Social Anxiety

Contents lists available at ScienceDirect

Journal of Anxiety Disorders

journal homepage: www.elsevier.com/locate/janxdis

Long-term effectiveness of cognitive behavioral therapy for youth with anxiety disorders☆

Arne Kodala,b,e,⁎, Krister Fjermestadb,i, Ingvar Bjellanda,b,e, Rolf Gjestadc, Lars-Göran Östb,j, Jon F. Bjaastadb,d,g, Bente S.M. Hauglandb,d, Odd E. Havikb,f, Einar Heiervangb,h, Gro Janne Wergelanda,b,d

a Department of Child and Adolescent Psychiatry, Division of Psychiatry, Haukeland University Hospital, N-5021 Bergen, Norway b Anxiety Research Network, Research Department, Division of Psychiatry, Haukeland University Hospital, N-5036 Bergen, Norway c Research Department, Division of Psychiatry, Haukeland University Hospital, N-5036 Bergen, Norway d Regional Center for Child and Youth Mental Health and Child Welfare, Uni Research Health, N-5008 Bergen, Norway e Department of Clinical Medicine, Faculty of Medicine, University of Bergen, N-5020 Bergen, Norway f Department of Clinical Psychology, Faculty of Psychology, University of Bergen, N-5020 Bergen, Norway g Division of Psychiatry, Stavanger University Hospital, N-4068 Stavanger, Norway h Institute of Clinical Medicine, University of Oslo, and Oslo University Hospital, N-0450 Oslo, Norway i Department of Psychology, University of Oslo, N-0373 Oslo, Norway j Department of Psychology, Stockholm University, SE-106 91 Stockholm, Sweden

A R T I C L E I N F O

Keywords: Anxiety disorders Long-term follow-up Youth Cognitive behavioral therapy Community clinic

A B S T R A C T

Cognitive behavioral therapy (CBT) has demonstrated favorable long-term outcomes in youth with anxiety disorders in efficacy trials. However, long-term outcomes of CBT delivered in a community setting are uncertain. This study examined the long-term outcomes of individual (ICBT) and group CBT (GCBT) in youth with anxiety disorders treated in community mental health clinics. A total of 139 youth (mean age at assessment 15.5 years, range 11–21 years) with a principal diagnosis of separation anxiety disorder (SAD), social anxiety disorder (SOP), and/or generalized anxiety disorder (GAD) were evaluated, on average, 3.9 years post-treatment (range 2.2–5.9 years). Outcomes included loss of all inclusion anxiety diagnoses, loss of the principal anxiety diagnosis and changes in youth- and parent-rated youth anxiety symptoms. At long-term follow-up, there was loss of all inclusion anxiety diagnoses in 53%, loss of the principal anxiety diagnosis in 63% of participants as well as significant reductions in all anxiety symptom measures. No statistical significant differences in outcome were obtained between ICBT and GCBT. Participants with a principal diagnosis of SOP had lower odds for recovery, compared to those with a principal diagnosis of SAD or GAD. In conclusion, outcomes of CBT for youth anxiety disorders delivered in community mental health clinics were improved at nearly 4 years post-treatment, and recovery rates at long-term follow-up were similar to efficacy trials.

1. Introduction

Cognitive behavioral therapy (CBT) is a well-established treatment for anxiety disorders in children and adolescents (hereafter youth) (Higa-McMillan, Francis, Rith-Najarian, & Chorpita, 2016). Meta-ana- lyses have shown that approximately 60% of youth recover from their anxiety disorders and experience significant symptom reduction fol- lowing treatment (James, James, Cowdrey, Soler, & Choke, 2013; Warwick et al., 2017). However, there has been less focus on the question of whether treatment outcomes are maintained in the long term. Relapse can lead to detrimental consequences at individual,

family, and societal levels, as early anxiety disorders predict later emotional, social, academic, and vocational problems (Copeland, Angold, Shanahan, & Costello, 2014; Kendall & Ollendick, 2004). Suc- cessful CBT treatment for youth anxiety disorders on the other hand, provides protection from later sequelae (Puleo, Conner, Benjamin, & Kendall, 2011; Wolk, Kendall, & Beidas, 2015). Furthermore, in- vestigating long-term outcomes is essential in establishing treatment efficacy in youth anxiety disorders (Chambless & Hollon, 1998).

Long-term follow-up is commonly defined as follow-up at least two years post-treatment (Gibby, Casline, & Ginsburg, 2017; Nevo & Manassis, 2009). To date, five studies based on separate samples have

https://doi.org/10.1016/j.janxdis.2017.11.003 Received 15 August 2017; Received in revised form 22 November 2017; Accepted 22 November 2017

☆ ClinicalTrials.gov identifier: NCT00586586. ⁎ Corresponding author at: Division of Psychiatry, Haukeland University Hospital, p.b 1400, 5021 Bergen, Norway. E-mail address: [email protected] (A. Kodal).

Journal of Anxiety Disorders 53 (2018) 58–67

Available online 26 November 2017 0887-6185/ © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

T

examined the long-term effects of CBT protocols in youth with mixed anxiety disorders in the form of separation anxiety disorder (SAD), social anxiety disorder (SOP), and/or generalized anxiety disorder (GAD) (Barrett, Duffy, Dadds, & Rapee, 2001; Benjamin, Harrison, Settipani, Brodman, & Kendall, 2013; Ginsburg et al., 2014; Kendall & Southam-Gerow, 1996; Kendall, Safford, Flannery-Schroeder, & Webb, 2004), over follow-up periods ranging from 2 to 19 years post-treat- ment (M = 7.9 years; Mdn = 6.2 years). These studies indicate that post-treatment outcomes were either maintained or improved at long- term follow-up, with 46.5–85.7% of study participants no longer ful- filling the diagnostic criteria for anxiety disorders (e.g. Barrett et al., 2001; Ginsburg et al., 2014). A recent review of long-term follow-up studies of youth treated for any anxiety disorder (with the exception of obsessive-compulsive disorder and post-traumatic stress disorder), with follow-up assessments a mean of 5.9 years post-treatment, found that 64.6% of youth were in remission. More specifically, 57.0% and 76.7% had lost all inclusion anxiety diagnoses and their primary anxiety di- agnosis, respectively (Gibby et al., 2017). In addition, different treat- ment formats in the form of individual CBT (ICBT) and group CBT (GCBT) in youth with anxiety disorder were examined by Saavedra et al.; the authors found no difference in long-term outcomes between ICBT and GCBT at a mean of 9.8 years post-treatment (Saavedra, Silverman, Morgan-Lopez, & Kurtines, 2010), consistent with previous meta-analyses of studies of short-term outcomes which showed similar effect sizes for both ICBT and GCBT (In-Albon & Schneider, 2006; Silverman, Pina, & Viswesvaran, 2008).

Long-term outcome studies differ considerably in reported outcome measures, e.g., absence of the principal inclusion anxiety diagnosis (Kendall et al., 2004), absence of all inclusion anxiety diagnoses (Barrett et al., 2001), or absence of all anxiety diagnoses (Benjamin et al., 2013). However, loss of one anxiety diagnosis does not ne- cessarily indicate the absence of further anxiety-related impairments. Furthermore, heterogeneity in reported outcomes makes comparisons across long-term follow-up studies difficult and hence challenges the generalizability of the study findings. Consequently, this calls for more detailed information on diagnostic outcomes following treatment, in- cluding loss of the principal anxiety diagnosis, all comorbid anxiety diagnoses, as well as symptom measure outcomes (Gibby et al., 2017; Warwick et al., 2017).

All of the above-cited studies are efficacy trials conducted at spe- cialized university clinics. Efficacy trials allow for high levels of methodological rigor and control, thus achieving high internal validity. However, to what extent findings from such studies are transferable to community clinical settings is unclear (Hunsley & Lee, 2007; Santucci, Thomassin, Petrovic, & Weisz, 2015). Factors that may influence treatment outcomes differentially in community clinics, compared to university clinics, include differing patient populations (e.g., different inclusion and exclusion criteria, greater population heterogeneity in the community setting), therapist-related factors (e.g., training, caseloads, access to expert supervision), treatment context (e.g., availability of research resources, treatment monitoring) and treatment content (e.g. potential less use of exposure exercises) (Smith et al., 2017). It is argued that these factors contribute to reduced effect sizes of treatment when efficacy-supported therapies are transferred to community clinics (Weisz et al., 2013).

To our knowledge, no study to date has examined the long-term outcomes of CBT for anxiety disorders in community mental health clinics, i.e., the effectiveness of long-term treatment. Several short-term effectiveness studies with follow-up assessments 3–15 months post- treatment (M = 9.8 months, Mdn = 9 months) reported recovery rates ranging from 52% to 78% (Barrington, Prior, Richardson, & Allen, 2005; Bodden et al., 2008; Chorpita et al., 2013; Lau, Chan, Li, & Au, 2010; Nauta, Scholing, Emmelkamp, & Minderaa, 2001; Nauta, Scholing, Emmelkamp, & Minderaa, 2003). Overall, the studies con- firmed the maintenance of treatment gains from post-treatment to follow-up, albeit with slightly lower recovery rates compared to those

obtained from efficacy trials. However, there is a need to examine ef- fectiveness of CBT for mixed anxiety disorders in youth beyond 15 months post-treatment.

It has been argued that the three main anxiety disorders SAD, SOP, and GAD are manifestations of the same underlying anxiety construct and therefore are amenable to treatment with the same CBT protocols (Crawley, Beidas, Benjamin, Martin, & Kendall, 2008; Silverman & Kurtines, 1996). However, recent short-term studies showed that chil- dren with SOP had poorer treatment outcomes from generic CBT pro- tocols, compared to those with GAD and/or SAD (Hudson et al., 2015; Reynolds, Wilson, Austin, & Hooper, 2012). Based on an efficacy trial, Kerns, Read, Klugman, and Kendall (2013) reported comparable out- comes for SOP, SAD, and GAD immediately following CBT but found youth with SOP were significantly less improved at 7.4-year follow-up. On the other hand, Barrett et al. (2001) found no evidence that pre- treatment diagnosis, including SOP, differentially affected long-term treatment outcomes. Thus, further studies on the long-term effects of CBT in youth with SOP are warranted.

The primary aim of the present study was to investigate the long- term outcomes of CBT in youth with anxiety disorders treated in community mental health clinics. Based on previous long-term efficacy studies and on short-term effectiveness studies, we expected that out- comes of CBT would be maintained or improved in the community setting, yet below comparative efficacy studies. The secondary aim was to investigate the effects of using different treatment formats (i.e., GCBT versus ICBT) on long-term outcomes. Based on existing evidence, we expected the effects of both treatment formats to be maintained during the follow-up period and to be equivalent at long-term follow-up. The third aim was to assess for disorder-specific differences in treatment outcomes, for which we predicted that outcomes in youth with a principal diagnosis of SOP would be inferior, compared to those with a principal diagnosis of GAD and/or SAD.

2. Method

2.1. Participants

Eligible participants were selected from a total of 179 youth who participated in a randomized controlled trial (RCT) investigating the effectiveness of ICBT and GCBT, compared to a waitlist control, in youth with mixed anxiety disorders treated in community mental health clinics (Wergeland et al., 2014). The study was conducted from 2008 to 2012. Age of participants ranged from 8 to 15 years at the time of recruitment. The inclusion criterion was a principal diagnosis of SAD, SOP, and/or GAD. The only exclusion criterion included pervasive de- velopmental disorder, psychotic disorder, severe conduct disorder, and/ or mental retardation. Participants were assessed pre- and post-treat- ment, and at 1-year follow-up. A detailed description of the original sample, method, and outcomes has been published elsewhere (Wergeland et al., 2014).

A total of 139 youth participated in the present study. Youth were assessed an average of 3.9 years post-treatment (SD = 0.8, range 2–6 years). Age of participants at long-term follow-up ranged from 11 to 21 years (M = 15.5, SD = 2.5), and 54.7% were female. Youth partici- pating in this long-term follow-up study (N = 139) were compared to those from the original RCT not participating in the present study (n = 40) in terms of pre-treatment socio-demographic characteristics (i.e., age, gender, ethnicity, parent occupational status) and pre-treat- ment clinical variables (i.e., clinical severity rating (CSR) of the prin- cipal anxiety diagnosis, anxiety and depressive symptoms, comorbidity, principal anxiety diagnosis present at post-treatment). There were no significant differences on any of these variables between youth parti- cipating and those not participating in the long-term follow-up study (see Table 1). Furthermore, no differences were found in post-treatment outcomes (loss of the principal diagnosis and loss of all inclusion an- xiety diagnoses, changes in symptom measures) between youth in the

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

59

present long-term study and those who attended post-treatment as- sessment but did not participate in the present long-term follow-up (n = 15).

2.2. Treatment, setting, and therapists

The treatment manual used in the RCT was FRIENDS for life, 4th edition (Barrett, 2005). This program stems from the Australian Coping Koala program (Barrett, Dadds, & Rapee, 1991) that was adapted from Kendall’s original Coping Cat manual (Kendall, 1990). Children aged 8–12 received the child version of the protocol, whereas youth aged 12–15 received the adolescent version. Youth aged 12 were treated using either the child (n = 34) or the adolescent version (n = 5), based on the clinician’s assessment of the youth's level of maturity. The ICBT protocol comprised ten 60-min sessions, and the GCBT protocol com- prised ten 90-min sessions. Two booster sessions were conducted, one and three months after the tenth session. Parents attended two of the ten sessions, the last 15 min of the remaining eight sessions, as well as two separate parent-only sessions.

The RCT was conducted at seven public child and adolescent mental

health outpatient clinics. Seventeen therapists participated, of whom five had completed a formal 2-year post-graduation CBT training and the remaining 12 had little or no previous training in CBT. All therapists received training in the FRIENDS for life protocol, as well as supervision throughout the treatment sessions by licensed FRIENDS therapists. All treatment sessions were delivered as part of the therapist's routine caseload, and all therapists administered both ICBT and GCBT, with two therapists participating in each GCBT session. Assessment was con- ducted by 16 assessors, all clinicians employed at the clinics. Assessors received specific training in the Anxiety Disorders Interview Schedule child and parent version (ADIS-C/P) in a 2-day workshop, and received supervision during the study period. Therapists’ treatment adherence and competence were assessed on a 7-point scale ranging from 0 (none/ poor skills) to 6 (thorough/excellent skills). Therapists’ mean scores for adherence to both treatment formats ranged from 3.97 to 5.42 (M = 4.60, SD = 0.88), and for competence 3.25–5.22 (M = 4.12, SD = 0.97), using the Competence and Adherence Scale for Cognitive Behavioral Therapy (CAS-CBT; Bjaastad et al., 2016). A predetermined score of 3.0 was set as the minimum threshold for adequate therapist adherence and competence. Further details on the protocol are as

Table 1 Comparison of pre- and post-treatment characteristics of participants and non-participants at long-term follow-up.

Participants (n = 139) Non-participants (n = 40)

M (SD) n % M (SD) n % p

Pre-treatment characteristics Gender 0.72 Female 76 54.7 20 50.0 Age, mean (SD) 11.46 2.11 11.85 2.03 0.16 Age group (years) 0.45 8–12 94 67.6 24 60.0 12–15 45 32.4 16 40.0

Family social classa 0.93 High 68 48.9 15 37.5 Middle 44 31.7 11 27.5 Low 15 10.8 2 5.0 Not reported 12 8.6 12 30.0

Principal diagnosis 0.10 SAD 49 35.3 9 22.5 SOP 64 46.0 20 50.0 GAD 26 18.7 11 27.5

Principal diagnosis CSR 7.01 1.10 6.92 1.19 0.60 Number of inclusion anxiety diagnoses 2.04 0.79 1.80 0.72 0.08 SCAS-C 36.77 16.87 34.61 15.58 0.50 SCAS-P 35.28 12.72 33.42 13.94 0.43 SMFQ-C 7.43 5.40 7.61 6.10 0.86 SMFQ-P 7.50 5.06 7.83 4.91 0.72 Other comorbidity 0.27 Other anxiety disorder 20 14.4 7 17.5 Depressiond 16 11.5 5 12.5 Externalizing disordersd 15 10.8 1 2.5 Tic disorderd 11 7.9 1 2.5 Anorexiad 0 0.0 2 5.0c

Post-treatment characteristics (n = 139) (n = 15b) Principal diagnosis CSR 4.72 2.37 4.53 2.83 0.77 Comorbid inclusion anxiety diagnoses 0.52 0.73 0.20 0.41 0.02c

Loss of principal diagnosisd 56 40.3 7 46.7 0.63 Loss of all inclusion anxiety diagnosesd 37 26.6 4 26.7 1.00 SCAS-C 24.15 16.00 23.42 12.70 0.34 SCAS-P 27.58 13.64 23.83 11.52 0.34 SMFQ-C 5.66 5.93 3.17 4.15 0.08 SMFQ-P 5.49 4.91 5.17 6.56 0.82

Note. Externalizing disorders = Attention deficit hyperactive disorder and/or Oppositional defiant disorder; GAD = Generalized Anxiety Disorder; GCBT = Group Cognitive Behavioral Therapy; ICBT = Individual Cognitive Behavioral Therapy; Other anxiety disorder = OCD, Panic Disorder, PTSD, Specific anxiety disorder; SAD = Separation Anxiety Disorder; SCAS = Spence Child Anxiety Scale, C = Child, P = Parent; SMFQ = Short Mood and Feelings Questionnaire, C = Child, P = Parent; SOP = Social Anxiety Disorder.

a Family social class was defined by the highest-ranking parent. Parents were classified into rank-ordered social classes in accordance with the Registrar General Social Class coding scheme (Currie et al., 2008).

b Fifteen of the 40 participants not attending the long-term follow-up were assessed at post-treatment. c Not significant after the modified Bonferroni procedure was applied. d Due to small sample sizes, Fisher’s exact test was used for these comparisons.

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

60

previously described (Wergeland et al., 2014).

2.3. Measures

2.3.1. The Anxiety Disorders Interview Schedule child and parent version (ADIS-C/P)

The ADIS-C/P (Silverman & Albano, 1996) was used to assess in- clusion diagnoses. The same modules assessing SAD, SOP and GAD which were used in the original RCT (pre-treatment, post-treatment, and 1-year follow-up) were also used in the long-term follow-up study, to provide comparable results across assessment points. The modules assessing DSM-IV criteria for the diagnoses of SAD, SOP, and GAD were used in youth aged 17 or younger (n = 107). Youth and parents were interviewed independently, with parents receiving the parent version of the Interview Schedule (ADIS-P) (n = 107). The CSR scale ranges from 0 to 8, with a minimum CSR score of 4 required for a clinical diagnosis to be assigned (Silverman & Albano, 1996). Diagnosis and CSR were assigned based on the youth and parent composite score. In cases of multiple anxiety diagnoses, the one causing the highest interference as measured by the CSR is considered to be the principal diagnosis. This interview was previously shown to have high interrater reliability (Silverman & Nelles, 1988) and retest reliability (Silverman, Saavedra, & Pina, 2001). For the purpose of the long-term follow-up study, the ADIS interview was slightly modified to also assess whether youth re- ceived additional treatment since completion of the treatment protocol.

2.3.2. The Anxiety Disorders Interview Schedule for DSM-IV (ADIS-IV-L) The ADIS-IV-L (Brown, Barlow, & DiNardo, 1994) was used to assess

DSM-IV criteria for SAD, SOP, and GAD in youth aged 18 or older (n = 32). Similar to the ADIS-C/P, the CSR scale ranges from 0 to 8, with a minimum CSR score of 4 required for a clinical diagnosis to be given. This interview method has previously demonstrated good to excellent reliability (Brown et al., 1994; Brown, Di Nardo, Lehman, & Campbell, 2001).

Diagnostic interviews at long-term follow-up were conducted face- to-face and video-recorded. Interviews conducted by phone (10.1%) were not recorded (see Section 2.4). A random selection of 20% of the video-recorded interviews was re-evaluated by expert raters blind to the assessors’ ratings. The interrater agreement was k = 0.94 on the ADIS-C and/or ADIS-P and the ADIS-IV-L. For the principal anxiety diagnosis, k values were: SAD = 1.00; SOP = 0.88; and GAD = 0.81. Regarding CSR scores for the anxiety diagnoses, intraclass correlations (ICCs) for the total sample was 0.97, whereas ICCs for the principal anxiety di- agnosis were: SAD = 1.00; SOP = 0.94; and GAD = 0.93.

2.3.3. The Spence Child Anxiety Scale child and parent version (SCAS-C/P) The SCAS-C/P (Spence, 1998) was used to assess youth anxiety

symptoms. The SCAS comprises 38 items rated on a 4-point scale (0 = never; 1 = sometimes; 2 = often; 3 = always), with a maximum sum score of 114. Validity, internal consistency, and adequate tes- t–retest reliability were previously demonstrated (Spence, 1998; Spence, Barrett, & Turner, 2003). Internal consistency for SCAS in the current sample was good to excellent (parent α = 0.92, child α = 0.89).

2.3.4. The Short Moods and Feelings Questionnaire child and parent version (SMFQ-C/P)

The SMFQ-C/P (Angold, Costello, Messer, & Pickles, 1995) was used to assess youth depressive symptoms. The SMFQ consists of 13 items rated on a 3-point scale (0 = not true; 1 = sometimes true; 2 = true), with higher scores indicating greater severity of symptoms. In a general population sample, a cutoff score of ≥ 8 was found to represent the best balance between sensitivity and specificity, compared to a diagnosis of depression (Angold et al., 1995). The SMFQ was previously shown to have excellent internal consistency and good test–retest reliability in children over a 2-week period (Costello & Angold, 1988). The SMFQ

differentiates well between psychiatric and nonpsychiatric subjects in a general population (Sharp, Goodyer, & Croudace, 2006). Internal con- sistency in the current sample was excellent (parent α = 0.92, child α = 0.93).

2.4. Procedure

At the time of inclusion in the original RCT, all 179 participants had consented to be contacted for long-term follow-up. At post-treatment, 154 youth completed the intervention and post-treatment assessment. Of these, 145 completed the 1-year follow-up assessment. At long-term follow-up, all 179 initial participants were contacted by telephone or mail. Among the treatment completers (n = 154), 15 did not wish to participate in the long-term follow-up. Thus, the present study sample comprised 139 youth, i.e., 77.7% of the total initial sample and 90.3% of treatment completers (See Fig. 1). Youth and families who agreed to participate were scheduled for separate youth and parent assessments. Most interviews were conducted face-to-face in the community out- patient clinics. Fourteen interviews were conducted by phone, due to participants having moved out of the region. Three certified ADIS-C/P interviewers (two psychologists and one child psychiatrist) conducted the interviews. The ADIS-C/P interview and questionnaires were com- pleted in the same session, whereas those who were interviewed by phone received and returned the questionnaires by mail. Interviewers were blind to the youth’s inclusion anxiety diagnoses, treatment format (ICBT or GCBT), and treatment outcome both at post-treatment and at 1-year follow-up. Participating youth and parents were each compen- sated with a gift card (worth US$60). The study was approved by, and conducted in accordance with the guidelines of Regional Committee for Medical and Health Research Ethics of Western Norway.

2.5. Data analysis

Analyses were based on the sample of 139 youth. There were no missing data regarding the diagnostic interviews. Missing data on items from the symptom measures SMFQ-C/P and SCAS-C/P ranged from 3.6% to 13.7%. Little’s missing completely at random test indicated that missing data on symptom measures occurred completely at random. Missing data on all continuous variables were accounted for by full information maximum likelihood (FIML) in Mplus (Wothke, 2000), such that a missing data point did not result in list-wise deletion.

Descriptive statistics and analyses of group differences between participants and non-participants were performed using SPSS 23. All other analyses were run using Mplus 7.4 (Muthén & Muthén, 2017). To account for nonnormality present in the data, all structural equation modelling (SEM) analyses were run using a Maximum Likelihood esti- mator with Robust standard errors (MLR), which is robust to violations of nonnormality (Muthén & Muthén, 2017). Seven clinics participated in the study. The mean number of participants at each of the seven clinics was 20 (range 10–25). The GCBT format consisted of 16 separate treatment groups, whereas youth treated with ICBT were grouped as one cluster at each clinic (hence seven clusters), resulting in a total of 23 clusters. The design was therefore partially clustered, and models were adjusted to counter potential clustering effects (Baldwin, Bauer, Stice, & Rohde, 2011). The Holm–Bonferroni method was applied to control for experiment-related error rate at 0.05, when conducting multiple tests. Analyses were controlled for age, gender, and pre- treatment CSR score for the principal anxiety diagnosis.

At long-term follow-up, 27.3% of participants (n = 38) reported having received additional treatment for anxiety following study treatment completion. To examine whether additional treatment in- fluenced study outcomes, all outcome analyses were performed with and without these participants. Since inclusion of participants receiving additional treatment was not found to alter the results, they were therefore included in all subsequent analyses (data not shown; available on request). Additional regression analyses of diagnostic treatment

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

61

response during follow-up and the use of additional treatment were performed.

Primary outcomes were loss of all inclusion anxiety diagnoses (SAD, SOP, and GAD), and loss of the principal diagnosis. Thus, complete recovery was defined as loss of all inclusion anxiety diag- noses and recovery was defined as loss of principal inclusion anxiety diagnosis. Secondary outcomes were change in youth- and parent- reported anxiety, change in depressive symptoms and change in CSR of the primary, secondary and tertiary diagnoses. Growth rates and intercepts on CSR scores and symptom measures varied considerably between pre- and post-treatment. As a consequence, these were analyzed using piecewise latent growth curve modeling (p-LGM; Wang & Wang, 2012). To account for individually varying times of observation, random slopes for participants were estimated (Muthén & Muthén, 2017).

Secondary analyses included comparisons of the outcomes of ICBT

versus GCBT and were estimated using logistic regression analyses for diagnostic outcomes, and p-LGM for symptom change. Given that a statistically non-significant result between the two treatment formats does not imply equivalence, we conducted analyses of equivalence between GCBT and ICBT. The analyses were conducted on the CSR of the principal anxiety diagnosis and youth- and parent-reported anxiety and depressive symptoms using the confidence interval (CI) method (Rogers, Howard, & Vessey, 1993). An equivalence interval of 15% around a difference of zero was defined, with GCBT outcomes as the reference group. Differences small enough to fall within this equiva- lence interval were considered to be of little clinical and/or practical importance.

To examine the impact of the principal anxiety diagnosis on out- comes, we used logistic regression analysis to estimate the odds ratio (OR) for loss of the principal diagnosis for each individual principal diagnoses. We then estimated the OR for loss of the principal diagnosis

Fig. 1. Participant flowchart.

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

62

and growth rates of the symptom measures using latent growth curve modelling for youth with a principal diagnosis of SOP or SAD, com- pared to those with a principal diagnosis of GAD.

3. Results

3.1. Primary research aim: diagnostic status at long-term follow-up

Of the 139 youth, 53% (n = 73) did not meet the criteria for any of their inclusion anxiety diagnosis at long-term follow-up, and 63% (n = 87) did not meet the criteria for their principal anxiety diagnosis. At post-treatment, for the same sample, the proportion of these parti- cipants was 27% (n = 37) and 40% (n = 56), respectively (see Table 2). Loss of principal and loss of all inclusion diagnoses differed between post-treatment and long-term follow-up, demonstrating a sig- nificant improvement (p < 0.05).

3.1.1. Diagnostic change during follow-up In total 19% (n = 27) of the 139 youth had lost all inclusion anxiety

diagnoses at post- and long-term follow-up, whereas 40% (n = 56) retained one or more inclusion anxiety diagnoses at post- or long-term follow-up. Of the 102 youth that had not lost all inclusion anxiety di- agnoses at post-treatment, 45% (n = 46) had recovered completely at long-term follow-up. Regression analysis revealed no significant asso- ciation between these 46 youth and the use of additional interim treatment (OR = 0.68, 95% CI [0.35, 1.32], p = 0.34). In total 31% (n = 43) of the 139 youth had lost their principal anxiety diagnosis at both post- and long-term assessment, whereas 28% (n = 39) retained their principal anxiety diagnosis at both assessment points. Among the 83 youth that did not lose their principal anxiety diagnosis at post- treatment, 53% (n = 44) recovered at long-term follow-up. Regression analysis revealed no significant association between these 44 youth and the use of additional interim treatment (OR = 0.87, 95% CI [0.48, 1.56], p = 0.69).

3.1.2. Clinical severity rating and symptom measures Table 3 displays the severity ratings of the principal, secondary, and

tertiary anxiety diagnoses and symptom measures at pre- and post- treatment and long-term follow-up. There was a significant reduction in CSR for the principal, secondary, and tertiary anxiety diagnoses (p < 0.05), as well as a significant reduction in symptom scores for all symptom measures from post-treatment to long-term follow-up (p < 0.05). Only youth self-reported depressive symptoms remained unchanged at long-term follow-up (p = 0.54). As found for post-treat- ment, analyses revealed no significant differences between the ICBT and GCBT conditions at long-term follow-up. No significant interaction effect between time and treatment format was found. Thus, the rate of symptom reduction during long-term follow-up was statistically similar for both treatment formats (i.e., ICBT and GCBT), and the improvement range for both treatment formats was also similar across time.

3.1.3. Relationship between treatment response at post-treatment and long- term follow-up

We examined the relationship between outcomes for the

participants’ principal anxiety diagnosis based on CSR scores at post- treatment and long-term follow-up, as detailed in Table 4. Based on criteria for clinically significant change by Jacobsen and Truax (1991), the participants’ clinical status at post-treatment and at long-term follow-up, using their CSR scores in comparison with their pre-treat- ment CSR scores, was classified into four categories: deterioration: CSR score increased by ≥ 2 points; no change: CSR score changed by ± 1 point; response: CSR score decreased by ≥ 2 points; and recovery: CSR score decreased by ≥ 2 points and the score was ≤ 3. Of the 56 par- ticipants who were classified as recovered at post-treatment, 43 (77% of recovered participants, or 31% of the entire sample, n = 139) were still classified as recovered at long-term follow-up, whereas the remaining 13 (23%) had worsened to some degree and re-qualified for their principal diagnosis. Of the 26 participants classified as treatment re- sponders at post-treatment, 17 (65%) had recovered at long-term follow-up, five maintained their treatment response status, whereas four had worsened and demonstrated no change from pre-treatment. Of the 53 participants classified as non-responders at post-treatment, 26 (49%) had recovered at long-term follow-up, eight (15%) were re- sponders, 18 (34%) remained unchanged from pre-treatment, and one participant had worsened. Finally, of the four participants classified as deteriorated at post-treatment, one had fully recovered at long-term follow-up, whereas two demonstrated only marginal change and thus were classified as no change, and one remained classified as deterio- rated. Summing up the number of participants who demonstrated maintenance, improvement, or worsening of their outcome status from post-treatment to long-term follow-up, 67 participants (48%) main- tained their classification, whereas 54 demonstrated further improve- ment (39%), with 44 who fully recovered, and 18 participants (13%) worsened by re-qualifying for their principal diagnosis. Regression analysis of the presented changes in CSR scores, demonstrated a sta- tistical significant relationship between treatment response at post- treatment and response at long-term follow-up (z = 3.0, p < 0.01), indicating further improvement during the follow-up period. Among youth recovering during follow-up (n = 44), chi-square analyses re- vealed no significant association with the use of interim mental health treatment and recovery (p = 0.22).

3.2. Secondary research aims

3.2.1. Individual versus group treatment At long-term follow-up, logistic regression analyses demonstrated

no statistical significant differences in loss of all inclusion anxiety di- agnosis and loss of principal inclusion anxiety diagnoses between ICBT and GCBT, with GCBT as the reference (49% versus 56%, OR = 1.26, 95% CI [0.80, 2.01], p = 0.41), or loss of principal inclusion anxiety diagnoses (59% versus 66%, OR = 1.29, 95% CI [0.85, 1.84], p = 0.50). p-LGM analyses showed no statistical significant differences between ICBT and GCBT in youth-rated anxiety symptoms (z = 0.86, p = 0.39), parent-rated youth anxiety symptoms (z = 0.49, p = 0.63), youth-rated depressive symptoms (z = 0.10, p = 0.92), or parent-rated youth depressive symptoms (z = −1.28, p = 0.20).

Equivalency between ICBT and GCBT at long-term follow-up was established for the CSR outcome score of the participant’s principal

Table 2 Loss of inclusion diagnoses following ICBT or GCBT and comparison between treatment formats.

ADIS-C/P combined diagnosis Long-term follow-up Post-treatment

Total sample (N = 139) ICBT (n = 70) GCBT (n = 69) χ2 ICBT (n = 70) GCBT (n = 69)

n % n % n % n % n %

Loss of all inclusion anxiety diagnoses 73 53.0 39 55.7 34 49.3 0.59 22 31.4 15 23.3 Loss of principal anxiety diagnosis 87 63.0 46 65.7 41 59.4 0.59 30 42.9 26 37.7

Note. GCBT = Group Cognitive Behavioral Therapy; ICBT = Individual Cognitive Behavioral Therapy.

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

63

anxiety diagnosis ( ± 0.94; 90% CI [−0.46, 0.71]). In contrast, equivalency was not established between the two treatment formats for SCAS-C ( ± 3.5; 90% CI [−6.25, 3.21]), SCAS-P ( ± 3.3; 90% CI [−5.38, 2.60]), SMFQ-C ( ± 0.84; 90% CI [−1.94, 1.56]), or SMFQ-P ( ± 0.62; 90% CI [−2.19, 1.08]).

3.2.2. Diagnosis-specific outcomes Differences in outcomes for the three principal inclusion anxiety

diagnoses were examined using multiple logistic regressions for diag- nostic outcomes, with lack of diagnosis as the reference. p-LGM models

were estimated for all symptom measures, and both types of analyses were controlled for age, gender, and pre-treatment CSR of the partici- pant’s principal disorder. The ORs for loss of the respective principal anxiety diagnoses at long-term follow-up were: SOP, OR = 0.27 (95% CI [0.15, 0.51], p < 0.01); SAD, OR = 1.88 (95% CI [0.99, 3.57], p = 0.11); and GAD, OR = 4.03 (95% CI [1.54, 10.56], p = 0.02). At post-treatment, the corresponding values were: SOP, OR = 0.63 (95% CI [0.35, 1.13], p = 0.19); SAD, OR = 0.60 (95% CI [0.33, 1.13], p = 0.18); and GAD, OR = 4.44 (95% CI [2.07, 9.52], p < 0.001).

With GAD as the reference, logistic regression analyses estimated that the OR for loss of all inclusion anxiety diagnoses at long-term follow-up was not statistically significant neither for SOP (OR = 0.47, 95% CI [0.21, 1.05], p = 0.12) nor for SAD (OR = 0.60, 95% CI [0.26, 1.40], p = 0.32). Following the same analysis procedure, the OR for loss of the principal diagnosis at long-term follow-up was lower for SOP (OR = 0.16, 95% CI [0.06, 0.44], p < 0.01), but not for SAD (OR = 0.47, 95% CI [0.16, 1.34], p = 0.23). There were no significant differences in the measures of anxiety (SCAS-C/P) and depressive (SMFQ-C/P) symptom reduction between SAD and SOP, compared to GAD, at long-term follow-up (data not shown).

4. Discussion

The present study is the first to assess the long-term effectiveness of a CBT protocol for youth with mixed anxiety disorders (SAD, SOP, and/ or GAD) delivered in community mental health clinics. Our findings confirmed loss of all inclusion anxiety diagnoses in 53% of participants, and loss of the principal anxiety diagnosis in 63%. Significant

Table 3 Severity ratings and symptom measures at post-treatment and long-term follow-up. Main effects of group and time and group by time interaction.

Measure All patients ICBT (n = 70) GCBT (n = 69) Pre-post Post-LTFU

Mean (SD) Mean (SD) Mean (SD) β z β z

CSR, principal anxiety diagnosis Pre-treatment (n = 139) 6.90 (1.14) 6.93 (1.09) 6.87 (1.19) G 0.05 0.19 −0.12 −0.33 Post-treatment 4.72 (2.37) 4.74 (2.47) 4.71 (2.28) T −0.31 −10.57*** −0.05 −8.85***

LTFU 2.29 (3.10) 2.16 (3.11) 2.43 (3.11) I 0.02 0.31 0.00 −0.31

CSR, secondary anxiety diagnosis Pre-treatment (n = 99) 6.31 (1.07) 6.23 (1.01) 6.39 (1.12) G 0.05 0.18 −0.22 −0.61 Post-treatment 4.01 (2.47) 4.15 (2.55) 3.88 (2.41) T −0.13 −3.64*** −0.05 −6.24***

LTFU 1.27 (2.63) 0.92 (2.14) 1.61 (2.99) I 0.06 0.90 −0.01 −0.81

CSR, tertiary anxiety diagnosis Pre-treatment (n = 45) 5.56 (1.22) 5.36 (1.09) 5.74 (1.32) G −0.10 −0.23 0.10 0.16 Post-treatment 3.96 (2.41) 4.05 (2.30) 3.87 (2.56) T −0.13 −3.64*** −0.05 −6.24***

LTFU 1.44 (2.81) 1.45 (2.84) 1.43 (2.84) I 0.01 0.34 0.00 −0.31

SCAS-C Pre-treatment (n = 131) 36.77 (16.87) 36.19 (15.97) 37.73 (17.87) G −1.34 −0.54 0.35 0.16 Post-treatment 27.65 (15.52) 27.23 (16.39) 28.07 (14.72) T −1.43 −6.58*** 0.09 −2.43*

LTFU 24.15 (16.00) 24.9 (14.13) 23.39 (17.78) I 0.06 0.15 0.06 0.85

SCAS-P Pre-treatment (n = 133) 35.28 (12.28) 34.78 (13.35) 35.83 (11.08) G 0.23 0.12 1.45 0.70 Post-treatment 27.58 (13.64) 28.15 (15.06) 27.02 (12.17) T −1.24 −7.24*** −0.15 −6.37***

LTFU 21.63 (13.26) 22.32 (16.69) 20.93 (13.90) I 0.06 0.15 0.06 0.85

SMFQ-C Pre-treatment (n = 128) 7.43 (5.40) 7.33 (5.19) 7.54 (5.65) G −0.23 −0.30 −0.03 −0.03 Post-treatment 5.97 (5.34) 5.90 (5.10) 6.04 (5.60) T −0.22 −2.95** −0.01 −0.62 LTFU 5.66 (5.93) 5.76 (5.67) 5.57 (6.22) I −0.03 −0.16 0.00 0.13

SMFQ-P Pre-treatment (n = 134) 7.50 (5.06) 7.43 (4.64) 7.57 (5.51) G 0.37 0.71 0.68 0.93 Post-treatment 5.49 (5.49) 5.95 (5.06) 5.03 (4.76) T −0.30 −4.34*** −0.03 −2.80**

LTFU 4.35 (4.35) 4.57 (5.25) 4.13 (4.82) I 0.33 2.03 −0.03 −1.70

Note. CSR = Clinical Severity Rating; G = Group; I = Interaction (all results obtained from Latent Growth Modeling in Mplus); LTFU = Long-Term Follow-Up; SCAS = Spence Children’s Anxiety Scale, C = Child, P = Parent; SMFQ = Short Mood and Feelings Questionnaire, C = Child, P = Parent; T = Time.

* p < 0.05. ** p < 0.01. *** p < 0.001.

Table 4 Comparison of clinical improvement rates for principal diagnosis from pre- to post- treatment and long-term follow-up.

Status at long-term follow-up

Status at post- treatment

Recovery Response No change Deterioration Total

Recovery 43 4 6 3 56 Response 17 5 4 0 26 No change 26 8 18 1 53 Deterioration 1 0 2 1 4 Total 87 17 30 5 139

Note. Definition of the four categories relates to change in the clinical severity rating (CSR) of the principal diagnosis at post-treatment and at long-term follow-up in com- parison with the pre-treatment CSR score. Definitions: deterioration = CSR increased by ≥ 2 points; no change = CSR changed ± 1 point; response = CSR decreased by ≥ 2 points; recovery = CSR decreased by ≥ 2 points and the score was ≤ 3.

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

64

improvement was also evident in the anxiety symptom measure, but not in youth-rated depressive symptoms. No differences in diagnostic and symptom outcome measures were found between ICBT and GCBT. Furthermore, our results indicated that, compared to GAD, the chance of loss of the principal diagnosis was significantly lower for youth with a principal diagnosis of SOP at inclusion.

Analysis of diagnostic treatment outcomes indicated a significant improvement from post-treatment to long-term follow-up. Almost 50% of youth that retained their principal and/or all inclusion anxiety di- agnoses at post-treatment lost these diagnoses at long-term follow-up. Also, significant reduction in anxiety symptoms was evident. Analysis of CSR response of youth's principal diagnosis between post-treatment and long-term follow-up, generally indicated maintenance or im- provement of the initial response at long-term follow-up. Thus, the results largely confirm improvement during the follow-up period. In comparison, other long-term outcome studies confirm maintenance of post-treatment results but not improvement (Ginsburg et al., 2014; Kendall et al., 2004; Saavedra et al., 2010). Several explanations for this continued improvement may apply. Outcomes at post-treatment were in the lower range compared to other effectiveness and efficacy trials (Bodden et al., 2008; James, James, Cowdrey, Soler, & Choke, 2015; Lau et al., 2010; Warwick et al., 2017), leaving more room for im- provement. Furthermore, clinical severity rating (CSR) of the principal anxiety diagnosis, and anxiety symptom change demonstrated steeper reductions from pre to- post, compared to post to long-term follow-up. Thus, the improvements may also relate to a delayed treatment effect, which may stem from a prolonged consolidation of acquired skills among youth and their parents (Ishikawa, Okajima, Matsuoka, & Sakano, 2007). Of notice; the treatment program used in this study consisted of 10 weekly sessions, whereas most other programs have 12–16 sessions (Bodden et al., 2008; Kendall et al., 1997). Thus, available time to conduct important CBT components such as exposure in this treatment protocol was limited. One may speculate whether youth had more time to implement and apply these skills after treat- ment, resulting in improved diagnostic outcomes. A final explanation may also be the effect of some spontaneous recovery among youth with anxiety disorders, as indicated by Nevo et al. (2014).

Our expectation that the long term outcomes in our effectiveness study would be below long term outcomes in efficacy studies was not consistently supported. In previous long-term efficacy studies reporting loss of all inclusion anxiety disorders, outcomes ranged from 47% to 86%, with an average mean follow-up of 9.5 years post-treatment and a weighted mean outcome of 53% (Barrett et al., 2001; Benjamin et al., 2013; Ginsburg et al., 2014). Our study result of 53% falls within the cited range and is consistent with the mean outcome from these studies. Regarding loss of the principal diagnosis, previous efficacy studies on long-term outcomes reported rates ranging between 65% and 90% (Ginsburg et al., 2014; Kendall et al., 2004), whereas the 63% loss of principal diagnosis in our study falls below the reported range. For long-term outcomes in youth with any anxiety disorder (with the ex- ception of post-traumatic stress disorder and obsessive-compulsive disorder), treated with a broad scope of CBT formats, loss of all inclu- sion anxiety diagnoses has been reported with a mean outcome at 57% (range 47–68%), and loss of the principal anxiety diagnosis with a mean outcome of 77% (range 48–93%) (Gibby et al., 2017). The present re- sults fall within these ranges, yet below the mean outcomes.

Several reasons might explain why the long-term improvement rates obtained in the present study are somewhat lower, compared to pre- vious efficacy studies. Compared to patients from university clinics, youth with anxiety disorders treated in community clinics have been reported to have higher levels of anxiety symptoms and higher severity ratings of their principal anxiety diagnosis (Villabø, Cummings, Gere, Torgersen, & Kendall, 2013). In addition, in our sample, severity ratings at baseline for the principal diagnosis were greater than the CSR values reported in most other published studies in the field (mean CSR = 7.01) (Wergeland et al., 2014). Lower severity at pre-treatment

may reflect less severe anxiety disorders, which are more amenable to therapy and thus remain in remission once effectively treated (Ginsburg et al., 2011). Moreover, 46% of our study participants presented with a principal diagnosis of SOP, which is more than the percentage of youth with SOP included in other long-term follow-up studies, e.g., 27.3% in Benjamin et al. (2013) and 21.2% in Barrett et al. (2001). Several previous studies of generic CBT protocols have associated the diagnosis of SOP with poorer treatment outcomes (Hudson et al., 2015; Reynolds et al., 2012). Finally, whereas university clinics are dedicated to clinical research, the primary mandate of community clinics is the provision of healthcare services to the community. Thus, therapists in research clinics commonly have more extensive training in the particular treat- ments provided and usually have more focused caseloads, thus allowing for the development of greater competency in delivering the specific treatment (Weisz, Krumholz, Santucci, Thomassin, & Ng, 2015). Most therapists who participated in our study had little or no experience in CBT prior to their participation. This may also have contributed to the lower long-term improvement rates obtained.

Our study showed no differences between ICBT and GCBT in long- term outcomes. This finding is in line with findings from meta-analyses of studies on short-term outcomes (In-Albon & Schneider, 2006; Silverman et al., 2008), as well as the long-term outcome efficacy study by Saavedra et al. (2010). Of interest on this point, comparable out- comes were also reported for ICBT and family CBT (Kendall, Hudson, Gosch, Flannery-Schroeder, & Suveg, 2008). Our results add to the re- search literature by demonstrating the long-term outcomes of CBT for anxiety disorders in a community mental health setting and by in- cluding an analysis of equivalence. Equivalency between the two treatment formats of ICBT and GCBT was established in terms of change in CSR for the principal anxiety diagnosis, although not for the other diagnostic and symptom outcome measures. Thus, our study findings partially confirm our prediction of equivalence between treatment outcomes for ICBT versus GCBT.

Our results indicated that the chance of recovery, i.e. loss of the principal diagnosis, was significantly lower for youth with a principal diagnosis of SOP. This is in line with findings reported by Hudson et al. (2015) and Crawley et al. (2008) demonstrating lower post-treatment outcomes following CBT in youth with SOP, compared to those with GAD and SAD. Interestingly, the ORs for recovery varied between post- treatment and long-term follow-up, depending on the pre-treatment principal diagnosis. Whereas youth with GAD and SAD showed in- creased odds for loss of principal diagnosis from post-treatment to long- term follow-up, SOP carried a lower chance of recovery during this period. This result was not affected by gender, age, or pre-treatment CSR of the principal diagnosis. Thus, despite an initial positive response to treatment, the post-treatment outcome in participants with a prin- cipal diagnosis of SOP waned over time. This finding is in agreement with that of Kerns et al. (2013). Hudson et al. (2015) suggested that generic CBT protocols might not be adequate to address the more specific characteristics associated with SOP, such as negative self- statements and social expectations (Spence & Rapee, 2016). This could mean that youth with SOP need more extensive treatment, allowing for more sessions of individualized exposure and consolidation of acquired skills than what was possible with the current protocol. In this regard, a shift towards more exposure at an earlier time during the course of treatment may be conducive to outcome improvement (Ale, McCarthy, Rothschild, & Whiteside, 2015).

The current study features notable strengths. It is the largest study to date to examine the effectiveness of a CBT protocol in youth with mixed anxiety disorders in a community mental health setting. Participants were routine referrals to community clinics, and only few exclusion criteria were applied. Recruitment, assessment, and treatment were undertaken by clinicians working at the participating clinics. The study achieved a high rate of participation, a low rate of missing data and the inclusion of four assessment points allowed for the use of ad- vanced statistical models. Taken together, these factors contribute to a

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

65

high degree of generalizability to other community mental health clinics.

The present study also has limitations. First, the use of the ADIS-C/P was limited to the SAD, SOP and GAD modules. Assessment of co- morbidity was based on the DAWBA interview, which was used only at inclusion (see Wergeland et al., 2014, for further details). Conse- quently, differential diagnoses and comorbidity based on the full ADIS- C/P could not be assessed, and the development of comorbid disorders through the trial and follow-up could therefore not be tracked. Func- tional outcomes were not included. The current assessment was cross- sectional, providing a brief glimpse of current symptoms and impair- ment. Although the study included a one-year follow-up assessment, data on the remission or recurrence of symptoms and diagnoses during follow-up was not available.

Another limitation of our study is the lack of a control group. This is an inherent limitation in most long-term follow-up studies, which prohibits causal associations between treatment and long-term out- come, given the influence of several likely confounding variables (Nevo & Manassis, 2009; Rith-Najarian et al., 2017). In light of this, we cannot exclude spontaneous remission or maturational effects as possible contributing factors to the continued outcome improvements reported in our long-term follow-up study.

The study took place in Norway, which may influence the gen- eralizability of the results. Whereas community mental health care services are free of charge for youth in Norway, many university based research clinics charge a service-fee based on household income (Villabø et al., 2013). This may contribute to systematic differences in who chooses to seek treatment and comparability between studies. Typical of many Norwegian community samples, the sample population was primarily Caucasian and the findings may therefore not apply to other ethnic groups (Nilsen, Eisemann, & Kvernmo, 2013). Norway is also commonly characterized by high living standards and fairly homogenous economic and social stratification (Heiervang et al., 2007). Thus the findings may not apply to youth groups within other socio-economic strata.

In conclusion, our study demonstrates long term effectiveness of CBT for mixed anxiety disorders in youth. These findings provide en- couragement for the wider use and implementation of CBT in com- munity mental health clinics. Furthermore, the comparable outcomes for ICBT and GCBT allow practitioners flexibility in the choice of treatment format when managing youth with SAD, SOP and/or GAD. Choice of treatment format is not dependent upon youth clinical fac- tors, but can be based on patient or parent preferences, community clinic resources or referral rates. While a community setting differs from a university clinic setting in many respects, and thus can be assigned responsibility for lower treatment outcomes, from a long-term per- spective, our study results indicate that the impact of these differences on treatment outcomes might be overestimated. However, while our findings lend support to the suitability of the treatment approach in the community setting, a significant number of youth in our study still met the criteria for an anxiety diagnosis at long-term follow-up. In con- junction, poorer outcomes for youth with a principal diagnosis of SOP were found. Identification of long-term outcome predictors and mod- erators to allow for more individual tailoring of treatment for youth with a principal diagnosis of SOP in particular, but also for youth with mixed anxiety disorders in general, is an important subject for future research.

Conflicts of interest

All authors declare no conflicts of interest.

Acknowledgments

This work was supported by the Western Norway Regional Health Authority [grant number 911366, 911253, and 911840]. We are

grateful to all the hardworking clinicians and administrative staff at the participating mental health clinics. We also express our deep gratitude to all the youth and their families who participated in the study and helped make this research possible.

References

Ale, C. M., McCarthy, D. M., Rothschild, L. M., & Whiteside, S. P. (2015). Components of cognitive behavioral therapy related to outcome in childhood anxiety disorders. Clinical Child and Family Psychology Review, 18, 240–251.

Angold, A., Costello, E. J., Messer, S. C., & Pickles, A. (1995). Development of a short questionnaire for use in epidemiological studies of depression in children and ado- lescents. International Journal of Methods in Psychiatric Research, 5, 237–249.

Baldwin, S. A., Bauer, D. J., Stice, E., & Rohde, P. (2011). Evaluating models for partially clustered designs. Psychological Methods, 16, 149–165.

Barrett, P. M., Dadds, M. R., & Rapee, R. M. (1991). Coping koala workbook. Nathan, Australia: School of Applied Psychology. Griffith University Unpublished manuscript.

Barrett, P. M., Duffy, A. L., Dadds, M. R., & Rapee, R. M. (2001). Cognitive-behavioral treatment of anxiety disorders in children: Long-term (6-year) follow-up. Journal of Consulting and Clinical Psychology, 69, 135–141.

Barrett, P. M. (2005). FRIENDS for life: Group leaders manual for youth. Brisbane: Australian Academic Press.

Barrington, J., Prior, M., Richardson, M., & Allen, K. (2005). Effectiveness of CBT versus standard treatment for childhood anxiety disorders in a community clinic setting. Behaviour Change, 22, 29–43.

Benjamin, C. L., Harrison, J. P., Settipani, C. A., Brodman, D. M., & Kendall, P. C. (2013). Anxiety and related outcomes in young adults 7 to 19 years after receiving treatment for child anxiety. Journal of Consulting and Clinical Psychology, 81, 865–876. http:// dx.doi.org/10.1037/a0033048.

Bjaastad, J. F., Haugland, B. S. M., Fjermestad, K. W., Torsheim, T., Havik, O. E., Heiervang, E. R., et al. (2016). Competence and Adherence Scale for Cognitive Behavioral Therapy (CAS-CBT) for anxiety disorders in youth: Psychometric prop- erties. Psychological Assessment, 28, 908–916. http://dx.doi.org/10.1037/ pas0000230.

Bodden, D. H., Bögels, S. M., Nauta, M. H., De Haan, E., Ringrose, J., Appelboom, C., et al. (2008). Child versus family cognitive-behavioral therapy in clinically anxious youth: An efficacy and partial effectiveness study. Journal of the American Academy of Child & Adolescent Psychiatry, 47, 1384–1394.

Brown, T. A., Barlow, D. H., & Di Nardo, P. A. (1994). Anxiety Disorders Interview Schedule for DSM-IV (ADIS-IV): Client interview schedule. NY: Graywind Publications Incorporated.

Brown, T. A., Di Nardo, P. A., Lehman, C. L., & Campbell, L. A. (2001). Reliability of DSM- IV anxiety and mood disorders: Implications for the classification of emotional dis- orders. Journal of Abnormal Psychology, 110, 49–58.

Chambless, D. L., & Hollon, S. D. (1998). Defining empirically supported therapies. Journal of Consulting and Clinical Psychology, 66, 7–18.

Chorpita, B. F., Weisz, J. R., Daleiden, E. L., Schoenwald, S. K., Palinkas, L. A., Miranda, J., et al. (2013). Long-term outcomes for the Child STEPs randomized effectiveness trial: A comparison of modular and standard treatment designs with usual care. Journal of Consulting and Clinical Psychology, 81, 999–1009. http://dx.doi.org/10. 1037/a0034200.

Copeland, W. E., Angold, A., Shanahan, L., & Costello, E. J. (2014). Longitudinal patterns of anxiety from childhood to adulthood: The Great Smoky Mountains Study. Journal of the American Academy of Child & Adolescent Psychiatry, 53, 21–33.

Costello, E. J., & Angold, A. (1988). Scales to assess child and adolescent depression: Checklists, screens, and nets. Journal of the American Academy of Child & Adolescent Psychiatry, 27, 726–737.

Crawley, S. A., Beidas, R. S., Benjamin, C. L., Martin, E., & Kendall, P. C. (2008). Treating socially phobic youth with CBT: Differential outcomes and treatment considerations. Behavioural and Cognitive Psychotherapy, 36, 379–389. http://dx.doi.org/10.1017/ S1352465808004542.

Currie, C., Molcho, M., Boyce, W., Holstein, B., Torsheim, T., & Richter, M. (2008). Researching health inequalities in adolescents: The development of the Health Behavior in School-Aged Children (HBSC) and family affluence scale. Social Science Medicine, 66, 1429–1436.

Gibby, B. A., Casline, E. P., & Ginsburg, G. S. (2017). Long-term outcomes of youth treated for an anxiety disorder: A critical review. Clinical Child and Family Psychology Review, 20, 201–225. http://dx.doi.org/10.1007/s10567-017-0222-9.

Ginsburg, G. S., Kendall, P. C., Sakolsky, D., Compton, S. N., Piacentini, J., Albano, A. M., et al. (2011). Remission after acute treatment in children and adolescents with an- xiety disorders: Findings from the CAMS. Journal of Consulting and Clinical Psychology, 79, 806–813.

Ginsburg, G. S., Becker, E. M., Keeton, C. P., Sakolsky, D., Piacentini, J., Albano, A. M., et al. (2014). Naturalistic follow-up of youths treated for pediatric anxiety disorders. JAMA Psychiatry, 71, 310–318. http://dx.doi.org/10.1001/jamapsychiatry.2013. 4186.

Heiervang, E., Stormark, K. M., Lundervold, A. J., Heimann, M., Goodman, R., Posserud, M. B., et al. (2007). Psychiatric disorders in Norwegian 8-to 10-year-olds: An epi- demiological survey of prevalence, risk factors, and service use. Journal of the American Academy of Child & Adolescent Psychiatry, 46(4), 438–447.

Higa-McMillan, C. K., Francis, S. E., Rith-Najarian, L., & Chorpita, B. F. (2016). Evidence base update: 50 years of research on treatment for child and adolescent anxiety. Journal of Clinical Child & Adolescent Psychology, 45(2), 91–113.

Hudson, J. L., Rapee, R. M., Lyneham, H. J., McLellan, L. F., Wuthrich, V. M., &

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

66

Schniering, C. A. (2015). Comparing outcomes for children with different anxiety disorders following cognitive behavioural therapy. Behaviour Research and Therapy, 72, 30–37. http://dx.doi.org/10.1016/j.brat.2015.06.007.

Hunsley, J., & Lee, C. M. (2007). Research-informed benchmarks for psychological treatments: Efficacy studies, effectiveness studies, and beyond. Professional Psychology: Research and Practice, 38, 21–33.

In-Albon, T., & Schneider, S. (2006). Psychotherapy of childhood anxiety disorders: A meta-analysis. Psychotherapy and Psychosomatics, 76, 15–24.

Ishikawa, S. I., Okajima, I., Matsuoka, H., & Sakano, Y. (2007). Cognitive behavioural therapy for anxiety disorders in children and adolescents: A meta-analysis. Child and Adolescent Mental Health, 12(4), 164–172. http://dx.doi.org/10.1111/j.1475-3588. 2006.00433.x.

Jacobsen, N. S., & Truax, P. (1991). Clinical Significance: A statistical approach to de- fining meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59, 12–19.

James, A. C., James, G., Cowdrey, F. A., Soler, A., & Choke, A. (2013). Cognitive beha- vioural therapy for anxiety disorders in children and adolescents. Cochrane Database of Systematic Reviews, 6, CD004690. http://dx.doi.org/10.1002/14651858. CD004690.pub3.

James, A. C., James, G., Cowdrey, F. A., Soler, A., & Choke, A. (2015). Cognitive beha- vioural therapy for anxiety disorders in children and adolescents. Cochrane Database Of Systematic Reviews, 2.

Kendall, P. C., & Ollendick, T. H. (2004). Setting the research and practice agenda for anxiety in children and adolescence: A topic comes of age. Cognitive and Behavioral Practice, 11, 65–74. http://dx.doi.org/10.1016/S1077-7229(04)80008-7.

Kendall, P. C., & Southam-Gerow, M. A. (1996). Long-term follow-up of a cognitive–be- havioral therapy for anxiety-disordered youth. Journal of Consulting and Clinical Psychology, 64, 724–730. http://dx.doi.org/10.1037/0022-006X.64.4.724.

Kendall, P. C., Flannery-Schroeder, E., Panichelli-Mindel, S. M., Southam-Gerow, M., Henin, A., & Warman, M. (1997). Therapy for youths with anxiety disorders: A second randomized clincal trial. Journal of Consulting and Clinical Psychology, 65(3), 366–380. http://dx.doi.org/10.1037/0022-006X.65.3.366.

Kendall, P. C., Safford, S., Flannery-Schroeder, E., & Webb, A. (2004). Child anxiety treatment: Outcomes in adolescence and impact on substance use and depression at 7.4-year follow-up. Journal of Consulting and Clinical Psychology, 72, 276–287. http:// dx.doi.org/10.1037/0022-006X.72.2.276.

Kendall, P. C., Hudson, J. L., Gosch, E., Flannery-Schroeder, E., & Suveg, C. (2008). Cognitive-behavioral therapy for anxiety disordered youth: A randomized clinical trial evaluating child and family modalities. Journal of Consulting & Clinical Psychology, 76. http://dx.doi.org/10.1037/0022-006x.76.2.282.

Kendall, P. C. (1990). The C.A.T project therapist manual. Ardmore, PA: Workbook Publishing.

Kerns, C. M., Read, K. L., Klugman, J., & Kendall, P. C. (2013). Cognitive behavioral therapy for youth with social anxiety: Differential short and long-term treatment outcomes. Journal of Anxiety Disorders, 27, 210–215. http://dx.doi.org/10.1016/j. janxdis.2013.01.009.

Lau, W., Chan, C. K., Li, J. C., & Au, T. K. (2010). Effectiveness of group cognitive-be- havioral treatment for childhood anxiety in community clinics. Behaviour Research and Therapy, 48, 1067–1077.

Muthén, L. K., & Muthén, B. O. (1998–2017). Mplus User’s Guide. (8th ed.). Los Angeles, CA: Muthén & Muthén.

Nauta, M. H., Scholing, A., Emmelkamp, P. M., & Minderaa, R. B. (2001). Cognitive be- havioural therapy for anxiety disordered children in a clinical setting: Does addi- tional cognitive parent training enhance treatment effectiveness? Clinical Psychology & Psychotherapy, 8, 330–340.

Nauta, M. H., Scholing, A., Emmelkamp, P. M., & Minderaa, R. B. (2003). Cognitive- behavioral therapy for children with anxiety disorders in a clinical setting: No ad- ditional effect of a cognitive parent training. Journal of the American Academy of Child & Adolescent Psychiatry, 42, 1270–1278.

Nevo, G. A., & Manassis, K. (2009). Outcomes for treated anxious children: A critical review of long-term-follow-up studies. Depression and Anxiety, 26, 650–660. http:// dx.doi.org/10.1002/da.20584.

Nevo, A., Gili, W., Avery, D., Fiksenbaum, L., Kiss, A., Mendlowitz, S., et al. (2014). Eight years later: Outcomes of CBT-treated versus untreated anxious children. Brain and Behavior, 4(5), 765–774. http://dx.doi.org/10.1002/brb3.274.

Nilsen, T. S., Eisemann, M., & Kvernmo, S. (2013). Predictors and moderators of outcome in child and adolescent anxiety and depression: A systematic review of psychological treatment studies. European Child & Adolescent Psychiatry, 22(2), 69–87. http://dx. doi.org/10.1007/s00787-012-0316-3.

Puleo, C. M., Conner, B. T., Benjamin, C. L., & Kendall, P. C. (2011). CBT for childhood anxiety and substance use at 7.4-year follow-up: A reassessment controlling for known predictors. Journal of Anxiety Disorders, 25(5), 690–696. http://dx.doi.org/10. 1016/j.janxdis.2011.03.005.

Reynolds, S., Wilson, C., Austin, J., & Hooper, L. (2012). Effects of psychotherapy for

anxiety in children and adolescents: A meta-analytic review. Clinical Psychology Review, 32, 251–262.

Rith-Najarian, L. R., Park, A. L., Wang, T., Etchison, A. I., Chavira, D. A., & Chorpita, B. F. (2017). Applying new evidence standards to youth cognitive behavioral therapies: A review. Behaviour Research and Therapy, 90, 147–158. http://dx.doi.org/10.1016/j. brat.2016.12.011.

Rogers, J. L., Howard, K. I., & Vessey, J. T. (1993). Using significance tests to evaluate equivalence between two experimental groups. Psychological Bulletin, 113, 553–565.

Saavedra, L. M., Silverman, W. K., Morgan-Lopez, A. A., & Kurtines, W. M. (2010). Cognitive behavioral treatment for childhood anxiety disorders: Long-term effects on anxiety and secondary disorders in young adulthood. Journal of Child Psychology and Psychiatry, 51, 924–934.

Santucci, L. C., Thomassin, K., Petrovic, L., & Weisz, J. R. (2015). Building evidence based interventions for the youth, providers, and contexts of real world mental health care. Child Development Perspectives, 9, 67–73. http://dx.doi.org/10.1111/cdep.12118.

Sharp, C., Goodyer, I. M., & Croudace, T. J. (2006). The Short Mood and Feelings Questionnaire (SMFQ): A unidimensional item response theory and categorical data factor analysis of self-report ratings from a community sample of 7- through 11-year- old children. Journal of Abnormal Child Psychology, 34, 365–377.

Silverman, W. K., & Albano, A. M. (1996). Anxiety Disorders Interview Schedule (ADIS-IV) parent interview schedules. Albany, NY: Greywind Publications.

Silverman, W. K., & Kurtines, W. M. (1996). Anxiety and phobic disorders: A pragmatic approach. NY: Springer Science & Business Media.

Silverman, W. K., & Nelles, W. B. (1988). The anxiety disorders interview schedule for children. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 772–778.

Silverman, W. K., Saavedra, L. M., & Pina, A. A. (2001). Test-retest reliability of anxiety symptoms and diagnoses with the anxiety disorders interview schedule for DSM-IV: Child and parent versions. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 937–944.

Silverman, W. K., Pina, A. A., & Viswesvaran, C. (2008). Evidence-based psychosocial treatments for phobic and anxiety disorders in children and adolescents. Journal of Clinical Child and Adolescent Psychology, 37, 105–130.

Smith, M. M., McLeod, B. D., Southam-Gerow, M. A., Jensen-Doss, A., Kendall, P. C., & Weisz, J. R. (2017). Does the delivery of CBT for youth anxiety differ across research and practice settings? Behavior Therapy, 48, 501–516. http://dx.doi.org/10.1016/j. beth.2016.07.004.

Spence, S. H., & Rapee, R. M. (2016). The etiology of social anxiety disorder: An evidence- based model. Behaviour Research and Therapy, 86, 50–67. http://dx.doi.org/10.1016/ j.brat.2016.06.007.

Spence, S. H., Barrett, P. M., & Turner, C. M. (2003). Psychometric properties of the Spence Children’s Anxiety Scale with young adolescents. Journal of Anxiety Disorders, 17, 605–625. http://dx.doi.org/10.1016/S0887-6185(02)00236-0.

Spence, S. H. (1998). A measure of anxiety symptoms among children. Behaviour Research and Therapy, 36, 545–566. http://dx.doi.org/10.1016/S0005-7967(98)00034-5.

Villabø, M. A., Cummings, C. M., Gere, M. K., Torgersen, S., & Kendall, P. C. (2013). Anxious youth in research and service clinics. Journal of Anxiety Disorders, 27, 16–24. http://dx.doi.org/10.1016/j.janxdis.2012.09.003.

Wang, J., & Wang, X. (2012). Structural equation modeling: Applications using Mplus. John Wiley & Sonshttp://dx.doi.org/10.1002/9781118356258.

Warwick, H., Reardon, T., Cooper, P., Murayama, K., Reynolds, S., Wilson, C., et al. (2017). Complete recovery from anxiety disorders following cognitive behavior therapy in children and adolescents: A meta-analysis. Clinical Psychology Review, 52, 77–91. http://dx.doi.org/10.1016/j.cpr.2016.12.002.

Weisz, J. R., Kuppens, S., Eckshtain, D., Ugueto, A. M., Hawley, K. M., & Jensen-Doss, A. (2013). Performance of evidence-based youth psychotherapies compared with usual clinical care: A multilevel meta-analysis. JAMA Psychiatry, 70, 750–761.

Weisz, J. R., Krumholz, L. S., Santucci, L., Thomassin, K., & Ng, M. Y. (2015). Shrinking the gap between research and practice: Tailoring and testing youth psychotherapies in clinical care contexts. Annual Review of Clinical Psychology, 11, 139–163. http://dx. doi.org/10.1146/annurev-clinpsy-032814-112820.

Wergeland, G. J., Fjermestad, K. W., Marin, C. E., Haugland, B. S.-M., Bjaastad, J. F., Oeding, K., et al. (2014). An effectiveness study of individual vs. group cognitive behavioral therapy for anxiety disorders in youth. Behaviour Research and Therapy, 57, 1–12. http://dx.doi.org/10.1016/j.brat.2014.03.007.

Wolk, C. B., Kendall, P. C., & Beidas, R. S. (2015). Cognitive-behavioral therapy for child anxiety confers long-term protection from suicidality. Journal of the American Academy of Child & Adolescent Psychiatry, 54(3), 175–179. http://dx.doi.org/10. 1016/j.jaac.2014.12.004.

Wothke, W. (2000). Longitudinal and multigroup modeling with missing data. In T. D. Little, K. U. Schnabel, & J. Baumert (Eds.). Modeling longitudinal and multilevel data (pp. 269–281). Mahwah, NJ: Lawrence Erlbaum. http://dx.doi.org/10.4324/ 9781410601940.

A. Kodal et al. Journal of Anxiety Disorders 53 (2018) 58–67

67