Predictors and moderators of treatment outcome of child-focused individual and group cognitive behavioral therapy for childhood anxiety disorders

Predictors and moderators of treatment outcome

of child-focused individual and group cognitive behavioral therapy

for childhood anxiety disorders

F. Drenthe

Master Thesis

Graduate School of Child Development and Education University of Amsterdam

Supervisor: Drs. L. Jongerden Second supervisor: Prof. Dr. S. M. Bögels

Predictors and moderators of treatment outcome

of child-focused individual and group cognitive behavioral therapy

for childhood anxiety disorders

F. Drenthe

Master Thesis

Graduate School of Child Development and Education University of Amsterdam

Supervisor: Drs. L. Jongerden

Second supervisor: Prof. Dr. S. M. Bögels Amsterdam, November 2013

Contents Abstract (English)………. 2 Abstract (Dutch/Nederlands)………... 2 Introduction………... 3 Methods……….. 8 Participants……….. 8 Materials……….. 11

Potential predictors and moderators……… 11

Treatment………. 11 Procedure………. 12 Statistical analysis……… 12 Results………. 14 Preliminary analysis………. 14 Predictor/moderator analysis……… 14

Explorative post hoc analyses ………. 16

Discussion……… 19

Abstract (English)

Objective: This study investigated predictors and moderators of treatment outcome of

child-focused individual and group cognitive behavioral therapy (CBT) for childhood anxiety disorders. Age, gender, social phobia and treatment format were examined. Method:

One-hundred nineteen clinically referred children (8 – 18 years) with anxiety disorders were assigned to individual CBT (ICBT) or group CBT (GCBT) and were assessed with a self-report measure of anxiety (SCARED-71) before, after and three months after treatment. Results: ICBT and social phobia predicted more post treatment improvement. ICBT and GCBT were equally effective in the long term. No moderators were detected. Children with social phobia report significantly more overall anxiety before, after and three months after treatment than children without social phobia. Conclusion: In clinical practice children regardless of age, gender and social phobia benefit from ICBT and GCBT. More predictor and moderator research of treatment outcome for childhood anxiety disorders is needed. Key Words: Childhood anxiety disorders, cognitive-behavioral therapy, group treatment, moderators, predictors, social phobia.

Abstract (Dutch/Nederlands)

Onderwerp: Predictoren and moderatoren van behandeluitkomst van kindgerichte individuele en

groeps cognitieve gedragstherapie (CGT) voor kinderen met een angststoornis. Leeftijd, sekse, sociale fobie en behandelconditie werden onderzocht. Methode: Honderdnegentien kinderen (8 – 18 jaar) met angststoornissen en hun ouders kregen individuele CGT (ICGT) of groeps CGT (GCGT). Voor, na en drie maanden na de behandeling vulden zij een vragenlijst die

angstsymptomen meet (SCARED-71) in. Resultaten: ICGT and sociale fobie voorspelden meer verbetering op de korte termijn. ICBT and GCBT bleken op de lange termijn even effectief. Er werden geen moderatoren gevonden. Kinderen met social fobie rapporteerden significant meer angstsymptomen voor, na en drie maanden na de behandeling dan kinderen zonder sociale fobie.

Conclusie: In de klinische praktijk profiteren kinderen ongeacht leeftijd, sekse en sociale fobie

van ICGT en GCGT. Meer onderzoek naar predictoren en moderatoren van behandeluitkomst voor kinderen met een angststoornis is nodig. Trefwoorden: Angststoornissen, cognitieve-gedragstherapie, groepsbehandeling, moderatoren, predictoren, sociale fobie.

Introduction

Anxiety disorders are some of the most prevalent disorders of childhood and adolescence

(Cartwright-Hatton, McNicol, & Doubleday, 2006; Graaf, Have, & Dorsselaer, 2010; Kessler,

Berglund, Demler, Jin, & Walters, 2005; Verhulst, Ende, Ferdinand, & Kasius, 1997). The median age of onset for anxiety disorders is found to be as early as 11 years (Kessler et al., 2005). In the Netherlands, every year an estimated 5 to 25% adolescents between 13 and 18 years meet the criteria for an anxiety disorder (Verhulst et al., 1997). For children below 12 years of age, the prevalence varies internationally from 3.05 to 23.90 %, the outliers excluded

(Cartwright-Hatton et al., 2006). Anxiety disorders are characterized by persistent, excessive and unrealistic fear (American Psychiatric Association, 2005) and may result into various other problems such as depression (Beesdo et al., 2007; Woodward & Fergusson, 2001), alcohol and drug abuse (Zimmerman et al., 2003; Woodward & Fergusson, 2001) and educational

underachievement (Woodward & Fergusson, 2001). The treatment of anxiety disorders in children and adolescents is therefore of great importance.

From an evidence-based perspective, cognitive-behavioral therapy (CBT) is currently the treatment of choice for childhood anxiety disorders (Compton et al., 2004). Thorough research by means of so-called Randomized Control Trial designs (RCTs) has shown the effectiveness of CBT in children and adolescents with anxiety disorders in the short and long term (In-Albon & Schneider, 2007; Silverman, Pina, & Viswesvaran, 2008). In the Netherlands, the child-focused cognitive behavioral therapy protocol “Discussing + Doing = Daring” (DDD) for children with an anxiety disorder other than Obsessive Compulsory Disorder (OCD) or Post Traumatic Stress Disorder (PTSD) (Bögels, 2008) has been proven to be efficacious (Bodden et al., 2008). DDD is offered both as an individual and group therapy (Bögels, 2008). Individual CBT (ICBT) and group CBT (GCBT) appear to be equally effective (Compton et al., 2004; Flannery-Schroeder & Kendall, 2000; In-Albon & Schneider, 2007; James, Soler, & Weatherall, 2005; Liber et al., 2008; Manassis et al., 2002; Muris, Mayer, Bartelds, Tierney, & Bogie, 2001; Silverman et al., 2008). After ICBT, 62% of the participating children and adolescents no longer met the criteria of their primary anxiety disorder. After GCBT, this was 54% (Liber et al., 2008). But ICBT and GCBT are not equally effective in all children. After all, no study has reported a 100% success rate (In-Albon & Schneider, 2007; Silverman et al., 2008). Research has shown that 47% of the

children and adolescents are not free of their primary diagnosis immediately after treatment using DDD. After three months, this is still 27% of the children (Bodden et al., 2008). Moreover, this percentage might be slightly higher, based on the results of an implementation study that evaluates the real world effectiveness of the protocol (Jongerden, Bögels, & Peijnenburg, 2011).

Given the high prevalence (Graaf et al., 2010; Kessler et al., 2005; Verhulst et al., 1997), the many negative consequences that unsuccessfully treated anxiety disorders can have (Beesdo et al., 2007; Woodward & Fergusson, 2001; Zimmerman et al., 2003) and the fact that there are still young people with anxiety disorders who do not or insufficient benefit from CBT (Bodden et al., 2008; In-Albon & Schneider, 2007; Jongerden et al., 2011; Silverman et al., 2008), it is important to improve the effectiveness of CBT. One way to improve the effectiveness of CBT is to uncover pre-treatment demographic and clinical variables that are predictive to treatment outcome measures, in order to be able to select the treatment likely to yield the greatest efficacy based on a child’s baseline characteristics (Kraemer, Wilson, Fairburn, & Agras, 2002).

There is still much confusion about which pre-treatment variables affect the treatment outcome of CBT in childhood anxiety disorder (Compton et al., 2004; Rapee, Schniering, & Hudson, 2009). Pre-treatment variables can be divided into predictors and moderators. Predictors of treatment outcome are variables that are present before treatment and have a main effect on treatment outcome regardless of treatment condition. Treatment moderators are variables that are also present before treatment but have an interactive effect with treatment condition on treatment outcome. A predictor answers the question of which variable (for example age) makes children belong to a subgroup (for example below the age of 12 or above the age of 12) that responds more or less favorably to treatment (for example CBT) in general. A moderator answers the question of which variable (for example gender) makes children belong to a subgroup (for example boys or girls) that will respond differentially to one treatment (for example ICBT) over another treatment (for example GCBT). A predictor can be considered an indicator of general prognosis. A moderator can be considered an indicator with prescriptive value (Kraemer et al., 2002).

Until now, more attention has been paid to overall effectiveness of treatments than to potential predictors and moderators of treatment outcome. Although moderator analysis of treatment outcome is considered to be valuable, there has been little emphasis on this type of analysis following RCTs (Kraemer et al., 2002; Compton et al., 2004). Compton et al. (2004)

found in their review of CBT for children and adolescents with anxiety and depressive disorders only 10 of 21 studies that reported on further analysis of the role of demographic characteristics and clinical variables in treatment outcome. Only four studies were found to compare child-focused ICBT and GCBT for childhood anxiety disorder and thus were able to a greater or lesser extent, to report about potential moderators of child-focused ICBT versus GCBT. All four studies conclude that more research is needed with larger samples sizes in which child-focused ICBT and GCBT are directly compared, to detect possible main and interaction effects

(Flannery-Schroeder & Kendall, 2000; Liber et al., 2008; Manassis et al., 2002; Muris et al., 2001). Considerably more research examined predictors of treatment outcome of CBT for childhood anxiety disorders in general. Still, compared to adult literature, limited research on predictors of outcome in child anxiety is available (Rapee et al., 2003; Rodebaugh, Holaway, & Heimberg, 2004).

In addition, the available research results are inconclusive. Gender seemed to be no predictor of treatment outcome (Berman et al., 2000; Flannery-Schroeder & Kendall, 2000; Liber et al., 2008; Manassis et al., 2002; Muris et al., 2001; Rappe, 2000; Southam-Gerow et al., 2001). In contrast, Mendlowitz et al. (1999) found girls to report significantly less anxiety after GCBT than boys (Mendlowitz et al., 1999), suggesting gender may be a predictor of GCBT response. Age is found to be a predictor in some studies, with younger child age predicting better outcomes (Bodden et al., 2008, Southam-Gerow et al., 2001). But other studies concluded age to be no predictor (Berman et al., 2000; Flannery-Schroeder & Kendall, 2000; Liber et al., 2008; Manassis et al., 2002; Muris et al., 2001; Rapee, 2000).

The level of child internalizing psychopathology at pretreatment is found to be a possible predictor with higher levels predicting less favorable treatment response (Southam-Gerow et al., 2001) while level of child externalizing behavior problems did not predict treatment response (Rapee, 2000; Southam-Gerow et al., 2001). Some investigations conclude that comorbid anxiety disorders or comorbid non-anxiety disorders do not predict treatment outcome (Kendall, Brady, & Verduin 2001; Rapee, 2003; Southam-Gerow et al., 2001). Berman et al. (2000) also found that comorbidity for other anxiety disorders and/or externalizing disorders were not significant predictors but comorbidity for depression did predict less favorable treatment outcome.

Also, the severity of the anxiety disorder may be a predictor (Layne et al., 2003; Liber et al., 2010; Southam-Gerow et al., 2001). Especially children's levels of trait anxiety might predict less favorable treatment outcome (Berman et al., 2000).

Finally, many parent and family characteristics seemed to be related to treatment outcome. Family dysfunction (Crawford & Manassis, 2001), low level of family cohesion (Victor, Bernat, Bernstein, & Layne, 2007), high levels of maternal emotional warmth (Liber et al., 2010), parental depression (Berman et al., 2000) and parental anxiety (Berman et al., 2000; Bodden et al., 2008), in particular paternal anxiety (Liber et al., 2008; Rapee, 2000) and paternal depression (Liber et al., 2008) appeared to be predictors that are related to less favorable

treatment outcome in children with anxiety disorders.

When it comes to possible moderators of child-focused ICBT versus GCBT for children and adolescents with anxiety disorders, the results are not univocal either (Flannery-Schroeder, & Kendall, 2000; Liber et al., 2008; Manassis et al., 2002; Muris et al., 2001). Demographic child characteristics like gender and age seemed to be no moderators in all four studies (Flannery-Schroeder & Kendall, 2000; Liber et al., 2008; Manassis et al., 2002; Muris et al., 2001). Flannery-Schroeder and Kendall (2000) and Muris et al. (2001) also found no other interaction effects that indicated potential moderators of ICBT versus GCBT (Flannery-Schroeder & Kendall, 2000; Muris et al., 2001). As a result of their experience, Muris et al. (2001) nevertheless suggest that for children with trauma or Attention Deficit Hyperactivity Disorder (ADHD), ICBT is possibly more appropriate. However, Manassis et al. (2002) found the level of hyperactivity to be no predictor or moderator. They found that children reporting higher levels of social anxiety, reported greater gains in ICBT than in GCBT, suggesting that the diagnosis social phobia might be considered a moderator (Manassis et al., 2002). In contrast, the results of Liber et al. (2008) suggest that, based on father report, children with social phobia benefit more in GCBT than in ICBT compared to children without social phobia (Liber et al., 2008).

In sum age, level of child internalizing psychopathology, comorbid depression, severity of anxiety disorder and many parent and family characteristics seem to be predictors of treatment outcome of CBT for childhood anxiety disorders in general. The studies that compare ICBT versus GCBT for childhood anxiety disorders are scarce and there has been little attention for the specific distinction between predictors and moderators so far. Few moderators of ICBT versus

GCBT have yet been discovered. The presence of social phobia and trauma may be moderators of treatment outcome of ICBT versus GCBT. But the results are contradictory. Therefore firm conclusions about predictors and moderators of ICBT versus GCBT cannot be made. More research on predictors and moderators is needed to determine what factors really play a role in the treatment outcome of ICBT and GCBT, to understand why in practice some children do and others do not benefit. This is important in order to be able to improve the effectiveness of

treatment for anxious children and adolescents who still do not benefit from CBT (Berman et al., 2000; Compton et al., 2004; In-Albron & Sneider, 2007). Knowledge of factors associated with treatment outcome can enhance our understanding of the working mechanisms of treatment. It can help to explain why hypothesized mechanisms, processes and causes of therapeutic change, operate differentially under differing circumstances or for different subgroups of individuals.In addition, it can shed light on developmental differences or similarities between different

subgroups of children and adolescents (Berman et al., 2000).This can serve as a guideline for developing additions or modifications to CBT to improve the effectiveness of CBT. Predictors are informative with regard to reasonable treatment expectations, treatment targets and may lead to hypotheses regarding potential developmental factors in the course of child and adolescent disorders. The identification of moderators is particularly useful for clinical practice because they suggest directions for differential treatment selection and planning. It can help to improve guidelines for the selection of treatments that are likely to achieve the greatest efficacy on the basis of characteristics of the child or adolescent prior to treatment (Kraemer et al., 2002; Wolitzky-Taylor et al., 2012).

This study therefore focuses on the following question: Which demographic and clinical child characteristics can be considered predictors and/or moderators of treatment outcome of child-focused ICBT versus GCBT for childhood anxiety disorders? Guided by the above work, clinical relevance and the available data the following pre-treatment variables were examined in the present investigation as potential predictors or moderators of treatment outcome: the

demographic child characteristics (age and gender) and the clinical child characteristic (social phobia). Age is expected to be a predictor with younger child age predicting better treatment outcome in general. Gender is expected to be no predictor or moderator. Social phobia is expected to be a moderator, with children having a social phobia showing less favorable

also examined as potential predictorof treatment outcome of CBT in order to be able to take any difference in effectiveness between ICBT and GCBT in consideration. Treatment format is expected to be no predictor.

Methods

To answer the research question, the data of the implementation study of DDD (Bögels, 2008) and the preliminary data (until 1. June 2013) from an ongoing research examining the

effectiveness of the group versus individual cognitive behavioral therapy DDD (Bögels, 2008) were used. Both studies were carried out by the University of Amsterdam (UvA) and led by Drs. L. Jongerden and Prof. Dr. S. M. Bögels.

Participants

In 2008 the manualized CBT DDD was published in book form and thus became accessible to anyone who wanted to work with it. In the context of this publication national DDD workshops and lectures were offered. During these workshops and lectures institutions and therapists who wanted to work with DDD in practice were asked to participate in the implementation study. Therapists could register themselves by filling in a questionnaire and could then submit treatments. Children and adolescents between the age of 8 and 18 years and their parents who, according to their therapists met the criterion (which was impairing anxiety problems) to

participate in individual or group DDD, were approached to participate in the study. There were no further exclusion criteria to participate in the study. There is no information about the children who were not indicated for the protocol and what indication criteria institutions used.

So far, 157 children (implementation study n = 124; effectiveness study n = 33) were enrolled with the intention to be treated by DDD. Of those children, 34 were research dropouts (mainly due to a loss of motivation for completing the measurements after treatment) and 4 children were treatment dropouts (death of a parent, loss of motivation). Given the nature of the research question, research and treatment dropouts were not included in the analysis. Therefore the analyses of this study are based on the data of 119 children (implementation study n = 99; effectiveness study n = 20).

The 119 children and their families were referred to 17 different mental health

institutions or took the initiative themselves (primary care, n = 4; secondary care n = 115). They were treated by 20 different therapists, mainly orthopedagogists, psychologists and healthcare psychologists. Out of 119 children, 78 (65.5%) children (32 boys, 46 girls) were given ICBT and 41 (34.5%) children (16 boys, 25 girls) participated in the GCBT.Table 1 displays demographic and clinical child characteristics of the two treatment conditions. The mean age of the children in the ICBT was 11.54 years (SD 2.93) and 10.56 years (SD 2.01) in the GCBT. The institutions diagnosed the children prior to treatment themselves according to their own procedures. Of 10 children the diagnosis was unknown. The average number of anxiety diagnoses per child prior to treatment was 1.12 (SD 0.66) in the ICBT and 1.24 (SD 0.70) in the GCBT. In the ICBT 10 (12.8%) children had a comorbid anxiety diagnosis and 19 (24.4%) children had another comorbid diagnosis. In the GCBT 9 (21.9%) children had a comorbid anxiety diagnosis and 17 (67.1%) children had another comorbid diagnosis. It is possible that settings are inclined to diagnose only the most prominent anxiety disorder and perhaps do not register comorbid (anxiety) disorders. Questionnaires (SCARED-71) were completed by 119 families. The questionnaires were completed by both parents and child in 100 families, by just the parents in one family, by the child only in two cases and by the mother and the child in 16 families. Based on the child-reported Screen for Child Anxiety Related Emotional Disorders-71 (SCARED-71, see materials for description of this questionnaire) scores, 32 (71%) children in the ICBT and 13 (29%) children in the GCBT had a social phobia. Based on the parent-reported SCARED-71 scores 31 (60%) children in the ICBT and 21 (40%) in the GCBT had a social phobia.

There were no significant pretreatment differences between the children in the two different treatment formats with regard to gender and total number of anxiety diagnoses or comorbid diagnoses. There were also no significant differences between total and social phobia symptoms and the presence of the diagnoses social phobia according to the child and parents on the SCARED-71. The demographic child characteristic age was not comparable between the two treatment formats; children in the ICBT were significantly older than children in the GCBT. This was taken into account in the analysis. Table 1 displays the comparisons and effect sizes of the demographic and clinical child characteristics in the ICBT versus GCBT.

Table 1

Means, standard deviations, comparisons and effect sizes of demographic and clinical child characteristics in the individual cognitive-behavioral therapy versus group cognitive-behavioral therapy treatment condition.

ICBT N = 78 GCBT N = 41 t / χ2 Effect size a Girls (n, %) 46 (59%) 25 (61%) .05 .02 Child age (M, SD) 11.54 (2.93) 10.56 (2.01) 2.14* .37 Pretreatment total anxiety symptoms

(SCARED-71)

Child report (M, SD) 44.70 (18.25) 42.11 (20.88) .70 .13 Parent report (M, SD) 45.46 (18.09) 48.56 (16.11) -.92 -.18 Pretreatment social phobia symptoms

(SCARED-71)

Child report (M, SD) 6.76 (4.78) 6.11 (3.89) .76 .15 Parent report (M, SD) 7.42 (4.94) 7.76 (3.87) -.37 -.07 Pretreatment diagnosis social phobia

(SCARED-71)

Child report (n, %) 32 (71%) 13 (29%) .99 -.09 Parent report (n, %) 31 (60%) 21 (40%) 1.44 .11 Pretreatment number of anxiety diagnosis

(M, SD)

1.12 (0.66) 1.24 (0.70) -.95 -.18

Pretreatment number of comorbid diagnosis (M, SD)

0.32 (0.56) 0.49 (0.64) -1,41 -.29

a _{phi coefficient as an effect size for categorical variables, Cohen’s d as an effect size for}

continuous variables.

Materials

Outcome measure

SCARED-71 Child anxiety symptoms were measured with the Dutch version of Screen for Child

Anxiety Related Emotional Disorders-71 (SCARED-71, Bodden, Bögels, & Muris, 2009) child and parent about child report. This questionnaire consists of 71 questions (e.g. “I am very afraid of an animal that actually is not dangerous”) and is appropriate for children from 8 to 18 years. Responses of the child and parents about the child are rated on a three-point Likert scale (0 = (almost) never, 1 = sometimes, 2 = often). Cut-off scores distinguish clinically anxious children from normal anxious children (Bodden et al., 2009). As outcome measure the total sum score of the child and the total sum score of the parents (mean of the sum of the total score of the father and the mother) were used. The validity and reliability of the SCARED-71 is satisfactory

(Bodden et al., 2009). In this sample the internal consistencies (Cronbach’s α) of the total scale at baseline are excellent: .92 (child), .91 (mother), .91(father).

Potential predictors and moderators

Age and gender Therapists provided information on age and gender of the child at the start of

treatment.

Social phobia Subscales of the SCARED-71 give an indication of the possible presence of the

most important DSM-IV anxiety disorders including social phobia (cut-off score for clinical relevance 8). The internal consistency of the social phobia subscale is proven to be high (Bodden et al., 2009). In this sample the internal consistencies (Cronbach’s α) of the subscale social phobia at baseline are good: .82 (child), .88 (mother), .85 (father). For the predictor/moderator analysis the pretreatment social phobia subscale sum score of the child and the pretreatment social phobia subscale sum score of the parents (mean of the social phobia subscale sum score of the father and mother) were used.

Treatment

Participating children and adolescents received either ICBT using DDD or GCBT using DDD. The ICGT DDD consists of 12 weekly sessions of one hour with the child. In session 1, 4 and 12 the parents are partially present. There are three separate optional parent sessions of 1 hour (Bögels, 2008). The GCGT DDD consists of 8 weekly child sessions of one and a half hours. In

session 8, the parents are partially present. In addition, there are three separate parent sessions in a group of one and a half hours (Peijnenburg & Bögels, 2008). In both treatment formats the children and parents use the same workbook and for the therapists there was a treatment protocol available. Also in both treatment variants the same therapy components are used namely,

psychoeducation, coping strategies, challenging anxious thoughts and formulating helpful thoughts, exposure in vivo, experiments and relapse prevention (Bögels, 2008).

Procedure

Quasi-experimental design Pre-existing groups are compared on the basis of a pre-test, post-test immediately after treatment and 3-month follow-up. Participants and therapists were free to choose between ICBT using DDD or GCBT using DDD. All children aged 12+ and parents signed an informed consent form when they decided to participate in the study. Father, mother and child completed paper or electronic questionnaire booklets before, immediately after and at 3 months follow-up. This took parents and children 15 to 20 minutes each time. If necessary, they received help of a research assistant in completing the questionnaires. The SCARED-71 data of all three measurements were used as outcome measures. For the predictor/moderator analysis age and gender were determined at the pre-test. The social phobia subscale of the SCARED-71 of the pre-test was used to determine the level of social phobia and the presence of the diagnosis social phobia of the child before treatment.

Statistical analysis

All analyses were conducted using SPSS 20.0. The data analyses are based on treatment completers of whom at least post treatment data were available. Missing value analyses on the questionnaires, missing less than 10% of the items, revealed that all little’s MCAR tests were non-significant. The missing items were therefore replaced by the mean of the subscale.

Correlation between improvement on the father- and mother-reported SCARED-71 scores from pretreatment to post treatment was small (r = .293, p < 0.05) and pretreatment compared to three months follow-up was high (r =. 641, p < 0.01). Therefore father and mother scores were

aggregated into one parent score. If only one parent participated the parent score consisted of one instead of two parents. Little’s MCAR tests were non-significant for the questionnaires that were missing as a whole. These were imputed using MVA Estimation Maximization.

For the predictor/moderator analysis hierarchical multiple regression was used. The dependent variable was defined as the improvement of total anxiety pretreatment compared to post treatment and pretreatment compared to 3 months after treatment (follow-up) according to children and parents separately. The model was thus run four times with four different

improvement scores as outcome measure. Model 1 (Children post treatment) was ran with the improvement scores of child-reported total anxiety (SCARED-71) pretreatment compared to post treatment as outcome measure. Model 2 (Children follow-up) was ran with the improvement scores of child-reported total anxiety (SCARED-71) pretreatment compared to follow-up as outcome measure. Model 3 (Parents post treatment) was ran with the improvement scores of parent-reported total anxiety (SCARED-71) pretreatment compared to post treatment as outcome measure. Model 4 (Parents follow-up) was ran with the improvement scores of parent-reported total anxiety (SCARED-71). The following independent variables were entered: (1) age, (2) treatment format (group versus individual), (3) gender, (4) social phobia (diagnosis) and the interaction term (5) treatment format (group versus individual) * social phobia symptoms (severity). When the improvement scores of children were used as outcome measure, social phobia diagnosis and level were also based on the child-reported scores. When the improvement scores of parents were used as outcome measure, social phobia diagnosis and level were also based on the parent-reported scores. To control for the detected difference in child age between the two treatment formats (ICBT versus GCBT), age was entered first at step 1 into the model followed by the others at step 2.

Preliminary analyses with Explore revealed there were no extreme outliers in the

independent and dependent variables that extended more than 3 box-lengths from the edge of the box. But 1.5% of the SCARED-71 total and social anxiety scores of children and parents were scores that extended more than 1.5 box-length from the edge of the box and in the ICBT 3 of the 78 children (3.85%) had ages (2 children age 18 and 1 age 19) that extended more than 1.5 box-length from the edge of the box. Multiple regression is very sensitive for outliers but by

trimming these outliers meaningful information of actual patients would get lost. Therefore so-called winsorizing was used meaning that the outliers were replaced by the next highest score that was not an outlier prior to the predictor/moderator analysis.

Results

Preliminary analyses

Hierarchical multiple regression was used to assess the ability of four control measures ((2) treatment format (group versus individual), (3) gender, (4) social phobia (diagnosis), (5) treatment format (group versus individual) * social phobia (level) to predict treatment outcome (improvement on SCARED total anxiety) immediately after treatment (post treatment) and three months after treatment (follow-up) according to children and parents after controlling for the influence of (1) age. The tolerance values for each independent variable (possible

predictor/moderator) were all above .10. The multicollinearity assumption was therefore not violated. This was also supported by the VIF values which were all below the cut-off of 10. Correlations between the independent variables (possible predictors/moderators) were all small (below .30). However, independent variables (2) treatment format and (5) treatment format * social phobia were quite highly intercorrelated (.795 child outcome measures, .845 for the parent outcome measures). The assumptions of normality, linearity and homoscedasticity were not violated.

Predictor/moderator analyses

Child reported pre to post treatment improvement Age was entered at Step 1, explaining 0% of

the variance in treatment outcome according to children at post treatment. After entry of the other predictors/moderators at Step 2 the total variance explained by the model as a whole at post treatment according to children was a significant 14%, F (5, 113) = 3.80, p = .003. In the final model, post treatment outcome according to children was significantly predicted by (2)

Treatment format and (4) Social phobia, with (2) Treatment format recording a higher beta value (beta = -.36, p = .027) than (4) Social phobia (beta = .25, p = .023). ICBT predicted more post treatment improvement than GCBT according to children. Having the diagnosis social phobia at pretreatment predicted more post treatment improvement according to children. The interaction (5) treatment format * social phobia was non-significant. This indicates (4) Social phobia was a predictor and not a moderator of post treatment outcome according to children, meaning that social phobia predicted more post treatment improvement according to children regardless of

treatment format. Table 2 displays the linear model of predictors of post treatment outcome according to children.

Table 2

Hierarchical linear regression model of predictors of child reported post treatment improvement on the SCARED total anxiety scale.

B SE B ẞ Sig Step 1 Constant 10.80 (-1.95, 23.54) 6.44 .096 Age 0.12 (-0.99, 1.23) 0.56 .20 .829 Step 2 Constant 18.18 (4.53, 31.82) 6.89 .009* (1) Age -0.66 (-1.82, 0.49) 0.58 -.11 .258 (2) Treatment format (0 = individual, 1 = group) -11.81 (-22.25, -1.36) 5.27 -.36 .027* (3) Gender (0 = boys, 1 = girls) 1.35 (-4.43, 7.13) 2.92 .04 .644 (4) Social phobia

(0 = no social phobia, 1 = social phobia)

7.99 (1.12, 14.86)

3.47 .25 .023*

(5) Treatment format * Social phobia 0.78 (-.630, 2.19)

0.71 .18 .275

* _{p < .05 (two-tailed)}

Note. R2 _{= .00 for Step 1; ∆R}2 _{=.14 for Step 2 (ps < . 05)}

Child reported pre to three months follow-up treatment improvement At follow-up according to

explained a non- significant 8% (F (5, 113) = 2.03, p = .080) of the total variance. Because the model as a whole did not significantly explain the variance in treatment outcome at follow-up, the predictive value of the individual variables was not interpreted.

Parent reported pre to post treatment improvement Age was entered at Step 1, explaining 2% of

the variance in treatment outcome according to parents at post treatment. After controlling for age, the four control measures explained an additional 3% of the variance in treatment outcome, R squared change = .03, F change (4, 113) = 1.03, p = .398. At post treatment according to parents the variance explained by the model as a whole was a non-significant 5%, F (5, 113) = 1.20, p = .313. Because the model as a whole did not significantly explain the variance in post treatment outcome, the predictive value of the individual variables was not interpreted.

Parent reported pre to three months follow-up treatment improvement Age explained 0% of the

variance in treatment outcome according to parents at follow-up. The total variance explained by the model as a whole was a non-significant 2%, F (5, 113) = 0.43, p = .824. Because the model as a whole did not significantly explain the variance in treatment outcome at follow-up, the predictive value of the individual variables was not interpreted.

Explorative post hoc analyses

Predictor Treatment format ICBT predicted more post treatment improvement than GCBT

according to children. In order to be able to interpret this result correctly, a closer look was taken at the course of the average total anxiety symptoms of children on the SCARED-71 prior to treatment, post treatment and three months after treatment at follow-up. Independent t-tests revealed that the mean total anxiety symptoms did not significantly differ between ICBT and GCBT at all three measurements. But, as can be seen in figure 1, the mean total anxiety

symptoms at post treatment compared to pretreatment had dropped more in ICBT than in GCBT. At follow-up children in ICBT and GCBT reached approximately the same level of total anxiety symptoms below the clinical cut-off point of 30. The decrease in total anxiety symptoms after three months of treatment in comparison to pretreatment was approximately the same.

Two Independent t-tests of the improvement scores of child-reported total anxiety symptoms (SCARED-71) pretreatment compared to post treatment and pretreatment compared to follow-up of children in the ICBT and children in the GCBT were conducted. The results were in

of figure 2. Immediately after treatment, children in the ICBT reported significantly greater reduction of total anxiety symptoms compared to pretreatment (M = 14.63, SD = 15.45) than children in the GCBT (M = 7.45, SD = 15.53, t (117) = 2.40, p = .02, d = .47). At follow-up, no differences (t (117) = 0.67, p = .50) occurred between ICBT (M = 18.09, SD = 17.59) versus GCBT (M = 15.74, SD = 18.96) with regard to the reduction of total anxiety symptoms compared to pretreatment.

Predictor Social phobia Having the diagnosis social phobia prior to treatment, predicted more

post treatment improvement outcome in general, based on child report. In order to be able to interpret this result correctly, children were split into two groups: children diagnosed with social phobia based on the child-reported SCARED-71 subscale social phobia (n = 45) versus children without the diagnosis social phobia based on the child-reported SCARED-71 subscale social phobia (n = 74).

First, three independent t-tests were conducted to compare the total anxiety symptoms reported by children with social phobia and children without social phobia at the three different measurements. There was a significant difference on total SCARED scores for children with social phobia (M = 57.26, SD = 15.52) and children without social phobia (M = 35.63, SD = 16.37, t (117) = -7.22, p =.000, two-tailed, d = 1.35) prior to treatment. Children with social phobia reported significantly more anxiety symptoms than children without social phobia.

Figure 1. Mean total anxiety symptoms (child report) on the SCARED-71 at pretreatment, post treatment and follow-up in individual cognitive-behavioral therapy (ICBT) versus group cognitive-behavioral therapy (GCBT). 15 20 25 30 35 40 45 50

Pretreatment Post treatment Follow-up

GCBT ICBT Total Anxiety symptoms (SCARED-71)

Measurement

Immediately after treatment, children with social phobia still reported significantly more total anxiety symptoms (M = 39.38, SD = 18.08) than children without social phobia (M = 26.96, SD = 16.96, t (117) = -3.72, p = .000, d = .71). Also three months after treatment at follow-up children with social phobia (M = 35.57, SD = 21.42) still reported significantly more total

anxiety symptoms than children without social phobia (M = 21.04, SD = 17.13, t (117) = -4.08, p = .000, d = .77).

Second, two independent t-tests were conducted to compare the improvement scores of child-reported total anxiety symptoms (SCARED-71) pretreatment compared to post treatment and pretreatment compared to follow-up of children with and without social phobia. As can be seen in figure 2, the total anxiety symptoms of children with the diagnosis social phobia had dropped relatively more than the total anxiety symptoms of children without the diagnosis social phobia from pre to post treatment. This difference was significant (t (117) = -3.21, p = 002, d = .61). Immediately after treatment, children with social phobia reported a significantly greater reduction in total anxiety symptoms compared to pretreatment (M = 17.88, SD = 17.23) than children without social phobia (M = 8.67, SD = 13.83). The pre to three months follow-up improvement of child-reported total anxiety symptoms (SCARED-71) did not significantly differ between children with and without social phobia (t (71.418) = -1.95, p = .06). At follow-up children with social phobia did not report significantly greater reduction in total anxiety

Figure 2. Average total anxiety symptoms according to the children on the SCARED-71 at pretreatment, post treatment and follow-up of children with the diagnosis social phobia and children without the diagnosis social phobia.

15 20 25 30 35 40 45 50 55 60

Pretreatment Post treatment Follow-up

Children without diagnosis social phobia

Children with diagnosis social phobia

Total anxiety SCARED-71

Measurement

symptoms (M = 21.69, SD = 21.32) than children without social phobia (M = 14.59, SD = 15.24). In addition it is important to notice that children with social phobia stayed above the cut-off point of 30 immediately after treatment and at follow-up, while children without social phobia were not clinically anxious anymore already immediately after treatment.

Discussion

The current study investigated whether treatment format predicts treatment outcome of child-focused CBT and whether the demographic child characteristics age and gender and the clinical child characteristic social phobia can be considered predictors and/or moderators of treatment outcome of child-focused ICBT versus GCBT for childhood anxiety disorders. The main results can be summarized as follows. First, contrary to what was expected, treatment format is found to be a predictor of post treatment outcome according to children, with ICBT predicting more improvement immediately after treatment than GCBT. In the long term however, in line with the hypotheses, ICBT and GCBT are proven to be equally effective. Second,social phobia is found to be a predictor of post treatment effectiveness according to children, although it was expected to be a moderator. Children with social phobia reported a significant greater reduction of anxiety symptoms post treatment compared to pretreatment than children without social phobia

regardless of treatment format. But post hoc analyses showed that the social phobic children had higher anxiety levels in general at pretreatment. Their anxiety levels did not drop enough

immediately after treatment to have anxiety levels under the clinical cut off while those of children without social phobia did. At follow-up compared to pretreatment there was no

significant difference in reduction of total anxiety symptoms between children with and without social phobia anymore. But children with social phobia still did not drop enough three months after treatment to have anxiety levels under the clinical cut off. Third, age was, contrary to the hypothesis no predictor of treatment outcome. Fourth, in line with the hypotheses, gender was neither a predictor nor moderator. Finally, the model as a whole only says something about the variance in treatment outcome according to children at post treatment. Age, gender, treatment format and social phobia all together appear to say nothing about the variance in treatment outcome according to children three months after treatment and according to parents at post treatment and three months after treatment.

This study, unlike several other studies (e.g. Liber et al., 2008; Muris et al., 2001), shows that ICBT in clinical practice is more effective in the short term than GCBT. Children who received individual therapy improved more on overall anxiety symptoms than children who were treated in a group. First, this might be the consequence of a ‘time-effect’. Although group

sessions were one and a half times as long as individual sessions (90 versus 60 minutes), GCBT consisted of only 8 sessions and ICBT of 12 sessions. This means that the so-called treatment minutes in ICBT were spread over a greater period of time. At post-test, children who received ICBT thus have had more time to improve than children who received GCBT. This explanation is consistent with the fact that three months after treatment at follow-up the difference in effectiveness has disappeared. Second, the difference in effectiveness in the short term between ICBT and GCBT might also be the consequence of a ‘dose-effect’. Due to the difference in number of sessions between ICBT and GCBT (12 versus 8 sessions), children who received individual therapy saw there therapist more often. In addition children who received ICBT did not have to share the treatment minutes with other children. This has manifested itself especially in more individual monitoring of exposure steps and homework in ICBT than in GCBT. Previous research shows that children in ICBT are better at recalling what they have learned during

treatment than children in the GCBT, maybe because children become more one-to-one

instruction and attention in ICBT than in GCBT (Flannery-Schroeder & Kendall, 2000). Taken all this together it is not inconceivable that children who receive ICBT realize a slightly faster progress than children who receive GCBT.

Although this study shows that in the short term ICBT is more effective, it is important to emphasize that the results of this study are consistent with previous studies showing that ICBT and GCBT are equally effective in the long term (Flannery-Schroeder & Kedall, 2000; In Albon-Schneider, 2007). Children in both treatment formats reached approximately the same level of anxiety symptoms below the clinical cutoff three months after treatment.

Manassis et al. (2002) found that children with social phobia reported substantially more total anxiety than children without social phobia and argued that children with social phobia benefit more from ICBT than GCBT. The results of the current study confirm that children with social phobia report more total anxiety but contradict that they would make greater gains when treated individually. Children with social phobia appeared to achieve a greater reduction of total anxiety symptoms immediately after treatment than children without social phobia regardless of

whether they were treated individually or in a group. Social phobia is proven to be a persistent condition with a remarkably high degree of comorbid conditions like depressive disorders (Brunello et al., 2000; Fehm, Pelissolo, Furmark, & Wittchen, 2005; Merikangas, Avenevoli, Acharyya, Zhang, & Angst, 2002).Comorbidity for depression is found to predict less favorable treatment outcome (Berman et al., 2000). Moreover, previous research has shown that severer anxiety predicts less favorable treatment response (Layne et al., 2003; Liber et al., 2010; Southam-Gerow et al., 2001). It is therefore surprising that in clinical practice children with social phobia, who in general reported significantly more total anxiety symptoms benefit more from CBT than children without social phobia immediately after treatment.

Although children with social phobia achieved a greater reduction in overall anxiety symptoms in the short term, they remained clinically anxious immediately after treatment while at that time children without social phobia already reached normal anxiety levels. This is in line with previous research that showed that children with higher levels of pretreatment severity are less likely to be recovered at post treatment (Layne et al., 2003; Liber et al., 2010; Southam-Gerow et al., 2001) but do report greater pre- to post treatment change (Liber et al., 2010).

However, the children with social phobia still did not achieve a normal anxiety level at three months follow-up. The fact that children with a social phobia experience more anxiety symptoms before, after and 3 months after treatment needs additional explanations. There is consistent evidence that there is an association between higher levels of behavioral inhibition and having the diagnosis social phobia in children (Essex, Klein, Slattery, Goldsmith, & Kalin, 2010; Gladstone, Parker, Mitchell, Wilhelm, & Malhi, 2005). Children with social phobia may

therefore suffer more from trait-anxiety than children with other anxiety disorders. And higher levels of trait anxiety are found to predict less favorable treatment outcome in children with anxiety disorders (Berman et al., 2000). The results of the current study suggest that children with social phobia who also report a high degree of total anxiety might simply need more time, maybe because the fear is so interwoven in their personality.

The results of this study are also consistent with other intervention studies for childhood anxiety disorders, showing that overall patterns of a large pre- to post treatment change are usually followed by gradual but continued improvement until follow-up (Bodden et al., 2008; Compton 2004; In-Albron & Sneider, 2007; Jongerden et al., 2011). After treatment, the improvement of children with social phobia continued gradually. At follow-up children with

social phobia did not significantly differ from children without social phobia in term of reduction in total anxiety symptoms. This supports the statement that children with social phobia who also report a high degree of total anxiety might simply need more time.

The regression model with age, gender, treatment format and social phobia did not significantly explain the variance in treatment outcome in the short and long term based on child and parents reports. Even though this is disappointing it is not entirely surprising given the fact that there are many other variables that have not been included in the present study that are proven to play a significant role in treatment outcome for childhood anxiety disorders like parent (Berman et al., 2000; Bodden et al., 2008; Liber et al., 2008, 2010; Rapee, 2000) and family characteristics (Crawford & Manassis, 2001; Victor et al., 2007). In addition, the therapeutic relationship probably explains some of the variance and should therefore be considered as a possible predictor and/or moderator. Findings from a meta-analytic review indicate a modest association between therapeutic relationship variables and treatment outcome in child and adolescent therapy (Shirk & Karver, 2003). Finally there are probably variables that we are not yet aware of, that play a role in treatment outcome, given the fact that to date, only a limited number of predictors and moderators of treatment outcome for childhood anxiety disorders have been detected (Compton et al., 2004; Rodebaugh et al., 2004).

The present study has several limitations. First, anxiety was measured with

questionnaires which gives an indication but not actual measuring of the presence of an anxiety disorder. This may have led to the unjustified assigning of children to either the social phobia or non-social phobia group. In addition, in the current analysis, the predictor/moderator social phobia is based on a subscale of the outcome measure. It would be desirable for the

predictor/moderator to use another instrument. Second, there are some methodological limitations regarding the study design: the absence of a control condition and the nonrandom allocation of children to the individual or group treatment which makes the results less reliable and valid. Other limitations of the study design relate to the uncontrollability of a number of factors: it is unknown how institutions determined the diagnoses of the children which may have led to underreporting of comorbid (anxiety) disorders and little is known about dropouts and children who were not indicated for ICBT or GCBT. On the other hand, it is important to emphasize that the design of the current investigation can be referred to as highly clinical representative. Controlled conditions are not always comparable to the circumstances in clinical

practice (Shadish, Matt, Navarro, & Phillips, 2000). The chosen design is in line with the goal of this study which was to make a contribution to improve the effectiveness of treatment for

anxious children and adolescents who in practice still do not always benefit from CBT. The use of multiple informants can also be considered as a strength.

Several recommendations can be drawn from the findings and the limitations of this investigation for future research. First, the very limited explanatory power of the variance in treatment outcome of the variables examined in this study, stresses the essence to pay more attention in the future to predictors and moderators of treatment outcome for childhood anxiety disorders. Beside child characteristics also parent, family and therapeutic relationship

characteristics should be included. In addition, hypotheses generating research should focus on the question in what way anxious children who benefit from treatment differ from anxious children whose treatment was less successful in order to discover unknown predictors and moderators. N = 1 studies might be very informative and suitable for the study of treatments of outstanding individuals in clinical practice (Borckardt et al., 2008). Second, RCT’s with large sample sizes are necessary in order to increase the reliability, validity and generalizability of the results and to be able to analyze the predictive and/or moderating value of each variable

separately. Third, the results of this investigation also underline the importance of careful examination of what a better treatment outcome really means. In future research treatment outcome should not only be assessed in terms of improvement but also in actual recovery.

Therefore it is recommended to use diagnostic interviews such as the Anxiety Disorder Interview Schedule C/P (ADIS-C/P) (Siebelink & Treffers, 2001; Silverman & Albano, 1996). Using multiple instruments also prevents that predictors/moderators are based on a subscale of the outcome measure. Fourth, the findings indicate that children with social phobia experience more overall anxiety than children with other anxiety disorders and that this might affect their

treatment outcome. Future research should therefore not only examine predictors and moderators of treatment outcome for children and adolescents with heterogeneous anxiety disorders but also examine specific predictors and moderators of treatment outcome for children with social

phobia. Finally, given the result that ICBT and GCBT are equally effective three months after treatment, it would also be useful in future research to focus on cost effectiveness of ICBT versus GCBT for children with anxiety disorders.

Several clinical implications can be drawn from the findings as well. Since ICBT and GCBT appeared to be equally effective in the long term, in clinical practice the choice between ICBT and GCBT can be based on pragmatic considerations such as therapeutic resources and the preference of the parents and the child. When GCBT is selected as treatment of choice it is important to explain to parents and child, in order to avoid false expectations, that the overall anxiety directly after the treatment might still be somewhat high, but that improvement generally continues after therapy leading to normal levels of overall anxiety. Also for children with social phobia the choice for ICBT or GCBT can be based on pragmatic considerations, because

children with social phobia benefit equally from both treatment formats. However, children with social phobia who also report a high degree of total anxiety might be a subgroup of children who would benefit from a so-called stepped care approach (Bower & Gilbody, 2005) in which they receive additional treatment after the basis CBT program in order to reach a normal level of total anxiety. Finally, age and gender were no predictors or moderators. This suggests that in clinical practice, the manualized CBT DDD (Bögels, 2008) is an equally effective intervention in the short and longer term for both boys and girls and young children and older adolescents with anxiety disorders, regardless whether it is offered individually or in a group. This makes DDD (Bögels, 2008) a particularly interesting intervention for clinical practice.

This study proved, ICBT and GCBT to be equally effective in the long term. ICBT and social phobia were found to be predictors of more post treatment improvement. No moderators were detected. This study has also shown that it is important to not only look at treatment improvement but also at treatment recovery and that children with social phobia report significantly more overall anxiety before, after and three months after treatment than children without social phobia. Given the fact that this study has not been able to detect predictors and/or moderators in the long term, underlines that more thorough research on predictors and

moderators of treatment outcome for childhood anxiety disorders is necessary in the future. This is important to determine what factors really play a role in the treatment outcome of ICBT and GCBT and to understand why in practice some children do and others do not benefit. In this way the effectiveness of treatment of childhood anxiety disorders can further be improved. This study shows that in clinical practice children and adolescents regardless of age, gender and diagnosis social phobia benefit from ICBT and GCBT. The choice is up to them.

References

American Psychiatric Association. (2005). Diagnostic and statistical manual of mental disorders, fourth edition, text revision (DSM-IV-TR). Washington: APA.

Beesdo, K., Bittner, A., Pine, D. S., Stein, M. B., Höfler, M., Lieb, R., & Wittchen, H. U. (2007). Incidence of social anxiety disorder and the consistent risk for secondary depression in the first three decades of life. Archives of General Psychiatry, 64(8), 903-912. doi:

10.1001/archpsyc.64.8.903

Berman, S. L., Weems, C. F., Silverman, W. K., & Kurtines, W. M. (2000). Predictors of outcome in exposure-based cognitive and behavioral treatments for phobic and anxiety

disorders in children. Behavior Therapy, 31(4), 713-731.doi: 10.1016/S0005-789(00)80040-4 Bodden, D. H. M., Bögels, S. M., & Muris, P. (2009). The diagnostic utility of the Screen for

Child Anxiety Related Emotional Disorders-71 (SCARED-71). Behaviour Research and Therapy, 47(5), 418-425. doi: 10.1016/j.brat.2009.01.015

Bodden, D. H. M., Bögels, S. M., Nauta, M. H., Haan, E., de, Ringrose, J., Appelboom, C., Brinkman, A. G., & Appelboom-Geerts, K. C. M. M. J. (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(12), 1384– 139. doi: 10.1097/CHI.0b013e318189148e

Bögels, S. M. (2008). Behandeling van angststoornissen bij kinderen en adolescenten met het cognitief gedragstherapeutisch protocol Denken + Doen = Durven. [Treatment of anxiety disorders in children and adolescents with cognitive behavior therapy protocol Discussing + Doing = Daring.] Houten: Bohn Stafleu van Loghum.

Borckardt, J. J., Nash, M. R., Morphy, M. D., Moore, M., Shaw, D., & O’Neil, P. (2008). Clinical practice as natural laboratory for psychotherapy research: A guide to case-bases time series analyses. American Psychologist, 63(2), 77-95. doi: 10.1037/0003-066X.63.2.77 Bower, P., & Gilbody, S. (2005). Stepped care in psychological therapies: Access, effectiveness

and efficiency: Narrative literature review. The British Journal of Psychiatry, 186(1), 11-17. doi: 10.1192/bjp.186.1.11

Brunello, N., Boer, J. A., den, Judd, L. L., Kasper, S., Kelsey, J. E., Lader, M., ... Wittchen, H. U. (2000). Social phobia: Diagnosis and epidemiology, neurobiology and pharmacology,

comorbidity and treatment. Journal of Affective Disorders, 60(1), 61-74. doi: 10.1016/S0165-0327(99)00140-8

Cartwright-Hatton, S., McNicol, K., & Doubleday, E. (2006). Anxiety in a neglected population: Prevalence of anxiety disorders in pre-adolescent children. Clinical Psychology Review, 26(7), 817–833. doi: 10.1016/j.cpr.2005.12.002

Compton, S. N., March, J. S., Brent, D., Albano, A. M., Weersing, V. R., & Curry, J. (2004). Cognitive-behavioral psychotherapy for anxiety and depressive disorders in children and adolescents: An evidence-based medicine review. American Academy of Child and Adolescent Psychiatry, 43(8), 930-959. doi: 10.1097/01.chi.0000127589.57468.bf

Crawford, A. M., & Manassis, K. (2001). Familial predictors of treatment outcome in childhood anxiety disorders. Journal of the American Academy of Child & Adolescent Psychiatry, 40(10), 1182–1189. doi: 10.1097/00004583-200110000-00012

Essex, M. J., Klein, M. H., Slattery, M. J., Goldsmith, H. H., & Kalin, N. H. (2010). Early risk factors and developmental pathways to chronic high inhibition and social anxiety disorder in adolescence. American Journal of Psychiatry, 176(1), 40-46. doi:

10.1176/appi.ajp.2009.07010051

Fehm, L., Pelissolo, A., Furmark, T., & Wittchen, H. U. (2005). Size and burden of social phobia in Europe. European Neuropsychopharmacology, 15(4), 453-462. doi:

10.1016/j.euroneuro.2005.04.002

Flannery-Schroeder, E. C., & Kendall, P. C. (2000). Group and individual cognitive-behavioral treatments for youth with anxiety disorders: A randomized clinical trial. Cognitive Therapy and Research, 24(3), 251–278. doi: 10.1023/A:1005500219286

Gladstone, G. L., Parker, G. B., Mitchell, P. B., Wilhelm, K. A., & Malhi, G. S. (2005), Relationship between self-reported childhood behavioral inhibition and lifetime anxiety disorders in a clinical sample. Depression and Anxiety, 22(3), 103–113.

doi: 10.1002/da.20082

Graaf, R. de, Have, M. ten, & Dorsselaer, S., van. (2010). De psychische gezondheid van de Nederlandse bevolking NEMESIS-2: Opzet en eerste resultaten. [The mental health of the Dutch population NEMESIS-2: Design and first results.] Utrecht: Trimbos-instituut. In-Albon, T., & Schneider, S. (2007). Psychotherapy of childhood anxiety disorders: A

James, A., Soler, A., Weatherall, R. (2005). Cognitive behavioural therapy for anxiety disorders in children and adolescents. Cochrame Database of Systematic Reviews, 4, CD004690. doi: 10.1002/14651858.CD004690.pub2

Jongerden, L., Bögels, S. M., & Peijnenburg, D. (2011). Maar hoe werkt het echt? De

implementatie van Denken + Doen = Durven, een cognitieve gedragstherapie voor kinderen en jongeren met angststoornissen. [But how does it really work? The implementation of Discussing + Doing = Daring, a cognitive behavioral therapy for children and adolescents with anxiety disorders.] Gedragstherapie, 44(2011), 313-325.

Kendall, P. C., Brady, E. U., & Verduin, T. L. (2001). Comorbidity in childhood anxiety disorders and treatment outcome. Journal of American Academy of Child & Adolescent Psychiatry, 40(7), 787–794. doi: 10.1097/00004583-200107000-00013

Kessler, R. C., Berglund, P., Demler, O., Jin, R., & Walters, E. E. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the national comorbidity survey replication. Archives of General Psychiatry, 62(6), 593-602. doi:10.1001/archpsyc.62.6.593 Kraemer, H. C., Wilson, G. T., Fairburn, C. G., & Agras, W. S. (2002). Mediators and

moderators of treatment effects in randomized clinical trials. Archives of General Psychiatry, 59(10), 877-883. doi: 10.1001/archpsyc.59.10.877

Layne, A. E., Bernstein, G. A., Egan, E. A., & Kushner, M. G. (2003). Predictors of treatment response in anxious-depressed adolescents with school refusal. Journal of the American Academy of Child & Adolescent Psychiatry, 42(3), 319–326. doi: 10.1097/00004583-200303000-00012

Liber, J. M., Widenfelt, B. M., Goedhart, A. W., Utens, E. M. W. J., Leeden, A. J. M., van der, Markus, M. T., & Treffes, P. D. A. (2008). Parenting and parental anxiety and depression as predictors of treatment outcome for childhood anxiety disorders: Has the role of fathers been underestimated? Journal of Clinical Child & Adolescent Psychology, 37(4), 747-758. doi: 10.1080/15374410802359692

Liber, J. M., Widenfelt, B. M., Leeden, A. J. M., van der, Goedhart, A. W., Utens, E. M. W. J., & Treffers, P. D. A. (2010). The relation of severity and comorbidity to treatment outcome with cognitive behavioral therapy for childhood anxiety disorders. Journal of Abnormal Child Psychology, 38(5), 683-694. doi: 10.1007/s10802-010-9394-1

Liber, J. M.,Widenfelt, B. M., van, Utens, E. M. W. J., Ferdinand, R. F., Leeden, A. J. M., van der, Gastel, W., van, & Treffers, P. D. A. (2008). No differences between group versus individual treatment of childhood anxiety disorders in a randomised clinical trial. Journal of Child Psychology and Psychiatry 49(8), 886–893. doi:10.1111/j.1469-7610.2008.01877.x Manassis, K., Mendlowitz, S. L., Scapillato, D., Avery, D., Fiksenbaum, L.,

Freiere, M., … Owens, M. (2002). Group and individual cognitive-behavioral therapy for childhood anxiety disorders: A randomized trial. Journal of the American Academy of Child & Adolescent Psychiatry, 41(12), 1423–1430. doi: 10.1097/00004583-200212000-00013 Mendlowitz, S. L., Manassis, K., Bradley, S., Scapillato, D., Miezitis, S., & Shaw, B. E. (1999).

Cognitive‐behavioral group treatments in childhood anxiety disorders: The role of parental involvement. Journal of the American Academy of Child & Adolescent Psychiatry, 38(10), 1223–1229. doi: 10.1097/00004583-199910000-00010

Merikangas, K. R., Avenevoli, S., Acharyya, S., Zhang, H., & Angst, J. (2002). The spectrum of social phobia in the Zurich cohort study of young adults. Biological Psychiatry, 15(1), 81-91. doi: 10.1016/S0006-3223(01)01309-9

Muris, P., Mayer, B., Bartelds, E., Tierney, S., & Bogie, N. (2001). The revised version of the Screen for Child Anxiety Related Emotional Disorders (SCARED-R): Treatment sensitivity in an early intervention trial for childhood anxiety disorders. British Journal of Clinical Psychology, 40(3), 323–336. doi: 10.1348/014466501163724

Peijnenburg, T. M., & Bögels, S. M. (2008). Protocollaire groepsbehandeling voor kinderen en jongeren met angststoornissen: Denken + Doen = Durven. [Group treatment protocol for children and adolescents with anxiety disorders: Discussing + Doing = Daring.] In: C. Braet & S.M.Bögels (Eds.) Protocollaire behandelingen voor kinderen met psychische klachten [treatment protocols for children with mental health problems] (pp.325-350). Amsterdam: Boom.

Rapee, R. M. (2000). Group treatment of children with anxiety disorders: Outcome and predictors of treatment response. Australian Journal of Psychology, 52(3), 125-129. doi: 10.1080/00049530008255379

Rapee, R. M. (2003). The influence of comorbidity on treatment outcome for children and adolescents with anxiety disorders. Behaviour Research and Therapy. 41(1), 105–112. doi: 10.1016/S0005-7967(02)00049-9

Rapee, R. M., Schniering, C. A., & Hudson, J. L. (2009). Anxiety disorders during childhood and adolescence: Origins and treatment. Annual Review of Clinical Psychology, 2009(5), 311-341. doi: 10.1146/annurev.clinpsy.032408.153628

Rodebaugh, T. L., Holaway, R. M., & Heimberg, R. G. (2004). The treatment of social anxiety disorder. Clinical Psychology Review, 24(7), 883–908. doi:10.1016/j.cpr.2004.07.007 Shadish, W. R., Matt, G. E., Navarro, A. M., & Philips, G. (2000). The effects of psychological

therapies under clinically representative conditions: A meta-analysis. Psychological Bulletin, 126(4), 512-529. doi: 10.1037//0033-2909.126.4.512

Siebelink, B. M., & Treffers, Ph. D. A. (2001). ADIS-C Anxiety disorders interview schedule for DSM-IV (Dutch version), anxiety disorders interview schedule for DSM-IV. Lisse: Swets & Zeitlinger.

Silverman, W. K., & Albano, A. M. (1996). Anxiety disorders interview schedule for DSM IV child version. San Antonio, TX: Psychological Corporation.

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 & Adolescent Psychology, 37(1), 105-130. doi: 10.1080/15374410701817907

Southam-Gerow, M. A., Kendall, P. C., & Weersing, V. R. (2001). Examining outcome

variability: Correlates of treatment response in a child and adolescent anxiety clinic. Journal of Clinical Child & Adolescent Psychology, 30(3), 422-436. doi:

10.1207/S15374424JCCP3003_13

Verhulst, F. C., Ende, J., van der, Ferdinand, R. F., & Kasius, M. C. (1997). The prevalence of DSM-III-R diagnoses in a national sample of dutch adolescents. Archives of General Psychiatry, 54(4), 329-336. doi:10.1001/archpsyc.1997.01830160049008

Victor, A. M., Bernat, D. H., Bernstein, G. A., & Layne, A. E. (2007). Effects of parent and family characteristics on treatment outcome of anxious children. Journal of Anxiety Disorders, 21(6), 835–848. doi: 10.1016/j.janxdis.2006.11.005

Wolitzky-Taylor, K. B., Arch, J. J., Rosenfield, D., & Craske, M. G. (2012). Moderators and Non-specific predictors of treatment outcome for anxiety disorders: A comparison of

cognitive behavioral therapy to acceptance and Commitment Therapy. Journal of Consulting and Clinical Psychology, 80(5), 786-799. doi: 10.1037/a0029418

Woodward, L. J., & Fergusson, D. M. (2001). Life course outcomes of young people with anxiety disorders in adolescence. Journal of the American Academy of Child & Adolescent Psychiatry, 40(9), 1086–1093. doi: 10.1097/00004583-200109000-00018

Zimmerman, P., Wittchen, H. U., Höfler, M., Pfister, H., Kessler, R. C., & Lieb, R. (2003). Primary anxiety disorders and the development of subsequent alcohol use disorders: a 4-year community study of adolescents and young adults. Psychological Medicine, 33(7),