lescents. Psychol Bull. 1992;111:244-255.

366 www.jaacap.org Journal of the American Academy of Child & Adolescent Psychiatry
Volume 58 / Number 3 / March 2019
 CBT FOR ANXIETY AND SUBSEQUENT DEPRESSION

20. Pine DS, Cohen P, Gurley D, Brook J, Ma Y. The risk for early-adulthood anxiety and 38. Kendall PC, Hedtke K. Cognitive-Behavioral Therapy gor Anxious Children: Therapist
depressive disorders in adolescents with anxiety and depressive disorders. Arch Gen Manual. Ardmore, PA: Workbook Publishing; 2006.
Psychiatry. 1998;55:56-64. 39. Cohen JA, Deblinger E, Mannarino AP, Steer RA. A multisite, randomized controlled
21. Cummings CM, Caporino NE, Kendall PC. Comorbidity of anxiety and depression in trial for children with sexual abuse-related PTSD symptoms. J Am Acad Child Adolesc
children and adolescents: 20 years after. Psychol Bull. 2014;140:816. Psychiatry. 2004;43:393-402.
22. Kessler RC, Avenevoli S, Merikangas KR. Mood disorders in children and adolescents: an 40. Cohen JA, Mannarino AP, Knudsen K. Treating sexually abused children: 1 year follow-
epidemiologic perspective. Biol Psychiatry. 2001;49:1002-1014. up of a randomized controlled trial. Child Abuse Neglect. 2005;29:135-145.
23. Silberg J, Rutter M, Neale M, Eaves L. Genetic moderation of environmental risk for 41. Silverman WK, Albano AM. The Anxiety Disorders Interview Schedule for Children
depression and anxiety in adolescent girls. Br J Psychiatry. 2001;179:116-121. (DSM-IV). San Antonio, TX: Psychological Corporation; 1996.
24. James A, Soler A, Weatherall R. Cognitive behavioural therapy for anxiety disorders in 42. Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P. Schedule for Affective
children and adolescents. Cochrane Database Syst Rev. 2005;18:CD004690. Disorders and Schizophrenia for School-Age Children
Present and Lifetime Version (K-
25. Treatment for Adolescents With Depression Study Team. Fluoxetine, cognitive- SADSPL): initial reliability and validity data. J Am Acad Child Adolesc Psychiatry. 1997;
behavioral therapy, and their combination for adolescents with depression: Treatment 36:980-987.
for Adolescents with Depression Study (TADS) randomized controlled trial. JAMA. 43. RUPP Group. The Pediatric Anxiety Rating Scale (PARS): development and psycho-
2004;292:807-820. metric properties. J Am Acad Child Adolesc Psychiatry. 2002;41:1061-1069.
26. Walkup JT, Albano AM, Piacentini J, et al. Cognitive behavioral therapy, sertraline, or a 44. Caporino NE, Brodman DM, Kendall PC, et al. Defining treatment response and
combination in childhood anxiety. N Engl J Med. 2008;359:2753-2766. remission in child anxiety: signal detection analysis using the Pediatric Anxiety Rating
27. Reynolds S, Wilson C, Austin J, Hooper L. Effects of psychotherapy for anxiety in Scale. J Am Acad Child Adolesc Psychiatry. 2013;52:57-67.
children and adolescents: a meta-analytic review. Clin Psychol Rev. 2012;32: 45. Angold A, Costello EJ, Messer SC, Pickles A. Development of a short questionnaire for
251-262. use in epidemiological studies of depression in children and adolescents. Int J Methods
28. Weisz JR, McCarty CA, Valeri SM. Effects of psychotherapy for depression in children Psychiatr Res. 1995;5:1-12.
and adolescents: a meta-analysis. Psychol Bull. 2006;132:132-149. 46. Wood A, Kroll L, Moore A, Harrington R. Properties of the Mood and Feelings
29. Pine DS, Cohen E, Cohen P, Brook J. Adolescent depressive symptoms as predictors of Questionnaire in adolescent psychiatric outpatients: a research note. J Child Psychol
adult depression: moodiness or mood disorder? Am J Psychiatry. 1999;156:133-135. Psychiatry Allied Discip. 1995;36:327-334.
30. Benjamin CL, Harrison JP, Settipani CA, Brodman DM, Kendall PC. Anxiety and 47. Petersen AC, Crockett L, Richards M, Boxer A. A self-report measure of pubertal status:
related outcomes in young adults 7 to 19 years after receiving treatment for child anxiety. aeliability, validity, and initial norms. J Youth Adolesc. 1988;17:117-133.
J Consult Clin Psychol. 2013;81:865-876. 48. Wager TD, Davidson ML, Hughes BL, Lindquist MA, Ochsner KN. Prefrontal-
31. Wolk CB, Carper MM, Kendall PC, Olino TM, Marcus SC, Beidas RS. Pathways to subcortical pathways mediating successful emotion regulation. Neuron. 2008;59:
anxiety–depression comorbidity: a longitudinal examination of childhood anxiety dis- 1037-1050.
orders. Depress Anxiety. 2016;33:978-986. 49. Nilsen TS, Eisemann M, Kvernmo S. Predictors and moderators of outcome in child and
32. Hankin BL, Fraley RC, Lahey BB, Waldman ID. Is depression best viewed as a con- adolescent anxiety and depression: a systematic review of psychological treatment studies.
tinuum or discrete category? A taxometric analysis of childhood and adolescent depres- Eur Child Adolesc Psychiatry. 2013;22:69-87.
sion in a population-based sample. J Abnorm Psychol. 2005;114:96-110. 50. Rudolph KD. Gender differences in emotional responses to interpersonal stress during
33. Compas BE, Ey S, Grant KE. Taxonomy, assessment, and diagnosis of depression during adolescence. J Adolesc Health. 2002;30(4, Suppl):3-13.
adolescence. Psychol Bull. 1993;114:323-344. 51. Kessler RC. Epidemiology of women and depression. J Affect Disord. 2003;74:5-13.
34. Wolitzky-Taylor K, Dour H, Zinbarg R, et al. Experiencing core symptoms of anxiety 52. Breslau J, Gilman SE, Stein BD, Ruder T, Gmelin T, Miller E. Sex differences in recent
and unipolar mood disorders in late adolescence predicts disorder onset in early adult- first-onset depression in an epidemiological sample of adolescents. Transl Psychiatry.
hood. Depress Anxiety. 2014;31:207-213. 2017;7:e1139.
35. American Psychological Association. Diagnostic and Statistical Manual of Mental Dis- 53. Crawley SA, Kendall PC, Benjamin CL, et al. Brief cognitive-behavioral therapy for
orders. 4th ed. Washington, DC: American Psychological Association; 1994. anxious youth: feasibility and initial outcomes. Cogn Behav Pract. 2013;20:123-133.
36. Wechsler D. Wechsler Abbreviated Scale of Intelligence (WASI). San Antonio, TX: 54. Khanna MS, Kendall PC. Computer-assisted cognitive behavioral therapy for child
Harcourt Assessment; 1999. anxiety: results of a randomized clinical trial. J Consult Clin Psychol. 2010;78:737.
37. Silk JS, Tan PZ, Ladouceur CD, et al. A randomized clinical trial comparing individual 55. Pramana G, Parmanto B, Kendall PC, Silk JS. The SmartCAT: An m-health platform for
cognitive behavioral therapy and child-centered therapy for child anxiety disorders. J Clin ecological momentary intervention in child anxiety treatment. Telemed J e-Health. 2014;
Child Adolesc Psychol. 2016;47:1-13. 20:419-427.

Journal of the American Academy of Child & Adolescent Psychiatry www.jaacap.org 367
Volume 58 / Number 3 / March 2019