RCT

NEW RESEARCH

Results From the Child/Adolescent Anxiety Multimodal Extended Long-Term Study (CAMELS): Primary Anxiety Outcomes Golda S. Ginsburg, PhD, Emily M. Becker-Haimes, PhD, Courtney Keeton, PhD, Philip C. Kendall, PhD, ABPP, Satish Iyengar, PhD, Dara Sakolsky, MD, Anne Marie Albano, PhD, Tara Peris, PhD, Scott N. Compton, PhD, John Piacentini, PhD, ABPP

Objective: To report anxiety outcomes from the multisite Child/Adolescent Anxiety Multimodal Extended Long-term Study (CAMELS). Rates of stable anxiety remission (defined rigorously as the absence of all DSM-IV TR anxiety disorders across all follow-up years) and predictors of anxiety remission across a 4-year period, beginning 4 to 12 years after randomization to 12 weeks of medication, cognitive-behavioral therapy (CBT), their combination, or pill placebo were examined. Examined predictors of remission included acute treatment response, treatment assignment, baseline child and family variables, and interim negative life events.

Method: Data were from 319 youths (age range 10.9�25.2 years; mean age 17.12 years) originally diagnosed with separation, social, and/or generalized anxiety disorders and enrolled in the multi-site Child/Adolescent Anxiety Multimodal Study (CAMS). Participants were assessed annually by independent evaluators using the age-appropriate version of the Anxiety Disorders Interview Schedule and completed questionnaires (eg, about family functioning, life events, and mental health service use).

Results: Almost 22% of youth were in stable remission, 30% were chronically ill, and 48% were relapsers. Acute treatment responders were less likely to be in the chronically ill group (odds ratio ¼ 2.73; confidence interval ¼ 1.14�6.54; p < .02); treatment type was not associated with remission status across the follow-up. Several variables (eg, male gender) predicted stable remission from anxiety disorders.

Conclusion: Findings suggest that acute positive response to anxiety treatment may reduce risk for chronic anxiety disability; identified predictors can help tailor treatments to youth at greatest risk for chronic illness.

Clinical Trial Registration Information: Child and Adolescent Anxiety Disorders (CAMS). http://clinicaltrials.gov/; NCT00052078.

Key words: anxiety, treatment, follow-up, cognitive-behavior therapy, sertraline

J Am Acad Child Adolesc Psychiatry 2018;57(7):471–480.

P

Journal of t Volume 57

ediatric anxiety disorders are highly prevalent, are associated with severe disability, and confer high personal and economic costs.1-3 These illnesses

are chronic, as data from retrospective and prospective studies show high degrees of continuity for anxiety disorders as children reach adolescence and adulthood.4-6 Conse- quently, identifying factors that interrupt the progression or persistence of these illnesses is paramount. One key factor may be early and effective treatment. Two evidenced-based treatments, cognitive-behavioral therapy (CBT) and selec- tive serotonin reuptake inhibitor (SSRI) use, have been shown to reduce anxiety and related impairment in the short term.7 However, data on whether an acute positive response to these treatments reduces risk of anxiety disability years later are limited. A recent review of studies examining the long-term benefits of treatments for pediatric anxiety disorders concluded that the majority of youths

he American Academy of Child & Adolescent Psychiatry / Number 7 / July 2018

receiving cognitive-behavioral treatment no longer meet diagnostic criteria for their initial anxiety diagnosis at an average of 5.85 years after treatment.8 However, studies in this review were cross-sectional (ie, a one-time assessment at follow-up) and consequently may over- or underestimate the long-term benefits of treatment. Follow-up studies of medication or combination treatment are lacking.

Our study, called the Child/Adolescent Anxiety Multi- modal Extended Long-term Study (CAMELS),9 was designed to address this issue by examining the long-term outcomes of anxious youth across 4 consecutive years1 who were initially randomized in the landmark Child/Adolescent Anxiety Multimodal Study (CAMS)7 to 1 of 4 treatment conditions (ie, CBT, SSRI, combination of SSRI and CBT, or pill placebo). Youth entered the CAMELS study between 4 and 12 years after their initial randomization in the CAMS. After study entry, participants were expected to be assessed

www.jaacap.org 471

GINSBURG et al.

annually over 5 consecutive years. Specifically, each year participants were asked to complete one “long” and “short” visit. During each long visit, a diagnostic evaluation was completed, along with several self- and parent-report ques- tionnaires. During the “short” visits, study staff conducted telephone interviews to complete questions about mental health symptoms and service use. Data presented in this article were obtained during long visits only that were completed over 4 consecutive years. The current study examined the following: (1) rates of stable anxiety disorder remission (as well as relapse and chronicity), and (2) pre- dictors of stable remission across the follow-up period. Examined predictors included the following: acute treatment response (ie, clinically meaningful improvement measured 12 weeks after randomization) to one of the CAMS treatments; the CAMS treatment assignment (sertraline, CBT, their combination, or pill placebo); the CAMS baseline child and family variables (ie, demographics, child clinical characteris- tics, presence of study entry diagnoses, family/parent func- tioning); and interim negative life events between the CAMS and this follow-up study. Data examining the long-term outcomes of treated youth can provide important informa- tion about downstream anxiety disability and estimates and expectations for prognosis to families. These data can also inform possible preventive efforts. Examining predictors of youth outcomes over time can help identify needed modifi- cations for subgroups of anxious youth, regardless of treat- ment type, to enhance maintenance of treatment gains and to reduce relapse. Moreover, the specific predictors examined in this study were selected in light of literature indicating their relation to acute and/or long-term anxiety treatment response.10 Based on the recent review of long-term treat- ment outcomes for anxious youth,8 we hypothesized that approximately 60% of youth would be in stable remission. With respect to predictors, we hypothesized that the CAMS treatment responders (compared to nonresponders), regard- less of treatment type, would more likely be in remission. Given that youth who received combination treatment were more likely to be acute treatment responders, we hypothe- sized that they would be more likely to be in stable remission at the follow-up.

METHOD Participants Participants were 319 volunteer families previously enrolled in the CAMS (representing 65.3% of the original sample of 488). All youth in the CAMS at baseline were between 7 and 17 years of age and met criteria for social, generalized, and/or separation anxiety disorder according to the Diag- nostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV TR). The average age range

472 www.jaacap.org

of these youth at the time of their first CAMELS follow-up assessment was 17.12 years (range ¼ 10.9�25.2 years). Average length of time since initial randomization to a CAMS treatment condition and the first CAMELS follow- up assessment was 6.51 years (SD ¼ 1.65). The length of time varied by participant based on enrollment date in the CAMS and recruitment date into CAMELS. Table 1 dis- plays the demographic and clinical characteristics of those enrolled in CAMELS (n ¼ 319) and those CAMS partici- pants who did not participate in CAMELS (n ¼ 169).

Measures Anxiety Disorders. Presence of current anxiety disorder was assessed annually at each long visit using the age-appropriate version of the Anxiety Disorders Interview Schedule for DSM- IV (ADIS).11,12 The ADIS interviews are the gold standard assessment tools for determining anxiety disorders, were used in the CAMS, and have been well validated. In the current study, interrater diagnostic agreement (ie, k values) were all greater than 0.90 for anxiety diagnoses based on a randomly selected sample (n ¼ 90; 28%) of ADIS ad- ministrations. Diagnostic agreement was defined as matching on the presence or absence of a disorder and a Clinical Severity Rating within one point. Within-person trajectories of anxiety disorders over the follow-up period were assessed by creating three groups of youths who completed three or more annual ADIS assessments: (1) re- mitters, defined rigorously as youth who did not meet diagnostic criteria for any anxiety disorder (defined as any DSM-IV TR anxiety disorder, including post-traumatic stress disorder and obsessive-compulsive disorder) at any follow-up visit; (2) chronic, defined as youth who met diagnostic criteria for one or more anxiety disorder at every follow-up visit; and (3) relapsers, youth who were anxiety diagnosis free during at least one follow-up visit and met diagnostic criteria for one or more anxiety disorders at another follow-up visit.

This definition of remission was selected over a less rigorous criterion, such as absence of all three of the CAMS entry disorders (generalized, social, and separation anxiety) or absence of a primary disorder only because numerous studies have demonstrated that pediatric anxiety disorders are highly comorbid (at both the symptom and disorder level) and manifest differently over the course of development (eg, separation anxiety disorder predicts later panic disorder). Thus, an overly narrow definition may confound conclusions about the long-term outcomes following anxiety treatment.

Predictor Variables CAMS Treatments. As noted, eligible youth were random- ized to 12 weeks of one of the following treatments (described

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

TABLE 1 Comparisons Between the Child/Adolescent Anxiety Multimodal Study (CAMS) and the Child/Adolescent Anxiety Multimodal Extended Long-term Study (CAMELS) Sample

All CAMELS Participants vs. Nonparticipants Participated in 1�2 Assessment Time Points vs. 3þ Time Points

Non participants (n ¼ 169)

CAMELS Participants (n ¼ 319) Statistic

1�2 Time Points Only (n ¼ 95)

3þ Time Points

(n ¼ 224) Statistic % % % %

Female gender 39.1 55.2 11.48**a 52.6 56.3 .35a

Ethnicity, Hispanic 19.5 8.2 13.45***a 9.5 7.6 .32a

Racial minority 26.0 18.5 3.78a 20.0 17.9 .20a

Tx condition 0.86a 1.71a

Combination 28.4 28.8 33.7 26.8 Sertraline 25.4 28.2 26.3 29.0 CBT 29.0 28.2 27.8 28.6 Placebo 17.2 14.7 12.6 15.6 CAMS Responseb 60.9 66.4 1.28a 75.9 62.8 4.60*a

Mean SD Mean SD Mean SD Mean SD

Baseline age 10.81 2.81 10.69 2.80 0.67c 10.82 2.72 10.55 2.83 0.78c

Hollingshead SES 44.45 12.07 49.70 10.99 L4.85***c 47.27 12.53 50.72 10.13 L2.59*c

Baseline CGI-S 5.05 0.76 5.01 0.71 0.59c 5.05 .74 5.00 0.70 0.66c

Baseline CGAS 51.21 6.99 50.45 7.17 1.13c 50.93 7.26 50.24 7.13 0.78c

Baseline # Dx 2.89 1.28 3.00 1.19 L0.92c 3.02 1.25 2.99 1.17 0.21c

Baseline Parent BFAM

11.11 5.75 10.95 5.12 0.30c 11.61 4.76 10.69 5.24 1.40c

Parent BSI 0.50 0.46 0.47 0.40 0.76c 0.49 0.42 0.46 0.39 0.52c

Years since Randomizationd

— — — — — 7.29 1.94 6.19 1.39 5.69***

Note: BFAM ¼ Brief Family Assessment of Functioning Measure; BSI ¼ Brief Symptom Inventory; CBT ¼ cognitive-behavioral therapy; CGAS ¼ Children’s Global Assessment of Functioning Scale; CGI-S ¼ Clinical Global Impression�Severity Scale; Dx ¼ diagnosis; SES ¼ socioeconomic status. aFrom c2 test. bData presented on the CAMS posttreatment response status represent available data only. It should be noted that 31 youth who did not participate in CAMELS and 18 youth who did participate in CAMELS were missing data for their 12-week outcome assessment and were not included in this comparison. If the last observation carried forward (LOCF) data imputation method is used for this variable, then youth in the CAMELS are more likely to have been CAMS treatment responders than youth who did not participate in the CAMELS. cFrom t test. dNot applicable for non-CAMELS participants. *p < .05; **p < .01; ***p < .001.

LONG-TERM OUTCOMES FOR ANXIOUS YOUTH

in detailed elsewhere7): (1) CBT: CBT consisted of the age- appropriate versions of the Coping Cat protocol: Coping Cat for children and C.A.T. Project for Adolescents13,14. Both protocols included 12 individual child sessions (60 minutes each) and 2 parent-only sessions scheduled over 12 weeks. (2) Sertraline (SRT): SRT treatment consisted of eight 30- to 60-minute in-person sessions involving discussing anxiety symptoms, functioning, and adverse events within supportive clinical care. Medication was administered using a “fixed-flexible” dosing strategy that was linked to clinical response and side effects. In general, participants’ medication dose was adjusted upward in 50-mg/d increments if anxiety levels were moderate to high. The dose was held stable, or

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

adjusted downward, if the participant had few anxiety symptoms or if there were impairing side effects. (3) Com- bination treatment (COMB): COMB consisted of both CBT and SRT. (4) Pill placebo (PBO): PBO consisted of a double- blinded treatment that paralleled SRT. Youth assigned to PBO who were considered “nonresponders” to treatment were offered the CAMS CBT and/or medication.

CAMS Treatment Responder Status. CAMS treatment responder was defined using the Clinical Global Impression�Improvement Scale (CGI-I)15 assigned at posttreatment (ie, 12 weeks after randomization). The CGI- I is a 7-point scale (1 ¼ very much improved, 7 ¼ very

www.jaacap.org 473

GINSBURG et al.

much worse) completed by the independent evaluator. Youth with CGI-I scores of 1 (very much improved) or 2 (much improved) were categorized as treatment responders.

Negative Life Events. Negative life events were assessed at each completed CAMELS visit using the Life Events Scale (LES),16 which is a 35-item parent-report measure of exposure to a broad range of stressful events that cut across family, school, and social domains. Examples include death of a parent, change in income, school failure, illness, and moving. A total score was calculated reflecting the aggregate number of endorsed negative life events since CAMS and over the follow-up period.

CAMS Baseline Demographic Variables. Demographic variables included child age, gender, race/ethnicity (minor- ity versus nonminority), and socioeconomic status as measured by the Hollingshead Index,17 all reported by the participating parent.

CAMS Baseline Child Clinical Variables. Child clinical variables included a) anxiety symptom severity as measured by the Clinical Global Impression�Severity Scale (CGI-S).15 The CGI-S is a well-established measure that yields a global rating of anxiety severity ranging from 1 (not at all ill) to 7 (extremely ill) and b) comorbid disorders (internalizing and externalizing) based on the age-appropriate version of the ADIS. Internalizing disorders included any non�study entry anxiety or depressive disorder. Externalizing disorders included disruptive behavior disorders (eg, oppositional defiant disorder, attention-deficit/hyperactivity disorder).

CAMS Baseline Family/Parent Variables.. Brief Family Assessment Measure (BFAM)18 The BFAM is a 14-item measure completed by parents to assess overall family func- tioning (eg, “We take the time to listen to each other”). A total score was used, with higher scores indicating more family dysfunction (Cronbach’s a at the CAMS baseline was 0.84).

Brief Symptom Inventory (BSI)19 The BSI is a widely used 53-item parent-report measure of global psychiatric symptomatology (Cronbach’s a at the CAMS baseline was 0.95). For the present analyses, the Global Severity Index (BSI-GSI) provided a single composite score of current symptoms of somatization, obsessive-compulsive disorder, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychosis. Higher scores suggest greater severity of psychological distress.

Burden Assessment Scale (BAS)20 The BAS is a 21-item parent completed measure that assesses caregiver burden associated with having a child with a mental health disorder. Parents indicated how much the child’s anxiety disrupts family life, emotions, and routines on a scale ranging from

474 www.jaacap.org

1 (not at all) to 5 (very much). A total score was used, with higher score reflecting greater burden. Cronbach’s a at the CAMS baseline sample was 0.91.

Additional Control Variables To reduce bias associated with interim mental health service use, the Supplemental Services Form of the ADIS (SSF), assessed at each CAMELS long and short visit, was entered as a control variable in all analyses. The SSF was administered by a research assistant at each follow-up visit and assessed the use of a broad range of mental health services since the CAMS. Examples of these services include psychiatric hospitalization, psychotropic medication, and various kinds of psychotherapy (eg, family, supportive, psychodynamic). A dichotomous score (ie, yes/no for any use across all SSF items) was calcu- lated over the entire follow-up period; 238 youth (74.6%) reported receiving some sort of mental health service use over the follow-up. Additional control variables included treat- ment site (collected at the CAMS baseline) and time since the CAMS randomization (in years).

Procedures This was a naturalistic follow-up study conducted at six sites in the United States (Johns Hopkins University, University of California Los Angeles, University of Pittsburgh, Temple University, Duke University, and Columbia University) and approved by the respective institutional review boards. Prior to enrollment, and after complete description of the study, parents/guardians provided written informed consent and children provided assent. Recruitment for CAMELS began in 2011 and ended in 2015. Recruitment for the original treatment study, the CAMS, occurred between 2002 and 2007. In the CAMS, eligible youth were randomized into one of four treatment conditions: CBT only, SRT (sertraline medication) only, COMB (combination CBT and SRT), or PBO (placebo). At posttreatment (12 weeks after randomi- zation), youth were evaluated by independent evaluators who determined treatment response status. Youth assigned to pill placebo who were determined to be “nonresponders” were offered any active CAMS treatments. Receipt of subsequent treatment for anxiety was documented on the SSF and statistically controlled for in the current study.

The proposed CAMELS grant was a 5-year study. However, because of variations in study start-up and end dates at each recruitment site, sample sizes varied across the 5 years and resulted in a significantly lower sample size of participants completing all five annual assessments. 319 youth completed their first long CAMELS visit. The sample size of youth who completed their subsequent long visits were: year 2 ¼ 239, year 3 ¼ 220, year 4 ¼ 209; year 5 ¼

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

LONG-TERM OUTCOMES FOR ANXIOUS YOUTH

91; only 61 youth participated in all 5 long visits. Because of the low year 5 enrollment, the current study included only data obtained across 4 years.

Potentially eligible families for the current study were contacted via letter, social media outlets, and phone. Fam- ilies expressing interest provided written informed consent and then were expected to complete a maximum of five annual in-person evaluations, depending on when the family was enrolled. During these “long” evaluations, measures described above (and several others) were collected. Between each long visit (mid-way at 6 months), a phone evaluation was planned (referred to as a “short” visit), during which time only questionnaire data were collected. During the long visits, a trained independent evaluator administered the age-appropriate diagnostic interview and other measures used in this study. Short visits were con- ducted by research assistants. Each family was reimbursed $130 for completion of each long visit and $50 for each

FIGURE 1 Child/Adolescent Anxiety Multimodal Extended Long-

488 Enrolled in

140 Assigned to receive

sertraline and CBT

133 Assigned to receive

sertraline alone

319 Enrolled in CAMELS

92 (66% of 140)

Received sertraline and CBT

90 (68% of 133)

Received sertraline

alone

90 (65% of 139)

Received CBT alone

47 (62% of 76)

Received placebo

1 CAMELS visit: n = 18

2 CAMELS visits: n = 13

3 CAMELS visits: n = 8

4 CAMELS visits: n = 33

5 CAMELS visits: n = 20

1 CAMELS visit: n = 10

2 CAMELS visits: n = 13

3 CAMELS visits: n = 11

4 CAMELS visits: n = 37

5 CAMELS visits: n = 19

1 CAMELS visit: n = 9

2 CAMELS visits: n = 15

3 CAMELS visits: n = 15

4 CAMELS visits: n = 35

5 CAMELS visits: n = 16

Note: CAMS ¼ Child/Adolescent Anxiety Multimodal Study; CBT ¼ cognitive-behavio

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

short visit. Figure 1 provides the Consolidated Standards Of Reporting Trials (CONSORT) diagram.

Data Analysis Analyses included all youth enrolled in CAMELS, regardless of their original CAMS treatment assignment. Descriptive analyses determined rates of remission across each year in this follow-up study. Acute treatment response and initial treatment assignment were examined as predictors of remission in each year using logistic regressions, controlling for treatment site, supplemental service use, and baseline anxiety severity. To determine predictors of the course of anxiety disorders within youth over time, participants with three or more long follow-up visits (n ¼ 224) were classified into one of the three groups defined earlier (ie, remitters, chronically ill, relapsers). Multinomial regression analyses examined predictors of group membership (remitter, chronic, relapse). Predictor analyses included the following

term Study (CAMELS) Participant Flow

the CAMS

139 Assigned to receive CBT

alone

76 Assigned to receive placebo

169 Did not Participate

146 Could not contact/ were not interested

23 Declined further contact during

the CAMS

1 CAMELS visit: n = 3

2 CAMELS visits: n = 6

3 CAMELS visits: n = 10

4 CAMELS visits: n = 22

5 CAMELS visits: n = 6

ral therapy.

www.jaacap.org 475

GINSBURG et al.

covariates: study site, the CAMS baseline anxiety severity (CGI-S), time since the CAMS randomization (in years), and mental health service use (using the SSF dichotomous variable) since the CAMS. These predictors were analyzed across a series of 5 models: demographic predictors (model 1), baseline clinical predictors (model 2), presence/absence of study entry diagnoses at the CAMS baseline (model 3), parent and family predictors (model 4), and life events (model 5). Missing data analysis for this subgroup of 224 participants revealed that no predictor had more than 2% missing data and was therefore handled via listwise deletion.

RESULTS CAMS and CAMELS Sample Comparisons Table 1 displays the demographic and clinical characteristics between those enrolled in CAMELS (n ¼ 319) and those not enrolled in CAMELS (n ¼ 169) as well as between those who completed three or more, versus less than three CAMELS annuallong-visitassessments.Youthwhodidnotparticipatein CAMELS did not differ significantly from CAMELS partici- pants on baseline anxiety severity, number of disorders, or percent of the CAMS treatment responders (all p values >.05). However, CAMELS participants were more likely to be fe- male, of non-Hispanic ethnicity, and to families from higher socioeconomic (SES) backgrounds than non-CAMELS par- ticipants. CAMELS youth who completed three or more annual long study visits were less likely to have been the CAMS treatment responders and were of lower SES backgrounds compared to those who completed only one to twostudy visits.

Primary Analyses: Anxiety Remission Rates Figure 2 displays the raw data on anxiety disorder remission rates for the total CAMELS sample and by

FIGURE 2 Remission Rates Across Follow-up Assessment Time Points

0

10

20

30

40

50

60

CAMELS Y1 CAMELS Y2 CAMELS Y3 CAMELS Y4

Total Sample CAMS Responder CAMS Nonresponder

* * *

Note: Child/Adolescent Anxiety Multimodal Study (CAMS) Responder status associated with increased likelihood of remission. CAMELS ¼ Child/Adolescent Anxiety Multimodal Extended Long-Term Study. Please note color figures are available online. *p < .05.

476 www.jaacap.org

treatment responder status across each of the 4 years in the follow-up study. For the total sample, remission rates ranged from 42% to 48.8% across years 1 to 4. When rates of remission were examined within person across all 4 years, results indicated that 21.7% were in stable remission, 30% were chronically ill, and 48% were relapsers (Figure 3). Rates of remission in each year by treatment group ranged from 40% to 60% in CBT, 40% to 52% in SRT, 41% to 49% in COMB, and 25% to 47% in PBO.

Predictors of Remission Results from logistic regression analyses examining pre- dictors of remission at each follow-up year indicated that the CAMS treatment responders were more likely to be in remission at each time point, with the exception of year 4 (year 1: OR ¼ 1.81, 95% CI ¼ 1.07�3.07, p ¼ .03; year 2: OR ¼ 2.23, 95% CI ¼ 1.24�4.01, p ¼ .007; year 3: OR ¼ 1.92, 95% CI ¼ 1.04�3.53, p ¼ .04; year 4: OR ¼ 1.26, 95% CI ¼ 0.69�2.31, p ¼ .45). Logistic regressions examining whether treatment assignment was a predictor of remission at any year revealed no association.

Results from multinomial regression analyses predicting group membership (remitters, relapsers, or chronically ill) indicated that the CAMS treatment responders were more likely to be in the remitter group relative to the chronic group (OR ¼ 2.73, 95% CI ¼ 1.14�6.54, p ¼ .02). The CAMS treatment assignment was not associated with remission status (all p values >.05). Several factors predicted group membership (Table 2): specifically, male gender (OR ¼ 3.85, 95% CI ¼ 1.59�9.35), younger age (OR ¼ 0.82, 95% CI ¼ 0.69�0.98]), youth without social phobia anywhere in their diagnostic profile (OR ¼ 0.36, 95%

FIGURE 3 Percent Remitters, Chronic, and Relapsers Across Follow-up Period

0

10

20

30

40

50

60

Consistent Remitters Relapsers Chronic

Total Sample CAMS Responders CAMS NonResponders

*

Note: Child/Adolescent Anxiety Multimodal Study (CAMS) Responder status asso- ciated with increased likelihood of group membership. Please note color figures are available online. *p < .05.

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

TABLE 2 Predictors of Remission Groups in the Multisite Child/Adolescent Anxiety Multimodal Extended Long-term Study (CAMELS) from Multinomial Regression Analysis

Predictor

Chronic vs. Remissiona Relapsers vs. Remissionb

OR CI p OR CI p CAMS Response 2.73 (1.14e6.54) .02 1.05 (0.47e2.33) .59 CAMS Treatment Conditionc

Combination 1.47 (0.41e5.33) .56 0.99 (0.29e3.40) .99 Sertraline only 1.74 (0.49e6.14) .39 0.95 (0.29e3.14) .93 CBT only 2.26 (0.63e8.12) .21 1.02 (0.31e3.38) .98

Demographic Predictors Age 0.82 (0.69e0.98) .03 0.89 (0.76e1.05) .17 Male gender 3.85 (1.59e9.35) .003 4.51 (2.03e10.04) <.001 Minority status 0.86 (0.26e2.90) .81 1.03 (0.34e3.12) .96 SES 1.17 (0.68e1.99) .58 1.07 (0.66e1.74) .89

Baseline Clinical Predictors Anxiety severity (CGI-S) .80 (0.37e1.72) .57 0.81 (0.41e1.60) .55 Overall functioning (CGAS) 1.09 (1.01e1.19) .03 1.03 (0.97e1.10) .33 Comorbid internalizing 0.59 (0.24e1.46) .26 0.96 (0.35e2.58) .93 Comorbid externalizing 0.74 (0.25e2.17) .59 0.77 (0.35e1.73) .53

Presence of Baseline Study Entry Diagnoses Separation anxiety 0.99 (0.44e2.26) .98 0.58 (0.28e1.19) .14 Social phobia 0.36 (0.13e0.96) .04 0.39 (0.17e0.88) .02 Generalized Anxiety 1.61 (0.54e4.76) .39 0.99 (0.38e2.66) .99

Parent and Family Predictors Parental global distress (BSI) 1.86 (0.58e5.99) .30 1.47 (0.51e4.23) .48 Family functioning (BFAM) 0.94 (0.90e0.98) .006 0.99 (0.96e1.02) .56 Caregiver strain (BAS) 0.99 (0.96e1.03) .73 0.99 (0.96e1.02) .22 Negative Life Events 0.37 (0.19e0.69) .002 0.62 (0.36e1.08) .09

Note: All analyses controlled for baseline anxiety severity, treatment site, years since Child/Adolescent Anxiety Multimodal Study (CAMS) randomi- zation, and supplemental service use over the follow-up period. BAS ¼ Burden Assessment Scale; BFAM ¼ Brief Family Assessment of Functioning Measure; BSI ¼ Brief Symptom Inventory; CBT ¼ cognitive-behavioral therapy; CGAS ¼ Children’s Global Assessment of Functioning Scale; CGI-S ¼ Clinical Global Impression�Severity Scale; OR ¼ odds ratio; SES ¼ socioeconomic status. aOdds ratios represent the odds of being in the remission group relative to the chronic group. bOdds ratios represent the odds of being in the remission group relative to the relapse group. cResults show results of active treatment compared to placebo intervention.

LONG-TERM OUTCOMES FOR ANXIOUS YOUTH

CI ¼ 0.13�0.96]), higher global functioning (OR ¼ 1.09, 95% CI ¼ 1.01�1.19), and better family functioning (OR ¼ 0.94, 95% CI ¼ 0.90�0.98; all measured pre- treatment) and fewer interim negative life events (OR ¼ 0.37, 95% CI ¼ 0.19�0.69]) were all associated with belonging to the stable remission group relative to the chronically ill group (all p values <.05). Youth without social phobia at baseline (OR ¼ 0.39, 95% CI ¼ 0.17�0.88) and male gender (OR ¼ 4.51, 95% CI ¼ 2.03�10.04) were also associated with increased likelihood of being in the remission group relative to the relapser group (p vallues <.05). Minority status, SES, baseline anxiety severity (CGI-S scores), diagnostic comorbidity, parental psychopathology, and caregiver strain at the CAMS baseline were not associated with group membership across the follow-up period.

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

Post Hoc Analyses Given the number of significant predictors found across the five models, we ran an additional multivariate model including all the significant predictors outlined above in addition to the established covariates (treatment site, sup- plemental service use, baseline anxiety severity, and years since randomization) to determine which predictors remained after controlling for the other significant predictors. Most predictors remained significant, with male gender (OR ¼ 3.01, 95% CI ¼ 1.16�7.79), absence of baseline social phobia (OR ¼ 0.32, 95% CI ¼ 0.10�0.98]), better baseline family functioning (OR ¼ 0.95, 95% CI ¼0.91�0.99), and fewer negative life events (OR ¼ 0.35, 95% CI ¼ 0.18�0.70) associated with increased likelihood of belonging to the remission group relative to the chronic group; youth age and baseline

www.jaacap.org 477

GINSBURG et al.

functioning were no longer significant. Only male gender remained a predictor of remission group membership rela- tive to the relapser group (OR ¼ 3.90, 95% CI ¼ 1.69�9.00); social phobia diagnosis at baseline was reduced to a trend (OR ¼ 0.40, 95% CI ¼ 0.16�1.03).

Finally, because of the potential confounding effects of interim mental health service use, we also examined patterns of services use across treatment condition and remission status groups, controlling for treatment site. Rates of sup- plemental service use ranged from 69% (CBT) to 83% (placebo). Logistic regression analyses showed no differences in the percentage of youth receiving interim mental health services based on original treatment condition relative to placebo (combination: OR ¼ 1.17, 95% CI ¼ 0.36�3.44; CBT: OR ¼ 0.48, 95% CI ¼ 0.17�1.36; sertraline: OR ¼ 0.73, 95% CI ¼ 0.25�2.10). Rates of supplemental service use based on remission status across the follow up were 63% (remission group), 74% (relapse group), and 90% (chronic group). Multinomial regressions indicated that youth who did not receive supplemental services were more likely to be in the remission group relative to the chronic group (OR ¼ 4.14, 95% CI ¼ 1.56�10.97) and the relapse group (OR ¼ 2.26, 95% CI ¼ 1.10�4.66).

DISCUSSION The present study provides rarely obtained multiyear pro- spective data on the long-term outcomes for a large sample of youth with anxiety disorders who were treated in the context of a randomized controlled multisite trial comparing sertra- line, cognitive-behavioral therapy, their combination, and a pill placebo. In contrast to cross-sectional follow-up studies and our initial hypothesis, findings revealed that despite receiving high-quality evidenced-based treatments for 12 weeks, less than half of the sample met the rigorous criteria for stable remission (ie, no longer met diagnostic criteria for any anxiety disorder across the follow-up years), revealing the chronic nature of pediatric anxiety disorders for a large portion of anxious youth and the need for enhanced treat- ment and relapse prevention strategies. Indeed, only 21.7% youth were consistently “anxiety free” (absence of all anxiety disorders) over the course of the follow-up, whereas the remainder of youth showed a pattern of relapse or chronic illness. Few studies have followed youth for multiple years, and these data suggest that cross-sectional snapshots may overestimate the durability of treatments’ success. Our find- ings also send a clear and disquieting message that some anxious youth are in need of longer or better treatments to reduce the burden of these illnesses over time.

Fortunately, youths who made significant clinical improvement after 12 weeks of treatment, regardless of treatment type, were more likely to be in stable remission

478 www.jaacap.org

(versus chronically ill), suggesting there are long-term ben- efits to getting good treatment early in the course of these disorders. However, the magnitude of this benefit over time was modest, indicating room for improvements in the short- term efficacy even for our best evidenced-based treatments. With respect to the relative benefits of one specific treat- ment type (ie, cognitive-behavioral therapy, sertraline, or their combination), findings revealed no association be- tween type of treatment assigned and remission status over time, suggesting that enhancing access to any effective treatment may have comparable long-term benefits. Of note, youth originally assigned to the pill placebo condition faired similarly to those assigned to an active CAMS treat- ment. One reason may be that the CAMS participants who were assigned to the pill placebo condition and were deemed nonresponders at week 12 (ie, posttreatment) were offered their choice of the CAMS treatment after the posttreatment assessment. Indeed, the vast majority (86%) of those assigned to pill placebo reported receiving an active treatment at CAMELS Visit 1,9 which likely accounts for the similarities in outcomes at the long-term follow-up.

Identifying predictors of long-term outcomes can help determine which youths are most at risk for poor outcomes, increase the use of assessment tools to identify those at higher risk for poor outcomes, lead to the development of new treatments for these subgroups of individuals, and provide helpful prognostic data for patients. Specifically, our data indicate that youth who were younger at the start of treatment were more likely to be in stable remission (compared to chronically ill youth) over the course of the follow-up study. It may be that older youth had experienced anxiety for longer periods of time prior to seeking treatment and that their anxiety may be more recalcitrant to treat- ment, emphasizing the value of intervening early. Older youth had higher anxiety severity at the CAMS baseline,21

also suggesting these youth may have had more severe and less malleable anxiety. Alternatively, it may be that parents of younger participants were more involved in treatment, provided more supervision and support, and facilitated completing treatment requirements such as taking medica- tion regularly or completing therapy assignments compared to older youth who have greater autonomy about whether or how much to engage in treatment. Males were also more likely to be in stable remission over the 4 years of the follow- up study, although they did not show lower levels of anxiety at baseline. The elevated risk for females has been reported in long-term follow-up studies of adolescent depression22

and suggests that greater screening and gender-specific in- terventions may be needed to help young women benefit more from anxiety treatment. We note, however, that there were relatively fewer males than females in the CAMELS

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

LONG-TERM OUTCOMES FOR ANXIOUS YOUTH

sample, which may have influenced our findings showing poorer overall outcomes for females.

Several pretreatment child clinical factors also predicted stable remission, such as youth with higher general func- tioning as well as participants whose parents endorsed more positive family interactions such as open communication, a sense of trust, and fair punishments, suggesting that such family factors may have a long-term impact on anxiety remission. In addition, youth with social anxiety disorder at baseline (not only as primary anxiety disorder) were less likely to be in stable remission. The relation between social anxiety disorder and poorer long-term outcomes has been reported elsewhere.23 The relation between social anxiety disorder and poorer outcomes may be a result of the developmental timing of CAMELS assessments. The ma- jority of youth in the CAMS at baseline were young (mean age, 7 years), whereas the average age in CAMELS at visit 1 was 17 years—a particularly vulnerable time socially. Thus, youth with social anxiety disorder at baseline may be more vulnerable to the social stressors, demands, and transitions associated with adolescence and emerging adulthood—the time period captured during CAMELS assessments.

Finally, youth who experienced fewer negative life events in the time period between the original treatment study and the follow-up were more likely to be in stable remission across the follow-up period. A substantial litera- ture has documented the relation between stressful life events and pediatric anxiety disorders.24,25 Future studies are needed to examine the types and timing of these events to better characterize their impact on long-term outcomes. Clinically, the assessment of such events should be con- ducted at regular intervals to address their negative effects. Assessments of negative life events seems particularly important for females, as females reported more exposure to negative life events in this sample than males and females were less likely to be in stable remission.

Results from this naturalistic follow-up may be confounded by numerous threats to internal validity (eg, maturation, history effects), and causal relations cannot be claimed. The unique characteristics of the sample restricts to whom results can be generalized. Moreover, the retention rate of original participants of the CAMS was only 65%, raising questions about the generalizability of findings and potential bias (either over- or underestimating remission rates, for instance). In addition, although there were few differences in baseline clinical variables (eg, baseline anxiety severity) be- tween the CAMS and CAMELS samples, those enrolled in CAMELS were different demographically (eg, less likely to be female and from ethnic minorities), again raising potential concerns about generalizability. Despite these limitations, CAMELS represents the largest sample of anxious youth to be

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018

followed over time in the United States, and the retention rate is similar to those reported in other multi-site trials.22

In sum, findings from this study may be used to inform clinical research and practice by providing patients with estimates of prognosis. Conclusions suggest there may be long-term benefits from early effective treatment with ser- traline, cognitive-behavioral therapy, and/or their combi- nation, for anxiety. However, for the majority of pediatric patients, anxiety disorders are chronic, and additional treatment and relapse prevention approaches appear war- ranted. This follow-up study represents a unique and rich opportunity to gain insight into the natural trajectories of anxiety disorders following acute treatment. Further in- depth analyses of these data will follow for years to come, providing additional insights into how we can improve the long-term outcomes for anxious youth.

Accepted May 3, 2018.

Dr. Ginsburg is with the University of Connecticut School of Medicine, West Hartford. Dr. Becker-Haimes is with the University of Pennsylvania, Philadel- phia. Dr. Keeton is with Johns Hopkins University, Baltimore, MD. Dr. Kendall is with Temple University, Philadelphia, PA. Drs. Iyengar and Sakolsky are with the University of Pittsburgh, PA. Dr. Albano is with Columbia University, NY. Drs. Peris and Piacentini are with the University of California, Los Angeles. Dr. Compton is with Duke University, Durham, NC.

This research was supported by the National Institute of Mental Health (NIMH) grants MH064089 to Dr. Ginsburg, MH064003 to Dr. Sakolsky, MH64088 to Dr. Piacentini, MH64092 to Dr. Albano, MH64107 to Dr. Compton, and MH063747 to Dr. Kendall. The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Dr. Becker-Haimes served as the statistical expert for this research.

The authors thank the following individuals who worked so diligently to complete this study: Columbia University Medical Center: M. Rynn, MD, A. Sarubbi, PsyD, J. Guerry, PhD, O. Jablonka, MA, K. Camacho, MA, A. Rapp, MA, C. Heleniak, BA, M. Goldfine PhD, and E. Alvarez, MA; Duke University: J. Curry, PhD, D. Hood, BA, B. Glass-Thomas, BA, C. Mauro, PhD, N. Heilbron, PhD, J. Sapyta, PhD, R, Dingfelder, PhD, J. Compton, PhD, and P. Cuper, PhD; The Johns Hopkins University School of Medicine: B. Gibby, MA, J. Walkup, MD, M. Crosby-Budinger, MA, N. Affrunti, PhD, K. Drake, PhD, T. Drazdowski, PhD, J. Christofferson, MA, S. Makhzoumi, MA, R. Teetsel, MA, E. Santana, MA, R. Boyer, MA, N. Marie Dull, MA, A. Gorden-Hollingsworth, PhD, F. Ba’th, MA, and R. Bloom, MA; Temple University: C. Settipani, PhD, C. Wei, PhD, D. Brodman, BA, J. Peterman, BA, H. Makover, PhD, and J. Edmunds, BA; UCLA Semel Institute for Neuroscience and Human Behavior: J. McCracken, MD, S. Chang, PhD, J. Podell, PhD, R. Vivrette, MD, PhD, D. Solis, BA, M. Wu, PhD, E. Ricketts, PhD, S. Bai, PhD, MPH, and N. Abrahami, LCSW; Western Psychiatric Institute and Clinic: B. Birmaher, MD, N. Curcio, MS, J. Bender, MSW, H. Kumar, MS, and K. Monk, BSN.

Disclosure: Dr. Ginsburg has received support from NIMH and from the US Department of Education/Institute of Education Sciences. Dr. Becker-Haimes has received support from NIMH. Dr. Keeton has received support from NIMH. Dr. Kendall has received support from NIMH and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD). He has received royalties from the sales of materials related to the treatment of anxiety disorders in youth (eg, Guilford Press; Workbook Publishing; Gyldendal Norsk; Gyldendal Akademisk). Dr. Iyengar has received support from NIMH. Dr. Sakolsky has received support from NIMH and NARSAD. She has served as a consultant for LEK Consulting Inc. Dr. Albano has received royalties from Oxford University Press for the Anxiety Disorders Interview Schedule, Child and Parent Versions. She has received an editor’s honorarium from the American Psychological Association. Dr. Peris has received support from NIMH, the Society for Clinical Child and Adolescent Psychology, and the TLC Foundation for Repetitive Behavior Disorders. She has received royalties from Oxford University Press. Dr. Compton has received support from NIMH, NC

www.jaacap.org 479

4

GINSBURG et al.

GlaxoSmithKline Foundation, Pfizer, Neurocrine Biosciences, and Mursion, Inc. He has served as a consultant for Shire and Mursion, Inc. He has received honoraria from the Nordic Long-Term OCD Treatment Study Research Group and the Centre for Child and Adolescent Mental Health, Eastern and Southern Norway. He has served on the scientific advisory board of Tourette Association of America, Anxiety and Depression Association of America, and Mursion, Inc. He has presented expert testimony for Duke University. Dr. Piacentini has received support from NIMH, the TLC Foundation for Body-Focused Repeti- tive Behaviors, the Tourette Association of America, the Pettit Family Foun- dation, and Pfizer Pharmaceuticals through the Duke University Clinical Research Institute Network. He is a co-author of the Child OCD Impact Scale �Revised (COIS-R), the Child Anxiety Impact Scale�Revised (CAIS-R), the Parent Tic Questionnaire (PTQ), and the Premonitory Urge for Tics Scale (PUTS)

80 www.jaacap.org

assessment tools, all of which are in the public domain therefore no royalties are received. He has received royalties from Guilford Press and Oxford Uni- versity Press. He has served on the speakers’ bureau of the Tourette Associ- ation of America, the International Obsessive-Compulsive Disorder Foundation, and the TLC Foundation for Body-Focused Repetitive Behaviors.

Correspondence to Golda S. Ginsburg, PhD, Department of Psychiatry, Uni- versity of Connecticut School of Medicine, 65 Kane Street Room 2033 West Hartford, CT 06119; e-mail: Gginsburg@uchc.edu

0890-8567/$36.00/ª2018 American Academy of Child and Adolescent Psychiatry

https://doi.org/10.1016/j.jaac.2018.03.017

REFERENCES

1. Costello EJ, Mustillo S, Erkanli A, Keeler G, Angold A. Prevalence and development of

psychiatric disorders in childhood and adolescence. Arch Gen Psychiatry. 2003;60:837-844. 2. Kessler RC, Avenevoli S, Costello EJ, et al. Prevalence, persistence, and sociodemo-

graphic correlates of DSM-IV disorders in the National Comorbidity Survey Replication Adolescent Supplement. Arch Gen Psychiatry. 2012;69:372-380.

3. Bodden D, Dirksen C, Bogels S. Societal burden of clinically anxious youth referred for treatment: a cost of illness study. J Abnormal Child Psychol. 2008;36:487-497.

4. Beesdo K, Bittner A, Pine DS, et al. Incidence of social anxiety disorder and the consistent risk for secondary depression in the first three decades of life. Arch Gen Psychiatry. 2007;64:903-912.

5. Bittner A, Egger HL, Erkanli A, Jane Costello E, Foley DL, Angold A. What do childhood anxiety disorders predict? J Child Psychol Psychiatry. 2007;48:1174-1183.

6. Broeren S, Muris P, Diamantopoulou S, Baker JR. The course of childhood anxiety symptoms: developmental trajectories and child-related factors in normal children. J Abnorm Child Psychol. 2013;41:81-95.

7. Walkup JT, Albano AM, Piacentini J, et al. Cognitive behavioral therapy, sertraline, or a combination in childhood anxiety. N Engl J Med. 2008;359:2753-2766.

8. Gibby B, Casline E, Ginsburg GS. Long-term outcomes of youth treated for an anxiety disorder: a critical review. J Clin Child Fam Psychol Rev. 2017;20:201-225.

9. Ginsburg GS, Becker EM, Keeton CP, et al. Naturalistic follow-up of youths treated for pediatric anxiety disorders. JAMA Psychiatry. 2014;71:310-318.

10. Compton SN, Peris TS, Almirall D, et al. Predictors and moderators of treatment response in childhood anxiety disorders: results from the CAMS Trial. J Consult Clin Psychol. 2014;82:212-224.

11. Brown TA, DiNardo PA, Barlow DH. Anxiety Disorders Interview Schedule for DSM- IV. New York: Graywind Publications; 1994.

12. Silverman WK, Albano A. The Anxiety Disorders Interview Schedule for DSM-IV: Child and Parent Versions. New York: Plenum Press; 1996.

13. Kendall PC, Hedtke K. Cognitive-Behavioral Therapy for Anxious Children: Therapist Manual (3rd ed). Ardmore, PA: Workbook Publishing; 2006.

14. Kendall PC, Choudhury M, Hudson J, Webb A. The C. A. T. Project Therapist Manual. Ardmore, PA: Workbook Publishing; 2002.

15. Guy W. ECDEU Assessment Manual for Psychopharmacology—Revised. Rockville, MD: National Institute of Mental Health; 1976.

16. Coddington RD. The significance of life events as etiologic factors in the diseases of children—II a study of a normal population. J Psychosom Res. 1972;16:205-213.

17. Hollingshead AB. Commentary on the indiscriminate state of social class measurement. Social Forces. 1971;49:563-567.

18. Skinner HA, Steinhauer PD, Santa-Barbara J. Family Assessment Measure�Version III. Cheektowaga, NY: Multi-Health Systems Incorporated; 1995.

19. Derogatis LR, Melisaratos N. The Brief Symptom Inventory: an introductory report. Psychol Med. 1983;13:595-605.

20. Reinhard SC, Gubman GD, Horwitz AV, Minsky S. Burden assessment scale for families of the seriously mentally ill. Eval Program Plann. 1994;17:261-269.

21. Kendall PC, Compton SN, Walkup JT, et al. Clinical characteristics of anxiety disor- dered youth. J Anxiety Disord. 2010;24:360-365.

22. Curry J, Silva S, Rohde P, et al. Recovery and recurrence following treatment for adolescent major depression. Arch Gen Psychiatry. 2011;68:263-269.

23. Kerns CM, Read KL, Klugman J, Kendall PC. Cognitive behavioral therapy for youth with social anxiety: differential short and long-term treatment outcomes. J Anxiety Disord. 2013;27:210-215.

24. Allen JL, Rapee RM, Sandberg S. Severe life events and chronic adversities as antecedents to anxiety in children: a matched control study. J Abnorm Child Psychol. 2008;36: 1047-1056.

25. Grover RL, Ginsburg GS, Ialongo N. Childhood predictors of anxiety symptoms: a longitudinal study. Child Psychiatry Hum Dev. 2005;36:133-153.

Journal of the American Academy of Child & Adolescent Psychiatry Volume 57 / Number 7 / July 2018