Predictors and Moderators of Treatment Outcomes for Anxious Children Randomized to Computer-Assisted Cognitive Behavioral Therapy or Standard Community Care

Abstract

Introduction:

Computer-assisted cognitive behavioral therapy (CCBT) for childhood anxiety disorders may aid the dissemination of CBT, while maintaining treatment fidelity. Although CCBT is an effective intervention, not everyone benefits equally from treatment. Identifying patient characteristics that predict who will benefit from treatment and to what extent can help with matching patients to suitable interventions, and allow researchers and clinicians to modify, and individualize, their treatment formats more effectively. Such predictors and moderators have not yet been examined for CCBT outcomes in anxious children and studies of more traditional treatment formats have yielded inconsistent results.

Methods:

Using data from a randomized clinical trial evaluating CCBT for children with anxiety disorders, this study examined predictors and moderators of treatment outcomes in a sample of 100 children (age: mean [M] = 9.82, standard deviation [SD] = 1.82), randomized to either CCBT (n = 49) or standard community care (n = 51). Potential predictors and moderators were identified from the literature and examined in stepwise multiple linear regression models, using posttreatment anxiety severity and global impairment as outcomes.

Results:

Parent-rated internalizing symptoms predicted posttreatment anxiety severity for both treatment groups. High pretreatment levels of anxiety severity predicted higher global impairment at posttreatment for the group receiving community care, but not for the CCBT group.

Conclusion:

Further research is needed to clarify which patient characteristics are associated with CCBT outcomes in a consistent way. ClinicalTrials.gov identifier: NCT01416805.

Introduction

Anxiety disorders in children and adolescents are common, debilitating, and often chronic if not appropriately treated (Cummings et al., 2014; Kendall et al., 2010; Merikangas et al., 2010; Woodward and Fergusson, 2001). Pediatric anxiety disorders confer additional risks for sustained impairment and development of co-occurring mood or substance abuse disorders in adulthood (Benjamin et al., 2013; Kendall et al., 2004; Kessler et al., 2005). Symptoms of anxiety disorders have been shown to interfere with functioning across social, academic, and familial domains (Langley et al., 2014). Cognitive-behavioral therapy (CBT) is a recommended first-line treatment for anxiety disorders in children (James et al., 2013; Sigurvinsdóttir, 2018; Silverman et al., 2008). However, CBT dissemination remains challenging and many children and adolescents with anxiety disorders are provided with nonevidence-based psychotherapy or pharmacologic interventions (Kendall et al., 2006; Novins et al., 2013; Shafran et al., 2009).

To improve CBT dissemination, while maintaining treatment fidelity, computer-assisted and computer-based CBT programs have been developed and assessed (Amir and Taylor, 2012; Kendall et al., 2011), and found to be as effective and acceptable as more traditional treatment delivery formats (Khanna and Carper, 2022; Khanna and Kendall, 2010; Spence et al., 2011; Spence et al., 2006; Storch et al., 2015). Although both traditional and computer-assisted cognitive-behavioral therapy (CCBT) are effective interventions, not everyone benefits from treatment.

Studies have found a nonresponse rate of up to 40% in clinical trials evaluating CBT for anxiety in youth (McGuire et al., 2015; Silverman et al., 2008; Storch et al., 2015). It is therefore important to examine demographic and clinical features that could predict who will respond to CBT treatment and, when developing a new format of delivery, identify who will be most likely to benefit. Identifying predictors and moderators of clinical outcomes not only helps with matching patients to suitable treatment but also knowing who fails treatment, and why, allows researchers and clinicians to modify, and individualize, their treatment formats more effectively (Kazdin, 2007; Steketee and Chambless, 1992).

Predictors refer to baseline characteristics of participants that are related to posttreatment outcomes in a consistent way (i.e., both in terms of magnitude and direction), regardless of which treatment the participant received (Kraemer et al., 2002). Predictors are not specific to one treatment or another, and are considered particularly useful in identifying, at baseline, refractory subgroups of individuals who might require new or refined interventions (Compton et al., 2014; Kraemer et al., 2002). Moderators can also be baseline characteristics of participants, which are associated with posttreatment outcomes. However, for moderators, the association differs in magnitude or direction (or both) depending on the specific treatment. That is, moderators specify for whom an assigned treatment is likely to be effective (Compton et al., 2014). Such information can be highly useful for matching individuals to specific treatments.

Predictors of traditionally delivered CBT outcomes have been previously examined for pediatric anxiety. The most consistent predictor of treatment nonresponse is higher baseline symptom severity (Compton et al., 2014; Last et al., 1998; Liber et al., 2010). Comorbid mood and externalizing disorders (Berman et al., 2000; Hudson et al., 2015; Rapee et al., 2013) and a diagnosis of social anxiety disorder (Compton et al., 2014; Hudson et al., 2015) have also been associated with poorer treatment outcome. However, these potential predictors are not reliably identified between studies and results for their significance are mixed (Knight et al., 2014). Similarly, several studies have examined potential moderators of outcomes.

Demographic variables (age (Albano et al., 2018; Kendall et al., 1997; Silk et al., 2018; Walkup et al., 2008), gender (Compton et al., 2014; Kendall et al., 2008; Silk et al., 2018), race/ethnicity (Compton et al., 2014; Taylor et al., 2018), socioeconomic status (Compton et al., 2014)), child clinical characteristics (symptom severity (Compton et al., 2014; Ginsburg et al., 2020; Nilsen et al., 2013), principal diagnoses (Albano et al., 2018; Compton et al., 2014; Ginsburg et al., 2020; Manassis et al., 2002), comorbidity (Halldorsdottir et al., 2014; Maric et al., 2018; Ollendick et al., 2008; Silk et al., 2018)), and parental factors (parental anxiety (Compton et al., 2014; Kendall et al., 2008), global psychopathology (Berman et al., 2000; Gonzalez et al., 2015; Taylor et al., 2018), overprotection (Ollendick et al., 2015), and caregiver strain (Compton et al., 2014)), have all been proposed as potential moderators and examined in trials comparing CBT to various control conditions and other treatments.

Overall, a clear and replicable pattern of moderator variables for youth CBT have not emerged, with inconsistent results across informants, measures, and studies (Norris and Kendall, 2021). That said, recent systematic reviews and meta-analyses suggest that, of all predictors analyzed, higher baseline anxiety severity and the presence of a social anxiety disorder diagnosis appear to be most consistently associated with poorer CBT outcomes (Evans et al., 2021; Kunas et al., 2021).

Research on potential predictors for CCBT for child anxiety, where computer-based interventions are used in conjunction with clinician involvement (as in the current study), are limited. In a study examining family accommodation with 72 youth with anxiety disorder, who received CCBT, the impact of accommodation on child functioning predicted remission, but not treatment response, and neither accommodation frequency nor impact of parent accommodation predicted remission or treatment response (Salloum et al., 2018). A few studies have examined predictors for a similar intervention; internet-delivered CBT (iCBT) with minimal therapist contact, for children with anxiety disorders (Spence et al., 2020). For iCBT, results are also inconsistent, with some trials finding parental factors, such as support (Spence et al., 2019), mental health (Morgan et al., 2018), and relationship quality (Spence et al., 2020), predictive of outcome, but other studies finding no such effect (Stjerneklar et al., 2019; Vigerland et al., 2016).

In a trial of CCBT for adults, only patient-rated understanding of computerized therapy material was found to predict anxiety levels at posttreatment (Craske et al., 2009). It is unclear when and for who digital mental health interventions, with or without therapist support, may be indicated for children with anxiety disorders (Khanna and Carper, 2022). Given the fact that a portion of youth do not respond to current anxiety treatments, additional work is needed to better clarify which treatments work for whom and to move the field closer to personalized treatment assignment decisions (Norris and Kendall, 2021).

As extant studies on predictors and moderators for child anxiety treatment outcomes have yielded inconsistent results, and no previous study has examined predictors and moderators of treatment outcome in CCBT, other than family accommodation, this study aims to examine potential predictors and moderators in a cohort of children receiving either CCBT or standard community care for anxiety disorders.

Since family accommodation was previously examined among children from this sample, who were immediately randomized to CCBT (n = 45), and who received CCBT after nonremission following treatment as usual (TAU) (n = 27) (Salloum et al., 2018), it was not included in this study. Other potential predictor and moderator variables were chosen based on previous studies and the following hypotheses derived from those studies: That increased anxiety severity will predict poorer treatment outcomes regardless of treatment condition (Compton et al., 2014; Last et al., 1998; Liber et al., 2010), that a diagnosis of social anxiety will predict poorer outcomes for those receiving CCBT (Compton et al., 2014; Hudson et al., 2015) and that comorbid mood and externalizing symptoms will negatively impact outcomes, and that this effect will be more pronounced in those receiving standard community care (Berman et al., 2000; Hudson et al., 2015; Rapee et al., 2013).

Methods

Participants

One hundred twenty-three children were recruited and screened through the normal patient flow at three Florida outpatient community mental health centers (Panhandle, Central Coast, and Southeastern Florida). Twenty-three youths failed to meet inclusion/exclusion criteria, resulting in a final sample of 100 children (age: M = 9.82, SD = 1.82), randomized in a 1:1 ratio to CCBT (n = 49) and standard community care (or TAU, n = 51).

All participants met the following inclusion criteria: (1) Primary anxiety diagnosis of separation anxiety disorder, social phobia, generalized anxiety disorder (GAD), specific phobia, or panic disorder, as determined by the Anxiety Disorders Interview Schedule for DSM-4 Child and Parent Versions (ADIS-IV-C/P) (Silverman and Albano, 1996) with a Clinical Severity Rating (CSR) of four or more. An independent evaluator in consensus with a review team led by a senior clinician determined the primary and comorbid diagnoses based upon degree of impairment/distress and all other available clinical information; (2) minimum score of 10 on the Pediatric Anxiety Rating Scale (PARS) (Research Units on Pediatric Psychopharmacology Anxiety Study Group, 2002), which corresponds to significant anxiety (Caporino et al., 2013); and (3) reading ability of at least SS = 85, on the Word Reading section of the Wide Range Achievement Test 4th Edition (Wilkinson et al., 2006).

Exclusion criteria were as follows: (1) Receiving concurrent psychotherapy or other counseling services, specifically targeting anxiety for the CCBT arm; (2) initiation of an antidepressant or antipsychotic medication within 6 and 12 weeks of study enrollment, respectively (antidepressants and antipsychotics). Medications were stable for 6 and 8 weeks before screening, and stable throughout treatment, although dosage reductions due to side effects were allowed in the CCBT group; (3) active suicidality or engagement in suicidal behaviors within 6 months of screening; and (4) bipolar disorder, psychosis, or autism. Families randomized to TAU were free to initiate, continue, change, or refrain from receiving any psychotherapeutic or pharmacological intervention.

Participants ranged in age from 7 to 13 years (M = 9.82), 56% were male, 72% were Caucasian, and GAD was the most common primary diagnosis (41%). Sample characteristics are presented in detail in the original study (Storch et al., 2015). There was no baseline difference between the two treatment groups, except for age, as participants randomized to the TAU group were on average 0.8 years older than those randomized to the CCBT group. In the CCBT group, 20 children (40.8%) met diagnostic criteria for social anxiety disorder, as did 16 (31.4%) in the TAU group. After eligible participants were randomized, four CCBT and four TAU participants dropped out or were withdrawn before completion due to varied reasons (e.g., due to multiple no shows, unable to be contacted, or a higher level of care needed).

Procedures

All study procedures were approved by the local Institutional Review Board. Referrals were phone-screened by site-specific staff; children who appeared to meet eligibility criteria were scheduled for a screening, at which point written informed consent and assent from the parent and child were obtained. Thereafter, a blinded independent evaluator located at a university-based research center interviewed parents and children separately through a secure internet platform using a web camera, while the others completed questionnaires. Eligible families were randomized in a 1:1 ratio into CCBT or TAU. Posttreatment assessments were completed within 1 week after the final CCBT session. All assessments were audio-recorded; 20% of the recorded PARS interviews were rated by a second rater for quality assurance (intraclass correlation = 0.87). Upon completion of the posttreatment assessment, families randomized to TAU were offered CCBT treatment. At each assessment, families were compensated $30.00.

Treatment

Camp Cope-A-Lot is a dual in-person and computerized CBT program developed for childhood anxiety with good empirical support (Donovan and March, 2014; Storch et al., 2015). Based on the Coping Cat protocol (Kendall and Hedtke, 2006a, 2006b), children received 12 weekly 50- to 60-minute sessions provided at an outpatient community mental health center. Although therapists were present at every visit, the first six sessions were primarily delivered over the computer and provided standardized content targeting the development of coping skills for managing anxiety (i.e., affective education, relaxation training, misidentification and labeling of anxiety-related cognition, and problem solving) through an interactive computer interface involving the child participating in a virtual camping experience.

The Camp Cope-A-Lot program uses computer flash animation, audio, 2D animations, photographs, videos, schematics, a built-in reward system, self-check system, written text, and a cartoon character, to guide the user through the program (Khanna and Kendall, 2008). Sessions 7–12 were primarily therapist led and involved gradual exposure to feared stimuli. Parent sessions were held at visits 3 and 7 to provide psychoeducation and foster parental support. Further details regarding procedures and treatment are available in Storch et al. (2015).

Measures

Anxiety Disorders Interview Schedule for DSM-4 Child and Parent Versions

The ADIS-IV-C/P (Silverman and Albano, 1996) is a clinician-administered semistructured interview that assesses the presence and severity of DSM-4 childhood anxiety, mood, and externalizing disorders. On a 0–8 scale, diagnostic presence and severity are established using a CSR score. The ADIS-IV-C/P has demonstrated excellent psychometric properties, including interrater reliability (Lyneham et al., 2007), and was administered at screening and posttreatment.

Pediatric Anxiety Rating Scale

The PARS (2002) is a psychometrically sound clinician-rated scale measuring child anxiety symptom presence and severity over the past week. For assessment of pretreatment anxiety severity and as the outcome variable, the clinician-rated six-item PARS Severity Scale was used in this study (Research Units on Pediatric Psychopharmacology Anxiety Study Group, 2002). The PARS contains items that assess anxiety in terms of symptom frequency, severity, severity of physical symptoms, avoidance, interference at home, and interference outside of the home (e.g., school, peer relationships). The PARS was administered at baseline and posttreatment (baseline α = 0.61).

Clinical Global Impressions-Severity

The Clinical Global Impressions-Severity (CGI-S) is a single-item rating made by the IE reflecting the global severity, impairment, and/or distress of anxiety symptoms, with scores ranging from 0 (no illness) to 6 (extremely severe symptoms) (Guy, 1976). The CGI-S was rated at baseline and posttreatment.

Child Behavior Checklist

The Child Behavior Checklist (CBCL) is a 118-item, psychometrically sound (Achenbach and Rescorla, 2001) parent-reported scale assessing a child's internalizing and externalizing symptoms. The internalizing scale was used to assess internalizing symptoms and the externalizing scale was used to assess disruptive behavior (baseline α = 0.95).

Children's Depression Inventory

The Children's Depression Inventory (CDI) is a psychometrically sound (Kovacs, 1985; Saylor et al., 1984) child-report assessment of depressive symptoms with excellent predictive utility (baseline α = 0.90) (Timbremont et al., 2004).

Childhood Anxiety Impact Scale-Child/Parent

The Childhood Anxiety Impact Scale-Child/Parent (CAIS-C/P) is a psychometrically sound (Langley et al., 2014; Langley et al., 2004) child- and parent-report measure assessing the impact of the child's anxiety on his/her psychosocial functioning in certain situations (school, social, home/family, and globally) over the past month (baseline CAIS-C α = 0.92 and CAIS-P 0.91).

Caregiver Strain Questionnaire

Caregiver Strain Questionnaire (CGSQ) is a 21-item parent self-report measure that assesses stress associated with parenting youth with emotional, behavioral, or developmental disorders (Brannan et al., 2012; Brannan et al., 1997). It consists of three subscales (Objective Strain, Subjective Internalized Strain, and Subjective Externalized Strain), and a Global Score, which is the sum of all three subscale scores. It has sound psychometric properties (Khanna et al., 2012).

Analysis

First, interaction terms were calculated between all predictor variables and treatment group allocation (e.g., baseline PARS Severity*treatment group, age*treatment group, gender*treatment group, social anxiety diagnosis*treatment group, CDI scores*treatment group, CBCL internalizing*treatment group and CBCL externalizing*treatment group). All continuous variables were standardized before interaction terms were created to avoid potential multicollinearity. A linear regression technique developed by Fournier and colleagues (Fournier et al., 2009; Gabriel et al., 2017; Taylor et al., 2018) was employed. The two outcome variables were clinician-rated anxiety severity (PARS Severity Scale) and clinician-rated global impairment (CGI-S), at posttreatment.

Potential predictors and moderators were grouped into four domains: (1) Demographic/treatment (age, gender, and treatment group (CCBT vs. TAU), age*treatment group, and gender*treatment group), and (2) child-rated (CDI, CAIS-C, CDI*treatment group, and CAIS-C *treatment group), (3) parent-rated (CAIS-P, GCSQ, CBCL internalizing, CBCL externalizing, CAIS-P *treatment group, GCSQ*treatment group, internalizing*treatment group, and externalizing*treatment group), and (4) clinician-rated variables (baseline PARS, baseline CGI-S, social anxiety diagnosis, baseline PARS*treatment group, baselines CGI-S*treatment group, and social anxiety diagnosis*treatment group). For each domain, the subsequent stepwise algorithm proceeds along the following steps:

Test a full model, including all variables and their interactions with treatment group.

Repeat the analysis retaining predictors with p < 0.20 from Step 1.

Repeat the analysis retaining predictors with p < 0.10 from Step 2.

Repeat the analysis retaining predictors with p < 0.05 from Step 3.

Step 4 is repeated until only variables with p < 0.05 remain in the model. Finally, all predictors significant at p < 0.05 in Step 4 across all domains are included in a combined model and all four steps are applied to the combined model. Variables that significantly interacted with treatment group allocation were examined as moderators. If at any step, an interaction term was significant (indicating a potential moderation effect), but not the corresponding main effect, the main effect was still retained. The merits of this analytic approach lie in its ability to limit the number of statistical tests and thus decrease unnecessary type I error accumulations and in its ability to avoid multicollinearity by inclusion of only a few variables from each domain into the final model (Gabriel et al., 2017; Powers et al., 2014).

The variable inflation factor (VIF) and tolerance index were examined for all predictor and moderator variables to examine potential multicollinearity. All variables included had a VIF lower than three and a tolerance above the recommended limit of 0.2, indicating minimal multicollinearity (Senaviratna and Cooray, 2019; Shrestha, 2020).

There was no more than 12% of data missing on any measure. As data were assumed to be missing at random, missing values for all included variables were handled with multiple imputations by chained equations. One hundred imputed datasets were generated, which were combined for regression analyses following Rubin's rules (Rubin, 2004).

Results

Multiple linear regression results with anxiety severity (PARS) as outcome

The treatment group (CCBT vs. TAU) was the only significant predictor of posttreatment PARS scores at step 4 from the demographic/treatment domain and was included in the combined model. For child-rated domain variables, none was significant beyond the step 1 threshold of p < 0.20. From the parent-rated domain, CBCL internalizing scores and the CAIS-P*treatment group interaction were significant predictors at step 4, and therefore included in the combined model. From the clinician-rated domain, only the social phobia diagnosis*treatment group interaction remained significant through step 4 and was included in the combined model.

The combined model consisted of treatment group, CBCL internalizing, CAIS-P*treatment group, and social phobia diagnosis*treatment group as potential predictor and moderators. Only CBCL internalizing was a significant predictor of posttreatment PARS scores beyond the step 1 threshold of p < 0.20 and was therefore the only variable included in the final model (Table 1). The final model explains 18.5% of the variance in posttreatment anxiety severity.

Table 1.

Multivariate Linear Regression Final Combined Model Results for Anxiety Severity (Pediatric Anxiety Rating Scale) as Outcome

Model summary
R	R square	Adjusted R square	Standard error of the estimate
0.430a	0.185	0.176	5.489

Coefficients
	B	Standard error	t	Sig.
Constant	12.599	1.418	8.887	0.000
CBCL internalizing	0.267	0.068	3.905	<0.001

Significant p-values are bolded.

CBCL, Child Behavior Checklist.

Multiple linear regression results with global impairment (CGI-S) as outcome

From the demographic/treatment domain, treatment group (CCBT vs. TAU) was the only significant predictor of posttreatment CGI-S scores at step 4. No variable from the child-rated domains was included in the final model as none was a significant predictor, beyond the p < 0.20 threshold in step 1. From the parent-rated domain, the CAIS-P *treatment group interaction was included in the final model as it was significant at step 4. From the clinician-rated domain, the baseline PARS*treatment group interaction was also significant at step 4 and therefore included in the final model.

The combined model consisted of treatment group, CAIS-P *treatment group, and PARS*treatment group. Only the baseline PARS*treatment group interaction was significant, resulting in a final model where it was included along with the main effects of baseline PARS and treatment group (Table 2). The final model explains 25.8% of the variance in posttreatment global impairment.

Table 2.

Multivariate Linear Regression, Final Combined Model Results for Global Impairment (Clinical Global Impressions-Severity) as Outcome

Model summary
R	R square	Adjusted R square	Standard error of the estimate
0.508a	0.258	0.232	0.861

Coefficients
	B	Standard error	t	Sig.
Constant	2.533	1.096	2.310	0.021
Baseline PARS	−0.001	0.052	−0.026	0.979
PARS^*treatment group	0.676	0.327	2.063	0.039
Treatment group (CCBT vs. TAU)	0.460	0.239	1.922	0.055

Significant p-values are bolded.

CCBT, computer-assisted cognitive behavioral therapy; PARS, Pediatric Anxiety Rating Scale; TAU, treatment as usual.

The baseline PARS*treatment group interaction term is significant, indicating a moderation effect. To examine this effect visually, the ModGraph-I program was used to produce a graphical display (Fig. 1) (Jose, 2013).

FIG. 1.

The impact of baseline anxiety severity (PARS) on posttreatment global impairment (CGI-S) by treatment groups. CGI-S, Clinical Global Impressions-Severity; PARS, Pediatric Anxiety Rating Scale.

For the TAU group, higher-baseline PARS severity scores are associated with higher CGI-S scores at posttreatment, while this effect is not found for the CCBT group.

Discussion

This study examined predictors and moderators of treatment outcomes in a sample of one hundred children randomized to either CCBT (n = 49) or standard community care (n = 51), from a treatment trial where CCBT was previously found to be superior to standard community care (Storch et al., 2015). As this is among the first studies to examine predictors and moderators of outcomes for CCBT for children with anxiety disorders, we included variables identified in earlier CBT studies as potential predictors and moderators even if evidence for their significance was inconsistent. Dissemination of evidence-based psychotherapy remains challenging and computer-assisted treatment delivery formats offer an accessible and effective alternative to traditional formats (Khanna and Carper, 2022; Khanna and Kendall, 2010; Storch et al., 2015). Our aim was to identify prognostic variables for such a treatment format, as well as variables that could be used to make prescriptive recommendations.

We found that age, gender, mood disorder symptoms, externalizing symptoms, functional impairment, and caregiver strain neither predicted nor moderated outcomes in this sample. As previous research has shown inconsistent results regarding the effects of these variables on treatment outcomes for children with anxiety, this was not surprising. Both treatment arms also provided flexibility regarding seeking interventions for co-occurring problems, which may have minimized between group differences. The overall low levels of externalizing problems among the sample may also have prevented us from observing associations with treatment outcomes due to limited variance.

Pretreatment anxiety severity is among the most frequently identified predictors of outcome in traditional CBT delivery formats for pediatric anxiety (Knight et al., 2014), although, as with most predictors, results are mixed (Berman et al., 2000; Compton et al., 2014; Hudson et al., 2015). We hypothesized that baseline anxiety severity would predict outcomes across both treatment conditions.

We found this to be the case only for the community care treatment arm when global impairment was the outcome. For the children in the CCBT arm, no association is found between baseline anxiety severity and posttreatment global impairment scores. This result might indicate that the active treatment condition, CCBT, has the strongest effect on those most severely anxious at baseline as a buffer for further impairment. It also indicates that usual care may be sufficient for children with less severe anxiety, although more substantial symptoms require care based on gold standard CBT principles.

Our hypothesis that a diagnosis of social anxiety would predict poorer outcomes for those receiving CCBT was not supported. A diagnosis of social anxiety disorder was not a significant predictor of outcome for either PARS or CGI-S in the overall sample, nor was it a significant moderator. A number of studies have found that children with social anxiety disorder tend to have inferior outcomes in CBT treatment (Knight et al., 2014). Furthermore, it has been suggested that children with social anxiety disorder do not respond as well to generic treatments targeting a heterogenous group of participants and diagnoses, as those with other forms of anxiety (Hudson et al., 2015).

However, these results contrast with this suggestion, indicating no significant difference in outcomes for children with social anxiety in either treatment arm. Over a third of the sample met diagnostic criteria for social anxiety disorder, so this finding can hardly be explained by low prevalence. It is worth noting that studies on iCBT (a similar treatment format to CCBT) have also found social anxiety is not predictive of outcomes (Spence et al., 2020; Stjerneklar et al., 2019; Vigerland et al., 2016). This might indicate that children with varied types of anxiety disorders respond well to computer-assisted and iCBT formats (Spence et al., 2020).

Finally, we hypothesized that comorbid mood and externalizing symptoms will negatively impact outcomes, and that this effect will be more pronounced in those receiving standard community care. This was partially supported as CBCL internalizing scores predicted higher anxiety severity at posttreatment, but it did so for the overall sample. The CBCL internalizing subscale includes symptoms of both mood and anxiety disorders and is therefore not a specific measure of mood disorders.

A more specific mood disorder measure, the CDI, was not found to be a significant predictor in this sample. This might indicate that nonspecific underlying internalizing psychopathology, such as that targeted in transdiagnostic treatment models for emotional disorders (Ehrenreich-May et al., 2017; Weersing et al., 2012), might be the factor predictive of treatment outcomes rather than specific mood or anxiety symptoms. This is further supported by the result that anxiety severity was by itself not a unique predictor of outcome in the regression models.

Several study limitations should be noted. First, although participants were recruited from three geographically diverse clinics, most of the sample was Caucasian. Second, although the TAU condition allowed participants to seek treatment, only 55.3% received active intervention, and it is unknown whether those interventions conformed to evidence-based standards.

Third, although comorbid mental health conditions were assessed, environmental and home stressors (e.g., changing schools, family dysfunction, and trauma history) were not, which may have impacted treatment outcomes.

Conclusion

This study examined potential predictors or moderators of treatment outcomes, for children receiving either CCBT or standard community care. Results indicated that parent-rated internalizing symptoms predicted anxiety severity at posttreatment for the whole sample and that pretreatment levels of anxiety predicted higher global impairment at posttreatment for the group receiving community care, but not for the CCBT group.

As this is the first study to examine these predictors and moderators for CCBT treatment, and previous studies of traditional CBT delivery formats have yielded highly inconsistent results, these results should be interpreted as preliminary. Further research is needed to clarify which patient characteristics, if any, are consistently associated with CCBT outcomes.

Clinical Significance

Clinicians should be aware that higher internalizing symptoms are associated with attenuated treatment response for children with anxiety disorders. For this sample, outcomes for those with social anxiety are similar to those with other disorders. For those receiving CCBT, higher anxiety levels did not predict more impairment at posttreatment, like it did for those receiving standard community care, indicating the importance of evidence-based intervention for highly anxious children.

Footnotes

Acknowledgments

The contributions of Tyne Pierce, Amanda Krucke, Christin Cooper, Wendy Kubar, Stephanie Dobband, April Lott, Ashley Holden, Elise Ward, Sonya Hernandez, Bhagirathy Sahasranaman, Nathalie Miniscalco, Pamela Galan, Tanya White, Lori Olsen, Shannon Massingale, Ruqayyah Gaber, John Bilbrey, Carol Clark, Shaun Dahle, Ed Mobley, Larry Williams, Michael Sulkowski, Elysse Arnold, Alessandro de Nadai, Joshua Nadeau, Anna Jones, Brittany Kugler, Joseph McGuire, Tanya Murphy, Danielle Ung, Jennifer Park, Marie McPherson, Robert Constantine, Adam Lewin, Nick Dewan, Muniya Khanna, and Nicole McBride, are gratefully acknowledged.

Authors' Contributions

O.S.: Conceptualization, data curation, formal analysis, and writing—original draft. A.G.G.: Writing—review and editing and formal analysis (supporting). W.K.G.: Writing—review and editing. A.S.: Writing—review and editing. E.A.S.: Conceptualization, supervision, funding acquisition, project administration, and writing—review and editing.

Disclosures

O.S. has no disclosure to report.

References

Achenbach

, Rescorla

. Manual for the ASEBA School-Age Forms & Profiles. University of Vermont, Research Center for Children, Youth, and Families: Burlington, VT; 2001.

Albano

, Comer

, Compton

, et al. Secondary outcomes from the child/adolescent anxiety multimodal study: Implications for clinical practice. Evid Based Pract Child Adolesc Mental Health, 2018; 3:30–41.

Amir

, Taylor

. Combining computerized home-based treatments for generalized anxiety disorder: An attention modification program and cognitive behavioral therapy. Behav Ther, 2012; 43:546–559.

Benjamin

, Harrison

, Settipani

, et al. Anxiety and related outcomes in young adults 7 to 19 years after receiving treatment for child anxiety. J Consult Clin Psychol, 2013; 81:865–876.

Berman

, Weems

, Silverman

, et al. Predictors of outcome in exposure-based cognitive and behavioral treatments for phobic and anxiety disorders in children. Behav Ther, 2000; 31:713–731.

Brannan

, Athay

, De Andrade

ARV

. Measurement quality of the Caregiver Strain Questionnaire-short form 7 (CGSQ-SF7). Admin Policy Mental Health Mental Health Serv Res, 2012; 39:51–59.

Brannan

, Heflinger

, Bickman

. The Caregiver Strain Questionnaire: Measuring the impact on the family of living with a child with serious emotional disturbance. J Emot Behav Disord, 1997; 5:212–222.

Caporino

, Brodman

, Kendall

, et al. Defining treatment response and remission in child anxiety: Signal detection analysis using the Pediatric Anxiety Rating Scale. J Am Acad Child Adolesc Psychiatry, 2013; 52:57–67.

Compton

, Peris

, Almirall

, et al. Predictors and moderators of treatment response in childhood anxiety disorders: Results from the CAMS trial. J Consult Clin Psychol, 2014; 82:212–224.

10.

Craske

, Rose

, Lang

, et al. Computer-assisted delivery of cognitive behavioral therapy for anxiety disorders in primary-care settings. Depress Anxiety, 2009; 26:235–242.

11.

Cummings

, Caporino

, Kendall

. Comorbidity of anxiety and depression in children and adolescents: 20 years after. Psychol Bull, 2014; 140:816–845.

12.

Donovan

, March

. Computer-based treatment programs for youth anxiety: A systematic review. Psychopathol Rev, 2014; 1:130–156.

13.

Ehrenreich-May

, Kennedy

, Sherman

, et al. Unified Protocols for Transdiagnostic Treatment of Emotional Disorders in Children and Adolescents: Therapist Guide. Oxford University Press: New York, NY; 2017.

14.

Evans

, Clark

, Leigh

. Are young people with primary social anxiety disorder less likely to recover following generic CBT compared to young people with other primary anxiety disorders? A systematic review and meta-analysis. Behav Cogn Psychother, 2021; 49:352–369.

15.

Fournier

, Derubeis

, Shelton

, et al. Prediction of response to medication and cognitive therapy in the treatment of moderate to severe depression. J Consult Clin Psychol, 2009; 77:775.

16.

Gabriel

, Jakubovski

, Taylor

, et al. Predictors of treatment response and drop out in the Treatment of Early-Onset Schizophrenia Spectrum Disorders (TEOSS) study. Psychiatry Res, 2017; 255:248–255.

17.

Ginsburg

, Pella

, Pikulski

, et al. School-based treatment for anxiety research study (STARS): A randomized controlled effectiveness trial. J Abnorm Child Psychol, 2020; 48:407–417.

18.

Gonzalez

, Peris

, Vreeland

, et al. Parental anxiety as a predictor of medication and CBT response for anxious youth. Child Psychiatry Hum Dev, 2015; 46:84–93.

19.

Guy

ECDEU Assessment Manual for Psychopharmacology. Government Printing Office: Washington, DC; 1976.

20.

Halldorsdottir

, Ollendick

, Ginsburg

, et al. Treatment outcomes in anxious youth with and without comorbid ADHD in the CAMS. J Clin Child Adolesc Psychol, 2015; 44(6):985–991.

21.

Hudson

, Keers

, Roberts

, et al. Clinical predictors of response to cognitive-behavioral therapy in pediatric anxiety disorders: The genes for treatment (GxT) study. J Am Acad Child Adolesc Psychiatry, 2015; 54:454–463.

22.

James

, James

, Cowdrey

, et al. Cognitive behavioural therapy for anxiety disorders in children and adolescents. Cochrane Database Syst Rev, 2013; 6:CD004690.

23.

Jose

PE.

ModGraph-I: A programme to compute cell means for the graphical display of moderational analyses: The internet version, Version 3.0. Victoria University of Wellington: Wellington, NZ; 2013.

24.

Kazdin

AE.

Mediators and mechanisms of change in psychotherapy research. Annu Rev Clin Psychol, 2007; 3:1–27.

25.

Kendall

, Compton

, Walkup

, et al. Clinical characteristics of anxiety disordered youth. J Anxiety Disord, 2010; 24:360–365.

26.

Kendall

, Flannery-Schroeder

, Panichelli-Mindel

, et al. Therapy for youths with anxiety disorders: A second randomized clinical trial. J Consult Clin Psychol, 1997; 65:366–380.

27.

Kendall

, Hedtke

. Cognitive-Behavioral Therapy for Anxious Children: Therapist Manual. Workbook Publishing: Ardmore, PA; 2006a.

28.

Kendall

, Hedtke

. Coping Cat Workbook. Workbook Publishing: Ardmore, PA; 2006b.

29.

Kendall

, Hudson

, Gosch

, et al. Cognitive-behavioral therapy for anxiety disordered youth: A randomized clinical trial evaluating child and family modalities. J Consult Clin Psychol, 2008; 76:282–297.

30.

Kendall

, Khanna

, Edson

, et al. Computers and psychosocial treatment for child anxiety: Recent advances and ongoing efforts. Depress Anxiety, 2011; 28:58–66.

31.

Kendall

, Safford

, Flannery-Schroeder

, et al. Child anxiety treatment: Outcomes in adolescence and impact on substance use and depression at 7.4-year follow-up. J Consult Clin Psychol, 2004; 72:276–287.

32.

Kendall

, Suveg

, Julie

. Facilitating the dissemination and application of cognitive-behavioral therapy for anxiety in youth. J Cogn Psychol, 2006; 20(3):243–245.

33.

Kessler

, Berglund

, Demler

, et al. Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication.[Erratum appears in Arch Gen Psychiatry. 2005;62(7):768 Note: Merikangas, Kathleen R [added]]. Arch Gen Psychiatry, 2005; 62:593–602.

34.

Khanna

, Carper

. Digital mental health interventions for child and adolescent anxiety. Cogn Behav Pract, 2022; 29:60–68.

35.

Khanna

, Kendall

. Computer-assisted CBT for child anxiety: The coping cat CD-ROM. Cogn Behav Pract, 2008; 15:159–165.

36.

Khanna

, Kendall

. Computer-assisted cognitive behavioral therapy for child anxiety: Results of a randomized clinical trial. J Consult Clin Psychol, 2010; 78:737–745.

37.

Khanna

, Madhavan

, Smith

, et al. Psychometric properties of the Caregiver Strain Questionnaire (CGSQ) among caregivers of children with autism. Autism, 2012; 16:179–199.

38.

Knight

, Mclellan

, Jones

, et al. Pre-treatment predictors of outcome in childhood anxiety disorders: A systematic review. Psychopathol Rev, 2014; 1:77–129.

39.

Kovacs

The Children's Depression, Inventory (CDI). Psychopharmacol Bull, 1985; 21:995–998.

40.

Kraemer

, Wilson

, Fairburn

, et al.. Mediators and moderators of treatment effects in randomized clinical trials. Arch Gen Psychiatry, 2002; 59:877–883.

41.

Kunas

, Lautenbacher

, Lueken

, et al. Psychological predictors of cognitive-behavioral therapy outcomes for anxiety and depressive disorders in children and adolescents: A systematic review and meta-analysis. J Affect Disord, 2021; 278:614–626.

42.

Langley

, Bergman

, Mccracken

, et al. Impairment in childhood anxiety disorders: Preliminary examination of the child anxiety impact scale-parent version. J Child Adolesc Psychopharmacol, 2004; 14:105–114.

43.

Langley

, Falk

, Peris

, et al. The child anxiety impact scale: Examining parent- and child-reported impairment in child anxiety disorders. J Clin Child Adolesc Psychol, 2014; 43:579–591.

44.

Last

, Hansen

, Franco

. Cognitive-behavioral treatment of school phobia. J Am Acad Child Adolesc Psychiatry, 1998; 37:404–411.

45.

Liber

, Van Widenfelt

, Van Der Leeden

, et al. The relation of severity and comorbidity to treatment outcome with cognitive behavioral therapy for childhood anxiety disorders. J Abnorm Child Psychol, 2010; 38:683–694.

46.

Lyneham

, Abbott

, Rapee

. Interrater reliability of the anxiety disorders interview schedule for DSM-IV: Child and parent version. J Am Acad Child Adolesc Psychiatry, 2007; 46:731–736.

47.

Manassis

, Mendlowitz

, Scapillato

, et al. Group and individual cognitive-behavioral therapy for childhood anxiety disorders: A randomized trial. J Am Acad Child Adolesc Psychiatry, 2002; 41:1423–1430.

48.

Maric

, Van Steensel

, Bögels

. Parental involvement in CBT for anxiety-disordered youth revisited: Family CBT outperforms child CBT in the long term for children with comorbid ADHD symptoms. J Attention Disord, 2018; 22:506–514.

49.

Mcguire

, Piacentini

, Lewin

, et al. A meta-analysis of cognitive behavior therapy and medication for child obsessive-compulsive disorder: Moderator of treatment efficacy, response, and remission. Depression Anxiety, 2015; 32:580–593.

50.

Merikangas

, He

J-P

, Brody

, et al. Prevalence and treatment of mental disorders among US children in the 2001–2004 NHANES. Pediatrics, 2010; 125:75–81.

51.

Morgan

, Rapee

, Salim

, et al. Predicting response to an internet-delivered parenting program for anxiety in early childhood. Behav Ther, 2018; 49:237–248.

52.

Nilsen

, Eisemann

, Kvernmo

. Predictors and moderators of outcome in child and adolescent anxiety and depression: A systematic review of psychological treatment studies. Eur Child Adolesc Psychiatry, 2013; 22:69–87.

53.

Norris

, Kendall

. Moderators of outcome for youth anxiety treatments: Current findings and future directions. J Clin Child Adolesc Psychol, 2021; 50:450–463.

54.

Novins

, Green

, Legha

, et al. Dissemination and implementation of evidence-based practices for child and adolescent mental health: A systematic review. J Am Acad Child Adolesc Psychiatry, 2013; 52:1009–1025.e18.

55.

Ollendick

, Halldorsdottir

, Fraire

, et al. Specific phobias in youth: A randomized controlled trial comparing one-session treatment to a parent-augmented one-session treatment. Behav Ther, 2015; 46:141–155.

56.

Ollendick

, Jarrett

, Grills-Taquechel

, et al. Comorbidity as a predictor and moderator of treatment outcome in youth with anxiety, affective, attention deficit/hyperactivity disorder, and oppositional/conduct disorders. Clin Psychol Rev, 2008; 28:1447–1471.

57.

Powers

, Warren

, Rosenfield

, et al. Predictors of PTSD symptoms in adults admitted to a Level I trauma center: A prospective analysis. J Anxiety Disord, 2014; 28:301–309.

58.

Rapee

, Lyneham

, Hudson

, et al. Effect of comorbidity on treatment of anxious children and adolescents: Results from a large, combined sample. J Am Acad Child Adolesc Psychiatry, 2013; 52:47–56.

59.

Research Units on Pediatric Psychopharmacology Anxiety Study Group. The Pediatric Anxiety Rating Scale (PARS): Development and psychometric properties. J Am Acad Child Adolesc Psychiatry, 2002; 41:1061–1069.

60.

Rubin

DB.

Multiple Imputation for Nonresponse in Surveys. Wiley Inter-Science: New York; 2004.

61.

Salloum

, Andel

, Lewin

, et al. Family accommodation as a predictor of cognitive-behavioral treatment outcome for childhood anxiety. Fam Soc, 2018; 99:45–55.

62.

Saylor

, Finch

, Spirito

, et al. The children's depression inventory: A systematic evaluation of psychometric properties. J Consult Clin Psychol, 1984; 52:955–967.

63.

Senaviratna

, Cooray

. Diagnosing multicollinearity of logistic regression model. Asian J Probabil Stat, 2019; 5:1–9.

64.

Shafran

, Clark

, Fairburn

, et al. Mind the gap: Improving the dissemination of CBT. Behav Res Ther, 2009; 47:902–909.

65.

Shrestha

Detecting multicollinearity in regression analysis. Am J Appl Mathemat Stat, 2020; 8:39–42.

66.

Sigurvinsdóttir

AL.

Effectiveness of Cognitive Behavioral Therapy (CBT) for Child and Adolescent Anxiety Disorders Across Different CBT Modalities and Comparisons: A Systematic Review and Meta-Analysis. University of Iceland: Reykjavik, Iceland; 2018.

67.

Silk

, Tan

, Ladouceur

, et al. A randomized clinical trial comparing individual cognitive behavioral therapy and child-centered therapy for child anxiety disorders. J Clin Child Adolesc Psychol, 2018; 47:542–554.

68.

Silverman

, Albano

. Anxiety Disorders Interview Schedule for DSM-IV: Child Version. The Psychologial Corporation/Graywind Publications: San Antonio, TX; 1996.

69.

Silverman

, Pina

, Viswesvaran

. Evidence-based psychosocial treatments for phobic and anxiety disorders in children and adolescents. J Clin Child Adolesc Psychol, 2008; 37:105–130.

70.

Spence

, Donovan

, March

, et al. A randomized controlled trial of online versus clinic-based CBT for adolescent anxiety. J Consult Clin Psychol, 2011; 79:629–642.

71.

Spence

, Holmes

, March

, et al. The feasibility and outcome of clinic plus internet delivery of cognitive-behavior therapy for childhood anxiety. J Consult Clin Psychol, 2006; 74:614–621.

72.

Spence

, March

, Donovan

. Social support as a predictor of treatment adherence and response in an open-access, self-help, internet-delivered cognitive behavior therapy program for child and adolescent anxiety. Internet Interv, 2019; 18:100268.

73.

Spence

, Prosser

, March

, et al. Internet-delivered cognitive behavior therapy with minimal therapist support for anxious children and adolescents: Predictors of response. J Child Psychol Psychiatry, 2020; 61:914–927.

74.

Steketee

, Chambless

. Methodological issues in prediction of treatment outcome. Clin Psychol Rev, 1992; 12:387–400.

75.

Stjerneklar

, Hougaard

, Thastum

. Guided internet-based cognitive behavioral therapy for adolescent anxiety: Predictors of treatment response. Internet Interven, 2019; 15:116–125.

76.

Storch

, Salloum

, King

, et al. A randomized controlled trial in community mental health centers of computer-assisted cognitive behavioral therapy versus treatment as usual for children with anxiety. Depress Anxiety, 2015; 32:843–852.

77.

Taylor

, Lebowitz

, Jakubovski

, et al. Monotherapy insufficient in severe anxiety? Predictors and moderators in the child/adolescent anxiety multimodal study. J Clin Child Adolesc Psychol, 2018; 47:266–281.

78.

Timbremont

, Braet

, Dreessen

. Assessing depression in youth: Relation between the Children's Depression Inventory and a structured interview. J Clin Child Adolesc Psychol, 2004; 33:149–157.

79.

Vigerland

, Lenhard

, Bonnert

, et al. Internet-delivered cognitive behavior therapy for children and adolescents: A systematic review and meta-analysis. Clin Psychol Rev, 2016; 50:1–10.

80.

Walkup

, Albano

, Piacentini

, et al. Cognitive behavioral therapy, sertraline, or a combination in childhood anxiety. New Engl J Med, 2008; 359:2753–2766.

81.

Weersing

, Rozenman

, Maher-Bridge

, et al. Anxiety, depression, and somatic distress: Developing a transdiagnostic internalizing toolbox for pediatric practice. Cogn Behav Pract, 2012; 19:68–82.

82.

Wilkinson

, Robertson

, Psychological Assessment

Resources

, Inc. 2006. Wrat 4: Wide Range Achievement Rest; Professional Manual, 4th ed. Psychological Assessment, Resources: Lutz, FL.

83.

Woodward

, Fergusson

. Life course outcomes of young people with anxiety disorders in adolescence. J Am Acad Child Adolesc Psychiatry, 2001; 40:1086–1093.