Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Review - Intervention

Cognitive behavioural therapy for anxiety disorders in children and adolescents

Authors' declarations of interest

Version published: 16 November 2020 Version history

https://doi.org/10.1002/14651858.CD013162.pub2
Collapse all Expand all

Background

Previous Cochrane Reviews have shown that cognitive behavioural therapy (CBT) is effective in treating childhood anxiety disorders. However, questions remain regarding the following: up‐to‐date evidence of the relative efficacy and acceptability of CBT compared to waiting lists/no treatment, treatment as usual, attention controls, and alternative treatments; benefits across a range of outcomes; longer‐term effects; outcomes for different delivery formats; and amongst children with autism spectrum disorders (ASD) and children with intellectual impairments.

Objectives

To examine the effect of CBT for childhood anxiety disorders, in comparison with waitlist/no treatment, treatment as usual (TAU), attention control, alternative treatment, and medication.

Search methods

We searched the Cochrane Common Mental Disorders Controlled Trials Register (all years to 2016), the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and PsycINFO (each to October 2019), international trial registries, and conducted grey literature searches.

Selection criteria

We included randomised controlled trials of CBT that involved direct contact with the child, parent, or both, and included non‐CBT comparators (waitlist/no treatment, treatment as usual, attention control, alternative treatment, medication). Participants were younger than age 19, and met diagnostic criteria for an anxiety disorder diagnosis. Primary outcomes were remission of primary anxiety diagnosis post‐treatment, and acceptability (number of participants lost to post‐treatment assessment), and secondary outcomes included remission of all anxiety diagnoses, reduction in anxiety symptoms, reduction in depressive symptoms, improvement in global functioning, adverse effects, and longer‐term effects.

Data collection and analysis

We used standard methodological procedures as recommended by Cochrane. We used GRADE to assess the quality of the evidence.

Main results

We included 87 studies and 5964 participants in quantitative analyses.

Compared with waitlist/no treatment, CBT probably increases post‐treatment remission of primary anxiety diagnoses (CBT: 49.4%, waitlist/no treatment: 17.8%; OR 5.45, 95% confidence interval (CI) 3.90 to 7.60; n = 2697, 39 studies, moderate quality); NNTB 3 (95% CI 2.25 to 3.57) and all anxiety diagnoses (OR 4.43, 95% CI 2.89 to 6.78; n = 2075, 28 studies, moderate quality).

Low‐quality evidence did not show a difference between CBT and TAU in post‐treatment primary anxiety disorder remission (OR 3.19, 95% CI 0.90 to 11.29; n = 487, 8 studies), but did suggest CBT may increase remission from all anxiety disorders compared to TAU (OR 2.74, 95% CI 1.16 to 6.46; n = 203, 5 studies).

Compared with attention control, CBT may increase post‐treatment remission of primary anxiety disorders (OR 2.28, 95% CI 1.33 to 3.89; n = 822, 10 studies, low quality) and all anxiety disorders (OR 2.75, 95% CI 1.22 to 6.17; n = 378, 5 studies, low quality).

There was insufficient available data to compare CBT to alternative treatments on post‐treatment remission of primary anxiety disorders, and low‐quality evidence showed there may be little to no difference between these groups on post‐treatment remission of all anxiety disorders (OR 0.89, 95% CI 0.35 to 2.23; n = 401, 4 studies)

Low‐quality evidence did not show a difference for acceptability between CBT and waitlist/no treatment (OR 1.09, 95% CI 0.85 to 1.41; n=3158, 45 studies), treatment as usual (OR 1.37, 95% CI 0.73 to 2.56; n = 441, 8 studies), attention control (OR 1.00, 95% CI 0.68 to 1.49; n = 797, 12 studies) and alternative treatment (OR 1.58, 95% CI 0.61 to 4.13; n=515, 7 studies).

No adverse effects were reported across all studies; however, in the small number of studies where any reference was made to adverse effects, it was not clear that these were systematically monitored.

Results from the anxiety symptom outcomes, broader outcomes, longer‐term outcomes and subgroup analyses are provided in the text.

We did not find evidence of consistent differences in outcomes according to delivery formats (e.g. individual versus group; amount of therapist contact time) or amongst samples with and without ASD, and no studies included samples of children with intellectual impairments.

Authors' conclusions

CBT is probably more effective in the short‐term than waiting lists/no treatment, and may be more effective than attention control. We found little to no evidence across outcomes that CBT is superior to usual care or alternative treatments, but our confidence in these findings are limited due to concerns about the amount and quality of available evidence, and we still know little about how best to efficiently improve outcomes.

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Cognitive behavioural therapy for anxiety in children and young people

Why is this review important?

Many children and young people experience problems with anxiety. Children and young people with anxiety disorders are more likely than their peers to have difficulty with friendships, family life, and school, and to develop mental health problems later in life. Therapies such as cognitive behavioural therapy (CBT) can help children and young people to overcome difficulties with anxiety by using new ways of thinking and facing their fears.

Who will be interested in this review?

Parents, children, and young people; people working in education and mental health services for children and young people; and general practitioners.

What questions does this review aim to answer?

This review updates and replaces previous Cochrane Reviews from 2005 and 2015, which showed that CBT is an effective treatment for children and young people with anxiety disorders.

This review aimed to answer the following questions:

• Is CBT more effective than a waiting list or no treatment?

• Is CBT more effective than other treatments and medication?

• Does CBT help to reduce anxiety for children and young people in the longer term?

• Are some types of CBT more effective than others? (e.g. individual versus group therapy)

• Is CBT effective for specific groups? (e.g. children with autism)

Which studies were included in the review?

We searched the databases to find all studies of CBT for anxiety disorders in children and young people published up to October 2019. In order to be included in the review, studies had to be randomised controlled trials (a type of study in which participants are assigned to one of two or more treatment groups using a random method) and had to include young people under 19 years of age with an anxiety disorder diagnosis. We included 87 studies with a total of 5964 participants in the analysis.

What does the evidence from the review tell us?

We rated the overall quality of the evidence as 'moderate’ or 'low'. There is evidence that CBT is more effective than a waiting list or no treatment in reducing anxiety in children and young people, although the findings did vary across studies. There is no clear evidence that CBT is more effective than other treatments. A small number of studies looked at outcomes six months after CBT was given and showed that reductions in anxiety continued. We found no clear evidence that one way of providing CBT is more effective than another (e.g. in a group, longer treatments, with parents) or that CBT is more or less effective for any specific group of children (e.g. children with autism spectrum disorders).

What should happen next?

Future research should compare CBT to alternative treatments and medication; identify who does and does not benefit from CBT and what those who do not benefit need; establish how to make CBT more accessible; and give far more consideration to neglected populations, including children and young people from low‐ and middle‐income countries.

Authors' conclusions

Implications for practice

This review reinforces conclusions from previous reviews finding moderate‐quality evidence that cognitive behavioural therapy (CBT) is more effective in improving diagnostic outcomes in the short term than a waiting list or no treatment. The advantage of CBT compared to waitlist/no treatment is evident across a range of other short‐term outcomes, although variation in findings or limited available data, or both, reduces certainty in these results. Clear and consistent evidence of differences between the formats of individual, group, and parent/family CBT continues to be lacking. Available, low‐quality evidence supports a benefit of CBT compared to attention control, but the evidence base remains small, and there continues to be insufficient and inclusive evidence that CBT is superior to either treatment as usual or alternative treatments.

Pharmacological interventions are typically not recommended as a first‐line treatment due to patient preference (e.g. Brown 2007), relative high rates of attrition, and concerns about potential side effects (Wang 2017). However, we know very little about the comparative efficacy of pharmacological interventions and CBT, with evidence still limited to one, albeit large, multicentre study (Walkup 2008). Clearly further work is needed to guide practice about the circumstances in which pharmacological approaches might be most acceptable and effective for the treatment of anxiety disorders in children and young people.

A major challenge for clinical practice is the very limited access that children with anxiety disorders have to evidence‐based psychological interventions, such as CBT. In a recent study, only 2% of children with anxiety disorders identified in the community in England had received CBT (Reardon 2020). These findings are consistent with previous studies indicating low rates of service use in the UK, Australia, and the USA (Green 2004; Lawrence 2015; Merikangas 2011). Of importance to this clinical challenge is our finding that there is currently no evidence of differences in outcome on the basis of therapist contact time, with similar effects being achieved when therapists provided less than 10 hours of clinical contact compared to higher amounts of clinical contact in terms of remission of the primary anxiety disorder and child‐ and parent‐reported anxiety symptoms. Notably, results were less strong in terms of remission of all anxiety disorders; however, there is evidence to suggest that benefits might generalise to other disorders beyond the end of (brief) treatment (e.g. Ollendick 2010). These findings indicate that access could be improved through the provision of stepped‐care models of treatment delivery in which relatively brief interventions are offered as a first‐line approach, and more intensive interventions are saved for those who need them (Bower 2005). A recent systematic evaluation of stepped‐care treatment for child anxiety disorders provided promising evidence for a benefit of a stepped‐care approach in terms of clinical time saved and reduced societal costs (Chatterton 2019).

Implications for research

There is now sufficient evidence to conclude that CBT is an effective treatment for anxiety disorders in children and young people when compared to waiting lists/no treatment.

A number of reviews have concluded that overall about 50% of children and young people recover from their primary anxiety disorder following a course of CBT, but still very little is known about how these outcomes can be improved and what the other 50% of patients need. A better understanding of what the core mechanisms of effective interventions are and how this varies according to individual characteristics or contexts is needed. For example, currently very little is known about how treatments should best be delivered according to child age and developmental status, the presence of comorbid conditions, or the nature of the anxiety disorders experienced. Almost three‐quarters of the studies in this review included children with a broad range of anxiety disorders, and those that focused on specific disorders were spread across studies of, for example, social anxiety disorder, selective mutism, panic disorder, generalised anxiety disorder, and specific phobia, thereby limiting our ability to draw firm conclusions about any particular approach for any particular disorder, let alone how that might vary according to, for example, child age. Larger studies that are sufficiently powered to examine predictors, mediators, and moderators of treatment outcomes are now needed to identify how to improve treatment outcomes and develop more personalised models of care.

Individual patient data meta‐analyses may be useful to help us better understand the extensive heterogeneity between study outcomes; however, meaningful conclusions will rely on clear and consistent reporting across studies so that data can be linked in sensible ways. In this review we focused on remission of the primary and all anxiety diagnoses, but a number of studies were excluded from these analyses as they reported other kinds of data from diagnostic assessments. This issue was highlighted in a recent position paper which set out a number of recommendations for reporting and terminology to promote greater consistency going forwards (Creswell 2020a). It is important to note a number of other areas where the field would benefit from greater clarity and consistency in reporting by researchers. For example, in treatment as usual studies, it was not always clear what 'treatment as usual' actually consisted of and to what extent it differed from the index treatment or a no treatment control. Indeed, categorising the format of any treatment typically relied on only a brief description, which may have led to faulty conclusions. We encourage researchers to make good use of online supplementary materials to ensure clear descriptions of the content and format of interventions, using consistent terminology (Creswell 2020a).

There continues to be a lack of evidence that CBT is superior to other credible alternative treatments, so it will be essential to identify what the core components of effective treatments are, for example through experimental and treatment dismantling studies (e.g. Whiteside 2015), so that outcomes can be optimised and treatments can be targeted to maximise their efficiency. Head‐to‐head comparisons of different active (and credible) interventions will be required, but these will inevitably require larger sample sizes, as they will need to be set up to detect smaller effects than those found with waitlist/no treatment comparisons; however, these designs will also enable studies to compare longer‐term outcomes between groups. Given the risks for longer‐term adjustment of anxiety disorders in childhood and adolescence, controlled studies with longer‐term outcomes are essential to ensure that children’s life trajectories are being optimally supported through early intervention.

As noted above, increasing access to effective treatments is an urgent challenge for our field, and it will be essential that studies consider accessibility of treatment and cost‐effectiveness going forwards. We did not consider outcomes in terms of cost‐effectiveness in this review; however, it is clear that to date very few studies have taken cost‐effectiveness into account when evaluating CBT for child anxiety disorders. We anticipate that cost‐effectiveness data will have a marked impact on the clinical implications that can be drawn from the available evidence, particularly given our finding that clinician contact time did not seem to be related to treatment outcome. Equally, it is often assumed that group treatments may be cost‐effective; however, this has not been supported when evaluated in other contexts (e.g. Mavranezouli 2015), and some of the group treatments identified in this review involved very high levels of therapist involvement and supported activities. Future studies would benefit from including a cost‐effectiveness analysis, and considering the costs and benefits to both the health service and broader society including costs related to the child themselves and family members, whose work and home lives are often heavily impacted by child anxiety problems (e.g. Bodden 2008b).

Finally, it was disappointing to see that, whilst the number of trials evaluating CBT for childhood anxiety disorders has increased substantially, these have not extended to children in low‐ or middle‐income countries, children with intellectual disabilities or physical health conditions, the most severely affected children in inpatient settings, and there are still very few studies that have focused specifically on either young children or (strikingly given the prevalence of anxiety disorders) adolescents.

Summary of findings

Open in table viewer
Summary of findings 1. CBT compared with waitlist for children and adolescents with anxiety disorders

CBT compared with waitlist for children and adolescents with anxiety disorders

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: waitlist/no treatment

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Waitlist1

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

178 per 1000

541 per 1000
(458 to 622)

OR 5.45 (3.90 to 7.60)

2697
(39 studies)

⊕⊕⊕⊝
moderate2

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 8.57, df = 2, P = 0.01, I² = 76.7%)

  • No difference in outcomes for individual versus group formats (Chi² = 0.90, df = 1, P = 0.34; I² = 0%)

  • No difference in outcomes for interventions with different amounts of therapist contact (Chi² = 0.52, df = 2, P = 0.77; I² = 0%)

  • No difference in outcomes for participants with and without ASD (Chi² = 2.25, df = 1, P = 0.13; I² = 55.5%)

Acceptability (number of participants lost to post‐treatment assessment)

104 per 1000

112 per 1000
(90 to 141)

OR 1.09 (0.85 to 1.41)

3158
(45 studies)

⊕⊕⊝⊝
low3

Remission of all anxiety diagnoses post‐treatment (ITT)

191 per 1000

512 per 1000
(406 to 616)

OR 4.43 (2.89 to 6.78)

2075

(28 studies)

⊕⊕⊕⊝
moderate2

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 8.14, df = 2, P = 0.02, I² = 75.4%)

  • No difference in outcomes for individual versus group formats (Chi² = 0.35, df = 1, P = 0.56; I² = 0%)

  • Difference in outcomes for interventions with different amounts of therapist contact (Chi² = 10.75, df = 2, P= 0.005; I² = 81.4%)

  • Insufficient studies for subgroup analyses examining outcomes for participants with and without ASD

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.67standard deviations lower (0.88 to 0.47 lower).

Moderate effect size

2831
(45 studies)

⊕⊕⊝⊝
low4

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 14.67,df = 2, P < 0.001, I² = 86.4%)

  • Difference in outcomes for individual versus group formats (Chi² =6.47, df = 1, P = 0.01; I² = 84.5%)

  • No difference in outcomes for interventions with different amounts of therapist contact (Chi² = 3.33, df = 2, P = 0.19; I² = 39.9%)

  • No difference in outcomes for participants with and without ASD (Chi² = 0.02, df = 1, P = 0.88; I² = 0%)

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (parent report) in the CBT groups was 0.70standard deviations lower (0.90 to 0.51 lower).

Moderate effect size

2137
(35 studies)

⊕⊕⊝⊝
low4

Subgroup analyses:

  • No difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 3.43, df = 2, P = 0.18, I² = 41.8%)

  • Difference in outcomes for individual versus group formats (Chi² = 6.79,df = 1, P = 0.009, I² = 85.3%)

  • No difference in outcomes for interventions with different amounts of therapist contact (Chi² = 3.77, df = 2, P = 0.15; I² = 46.9%)

  • No difference in outcomes for participants with and without ASD (Chi² = 1.42, df = 1, P = 0.23; I² = 29.8%)

Reduction in depressive symptoms post‐treatment

The mean depressive symptoms in the CBT groups was 0.34 standard deviations lower (0.51 to 0.17 lower).

Small effect size

1157

(17 studies)

⊕⊕⊕⊝
moderate2

Improvement in global functioning post‐treatment

The mean global functioning in the CBT groups was 1.03 standard deviations higher (0.68 to 1.38 higher).

Large effect size

557

(11 studies)

⊕⊕⊝⊝
low5

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and waitlist/no treatment groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of waitlist groups.
2Downgraded one level due to moderate heterogeneity (inconsistency).
3Downgraded two levels due large variation in treatment effects across studies (inconsistency) and wide confidence intervals (imprecision).
4Downgraded two levels due to substantial heterogeneity (inconsistency).
5Downgraded two levels due to substantial heterogeneity (inconsistency) and assessed and reported in small number of eligible studies (study limitations).

Open in table viewer
Summary of findings 2. CBT compared with treatment as usual for anxiety disorders in children and adolescents

CBT compared with treatment as usual for anxiety disorders in children and adolescents

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: treatment as usual

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk1

Corresponding risk

Treatment as usual

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

408 per 1000

687 per 1000
(383 to 886)

OR 3.19 (0.90 to 11.29)

487

(8 studies)

⊕⊕⊝⊝
low2

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused versus child and parent) (Chi² = 10.90, df = 1, P < 0.001, I² = 90.8%)

  • Insufficient studies for subgroup analyses examining differences in outcomes for individual versus group formats and interventions with varyingamount of therapist contact time

  • Difference in outcomes for participants with and without ASD (Chi² = 5.71, df = 1, P = 0.02; I² = 82.5%)

Acceptability (number of participants lost to post‐treatment assessment)

93 per 1000

124 per 1000
(70 to 209)

OR 1.37 (0.73 to 2.56)

441

(8 studies)

⊕⊕⊝⊝
low3

Remission of all anxiety diagnoses post‐treatment (ITT)

414 per 1000

660 per 1000
(451 to 820)

OR 2.74 (1.16 to 6.46)

203

(5 studies)

⊕⊕⊝⊝
low3

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.15standard deviations lower (0.78 lower to 0.48 higher).

Cross 0

214

(6 studies)

⊕⊕⊝⊝
low2

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (parent report) in the CBT groups was 0.32standard deviations lower (0.70 lower to 0.06 higher).

Cross 0

228

(7 studies)

⊕⊕⊝⊝
low2

Reduction in depressive symptoms post‐treatment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Improvement in global functioning post‐treatment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and Treatment as Usual groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of treatment as usual groups.
2Downgraded two levels due to at least moderate heterogeneity (inconsistency) and wide confidence intervals and small number of events or participants (imprecision).
3Downgraded two levels due to large variation in treatment effects across studies (inconsistency) and wide confidence intervals and small number of events (imprecision).

Open in table viewer
Summary of findings 3. CBT compared with attention control for anxiety disorders in children and adolescents

CBT compared with attention control for anxiety disorders in children and adolescents

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: attention control

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Attention control1

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

293 per 1000

486 per 1000
(355 to 617)

OR 2.28 (1.33 to 3.89)

822
(10 studies)

⊕⊕⊝⊝
low2,

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused versus child and parent) (Chi² = 7.65, df = 1, P = 0.006; I² = 86.9%)

  • No difference in outcomes for individual versus group formats (Chi² = 1.22, df = 1, P = 0.27; I² = 17.7%)

  • Insufficient studies for subgroup analyses examining outcomes for interventions with varying amounts of therapist contact, and participants with and without ASD

Acceptability (number of participants lost to post‐treatment assessment)

201 per 1000

201 per 1000
(146 to 272)

OR 1.00 (0.68 to 1.49)

797
(12 studies)

⊕⊕⊝⊝
low3

Remission of all anxiety diagnoses post‐treatment (ITT)

185 per 1000

385 per 1000
(217 to 584)

OR 2.75 (1.22 to 6.17)

378
(5 studies)

⊕⊕⊝⊝
low2

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.31standard deviations lower (0.51 to 0.11 lower).

Small effect size

978
(15 studies)

⊕⊕⊕⊝
moderate4

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (parent report) in the CBT groups was 0.25standard deviations lower (0.61 lower to 0.11 higher).

Cross 0

638
(8 studies)

⊕⊕⊝⊝
low5

Reduction in depressive symptoms post‐treatment

The mean depressive symptoms in the CBT groups was 0.18standard deviations lower (0.45 lower to 0.09 higher).

Cross 0

613
(10 studies)

⊕⊕⊝⊝
low5

Improvement in global functioning post‐treatment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and attention control groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of attention control groups.
2Downgraded two levels due to at least moderate heterogeneity and large variation in treatment effects (inconsistency) and small number of events (imprecision).
3Downgraded two levels due to large variation in treatment effects across studies (inconsistency) and wide confidence intervals and small number of events (imprecision).
4Downgraded one level due to moderate heterogeneity (inconsistency).
5Downgraded two levels due to at least moderate heterogeneity (inconsistency) and wide confidence intervals (imprecision).

Open in table viewer
Summary of findings 4. CBT compared with alternative treatment for anxiety disorders in children and adolescents

CBT compared with alternative treatment for anxiety disorders in children and adolescents

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: alternative treatment

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Alternative treatment1

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

See comment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect or conduct subgroup analyses

Acceptability (number of participants lost to post‐treatment assessment)

115 per 1000

171 per 1000
(74 to 350)

OR 1.58 (0.61 to 4.13)

515
(7 studies)

⊕⊕⊝⊝

low2

Remission of all anxiety diagnoses post‐treatment (ITT)

607 per 1000

579 per 1000
(351 to 775)

OR 0.89 (0.35 to 2.23)

401
(4 studies)

⊕⊕⊝⊝

low2

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.09standard deviations lower (0.40 lower to 0.21 higher).

Cross 0

399
(6 studies)

⊕⊕⊝⊝
low3

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.13standard deviations lower (0.33 lower to 0.06 higher).

Cross 0

423
(6 studies)

⊕⊕⊝⊝
low4

Reduction in depressive symptoms post‐treatment

Not estimable

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Improvement in global functioning post‐treatment

Not estimable

Not estimable

‐

See comment

No evidence available to estimate effect

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and Treatment as Usual groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of alternative treatment groups.
2Downgraded two levels due to at least moderate heterogeneity (inconsistency) and wide confidence intervals and small number of events (imprecision).
3Downgraded two levels due to moderate heterogeneity (inconsistency) and wide confidence intervals (imprecision).
4Downgraded two levels due to large variation in treatment effects across studies (inconsistency) and wide confidence intervals (imprecision).

Background

Description of the condition

Anxiety disorders are amongst the most common psychiatric disorders, occurring in 6.5% of all children and adolescents (Polanczyk 2015). One of the diagnostic challenges in children and adolescents involves distinguishing normal, developmentally appropriate worries, fears, and shyness from anxiety disorders. For example, primary school‐age children commonly have worries about injury and natural events, whereas older children and adolescents typically have worries and fears related to school performance, social competence, and health issues (Beesdo 2009). Distinguishing features of pathological anxiety include severity, persistence, and associated impairment.

The International Classification of Diseases (ICD), WHO 1992, and Diagnostic and Statistical Manual of Mental Disorders (DSM‐5), APA 2013, diagnostic systems distinguish various types of anxiety disorders, including generalised anxiety disorder, panic disorder, social anxiety disorder, separation anxiety disorder, agoraphobia, specific phobias, and selective mutism. These anxiety disorders are often associated with significant impairment in personal, social, and academic functioning (Pine 2009). Comorbidities are common, in particular with other anxiety disorders in children and adolescents (Leyfer 2013), and depression in adolescents (Essau 2003).

The presentation of anxiety disorders varies with age. Separation anxiety disorders are more common in younger children than in adolescents, and difficulties with social anxiety are typically associated with greater disturbance in adolescence (Waite 2014). Anxiety disorders with an onset in childhood often persist into adolescence and early adulthood (Copeland 2014), and are also associated with adverse academic, health, and social functioning in adulthood (Copeland 2014; Essau 2014). Yet these often remain untreated, with an average delay of 9 to 23 years before anxiety disorders are first treated (Wang 2005). It is clear that anxiety disorders in this age group present serious ongoing health issues, therefore effective and readily accessible treatments are needed.

Description of the intervention

Current treatments for anxiety disorders in this age group include behavioural therapy, cognitive behavioural therapy (CBT), or medication, or a combination of some or all these. National Institute for Health and Care Excellence (NICE) guidelines are available for the treatment of social anxiety disorder, which recommend CBT that is specifically focused on social anxiety (NICE 2013). Given the prevalence of these disorders, the age of onset, and public views on the acceptability of pharmacological treatments, psychological treatments are often preferred as first‐line therapy (Brown 2007; Young 2006). CBT is a collaborative psychological treatment that can be delivered in various formats, individually or in groups, and with varying levels of parent or family involvement.

One of the first manualised CBT programmes was Coping Cat (Kendall 1994), which consists of psycho‐education, modification of negative cognitions, exposure, social competence training, coping behaviour, and self‐reinforcement sessions. Others have followed, including the Cool Kids programme (Hudson 2009; Rapee 2006), the Coping Koala programme (Barrett 1996), Skills for Academic and Social Success (SASS) (Masia‐Warner 2005), ACTION (Waters 2009), Intervention With Adolescents With Social Phobia (IAFS) (Sanchez‐Garcia 2009), the TAPS (Warner 2011), and Building Confidence programme (Galla 2012). Alternative programmes involve providing direct support to parents alone, guiding them to implement CBT strategies with their child (Guided Parent‐Delivered CBT (GPD‐CBT) (Lyneham 2006; Thirlwall 2013; Waters 2009). CBT programmes have typically adopted a generic approach and target a range of anxiety disorders (e.g. Coping Cat), but some are disorder‐specific, targeting, for example, social anxiety disorder (e.g. SASS). CBT programmes have been modified in various ways to make them appropriate for children with autism spectrum disorders (ASD), such as by including social stories, social coaching, visual aids, and structured worksheets (Ung 2015). Such programmes include the Multimodal Anxiety and Social Skills Intervention (MASSI) programme (White 2013), TAFF (Schneider 2011), Behavioural Interventions for Anxiety in Children with Autism (BIACA) (Wood 2009), and Facing Your Fears (FYF) (Reaven 2012).

How the intervention might work

CBT for anxiety disorders in children and adolescents typically involves helping the child to recognise anxious feelings and bodily or somatic reactions to anxiety, identify thoughts or cognitions in anxiety‐provoking situations (e.g. unrealistic or negative attributions and expectations), and modifying these anxiety‐provoking cognitions (e.g. testing out predictions based on anxious thoughts, modifying anxious self‐talk into coping self‐talk). A key CBT procedure is exposure (Creswell 2020b), which typically involves testing out and ‘facing’ fears, often in a gradually increasing hierarchy. Behavioural training strategies such as modelling and role playing are often applied as opportunities for developing coping skills (e.g. problem‐solving skills, social skills, relaxation training).

CBT for anxiety disorders in children and adolescents has traditionally begun with six to nine face‐to‐face sessions of anxiety management strategies (emotion identification, relaxation training, cognitive strategies), followed by exposure work (Barrett 1996; Kendall 2006). In one meta‐analysis (Reynolds 2012), this traditional format of anxiety management sessions followed by exposure was observed in 93% of studies. However, Ale 2015 found that treatment outcomes in CBT treatment trials for child and adolescent anxiety disorders were not related to the use of relaxation strategies or the timing of exposure work, and therefore suggested that relaxation training may not be an essential ingredient of CBT, and that it may not be necessary to delay exposure until after anxiety management sessions. Moreover, there is also some preliminary evidence that introducing exposure early in treatment, without any prior anxiety management sessions, could improve outcomes whilst requiring fewer appointments (Whiteside 2015).

Indeed, questions remain about the mechanism of change within CBT. Cognitive restructuring and exposure tasks have each been found to make substantial contributions to improvement in child and adolescent anxiety in line with CBT theory (Peris 2015). More time devoted to exposure has been linked to better outcomes, and greater time spent on more difficult exposure tasks has been shown to predict better outcomes (Peris 2017). Change in coping efficacy, but not anxious self‐talk, has been found to mediate change in anxiety symptoms associated with CBT, medication (sertraline), and their combination, compared to placebo control (Kendall 2016). Furthermore, therapists’ ratings of child compliance and mastery also predict better outcomes (Peris 2017). Cognitive development also plays an important role, and whilst targeting behavioural avoidance appears to be crucial for both children and adolescents, treatment that directly addresses interpretation biases may be particularly beneficial for adolescents, but less so for younger children (Waite 2015).

CBT has been adapted to include family and parents in treatment sessions; however, parents/family members have been included in a wide variety of ways, including, for example, providing education and information on the child‐focused CBT, and/or modifying parents’ beliefs and behaviours, and/or addressing parental anxiety. These different approaches are based on different conceptualisations of the mechanisms of change in child anxiety through parental involvement. It is important to note, however, that changes in parent and child responses are likely to be bidirectional. For example, one study found that child‐focused anxiety treatments resulted in improvements in non‐targeted parent symptoms and family functioning, particularly when children responded successfully to treatment (Keeton 2013).

It is generally assumed that CBT can be applied only after the child has reached a certain level of cognitive development. Kendall 1993 argued that the ability to measure a thought or belief against the notion of a rational standard and the ability to understand that a thought or belief can cause a person to behave and feel in a certain way were central to its proper use. The question arises: at what age does a child have the cognitive capacities to undertake these cognitive operations? One study reported positive effects of CBT in children younger than six years of age (Hirshfeld‐Becker 2010); however, it is not clear whether children this age are able to use the cognitive strategies that are included in traditional CBT protocols. In line with this, research suggests that young children may be more responsive to the behavioural than the cognitive elements of this approach (Essau 2004). With younger children, parental involvement appears particularly important. Indeed, the treatment of anxiety disorders in young children can be effective by applying CBT principles through working directly with parents alone (Cartwright‐Hatton 2011), although there are inconsistent findings in relation to whether child‐parent delivery format is superior to parent‐only or not (Monga 2015; Waters 2009).

Why it is important to do this review

Anxiety disorders in children and adolescents represent a considerable source of morbidity and are associated with later adult psychopathology and greater cost than any other mental health disorder (Fineberg 2013). However, despite high prevalence and substantial morbidity, anxiety disorders in childhood and adolescence can be difficult to diagnose, and may be under‐recognised (NICE 2013), and therefore undertreated (Pine 2009). It is widely reported that only a minority of children and adolescents with mental health problems receive treatment (Green 2004; Merikangas 2011); a recent UK survey reported that fewer than 40% of children with anxiety disorders receive any professional support, and less than 3% receive CBT (Reardon 2020). Limited service provision represents a key barrier to treatment access (Reardon 2017), highlighting the importance of maximising the efficiency of treatment delivery to help ensure that effective treatment is more readily available to children and young people when they need it.

The evidence base for the treatment of anxiety disorders in children and adolescents is growing. Initial trials of CBT were positive (Barrett 1996; Kendall 1994; Kendall 1997), and further randomised controlled trials and reviews followed. Several reviews suggest that CBT for anxiety disorders in this age group is effective (Ale 2015; Crowe 2017; James 2015; Reynolds 2012; Silverman 2008), including an overview of systematic reviews and a network meta‐analysis (Bennett 2016; Zhou 2019). Overall there was a moderate response rate (e.g. 59%; James 2015), and a recent review indicated that benefits extend to broader outcomes, including depressive symptoms and general functioning (Kreuze 2018). However, CBT has not been shown to be superior to active controls or treatment as usual (James 2015; Southam‐Gerow 2010). In some previous studies, the effect sizes associated with CBT for childhood anxiety disorders did not differentiate it from attention placebo, although CBT has been found to be more effective than waiting list control (Ale 2015). It remains unclear whether CBT is superior to alternative treatment approaches.

This review updates and replaces previous Cochrane Reviews of CBT for anxiety disorders in children and adolescents (James 2005; James 2015). The current review aims to provide comprehensive and up‐to‐date evidence on the efficacy and acceptability of CBT in the treatment of anxiety disorders in children and adolescents, with varying amounts of therapist contact time and differing delivery formats, including individual, group, with/without family/parent involvement, and parent‐led. Furthermore, we planned to examine the efficacy of CBT relative to treatment as usual, attention control, and alternative treatments. The question of the comparative efficacy of medication versus CBT and the combination of CBT and medication was also to be addressed.

A recent development is the issue of the reporting of remission of all anxiety diagnoses, as well as the primary anxiety diagnosis (Warwick 2016). Given the high level of comorbid anxiety disorders, this is an important issue, but surprisingly has been previously overlooked. Indeed, focusing solely on recovery from the primary anxiety diagnosis means that children with comorbid anxiety disorders that are present following treatment are often still classed as ‘recovered’. As such, this review distinguishes between and examines outcomes based on both the absence of the ‘primary anxiety disorder’ and the absence of all anxiety disorders. As well as providing evidence on anxiety outcomes (diagnoses and symptoms), we also planned to examine benefits in relation to broader outcomes, including depressive symptoms and global functioning.

We also aimed to assess whether treatment effects of CBT are maintained beyond post‐treatment and at longer‐term follow‐up. Whilst it would not be possible to determine the youngest age at which a child can benefit from CBT without individual patient data, we would identify the age of the youngest participants in trials of CBT for child and adolescent anxiety disorders. It is recognised that children and adolescents with ASDs have high rates of anxiety disorders (van Steensel 2013); however, studies of CBT for anxiety disorders in ASD have had mixed outcomes (Murphy 2017; Ung 2015). Furthermore, it is unclear how anxiety disorders are recognised or, indeed, treated in those with intellectual impairments, indicating a pressing need for work in this area. We planned to examine the efficacy of CBT in children and adolescents with ASD and those with intellectual impairments.

Previous Cochrane Reviews used a cutoff of nine sessions of CBT, based on the practice and thinking at the time (James 2005; James 2015). However, since the last Cochrane Review (James 2015), there have been several developments in the delivery of CBT for anxiety disorders. These include briefer or shorter interventions, in terms of not only the number of sessions, but also the total duration of treatment. As such, we would not set a threshold for the minimum number of sessions, and would explore how treatment effects differ with treatment duration. It is important to note that one potentially efficient means to deliver CBT is online or via digital devices (e.g. computerised CBT). But as online and digital CBT interventions for child anxiety disorders have been reviewed elsewhere (e.g. Pennant 2015), this review focused on face‐to‐face delivery models that include direct contact with either the child or parent alone, or the child and parent together.

Objectives

  • To carry out a meta‐analysis of identified studies to determine whether CBT leads to remission of 1) the primary child/adolescent anxiety disorder and 2) all anxiety diagnoses, and/or 3) a clinically significant reduction in anxiety symptoms in comparison with waiting list/no treatment, treatment as usual, attention control, alternative treatment, or medication.

  • To determine the comparative efficacy of CBT alone, and the combination of CBT and medication, versus drug placebo.

  • To determine if CBT is an acceptable treatment, relative to waiting list/no treatment, treatment as usual, attention control, alternative treatment, and medication.

  • To determine whether CBT for anxiety leads to a clinically significant reduction in depressive symptoms and/or improvements in global functioning.

  • To identify any adverse effects associated with CBT, relative to waiting list/no treatment, treatment as usual, attention control, alternative treatment, and medication.

  • To determine whether post‐treatment gains of CBT are maintained at longer‐term follow‐up.

  • To describe the age range of participants included in CBT trials in order to determine the age of the youngest participants.

  • To carry out subgroup analyses of different types of CBT according to 1) amount of therapist contact time; and 2) delivery format (individual and group; child‐focused, child and parent/family, and parent‐only).

  • To carry out a subgroup analysis of CBT for children and adolescents with ASD and for children and adolescents with intellectual impairments.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs), cross‐over trials, and cluster‐randomised trials.

Types of participants

Participant characteristics

Children and adolescents younger than 19 years.

Diagnosis

We included participants meeting diagnostic criteria of the DSM (DSM III, III‐R, IV, IV‐TR, 5), APA 1980; APA 1987; APA 1994; APA 2000; APA 2013, or the ICD‐9 and ICD‐10, WHO 1978; WHO 1992, for an anxiety disorder.

Diagnoses had to be made by reliable and valid structured interviews for DSM or ICD child and adolescent anxiety disorders. Disorders classified as anxiety disorders vary across different versions of the DSM, and we included participants meeting diagnostic criteria for one or more of the following disorders: generalised anxiety disorder or overanxious disorder, separation anxiety disorder, social phobia or social anxiety disorder, panic disorder, agoraphobia, simple or specific phobias, or selective mutism.

Comorbidity

We included all comorbidities allowable for anxiety disorders under the rules of DSM or ICD, such as ASD, intellectual impairment, depressive disorders, and physical disorders.

Settings

We included all settings, such as research settings (i.e. university outpatient clinics, inpatient services, community clinics, and schools).

Exclusion criteria

We excluded studies that only included participants with post‐traumatic stress disorder or obsessive compulsive disorder, or both, as they are covered by separate Cochrane Reviews (Gillies 2016; O'Kearney 2006), and are no longer classified as ‘anxiety disorders’ in the DSM‐5 (APA 2013).

Types of interventions

Experimental intervention

The intervention had to be manualised CBT, or modular CBT, alone or in combination with medication, following a documented, written protocol and provided by trained therapists. Since the last review (James 2015), where the number of sessions was arbitrarily fixed at nine, there are now several studies indicating that shorter treatment, in terms of number of sessions or duration of sessions, or both, may be effective. We therefore did not include a minimum number of sessions or duration of sessions as a requirement.

CBT had to be administered according to standard principles as a psychological model of treatment involving helping the child to recognise anxious feelings and somatic reactions to anxiety; identify cognitions in anxiety‐provoking situations; modify these anxiety‐provoking cognitions; and respond to behavioural training strategies with exposure in vivo or by imagination.

CBT could be delivered to children (child‐focused), to children and parents/family members (child and parent), or to parents alone (parent‐only). In child‐focused CBT, the child had direct face‐to‐face contact with a therapist, and the intervention was delivered to children only, with minimal or no parental involvement. CBT with direct parent/family involvement (child and parent) included face‐to‐face child sessions and conjoint or separate sessions with parents/family that included providing psycho‐education for parents or teaching parents to be co‐therapists. Parent‐delivered CBT (parent‐only) only involved direct face‐to‐face contact with parents, and provided support for parents to help them to implement CBT strategies in their child’s day‐to‐day life. Child‐focused, child and parent, and parent‐only CBT could be delivered individually or in a group format. We did not include CBT interventions delivered online or via digital devices (e.g. computerised CBT).

Where studies included medication for the treatment of anxiety (in combination with CBT or alone), no concurrent medication for the treatment of anxiety was to be administered naturalistically. Where studies did not include medication for the treatment of anxiety, any medications administered naturalistically needed to be stable before and during the study.

Comparator interventions
Where CBT was delivered alone

  • Waitlist and no treatment for anxiety during that period

  • Treatment as usual (usual treatment or care; where usual care includes therapy, it does not include elements of CBT)

  • Attention control (attention only, e.g. support or education, but with no elements of CBT)

  • Alternative treatment (one specific non‐pharmacological intervention for the treatment of anxiety that followed a documented protocol and did not include CBT elements)

  • Medication for the treatment of anxiety

  • Drug placebo

Where CBT was delivered in combination with medication for the treatment of anxiety

  • Drug placebo

Types of outcome measures

We included studies that met the above inclusion criteria and reported or provided data on at least one of the following outcomes.

Primary outcomes

  • Remission: the absence of the primary diagnosis of an anxiety disorder post‐treatment, made by reliable and valid structured interviews for DSM or ICD child and adolescent anxiety disorders, such as:

    • Anxiety Disorder Interview Schedule for Children – Child and Parent (ADIS‐C/P) (Silverman 1987);

    • Anxiety Disorder Interview Schedule for Children – Child (ADIS‐C) (Silverman 1987);

    • Anxiety Disorder Interview Schedule for Children – Parent (ADIS‐P) (Silverman 1987);

    • Diagnostic Interview Schedule for Children, Adolescents and Parents (DISCAP) (Holland 1995).

  • Acceptability: loss of participants to post‐treatment assessment

Secondary outcomes

  • Remission: defined as the absence of all diagnoses of an anxiety disorder post‐treatment, made by reliable and valid structured interviews for DSM or ICD child and adolescent anxiety disorders.

  • Reduction in anxiety symptoms post‐treatment: measured using psychometrically robust measures of anxiety symptoms that yield symptom scores on continuous scales (Myers 2002), such as:

    • Screen for Child Anxiety Related Emotional Disorders (SCARED) (Birmaher 1999);

    • Spence Children’s Anxiety Scale (SCAS) (Spence 1997);

    • Revised Children’s Anxiety and Depression Scale (RCADS) – Anxiety Scale (Chorpita 2000);

    • Revised Children’s Manifest Anxiety Scale (RCMAS) (Reynolds 1985);

    • Multidimensional Anxiety Scale for Children (MASC) (March 1997);

    • State‐Trait Anxiety Inventory for Children (STAI‐C) (Spielberger 1973);

    • Social Phobia and Anxiety Inventory for Children (SPAI‐C) (Beidel 1995);

    • Social Anxiety Scale for Adolescents (SAS‐A) (La Greca 1998).

These scales could be self‐report or completed by a parent or an independent rater. Multiple reporters are often used, but the reliability of each reporter is likely to vary with the child’s age (Evans 2017). We therefore determined reduction in anxiety symptoms separately for 1) self‐reported and 2) parent‐reported or independent rater, or both. Multiple measures are also often reported, and we included the most validated, best recognised, or most frequently used measures in the analysis.

A crucial issue is how well these measures discriminate between clinical and non‐clinical levels of anxiety. We prioritised symptom measures that were closely aligned with diagnostic categories, with strong discriminant validity (e.g. RCADS, SCAS, SCARED). We also prioritised broad measures of anxiety symptoms (e.g. SCAS, SCARED, MASC), rather than disorder‐specific symptom measures (e.g. SPAI‐C, SAS‐A). (See Appendix 1)

  • Reduction in depressive symptoms post‐treatment: measured using psychometrically robust measures of depressive symptoms that yielded symptom scores on continuous scales, such as:

    • Children’s Depression Inventory (Kovacs 1989);

    • Beck Depression Inventory (Beck 1996);

    • Revised Children’s Anxiety and Depression Scale (RCADS) – Depression Scale (Chorpita 2000);

    • Mood and Feelings Questionnaire (Angold 1995).

If multiple depressive symptom measures/reporters were used, we included the most validated, best recognised, or most frequently used measures in the analysis.

  • Improvement in global functioning post‐treatment: measured using psychometrically robust measures of global functioning that yield symptom scores on continuous scales, such as:

    • Children’s Global Assessment Scale (CGAS) (Shaffer 1983).

If multiple global functioning measures/reporters were used, we included the most validated, best recognised, or most frequently used measures in the analysis.

  • Adverse events: we determined adverse events outcomes by the number and type of reported adverse events during the trial from randomisation to post‐treatment assessment (e.g. deterioration in anxiety symptoms, deterioration in global functioning, rates of self‐harm, suicide attempts).

  • Remission defined by the absence of the primary anxiety disorder diagnosis at a series of follow‐up time points (≤ 6 months post‐treatment, > 6 months post‐treatment but ≤ 12 months post‐treatment, and > 12 months post‐treatment).

  • Remission defined as the absence of all diagnoses of an anxiety disorder at a series of follow‐up time points (≤ 6 months post‐treatment, > 6 months post‐treatment but ≤ 12 months post‐treatment, and >12 months post‐treatment).

  • Reduction in anxiety symptoms at a series of follow‐up time points (≤ 6 months post‐treatment, > 6 months post‐treatment but ≤ 12 months post‐treatment, and > 12 months post‐treatment).

Search methods for identification of studies

We identified eligible studies (RCTs) of CBT for anxiety disorders in children and adolescents from the Cochrane Common Mental Disorders Controlled Trials Register (CCMDCTR; all years to 2016) (Appendix 2).

Electronic searches

We also ran searches on the following databases using relevant keywords, subject headings (controlled vocabularies), and search syntax, appropriate to each resource (Appendix 3). Searches were initially conducted on 21 November 2018 and updated 10 October 2019.

  • Cochrane Central Register of Controlled Trials (CENTRAL; Issue 10 of 12, 2019) in the Cochrane Library

  • Ovid MEDLINE (2016 to 10 October 2019)

  • Ovid Embase (2016 to 2019 Week 40)

  • Ovid PsycINFO (all years to October Week 40 2019)

We applied no restriction on language or publication status to the searches.

We also searched the international trial registries (10 October 2019) (including US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (clinicaltrials.gov) and the World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch/)) to identify additional ongoing and unpublished studies.

Searching other resources

Grey literature

We searched the grey literature for dissertations and theses (all years to 14 October 2019) (Appendix 4), as follows.

Reference lists

We checked the reference lists of all included studies and relevant systematic reviews to identify additional studies missed from the original electronic searches (e.g. unpublished or in‐press citations).

Correspondence

We contacted study authors and subject experts for information on unpublished or ongoing studies, or to request additional data. Where studies included some eligible participants (i.e. some participants were younger than 19 years, or some participants met the diagnostic criteria for an anxiety disorder), we contacted the authors to request data on eligible participants.

Data collection and analysis

Selection of studies

Two review authors (AJ, TR) independently screened the titles and abstracts of all studies identified as a result of the search, coding them as ‘retrieve' (eligible or potentially eligible/unclear) or ‘do not retrieve'. We retrieved the full‐text study reports/publications, and two review authors (AJ, TR) independently screened the full texts to identify studies for inclusion, and identified and recorded reasons for exclusion of the ineligible studies. Any disagreements were resolved through discussion or by consulting a third review author (GJ, AS, CC) if required. We identified and excluded duplicate records and collated multiple reports that related to the same study so that each study, rather than each report, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and ‘Characteristics of excluded studies' table (Moher 2015). We used Covidence software for the screening process (Covidence).

Data extraction and management

We used a data collection form created in Covidence to extract study characteristics and outcome data that we had piloted on two studies in the review. Two review authors (AJ, TR) extracted study characteristics and outcome data from the included studies. We extracted the following study characteristics.

  • Methods: study design, total duration of study, details of any ‘run‐in' period, number of study centres and location, study setting, withdrawals, and date of study

  • Participants: number, mean age, age range, gender, severity of condition, diagnostic criteria, comorbid conditions, inclusion criteria, and exclusion criteria

  • Interventions: intervention, comparison, concomitant medications, excluded medications, delivery format, therapist contact time, who delivers intervention

  • Outcomes: primary and secondary outcomes specified and collected, and time points reported

  • Notes: funding for study, and notable conflicts of interest of study authors

We noted in the ‘Characteristics of included studies' table if outcome data were not reported in a useable way. Any disagreements were resolved by consensus or by involving a third review author (GJ, AS, CC). One review author (TR) transferred data into the Review Manager 5 file (Review Manager 2014). We double‐checked that data were entered correctly by comparing the data presented in the systematic review with the study reports. A second review author (AJ) spot‐checked study characteristics for accuracy against the study report.

Main comparisons

  • CBT compared with waiting list and no treatment controls

  • CBT compared with treatment as usual

  • CBT compared with attention control

  • CBT compared with alternative treatments

  • CBT compared with medication or drug placebo

  • CBT and medication combination compared with drug placebo

Assessment of risk of bias in included studies

Two review authors (AJ, TR) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). Any disagreements were resolved by discussion or by involving another review author (GJ, CC). We assessed risk of bias according to the following domains.

  • Random sequence generation

  • Allocation concealment

  • Blinding of participants and personnel

  • Blinding of outcome assessment

  • Incomplete outcome data

  • Selective outcome reporting

  • Other bias, including therapy integrity

We judged each potential source of bias as low, high, or unclear and provided a supporting quotation from the study report together with a justification for our judgement in the ‘Risk of bias' table. We summarised the 'Risk of bias' judgements across different studies for each of the domains listed.

Selection bias: we assessed the adequacy of the randomisation process in terms of the description of adequacy of sequence generation and the concealment of treatment group allocation.

Performance bias: given the nature of psychological interventions, blinding of either participants or personnel delivering the treatments could only be possible in studies involving CBT versus treatment as usual, attention control, or alternative treatment, therefore we were only able to assess attempts at blinding in those studies.

Detection bias: we evaluated whether study personnel carrying out outcome assessments were blinded to the treatment status of participants.

Attrition bias: we determined whether studies provided a description of withdrawals and dropouts.

Other bias: we also assessed therapy integrity, including therapist competence and adherence to treatment protocol, for all included studies.

We did not exclude studies from meta‐analysis on the basis of the ‘Risk of bias’ assessment. We conducted sensitivity analyses for the primary outcome, excluding trials with high or unclear risk of bias ratings for allocation concealment, and excluding trials with high or unclear risk of bias for blinding of outcome assessment if appropriate. We reported the remainder of the ‘Risk of bias' assessments for these trials, and included discussion of this assessment in the Results and Discussion sections.

Measures of treatment effect

In order to assess post‐treatment outcomes, we used dichotomous data on remission of primary anxiety diagnosis and all anxiety diagnoses, and continuous data on anxiety symptoms, depressive symptoms, and global functioning, with the use of standardised measures. We used data from the assessment administered immediately after treatment (or the assessment closest to the end of treatment) to assess post‐treatment outcomes. We also used these measures to evaluate the maintenance of treatment effects at a series of follow‐up time points (≤ 6 months post‐treatment, > 6 months post‐treatment but ≤ 12 months post‐treatment, and > 12 months post‐treatment). Where studies reported follow‐up data at multiple time points within one category (e.g. one‐month and three‐month follow‐up), we used data from the longer follow‐up period. To assess acceptability, we used frequency data on the numbers of participants who were lost to post‐treatment assessment. Adverse events were determined by the number and type of adverse events during the trial, from randomisation to the post‐treatment assessment.

Dichotomous data

We analysed dichotomous data as odds ratios (OR) and 95% confidence intervals (CI).

Continuous data

We analysed continuous data as mean difference (MD) or standardised mean difference (SMD). We entered data presented as a scale with a consistent direction of effect.

We narratively described skewed data reported as medians and interquartile ranges.

Unit of analysis issues

Cluster‐randomised trials

We included cluster‐randomised controlled trials based in schools. Cluster‐randomised trials can, in principle, be combined with individually randomised trials in the same meta‐analysis (Deeks 2017). We did not anticipate that there would be many cluster‐randomised trials, therefore we included identified cluster‐randomised trials in the meta‐analyses and sensitivity analyses that we planned to undertake to investigate the robustness of any conclusions drawn. To correct the influence of any cluster trials, we used an average intraclass correlation coefficient of 0.02 (Health Services Research Unit 2004).

The effective sample size of a single intervention group in a cluster‐randomised trial is its original sample size divided by the ‘design effect’. The design effect is 1 + (M − 1) ICC, where M is the average cluster size and ICC is the intracluster correlation coefficient (Rao 1992). For dichotomous data, we divided both the number of participants and the number experiencing the event by the same design effect. For continuous data, only the sample size was reduced; we did not alter means and standard deviations (SDs).

Cross‐over trials

We did not anticipate that there would be many cross‐over trials, and the data required to include a paired analysis in a meta‐analysis is often not reported (Higgins 2011). We therefore planned to include any identified cross‐over trials in the meta‐analysis, but only data from the first trial period (i.e. prior to the ‘cross‐over').

Studies with multiple treatment groups

Where multiple trial arms were reported in a single trial, we included only the relevant arms. Studies with more than two intervention arms can pose analytical problems in pair‐wise meta‐analysis. Where studies had two or more relevant active treatment arms to be compared against controls, we managed data as follows.

Continuous data

We divided the control group equally into two or more groups to compare the means and SDs of these groups against the means and SDs of the two treatment arms.

Dichotomous data

For trials with two or more active treatment arms and a control group, we split participants in the control arm group equally between the active treatment arms.

Dealing with missing data

We contacted investigators or study sponsors in order to verify key study characteristics and to obtain missing numerical outcome data where possible (e.g. when we identified a study as abstract only, when a study included a post‐treatment diagnostic assessment but did not report the required remission outcomes, or when a study did not report standard errors or SDs). We documented all correspondence with study authors and reported which study authors responded to our queries.

Missing statistics

In the first instance, we attempted to contact the original researchers for any missing data. If the study only reported standard errors, we calculated SDs.

Missing participants

We undertook intention‐to‐treat (ITT) analyses. When analysing dichotomous data, we assumed that all non‐completers in the CBT group were treatment failures, and non‐completers in the control group were treatment successes, thereby yielding the most conservative treatment estimate.

For dichotomous outcomes, we also undertook completer analysis, using only data from participants who completed post‐treatment assessments.

We did not use any statistical or other methods to impute missing data for continuous outcomes, as we did not have access to raw data.

Assessment of heterogeneity

We assessed clinical heterogeneity by comparing differences in the distribution of important participant factors between studies (e.g. age, gender, specific diagnosis, duration and severity of disorder, associated comorbidities). We assessed methodological heterogeneity by comparing trial factors (randomisation, concealment, blinding of outcome assessment, losses to follow‐up). We used the Chi² test, Deeks 2017, and the I² statistic, Higgins 2003, to assess heterogeneity. We set significance at P < 0.1. The Cochrane Handbook for Systematic Reviews of Interventions recommends using a range for the I² statistic and a guide to interpretation (Deeks 2017). If we found either moderate heterogeneity (I² in the range of 30% to 60%) or substantial heterogeneity (I² in the range of 60% to 90%), we performed subgroup and sensitivity analyses where possible.

Assessment of reporting biases

Where a minimum of 10 studies were included, we investigated publication bias using funnel plots, Sterne 2017, and subjected any asymmetry found to statistical investigation using Egger’s test (Stata 2012).

Data synthesis

We undertook meta‐analyses only where this was meaningful, that is if the treatments, participants, and the underlying clinical question were similar enough for pooling to make sense. We undertook ITT and completer analyses.

We planned to carry out separate analyses to identify whether CBT was more effective post‐treatment than waiting list/no treatment; treatment as usual, attention controls, alternative treatments, medication; and drug placebo; and also to identify whether CBT in combination with medication was more effective than drug placebo.

We used follow‐up data for each comparison to assess maintenance of treatment gains. If it was meaningful to do so, we pooled data separately for each follow‐up time point (≤ 6 months post‐treatment, > 6 months post‐treatment but ≤ 12 months post‐treatment, and > 12 months post‐treatment). Where studies reported follow‐up data at multiple time points within one category (e.g. 9‐month and 12‐month follow‐up), we included data from the longer follow‐up period.

Dichotomous data

We used ORs and 95% CIs based on the random‐effects model, with pooling of data via the inverse variance method of weighting. We set significance at P < 0.05. Where available, we used combined data from an interview with the child or adolescent and the parent; otherwise we used data from one interview (child/adolescent or parent interview). We calculated the number needed to treat for an additional beneficial outcome (NNTB) with 95% CIs (Stata 2012). We calculated a summary statistic of all those responding to treatment as a percentage of the total number of participants for each comparison.

Continuous data

We conducted analysis of continuous data based on the random‐effects model, with pooling of data via the inverse variance method of weighting. We used the SMD to pool continuous data measured in different ways across studies but conceptually the same (i.e. measuring anxiety or depressive symptoms or global functioning). For continuous data measuring anxiety symptoms, we pooled child/adolescent report and parental/clinician reports separately. Where both endpoint and change data were available for the same outcome, we presented the endpoint. We set significance at P < 0.05.

Tables and figures

We transferred data into Review Manager 5 (Review Manager 2014), and presented them graphically, so that the area to the left of the line of no effect (or right for remission outcomes) indicated a favourable outcome for CBT. We used tables to display characteristics of the included studies. We presented a brief list of excluded studies in a table with their reasons for exclusion. We summarised risk of bias in the included studies in a figure, and included a PRISMA flow chart (Moher 2015).

Subgroup analysis and investigation of heterogeneity

The critical need to improve access to treatment for child anxiety disorders means that it is particularly important to explore the efficacy of approaches that may help maximise treatment efficiency, including alternative delivery formats and briefer interventions that involve less therapist contact time than traditional approaches. As outlined above, there are also unanswered questions in relation to the benefits of CBT for children and adolescents with ASD and intellectual impairments. We therefore set out to explore the efficacy of different delivery formats and of briefer and shorter interventions, and to examine treatment effects amongst children and adolescents with ASD and those with intellectual impairment, using subgroup analyses. Specifically, where it was possible and meaningful to do so, we undertook subgroup analyses to examine differences between:

  • delivery formats (child‐focused, child and parent, and parent‐only; individual and group);

  • interventions with a varying amount of therapist contact time (< 10 hours, ≥ 10 hours and < 20 hours, ≥ 20 hours);

  • children and adolescents with and without ASDs;

  • children and adolescents with and without intellectual impairments.

To examine differences between age groups, we also undertook post hoc subgroup analyses evaluating the differences between studies where all participants were age 12 or younger (≤ 12 years); all participants were age 12 or older (≥ 12 years); and studies that included participants under and over age 12 (< 12 years and ≥ 12 years).

We undertook subgroup analyses for the primary outcome (remission of primary anxiety disorder) across comparisons. For the main comparison with the largest number of studies (CBT versus waitlist/no treatment), we also undertook subgroup analyses for the most frequently reported secondary outcomes (remission of all anxiety diagnoses post‐treatment and reduction in child‐ and parent‐reported anxiety symptoms post‐treatment). To ensure subgroup analyses were meaningful, we only undertook these analyses where there were data from at least three studies for each subgroup.

We assessed statistical heterogeneity for all analyses and between groups with the Chi2 test and the I2 statistic, and set significance at P < 0.1.

Sensitivity analysis

Sensitivity analysis is the study of how the uncertainty in the output of an analysis can be apportioned to different sources of uncertainty in its inputs. Sensitivity analyses can therefore be carried out to test the robustness of decisions made in the review process. We carried out sensitivity analyses where there was evidence of the following:

  • significant heterogeneity: we inspected forest plots and examined each study in turn to determine the source of any significant heterogeneity;

  • selection bias: we excluded those studies judged to be at high risk of selection bias from the main analysis;

  • allocation concealment: we excluded those studies judged to be at high risk of bias for allocation concealment from the main analysis.

Where appropriate, we undertook all of the above sensitivity analyses for the ITT and completer analyses.

Where meta‐analyses examining anxiety symptoms included broad measures of anxiety symptoms and disorder‐specific symptom measures, we also undertook sensitivity analyses excluding studies that only reported disorder‐specific symptom measures.

GRADE and ‘Summary of findings' table

We created a ‘Summary of findings' table including the following primary outcomes:

  • remission of primary anxiety diagnosis post‐treatment;

  • acceptability in terms of dropouts from randomisation to the post‐treatment assessment;

and the following secondary outcomes:

  • remission of all anxiety diagnoses post‐treatment;

  • reduction in anxiety symptoms (self‐reported and parent‐reported) post‐treatment;

  • reduction in depressive symptoms post‐treatment;

  • improvement in global functioning post‐treatment;

  • adverse events from randomisation to the post‐treatment assessment.

We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of a body of evidence as it relates to the studies that contribute data to the meta‐analyses for the prespecified primary and secondary outcomes. Two review authors (AJ, TR) independently assessed risk of bias, and in case of disagreement sought consensus between four review authors (AJ, TR, GJ, CC). We used the methods and recommendations described in Section 8.5, Higgins 2017, and Chapter 12, Schünemann 2017, of the Cochrane Handbook for Systematic Reviews of Interventions, employing GRADEpro GDT software (GRADEpro GDT). We justified all decisions to downgrade or upgrade the quality of studies using footnotes, and made comments to aid the reader's understanding of the review where necessary. We considered whether there was any additional outcome information that could not be incorporated into the meta‐analyses, noting this in the comments and stating if it supported or contradicted the information from the meta‐analyses.

Results

Description of studies

See Characteristics of included studies and Table 1.

Open in table viewer
Table 1. Summary of characteristics of included studies

Study

Comparison 1

Comparison 2

Comparison 3

Broad vs specific anxiety measure

Delivery mode

Delivery mode

ASD vs not ASD

Therapist contact time (hours)

Age

Setting

Disorder‐specific

Other comorbid conditions (required)

Afshari 2014

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

10 to 20

mix

clinic

Arendt 2016

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

20+

mix

clinic (university)

Barrett 1996

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and child+parent

Individual

Not ASD

10 to 20

mix

clinic (university)

Barrett 1998

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Child focused and child+parent

Group

Not ASD

20+

mix

clinic (university)

Barrington 2005

CBT vs TAU

Broad

Child+parent

Individual

Not ASD

10 to 20

mix

clinic (CAMHS)

Berge 2017

CBT vs Wait list / no treatment

Specific

Child focused

Individual

Not ASD

less than 10

mix

clinic ‐dental

Specific Phobia

Cartwright Hatton 2011

CBT vs Wait list / no treatment

Parent only

Group

Not ASD

20+

≤ 12

clinic‐university hospital

Chalfant 2007

CBT vs Wait list / no treatment

Broad

Child+parent

Group

ASD

20+

mix

clinic

Cheung 2016

CBT vs Wait list / no treatment

CBT vs Alternative treatment

Broad

Child focused

Individual

Not ASD

10 to 20

≤ 12

clinic‐CAMHS

Chiu 2013

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

≤ 12

school

Cobham 2017

CBT vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

less than 10

mix

clinic (university)

Cornacchio 2019

CBT vs Wait list / no treatment

Child+parent

Group

Not ASD

20+

≤ 12

clinic

Selective mutism

Creswell 2017

CBT vs Alertnative treatment

Broad

Parent only

Individual

Not ASD

less than 10

≤ 12

clinic (CAMHS)

Dadds 1997

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

10 to 20

mix

school

Flannery Schroeder 2000

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused

Individual and group

Not ASD

10 to 20 and 20+

mix

clinic (university)

Fujii 2013

CBT vs TAU

Child+ parent

Individual

ASD

20+

≤ 12

clinic

Gallagher 2004

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

less than 10

≤ 12

clinic

Social anxiety disorder

Ginsburg 2002

CBT vs Attention control

Broad

Child focused

Group

Not ASD

less than 10

≥ 12

school

Ginsburg 2012

CBT vs TAU

Broad

Child focused

Individual

Not ASD

less than 10

mix

school

Ginsburg 2019

CBT vs TAU

Child focused

Individual

Not ASD

less than 10

mix

school

Hancock 2018

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child +parent

Group

Not ASD

10 to 20

mix

clinic (university)

Herbert 2009

CBT vs Attention control

CBT vs Attention control

Specific

Child focused

Group and Individual

Not ASD

20+ and 10 to 20

≥ 12

clinic (university)

Social anxiety disorder

Hirshfeld Becker 2010

CBT vs Wait list / no treatment

Child + parent

Individual

Not ASD

20+

≤ 12

clinic (university)

Holmes 2014

CBT vs Wait list / no treatment

Broad

Child + parent

Group

Not ASD

10 to 20

≤ 12

clinic (university)

GAD

Hudson 2009

CBT vs Attention control

Broad

Child +parent

Group

Not ASD

20+

mix

clinic (university)

Ingul 2003

CBT vs Attention control

CBT vs Attention control

Broad

Child focused

Individual and Group

Not ASD

10 to 20

≥ 12

schools

Social anxiety disorder

Ishikawa 2019

CBT vs Wait list / no treatment

Broad

Child +parent

Individual

Not ASD

less than 10

mix

clinic

Kendall 1994

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

Kendall 1997

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

Kendall 2008

CBT vs Attention control

CBT vs Attention control

Broad

Child focused and child+parent

Individual

Not ASD

10 to 20

mix

clinic (university)

Kennedy 2009

CBT vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Khanna 2010

CBT vs Attention control

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

Kidd 2018

CBT vs Wait list / no treatment

Broad

Child + parent

Group

ASD

20+

≥ 12

clinic (university)

Last 1998

CBT vs Attention control

Broad

Child and parent

Individual

Not ASD

10 to 20

mix

clinic (university)

School refusal

Lau 2010

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

10 to 20

≤ 12

clinic

Lau 2017

CBT vs Wait list / no treatment

Broad

Child +parent

Group

Not ASD

10 to 20

≤ 12

child centres and preschools

Lebowitz 2019

CBT vs Alternative treatment

Broad

Child focused

Individual

Not ASD

10 to 20

≤ 12

clinic

Leutgeb 2012

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

less than 10

mix

university

Specific Phobia

Masia Warner 2005

CBT vs Wait list / no treatment

Specific

Child focused

Group focused

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

Masia Warner 2007

CBT vs Attention control

Specific

Child focused

Group focused

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

Masia Warner 2011

CBT vs Wait list / no treatment

Child focused

Individual

Not ASD

10 to 20

mix

primary care/special clinic

functional physical complaints

Masia Warner 2016

CBT vs Attention control

CBT vs Attention control

Specific

Child focused

Group focused

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

McConachie 2014

CBT vs Wait list / no treatment

Broad

Child + parent

Group

ASD

20+

mix

clinic (university)

McNally Keehn 2013

CBT vs Wait list / no treatment

Broad

Child focused

Individual

ASD

20+

mix

clinic (university)

Melfsen 2011

CBT vs Wait list / no treatment

Specific

Child and parent

Individual

Not ASD

20+

mix

clinic

Social anxiety disorder

Muris 2002

CBT vs Attention control

Broad

Child focused

Group

Not ASD

less than 10

≤ 12

school

Murphy 2017

CBT vs Alternative treatment

Broad

Child focused

Individual focused

ASD

10 to 20

≥ 12

CAMHS

O'Brien 2007

CBT vs TAU

Child and parent

Group

Not ASD

10 to 20

mix

clinic

Olivares 2005

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

≥ 12

school

Social anxiety disorder

Olivares 2014

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

≥ 12

school

Social anxiety disorder

Olivares 2019

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Child focused

Group

Not ASD

10 to 20

≥ 12

school

Social anxiety disorder

Ollendick 2009

CBT vs Wait list / no treatment

CBT vs Attention control

Broad

Child focused

Individual

Not ASD

less than 10

mix

clinic

Specific Phobia

Ost 2001

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and child+parent

Individual

Not ASD

less than 10

mix

referrals from clinic and schools

Specific Phobia

Perrin 2019

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

CAMHS

GAD

Pincus 2010

CBT vs Attention control

Broad

Child focused

Individual

Not ASD

less than 10

≥ 12

clinic (university)

Panic Disorder

Rapee 2005

CBt vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

less than 10

≤ 12

preschools

Rapee 2006

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Reaven 2012

CBT vs TAU

Child+parent

Group

ASD

10 to 20

mix

clinic (university)

Reigada 2015

CBT vs Alternative treatment

Specific

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

IBD

Rosa‐Alcazar 2009

Cbt vs Attention control

CBT vs Attention control

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

Salari 2018

CBT vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Salum 2018

CBT vs Attention control

CBT vs Alternative treatment

Broad

Child+parent

Group

Not ASD

10 to 20

≤ 12

clinic

Sanchez Garcia 2009

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

mix

School

Social anxiety disorder

Santucci 2013

CBT vs Wait list / no treatment

Broad

Child +parent

Group

Not ASD

20+

≤ 12

clinic (university)

Separation anxiety disorder

Schneider 2011

CBT vs Wait list / no treatment

Broad

Child +parent

Individual

Not ASD

10 to 20

≤ 12

clinic (university)

Separation anxiety disorder

Sciberras 2018

CBT vs TAU

Broad

Child+parent

Individual

Not ASD

10 to 20

≤ 12

clinic ADHD

ADHD

Shahnavaz 2016

CBT vs TAU

Specific

Child+parent

Individual

Not ASD

10 to 20

mix

dental clinics

Specific Phobia

Sharma 2017

CBT vs TAU

Broad

Child focused

Group

Not ASD

20+

mix

clinic (university)

Headaches

Shortt 2001

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Silk 2018

CBT vs Alternative treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

Silverman 1999a

CBT vs Attention control

Broad

Child+parent

Individual

Not ASD

10 to 20

mix

clinic (university)

Silverman 1999b

CBT vs Wait list / no treatment

Broad

Child + parent

Group

Not ASD

10 to 20

mix

clinic (university)

Simon 2011

CBt vsWait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and parent only

Group

Not ASD

less than 10 and 10 to 20

mix

School

Smith 2014

CBT vs Wait list / no treatment

Broad

Parent only

Individual

Not ASD

10 to 20

mix

clinic (university)

Southam Gerow 2010

CBT vs TAU

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinics

Spence 2000

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and child+parent

Group

Not ASD

20+ and 10 to 20

mix

clinic (university)

Social anxiety disorder

Spence 2006

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

10 to 20

mix

clinic (university)

Spence 2011

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

Not ASD

10 to 20

≥ 12

clinic (university)

Storch 2013

CBT vs TAU

Broad

Child+parent

Individual

ASD

20+

≤ 12

clinic (university)

Storch 2015

CBT vs TAU

Broad

Child+parent

Individual

ASD

20+

≥ 12

clinic (university)

Thirlwall 2013

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Parent only

Individual

Not ASD

less than 10

≤ 12

clinic

Villabo 2018

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused

Individual and Group

Not ASD

10 to 20

mix

clinic

Waters 2009

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Child+Parent and parent only

Group

Not ASD

10 to 20 and 20 +

≤ 12

clinic (university)

Wergelenad 2014

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child+parent

Individual and group

Not ASD

10 to 20

mix

clinics‐CAMHS

White 2013

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

ASD

20+

≥ 12

clinic (university)

Wood 2009

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

ASD

20+

mix

clinic (university)

Wood 2015

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

ASD

20+

mix

clinic (university)

ASD: autism spectrum disorders
CBT: cognitive behavioural therapy
GAD: generalised anxiety disorder
TAU: treatment as usual

Results of the search

The initial database search (21 November 2018) identified 6354 records, and an updated database search (10 October 2019) identified a further 2081 records. A search of the grey literature identified a further 473 records, and an additional 38 records were identified from searching reference lists and correspondence with study authors and experts in the field.

After removing duplicates, we screened a total of 5488 titles/abstracts, and excluded 4746 records. We then assessed 742 full‐texts for eligibility for inclusion in the review, and identified 88 eligible studies (130 articles) (Figure 1).

Figure 1
Study flow diagram.

Study flow diagram.

We contacted 58 study authors for clarification or additional data, or both, and 16 study authors provided relevant outcome data for 20 of the included studies.

Included studies

Eighty‐eight studies met the criteria for inclusion in the review, and 87 studies were included in meta‐analyses.

Design

All included studies were RCTs. Two studies used a cluster‐randomised design, where schools, Dadds 1997, and school‐based clinicians, Ginsburg 2019a, were randomised, rather than individual participants. No included studies used a cross‐over design.

Amongst the studies included in meta‐analyses, 20 studies had two CBT arms (Afshari 2014; Barrett 1996; Barrett 1998; Cheung 2016; Flannery Schroeder 2000; Hancock 2018; Herbert 2009; Ingul 2013; Kendall 2008; Masia Warner 2016; Olivares 2014; Olivares 2019; Ost 2001; Salum 2018; Simon 2011; Spence 2000; Thirlwall 2013; Villabo 2018; Waters 2009; Wergeland 2014), including one factorial design (Salum 2018), and another three studies included multiple comparison arms (Cheung 2016; Ollendick 2009; Rosa Alcazar 2009).

Fifty‐seven studies were included in CBT versus waitlist/no treatment comparisons (Afshari 2014; Arendt 2016; Barrett 1996; Barrett 1998; Berge 2017; Cartwright Hatton 2011; Chalfant 2007; Cheung 2016; Chiu 2013; Cobham 2017; Cornacchio 2019; Dadds 1997; Flannery Schroeder 2000; Gallagher 2004; Hancock 2018; Hirshfeld Becker 2010; Holmes 2014; Ishikawa 2019; Kendall 1994; Kendall 1997; Kennedy 2009; Kidd 2018; Lau 2010; Lau 2017; Leutgeb 2012; Masia Warner 2005; Masia Warner 2011; McConachie 2014; McNally Keehn 2013; Melfsen 2011; Olivares 2005; Olivares 2014; Olivares 2019; Ollendick 2009; Ost 2001; Perrin 2019; Rapee 2005; Rapee 2006; Rosa Alcazar 2009; Salari 2018; Sanchez Garcia 2009; Santucci 2013; Schneider 2011; Shortt 2001; Silverman 1999b; Simon 2011; Smith 2014; Spence 2000; Spence 2006; Spence 2011; Thirlwall 2013; Villabo 2018; Waters 2009; Wergeland 2014; White 2013; Wood 2009; Wood 2015). With the exception of three studies (Dadds 1997; Rapee 2005; Simon 2011), participants assigned to the control arm were offered CBT after a waitlist period. Seven studies included eligible follow‐ups that included CBT and waitlist/no treatment participants. Follow‐up assessments were conducted between 3 and 24 months post‐treatment.

Twelve studies compared CBT to treatment as usual (TAU) (Barrington 2005; Fujii 2013; Ginsburg 2012; Ginsburg 2019a; O'Brien 2007; Reaven 2012; Sciberras 2018; Shahnavaz 2016; Sharma 2017; Southam Gerow 2010; Storch 2013; Storch 2015). TAU varied considerably across studies, ranging from eclectic non‐CBT therapeutic approaches or strategies that were offered as part of routine care (Barrington 2005; Ginsburg 2012; Ginsburg 2019a; Southam Gerow 2010), to allowing participants to continue and seek any psychosocial or pharmacological interventions (Fujii 2013; Reaven 2012; Storch 2013; Storch 2015). Four CBT versus TAU studies included 6‐month, 12‐month, or both 6‐ and 12‐month follow‐ups.

Fifteen studies compared CBT to an attention control (Ginsburg 2002; Herbert 2009; Hudson 2009; Ingul 2013; Kendall 2008; Khanna 2010; Last 1998; Masia Warner 2007; Masia Warner 2016; Muris 2002; Ollendick 2009; Pincus 2010; Rosa Alcazar 2009; Salum 2018; Silverman 1999a), which was designed to act as an attention condition, and often included a psychoeducation component. Six CBT versus attention control studies included follow‐ups up to 12 months.

Seven studies included a CBT versus an alternative treatment comparison (Cheung 2016; Creswell 2017; Lebowitz 2019; Reigada 2015; Salum 2018; Silk 2018). Alternative treatments designed to target children's anxiety difficulties included: solution‐focused therapy (Creswell 2017), the parent intervention SPACE (Supportive Parenting for Anxious Childhood Emotions) (Lebowitz 2019), child‐centred therapy (Silk 2018), counselling (Murphy 2017), non‐directive supportive therapy (Reigada 2015), and ABMT (Attention Bias Modification Treatment) (Cheung 2016; Salum 2018). Four CBT versus alternative treatment studies included follow‐ups up to 12 months post‐treatment.

Only one study compared CBT to medication, and CBT in combination with medication to drug placebo (Walkup 2008), therefore it was not possible to conduct meta‐analyses for these comparisons.

Sample size

The 87 studies included in meta‐analyses involved a total of 3492 CBT participants and 2472 controls (waitlist/no treatment controls, N = 1537; TAU, N = 260; attention control, N = 441; alternative treatment, N = 234).

Study sample sizes ranged from 12 participants, Ginsburg 2002; Sciberras 2018, to 206 participants (corrected sample size; Ginsburg 2019a).

Setting

Two‐thirds of the included studies were conducted in either the USA or Australia. The remaining studies were conducted in the UK, Spain, Norway, the Netherlands, Germany, Sweden, Iran, Brazil, Hong Kong, India, Japan, Denmark, and Ireland.

Most studies were conducted in clinical settings, either university research clinics (specified in 41 studies), community mental health clinics (specified in 12 studies), or dental clinics (Berge 2017; Shahnavaz 2016). Sixteen studies recruited participants through schools (Chiu 2013; Dadds 1997; Ginsburg 2002; Ginsburg 2012; Ginsburg 2019a; Ingul 2013; Masia Warner 2005; Masia Warner 2007; Masia Warner 2016; Muris 2002; Olivares 2005; Olivares 2014; Olivares 2019; Rosa Alcazar 2009; Sanchez Garcia 2009; Simon 2011), and two through preschools (Lau 2017; Rapee 2005). No studies were conducted in inpatient settings.

Participants

The included studies involved participants of varied ages. Sixteen studies focused on adolescents and only included participants aged 12 years or over (Ginsburg 2002; Herbert 2009; Ingul 2013; Kidd 2018; Masia Warner 2005; Masia Warner 2007; Masia Warner 2016; Murphy 2017; Olivares 2005; Olivares 2014; Olivares 2019; Pincus 2010; Rosa Alcazar 2009; Spence 2011; White 2013; Wood 2015). Twenty‐five studies included participants up to age 12 years (Cartwright Hatton 2011; Cheung 2016; Chiu 2013; Cornacchio 2019; Creswell 2017; Fujii 2013; Gallagher 2004; Hirshfeld Becker 2010; Holmes 2014; Kennedy 2009; Lau 2010; Lau 2017; Lebowitz 2019; Muris 2002; Rapee 2005; Rapee 2006; Salari 2018; Salum 2018; Santucci 2013; Schneider 2011; Sciberras 2018; Shortt 2001; Storch 2013; Thirlwall 2013; Waters 2009), including 5 studies where participants were all under 8 years of age (Hirshfeld Becker 2010; Kennedy 2009; Lau 2017; Rapee 2005; Schneider 2011). The youngest participants in the included studies were two to three years of age (Cartwright Hatton 2011; Kennedy 2009; Lau 2017; Rapee 2005). The remaining studies included participants both under and over 12 years.

All participants met the diagnostic criteria for at least one anxiety disorder. In most studies, any anxiety disorder, or a subset of anxiety disorders (e.g. generalised anxiety disorder (GAD), social anxiety disorder, or separation anxiety disorder) was required for inclusion. In five studies (Cartwright Hatton 2011; Creswell 2017; Dadds 1997; Rapee 2005; Simon 2011), not all study participants met the diagnostic criteria for an anxiety disorder, but only those who did were included in this review.

Twenty‐four studies focused on a specific anxiety disorder where participants all met the diagnostic criteria for the target anxiety disorder. Disorder‐specific studies included participants with social anxiety disorder (Gallagher 2004; Herbert 2009; Ingul 2013; Masia Warner 2005; Masia Warner 2007; Masia Warner 2016; Melfsen 2011; Olivares 2005; Olivares 2014; Olivares 2019; Rosa Alcazar 2009; Sanchez Garcia 2009; Spence 2000), specific phobia (Berge 2017; Leutgeb 2012; Ollendick 2009; Ost 2001; Shahnavaz 2016), separation anxiety disorder (Santucci 2013; Schneider 2011), GAD (Holmes 2014; Perrin 2019), panic disorder with/without agoraphobia (Pincus 2010), and selective mutism (Cornacchio 2019).

In some studies comorbid conditions were required for inclusion. Twelve studies included participants with an ASD (Chalfant 2007; Fujii 2013; Kidd 2018; McConachie 2014; McNally Keehn 2013; Murphy 2017; Reaven 2012; Storch 2013; Storch 2015; White 2013; Wood 2009; Wood 2015). Other comorbidities required for inclusion in individual studies included attention‐deficit/hyperactivity disorder (ADHD) (Sciberras 2018), school refusal (Last 1998), headaches (Sharma 2017), functional physical complaints (Masia Warner 2011), and inflammatory bowel disease (IBD) (Reigada 2015). No studies included samples of children with intellectual impairments.

Intervention

The most commonly used or adapted treatment manuals were Coping Cat (Kendall 1994), and Cool Kids (Hudson 2009; Rapee 2006). Studies targeting specific anxiety disorders used disorder‐specific interventions (e.g. OST (One‐Session Treatment) for specific phobias, IAFS (Intervention in Adolescents with Social Phobia) for social anxiety disorder), and studies targeting children with ASD (or another comorbidity) used adapted interventions developed specifically for the target population (e.g. BIACA (Behavioral Interventions for Anxiety in Children with Autism), TAPS (Treatment of Anxiety and Physical Symptoms)).

Delivery formats varied across the included studies, including child‐focused interventions (45 studies), interventions delivered to both children and parents (40 studies), and interventions delivered only to parents (10 studies). Approximately half of the CBT interventions involved a group format, including child‐focused interventions (e.g. Gallagher 2004; Ginsburg 2002; Olivares 2005), interventions delivered to children and parents (e.g. Hudson 2009; McConachie 2014; Waters 2009), and parent‐only interventions (e.g. Cartwright Hatton 2011; Cobham 2017; Rapee 2005).

The amount of therapist contact time also varied considerably in the included studies, ranging from under 10 hours (interventions in 16 studies, e.g. Creswell 2017; Ishikawa 2019; Ollendick 2009) to more than 20 hours (interventions in 23 studies, e.g. Herbert 2009; Sharma 2017; Storch 2015). The remaining interventions involved between 10 and 20 hours of therapist contact time.

Measures

The majority of included studies (70 out of 87) included in meta‐analyses used a post‐treatment diagnostic assessment, and reported (or provided) data on remission from primary and/or all anxiety disorders. In most cases, the Anxiety Disorder Interview Schedule for Children (ADIS‐C) was used, although there was variation in how it was administered, including either separate child and parent interviews (e.g. Spence 2011; Thirlwall 2013; Wergeland 2014), joint child and parent interviews (Barrington 2005; Ishikawa 2019; Khanna 2010), only child interviews (e.g. Ginsburg 2002; Olivares 2005; Sanchez Garcia 2009), or only parent interviews (e.g. Lau 2017; McNally Keehn 2013; Waters 2009). Alternative post‐treatment diagnostic assessments included: the Kinder‐DIPS (Melfsen 2011; Schneider 2011), Kiddie Schedule for Affective Disorders and Schizophrenia (K‐SADS) (Hirshfeld Becker 2010; Last 1998; Lau 2010; Silk 2018), Diagnostic Interview Schedule for Children, Adolescents and Parents (DISCAP) (Shortt 2001), Diagnostic Interview Schedule for Children (Southam Gerow 2010), an adapted version of the Development and Well‐Being Assessment (DAWBA) (Shahnavaz 2016), and the Mini‐KID (Sharma 2017).

Anxiety symptoms were assessed using a range of child‐ and parent‐reported questionnaire measures of child anxiety symptoms, including the Spence Children's Anxiety Scale (SCAS‐C/P) (or the parent‐reported Preschool Anxiety Scale for younger children), Multidimensional Anxiety Scale for Children (MASC‐C/P), the Screen for Children Anxiety Related Emotional Disorders (SCARED‐C/P), State Trait Anxiety Inventory‐Trait Scale (STAIC‐T), the child‐reported Revised Children’s Manifest Anxiety Scale (RCMAS), the child‐reported Revised Child Anxiety and Depression Scale‐Anxiety Scale (RCADS‐A), the parent‐reported Child Behavior Checklist‐Anxiety Scale (CBCL‐A), and the parent‐reported Child and Adolescent Symptom Inventory‐4 ASD Anxiety Scale. Where broad measures of anxiety symptoms were not reported, the following disorder‐specific symptom questionnaire measures were used in the main analyses: Social Anxiety Scale for Adolescents (SAS‐A) (Herbert 2009; Masia Warner 2005; Masia Warner 2007; Olivares 2014; Rosa Alcazar 2009), Social Phobia and Anxiety Inventory for Children (SPAI) (Herbert 2009; Masia Warner 2005; Masia Warner 2007; Masia Warner 2016; Melfsen 2011; Olivares 2005; Rosa Alcazar 2009; Sanchez Garcia 2009), IBD‐Specific Anxiety Scale (IBD‐SAS) (Reigada 2015), Intra Oral Injection Fear Scale (IOIF‐s) (Berge 2017), and Child Survey Fear Schedule: Dental Subscale (Shahnavaz 2016).

Depressive symptoms were most frequently assessed using self‐reported questionnaires, and where parent‐report was used, it was always used alongside self‐report, so only self‐report depressive symptom measures were used in the analyses. The most frequently used measure of depressive symptoms was the Children's Depression Inventory; other measures included the Short Mood and Feelings Questionnaire (Arendt 2016; Kidd 2018; Thirlwall 2013; Wergeland 2014), Mood and Feelings Questionnaire (Perrin 2019), Revised Child Anxiety and Depression Scale‐Depression scale (RCADS‐D) (Muris 2002), and Beck Depression Inventory‐II (BDII) (Masia Warner 2007).

The included studies that assessed post‐treatment global functioning all used the clinician‐rated Children’s Global Assessment Scale (CGAS) or adapted, White 2013, or translated versions of the CGAS (Melfsen 2011).

Excluded studies

We excluded 593 full‐text articles; the reasons for exclusion are detailed in Characteristics of excluded studies.

We excluded studies if participants were not fully randomised to conditions (Bodden 2008a; Cobham 2012; Hayward 2000; Kujawa 2019; Mendlowitz 1999; Nauta 2003), or if the study did not use a standardised structured interview to determine the presence of baseline anxiety disorders (e.g. Gil Bernal 2009; Jansen 2012; Kerns 2016; Sevi Tok 2016; Sung 2011). We excluded studies where only some participants were under 19 years of age (e.g. Salzer 2018), or only some participants met the diagnostic criteria for anxiety disorder (e.g. Bernstein 2005; Ginsburg 2019b; Pereira 2014; Weisz 2012), if the study authors did not provide data on eligible participants. We also excluded studies if the intervention did not include both cognitive and behavioural elements (e.g. Beidel 2000; Beidel 2007; Bergman 2013; Cotton 2019; Ebrahiminejad 2016; Oerbeck 2014; Ollendick 2018; Ost 2015; Ozyurt 2018; Rudy 2017; Storch 2019); there were no face‐to‐face CBT sessions (e.g. Lyneham 2006); or where the control intervention included elements of CBT (e.g. Clementi 2019).

We excluded five studies because useable data were not provided for any relevant outcomes (Asbrand; Asbrand 2019; Baer 2005; Flatt 2010; NCT00576719).

Ongoing studies

We identified five ongoing studies that meet our inclusion criteria (see Characteristics of ongoing studies).

Studies awaiting classification

We identified trial registrations for 13 studies where insufficient information was available to determine eligibility, and either the trial was ongoing/not started or the trial data were not available (see Characteristics of studies awaiting classification).

Risk of bias in included studies

A summary of the risk of bias in the included studies is provided in Figure 2 and Figure 3.

Figure 2
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Over half of the included studies (47 out of 88) clearly described an adequate process of sequence generation, and just under half (39 out of 88) clearly described an adequate process of allocation concealment. In remaining studies, insufficient or no information about the sequence generation process was provided.

We performed a sensitivity analysis for the primary outcome (remission of primary anxiety disorder post‐treatment) amongst CBT versus waitlist controls, using only those studies that clearly described sequence generation, and found similar findings to the main analysis (odds ratio (OR) 4.84, 95% confidence interval (CI) 3.03 to 7.73, Z = 6.61, P < 0.001) (Analysis 1.32).

Blinding

The vast majority of studies (72 out of 88) either clearly detailed using blind assessors to administer post‐treatment diagnostic interviews, or only used a parent/child report questionnaire, therefore we judged the risk of bias from not blinding to be low.

We performed a sensitivity analysis examining remission of primary anxiety disorder diagnosis amongst CBT versus waitlist controls, where studies that did not specify using blind assessors were removed. The findings were similar to the main analysis (OR 5.54, 95% CI 3.76 to 8.18, Z = 8.63, P < 0.001) (Analysis 1.33).

Blinding of participants and personnel was only possible in some CBT versus TAU, active control, or alternative treatment studies. These studies described measures to ensure comparability of interventions (e.g. amount of contact time); however, few studies specifically detailed attempts to conceal 'test' versus 'control' intervention from participants.

Incomplete outcome data

Most studies described information on dropouts and missing outcome data. The amount of missing data was small, with a description of appropriate methods to manage missing data provided. Sixteen studies provided insufficient information on missing data, and in three studies missing data were not balanced across groups, with more missing data in the CBT group than in the control group (Afshari 2014; Last 1998; Salum 2018).

Selective reporting

With the exception of two studies where post‐treatment anxiety symptoms outcomes were not reported for CBT and controls (O'Brien 2007; Waters 2009), the study reports included expected outcomes relevant to this review.

Notably, however, six studies included a post‐treatment diagnostic assessment and reported outcomes from this assessment (e.g. remission of a subset of anxiety disorders, reduction in the number of anxiety disorders), but did not report or provide data on remission of primary anxiety disorder or remission of all anxiety disorders (Ginsburg 2012; Hancock 2018; Murphy 2017; Pincus 2010; Walkup 2008; Wood 2009), so these studies were not included in our analyses of diagnostic outcomes.

Other potential sources of bias

Therapy integrity, including therapist competence and adherence, was assessed across all studies. In the majority of studies (61 out of 88), trained and supervised therapists were used, and adherence was measured and reported. In the remaining studies insufficient information was provided on competence or adherence, or both.

The most frequently identified other potential source of bias related to the timing of the post‐treatment assessment. In seven studies (Flannery Schroeder 2000; Fujii 2013; Ishikawa 2019; Kendall 1994; Kendall 1997; Masia Warner 2011; Pincus 2010), the timing of the post‐treatment assessment varied across groups, with a shorter waitlist or control period than treatment period.

Effects of interventions

See: Summary of findings 1 CBT compared with waitlist for children and adolescents with anxiety disorders; Summary of findings 2 CBT compared with treatment as usual for anxiety disorders in children and adolescents; Summary of findings 3 CBT compared with attention control for anxiety disorders in children and adolescents; Summary of findings 4 CBT compared with alternative treatment for anxiety disorders in children and adolescents

Comparison 1: CBT versus waitlist/no treatment

See summary of findings Table 1.

Primary outcomes
1.1 Remission of primary anxiety diagnosis post‐treatment

We included 39 studies in the analyses examining remission of primary anxiety disorder diagnosis post‐treatment (Arendt 2016; Cartwright Hatton 2011; Cobham 2017; Cornacchio 2019; Dadds 1997; Flannery Schroeder 2000; Gallagher 2004; Hirshfeld Becker 2010; Holmes 2014; Ishikawa 2019; Kendall 1994; Kendall 1997; Kidd 2018; Lau 2010; Masia Warner 2005; Masia Warner 2011; McConachie 2014; McNally Keehn 2013; Melfsen 2011; Olivares 2005; Olivares 2014; Olivares 2019; Ollendick 2009; Ost 2001; Perrin 2019; Rapee 2006; Rosa Alcazar 2009; Sanchez Garcia 2009; Santucci 2013; Schneider 2011; Silverman 1999b; Spence 2000; Spence 2006; Spence 2011; Thirlwall 2013; Villabo 2018; Waters 2009; Wergeland 2014; Wood 2015).

Moderate‐quality evidence from ITT analysis revealed a response rate for remission of primary anxiety diagnosis of 49.4% (802 out of 1623) for CBT versus 17.8% (191 out of 1074) for controls (OR 5.45, 95% CI 3.90 to 7.60, Z = 9.96, P < 0.001) (Analysis 1.1). Over half of the studies (59%; 23 out of 39) and comparisons (52%; 25 out of 48) showed a clear benefit for CBT (95% CI does not cross 1) (Figure 4), with evidence of moderate heterogeneity (Chi² = 97.73, df = 47, P < 0.001; I² = 52%).

Figure 4
Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Subgroup analyses

We found a difference in outcomes for the different delivery formats (child‐focused, child and parent, parent only) (Chi² = 8.57, df = 2, P = 0.01), with substantial subgroup heterogeneity (I² = 76.7%) (Analysis 1.1). Inspection of the forest plot and the summary statistics suggest that studies evaluating child‐focused CBT produced greater remission of primary anxiety diagnosis in the CBT arms (OR 10.42, 95% CI 5.84 to 18.58) compared to waitlist/no treatment controls than studies evaluating CBT delivered to both children and parents (OR 4.08, 95% CI 2.72 to 6.11) or parents only (OR 2.83, 95% CI 1.12 to 7.16). Notably, however, there was also evidence of moderate heterogeneity amongst studies evaluating child‐focused CBT (Chi² = 43.41, df = 22, P = 0.004; I² = 49%).

We found no differences in remission of primary anxiety diagnosis outcomes for individual‐ versus group‐delivered CBT (Chi² = 0.90, df = 1, P = 0.34; I² = 0%); interventions with different amounts of therapist contact time (< 10 hours, ≥ 10 and < 20 hours, ≥ 20 hours) (Chi² = 0.52, df = 2, P = 0.77; I² = 0%); or studies with participants in different age groups (≤ 12 years, ≥ 12 years, < 12 years and ≥12 years) (Chi² = 1.59, df = 2, P = 0.45; I² = 0%) (Analysis 1.2; Analysis 1.3; Analysis 1.4).

Subgroup analyses revealed no significant difference in outcomes between studies including participants with and without ASD (Chi² = 2.25, df = 1, P = 0.13; I² = 55.5%) (Analysis 1.5); however, only four studies included children/adolescents with ASD, compared to 35 studies with participants without ASD. The four studies of children/adolescents with ASD revealed no significant difference between CBT and waitlist/no treatment controls in rates of remission of primary anxiety disorder diagnosis (OR 1.79, 95% CI 0.39 to 8.16, Z = 0.76, P = 0.45, n = 136, I² = 59%).

Completer analysis

The completer analysis included 1422 CBT participants and 984 controls, with a response rate for remission of primary anxiety diagnosis of 56.4% for CBT versus 11.7% for controls (OR 11.55, 95% CI 8.41 to 15.86, Z = 15.11, P < 0.001) (Analysis 1.6). Most of the studies (89%, 34 out of 38) and comparisons (81%; 38 out of 47) showed a clear benefit for CBT (95% CI does not cross 1). There was no heterogeneity across studies (Chi² = 61.53, df = 46, P = 0.06; I² = 25%).

1.2 Acceptability: loss of participants to post‐treatment assessment

Low‐quality evidence showed no difference between CBT and waitlist/no treatment controls in the rate of those lost to post‐treatment assessment (Analysis 1.7), indicating a similar degree of acceptability. The rate of loss to post‐treatment assessment was 13.5% (260 out of 1919) in the CBT group versus 10.4% (129 out of 1239) in the control group (OR 1.09, 95% CI 0.85 to 1.41, Z = 0.69, P = 0.49, n = 3158, k = 45, I² = 4%).

Secondary outcomes
1.3 Remission for all anxiety disorders post‐treatment

We included 28 studies in the analyses examining post‐treatment remission of all anxiety disorder diagnoses (Arendt 2016; Barrett 1996; Barrett 1998; Cartwright Hatton 2011; Chalfant 2007; Chiu 2013; Cobham 2017; Dadds 1997; Hirshfeld Becker 2010; Ishikawa 2019; Kennedy 2009; Lau 2017; McConachie 2014; Melfsen 2011; Olivares 2005; Perrin 2019; Rapee 2005; Rapee 2006; Rosa Alcazar 2009; Sanchez Garcia 2009; Shortt 2001; Simon 2011; Spence 2000; Spence 2006; Spence 2011; Thirlwall 2013; Villabo 2018; Waters 2009).

The ITT analysis included 1213 CBT participants and 862 controls, with a response rate for remission of all anxiety disorder diagnoses of 46.8% for CBT versus 19.1% for waitlist/no treatment controls (OR 4.43, 95% CI 2.89 to 6.78, Z = 6.85, P < 0.001) (Analysis 1.8). Sixty per cent of included studies (17 out of 28) and 51% of comparisons (18 out of 35) reported findings showing a clear benefit for CBT (95% CI does not cross 1). There was evidence of moderate heterogeneity (Chi² = 84.49, df = 34, P < 0.001; I² = 60%), and thus moderate‐quality evidence overall for the benefit of CBT compared to waitlist/no treatment controls for remission of all anxiety disorders.

Subgroup analyses

We found a difference in outcomes for different delivery formats (child‐focused, child and parent, parent‐only) (Chi² = 8.14, df = 2, P = 0.02), with substantial subgroup heterogeneity (I² = 75.4%). There was evidence of substantial heterogeneity amongst studies evaluating child‐focused CBT (I² = 70%), but we found that overall these studies produced greater remission of all anxiety disorder diagnoses following CBT compared to waitlist/no treatment controls (OR 8.52, 95% CI 2.97 to 24.39), than studies evaluating child‐and‐parent delivered CBT (OR 5.19, 95% CI 3.26 to 8.27), or studies evaluating parent‐delivered CBT (OR 1.89, 95% CI crosses 1).

We found no differences in remission of all anxiety disorder diagnoses outcomes for individual‐ versus group‐delivered CBT (Chi² = 0.35, df = 1, P = 0.56; I² = 0%) (Analysis 1.9); however, there was variation in these outcomes amongst interventions with differing amounts of therapist contact time (Chi² = 10.75, df = 2, P = 0.005; I² = 81.4%) (Analysis 1.10). Only five studies evaluated interventions involving < 10 hours therapist contact time (Cobham 2017; Ishikawa 2019; Rapee 2005; Simon 2011; Thirlwall 2013), and did not indicate a benefit for CBT compared to waitlist/no treatment control groups for remission of all anxiety disorders (OR 1.42, 95% CI 0.68 to 2.96, Z = 0.93, P = 0.35), whilst interventions with 10 to 20 hours or ≥ 20 hours contact time both showed a clear benefit for CBT (OR 6.59, 95% CI 3.62 to 12.01 and OR 5.03, 95% CI 2.55 to 9.93, respectively).

Subgroup analyses did not show differences in remission of all anxiety disorder diagnoses outcomes amongst studies with participants in different age groups (Chi² = 0.80, df = 2, P = 0.67; I² = 0%) (Analysis 1.11), and as only two studies included participants with ASD, meaningful subgroup analyses comparing studies with and without participants with ASD was not possible.

Completer analysis

The completer analysis included 1074 CBT participants and 797 controls, with a response rate for remission of any anxiety disorder diagnosis of 52.9% for CBT versus 14.2% for controls (OR 9.13, 95% CI 5.78 to 14.41, Z = 9.49, P < 0.001) (Analysis 1.12). More than three‐quarters of included studies (21 out of 27) and 65% of comparisons (22 out of 34) reported findings showing a clear benefit for CBT (95% CI does not cross 1). There was evidence of moderate heterogeneity (Chi² = 68.75, df = 33, P < 0.001; I² = 52%).

1.4 Reduction in anxiety symptoms (child report) post‐treatment

We included 45 studies in the analyses examining post‐treatment child‐reported anxiety symptoms (Afshari 2014; Arendt 2016; Barrett 1996; Berge 2017; Chalfant 2007; Cheung 2016; Chiu 2013; Cobham 2017; Dadds 1997; Flannery Schroeder 2000; Gallagher 2004; Hancock 2018; Holmes 2014; Ishikawa 2019; Kendall 1994; Kendall 1997; Kidd 2018; Lau 2010; Leutgeb 2012; Masia Warner 2005; McConachie 2014; McNally Keehn 2013; Melfsen 2011; Olivares 2005; Olivares 2014; Ollendick 2009; Ost 2001; Perrin 2019; Rapee 2006; Rosa Alcazar 2009; Salari 2018; Sanchez Garcia 2009; Santucci 2013; Schneider 2011; Shortt 2001; Silverman 1999b; Simon 2011; Smith 2014; Spence 2000; Spence 2006; Spence 2011; Thirlwall 2013; Villabo 2018; Wergeland 2014; Wood 2009). The standardised mean difference (SMD) between CBT groups and waitlist/no treatment controls was −0.67 (95% CI −0.88 to −0.47, Z = 6.36, P < 0.001, n = 2831) (Analysis 1.13; Figure 5). Substantial heterogeneity (Chi² = 352.54, df = 56, P < 0.001; I² = 84%) reduced our confidence in this result, and we rated this evidence as low‐quality.

Figure 5
Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.13 Reduction in anxiety symptoms (child report) post‐treatment.

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.13 Reduction in anxiety symptoms (child report) post‐treatment.

Indeed, including only studies that used a broad measure of child‐reported anxiety symptoms (38 studies, 2459 participants), the SMD between CBT groups and waitlist controls was reduced to −0.41 (95% CI −0.57 to −0.25, Z = 5.13, P < 0.001, I² = 68%) (Analysis 1.14).

Subgroup analyses

We found a significant difference between child‐focused, child‐and‐parent, and parent‐only CBT (Chi² = 14.67, df = 2, P < 0.001) on child‐reported anxiety symptom outcomes, with considerable subgroup heterogeneity (I² = 86.4%). There was no significant difference between CBT and waitlist/no treatment control groups in post‐treatment child‐reported anxiety symptoms for parent‐delivered CBT (SMD 0.04, 95% CI −0.38 to 0.46, Z = 0.19, P = 0.85, 5 studies). The forest plot and summary statistics indicated that child‐focused CBT produced a greater reduction on child‐reported anxiety symptoms amongst CBT groups compared to waitlist/no treatment controls (SMD −1.04, 95% CI −1.41 to −0.67, Z = 5.48, P < 0.001, 24 studies) than child‐and‐parent delivered CBT (SMD −0.45, 95% CI −0.67 to −0.23, Z = 3.99, P < 0.001, 20 studies), but with substantial heterogeneity within each of these subgroups (child‐focused: I² = 88%; child‐and‐parent: I² = 69%).

Subgroup analyses also revealed a significance difference between individual‐ and group‐focused CBT (Chi² = 6.47, df = 1, P = 0.01; I² = 84.5%) (Analysis 1.15). The SMD in post‐treatment child‐reported anxiety symptoms between CBT and waitlist/no treatment control groups was −0.91 (95% CI −1.22 to −0.60) in studies evaluating group‐focused CBT compared to −0.39 (95% CI −0.64 to −0.15) in studies evaluating individual‐focused CBT.

We found no differences between interventions with different amounts of therapist contact time (< 10 hours, ≥ 10 and < 20 hours, ≥ 20 hours) (Chi² = 3.33, df = 2, P = 0.19; I² = 39.9%) (Analysis 1.16), but child‐reported anxiety symptoms outcomes did vary amongst different age groups (Chi² = 8.27, df = 2, P = 0.02), with substantial subgroup heterogeneity (I² = 75.8%) (Analysis 1.17). Amongst studies of children aged ≥ 12 years, the SMD between CBT and waitlist/no treatment controls was −1.78 (95% CI −3.01 to −0.56) compared to −0.62 (95% CI −0.83 to −0.41) where studies included participants over and under 12 years of age, and −0.23 (95% CI cross 0) amongst studies of children ≤ 12 years, but with variation within each of these age groups (I² = 71% to 95%).

We did not find differences between studies including participants with and without ASD on post‐treatment child‐reported anxiety symptoms outcomes (Chi² = 0.02, df = 1, P = 0.88; I² = 0%) (Analysis 1.18).

1.5 Reduction in anxiety symptoms (parent report) post‐treatment

We included 35 studies in analyses examining post‐treatment parent‐reported anxiety symptoms (Afshari 2014; Arendt 2016; Chalfant 2007; Cheung 2016; Chiu 2013; Cobham 2017; Flannery Schroeder 2000; Holmes 2014; Ishikawa 2019; Kendall 1994; Kendall 1997; Kennedy 2009; Kidd 2018; Lau 2010; Lau 2017; Masia Warner 2005; McConachie 2014; McNally Keehn 2013; Olivares 2014; Perrin 2019; Rapee 2005; Rapee 2006; Rosa Alcazar 2009; Santucci 2013; Schneider 2011; Silverman 1999b; Smith 2014; Spence 2006; Spence 2011; Thirlwall 2013; Villabo 2018; Wergeland 2014; White 2013; Wood 2009; Wood 2015). The SMD between CBT and waitlist/no treatment control groups was −0.70 (95% CI −0.90 to −0.51, Z = 7.01, P < 0.001, n = 2137, low‐quality evidence), with substantial heterogeneity (Chi² = 171.97, df = 40, P < 0.001; I² = 77%) (Analysis 1.19; Figure 6.

Figure 6
Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.19 Reduction in anxiety symptoms (parent report) post‐treatment.

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.19 Reduction in anxiety symptoms (parent report) post‐treatment.

Including only studies that used a broad measure of parent‐reported anxiety symptoms (32 studies, 1952 participants), the SMD between CBT and waitlist groups was reduced to −0.59 (95% CI −0.77 to −0.41, Z = 6.33, P < 0.001, I² = 71%) (Analysis 1.20).

Subgroup analyses

Subgroup analyses revealed no significant differences between child‐focused, child and parent, and parent‐only on parent‐reported anxiety symptom outcomes (Chi² = 3.43, df = 2, P = 0.18) (Analysis 1.19). However, there was evidence of a difference amongst individual compared to group interventions (Chi² = 6.79, df = 1, P = 0.009), with substantial subgroup heterogeneity (I² = 85.3%) (Analysis 1.21). The SMD between CBT and waitlist/no treatment post‐treatment parent‐reported anxiety symptoms was −0.92 (95% CI −1.21 to −0.62) amongst group interventions, compared to −0.43 (95% CI −0.65 to −0.21) for individual interventions.

We did not find differences in parent‐reported anxiety symptom outcomes across interventions involving varying amount of therapist contact time (Chi² = 3.77, df = 2, P = 0.15; Analysis 1.22). On the other hand, there was evidence of a difference in these outcomes amongst different age groups (Chi² = 5.87, df = 2, P = 0.05), with substantial subgroup heterogeneity (I² = 65.9%) (Analysis 1.23). Amongst studies that included participants aged ≥ 12 years, the SMD between CBT and waitlist/no treatment groups was −1.07 (95% CI −1.78 to −0.37), compared to −0.76 (95% CI −1.04 to −0.49) in studies that included participants under and over 12 years of age, and −0.40 (95% CI −0.63 to −0.17) where participants were ≤ 12 years.

We found no significant differences in parent‐reported anxiety symptom outcomes amongst samples with and without ASD (Chi² = 1.42, df = 1, P = 0.23; Analysis 1.24).

1.6 Reduction in depressive symptoms post‐treatment

Seventeen studies reported post‐treatment depressive symptoms (Arendt 2016; Barrett 1996; Flannery Schroeder 2000; Gallagher 2004; Ishikawa 2019; Kendall 1994; Kendall 1997; Kidd 2018; Masia Warner 2005; Melfsen 2011; Ollendick 2009; Ost 2001; Perrin 2019; Salari 2018; Silverman 1999b; Thirlwall 2013; Wergeland 2014). The SMD between CBT and waitlist/no treatment control groups was −0.34 (95% CI −0.51 to −0.17, Z = 3.96, P < 0.001, n = 1157, moderate‐quality evidence), with moderate heterogeneity (Chi² = 39.62, df = 22, P = 0.01; I² = 44%) (Analysis 1.25).

1.7 Improvement in global functioning post‐treatment

Data pooled from 11 studies that reported post‐treatment global functioning produced an SMD between CBT and waitlist/no treatment control groups of 1.03 (95% CI 0.68 to 1.38, Z = 5.83, P < 0.001, n = 557, low‐quality evidence), with substantial heterogeneity (Chi² = 37.41, df = 11, P < 0.001; I² = 71%) (Analysis 1.26) (Cornacchio 2019; Holmes 2014; Masia Warner 2005; Masia Warner 2011; Melfsen 2011; Perrin 2019; Salari 2018; Santucci 2013; Spence 2011; Villabo 2018; White 2013).

1.8 Adverse events

No included study reported adverse events in both CBT and waitlist/no treatment groups, so meta‐analysis was not possible.

Remission of primary anxiety disorder diagnosis at follow‐up

Four studies reported remission of primary anxiety disorder at follow‐up assessments ≤ 6 months post‐treatment (Dadds 1997; Gallagher 2004; Olivares 2005; Rosa Alcazar 2009), with ITT analysis showing a rate for remission of primary anxiety diagnosis of 66.7% (48 out of 72) for CBT groups versus 34.6% (28 out of 81) for waitlist/no treatment controls (OR 10.94, 95% CI 2.33 to 51.41, Z = 3.03, P = 0.002, I² = 40%) (Analysis 1.27). Only 3 studies (166 participants) reported remission of primary anxiety disorder 12 months post‐treatment (Cartwright Hatton 2011; Dadds 1997; Rosa Alcazar 2009); pooled data did not suggest a benefit for CBT compared to waitlist/no treatment control groups (OR 2.80, 95% 0.24 to 33.19, Z = 0.82, P = 0.41, I² = 83%).

Only one study reported remission of primary anxiety disorder > 12 months post‐treatment, so meta‐analysis was not possible.

Completer analyses produced similar findings, with evidence of an intervention effect at follow‐up assessments ≤ 6 months post‐treatment but not at 12‐month follow‐up (Analysis 1.28)

Remission of all anxiety disorder diagnoses at follow‐up

Five studies reported data on remission of all anxiety disorders at follow‐up assessments (Cartwright Hatton 2011; Dadds 1997; Olivares 2005; Rosa Alcazar 2009; Simon 2011), including ≤ 6 months post‐treatment (Dadds 1997; Olivares 2005; Rosa Alcazar 2009), 12 months post‐treatment (Cartwright Hatton 2011; Dadds 1997; Rosa Alcazar 2009), and 24 months post‐treatment (Dadds 1997; Simon 2011). Neither ITT analyses (Analysis 1.29) nor completer analyses (Analysis 1.30) showed a significant difference between CBT and waitlist/no treatment at any of these follow‐up time points.

Reduction in anxiety symptoms at follow‐up

Six studies reported data on child‐reported anxiety symptoms at follow‐up assessments (Analysis 1.31) (Afshari 2014; Dadds 1997; Gallagher 2004; Olivares 2005; Rosa Alcazar 2009; Simon 2011). Child‐reported anxiety symptoms at ≤ 6 months post‐treatment produced an SMD of −1.92 (95% CI −2.95 to −0.89, Z = 3.65, P < 0.001, n = 179) between CBT and waitlist/no treatment groups (Afshari 2014; Dadds 1997; Gallagher 2004; Olivares 2005; Rosa Alcazar 2009), with substantial heterogeneity across studies (Chi² = 35.35, df = 5, P < 0.001; I² = 86%). Only two studies reported corresponding outcomes at 12‐month follow‐up (Dadds 1997; Rosa Alcazar 2009), and only one study at 24‐month follow‐up (Simon 2011), so meta‐analyses were not possible for these time points.

As only two studies reported data on parent‐reported anxiety symptoms at 6 months post‐treatment (Afshari 2014; Rosa Alcazar 2009), and one with corresponding data at 12 months post‐treatment (Rosa Alcazar 2009), meta‐analyses were not performed using these data.

Comparison 2: CBT versus treatment as usual

See summary of findings Table 2.

Primary outcomes
2.1 Remission of primary anxiety diagnosis post‐treatment

Eight studies provided uncertain evidence on post‐treatment remission of primary anxiety disorder diagnosis for CBT versus TAU (Fujii 2013; Ginsburg 2019a; Reaven 2012; Shahnavaz 2016; Sharma 2017; Southam Gerow 2010; Storch 2013; Storch 2015).

The ITT analysis included 281 CBT participants and 206 TAU participants, with no evidence of a significant difference in the rate of remission of primary anxiety diagnosis post‐treatment (44.1% and 40.8% respectively; OR 3.19, 95% CI 0.90 to 11.29, Z = 1.80, P = 0.07, low‐quality evidence) (Analysis 2.1). There was evidence of substantial heterogeneity (Chi² = 28.93, df = 7, P < 0.001; I² = 76%).

Subgroup analyses

There was a significant difference in remission of primary anxiety disorder outcomes amongst studies evaluating child‐focused CBT compared to child‐and‐parent delivered CBT (Chi² = 10.90, df = 1, P < 0.001; I² = 90.8%). Data pooled from five studies (172 participants) that evaluated a child‐and‐parent delivered intervention showed a clear benefit for CBT compared to TAU (OR 8.56, 95% CI 3.10 to 23.66, Z = 4.14, P < 0.001), whereas data from three studies (315 participants) evaluating child‐focused interventions did not (OR 0.61, 95% CI cross 1). None of the studies that compared CBT to TAU evaluated a parent‐only intervention.

Planned subgroup analyses comparing individual versus group interventions, interventions with varying amount of therapist contact time, and different age groups were not possible due to an insufficient number of studies. Only two studies evaluated group CBT; only one study used an intervention with < 10 hours therapist contact time; and five (out of eight) studies included pre‐adolescent and adolescent participants.

Subgroup analyses also indicated a difference in remission outcomes in studies with participants with ASD compared to studies with participants without ASD (Chi² = 5.71, df = 1, P = 0.02; I² = 82.5%) (Analysis 2.2). Pooled data from four ASD samples showed a higher rate of remission of primary anxiety disorder amongst CBT groups compared to TAU groups (OR 11.25, 95% CI 3.11 to 40.79, Z = 3.68, P < 0.001, n = 142, I² = 0%), but corresponding data from four samples without ASD showed no significant difference between CBT and TAU groups (OR 1.14, 95% CI 0.29 to 4.47, Z = 0.18, P = 0.85, n = 345, I² = 75%).

Completer analysis

The completer analysis included 248 CBT participants and 192 TAU participants, with a response rate for remission of primary anxiety diagnosis of 50.4% for CBT versus 36.5% for controls (OR 4.87, 95% CI 1.51 to 15.72, Z = 2.65, P = 0.008) (Analysis 2.3). Similar to the ITT analysis, there was substantial heterogeneity across the eight studies (Chi² = 21.30, df = 7, P = 0.003; I² = 67%).

2.2 Acceptability: loss of participants to post‐treatment assessment

We found no evidence of a difference between CBT and TAU in rate of those lost to post‐treatment assessment, illustrating a comparable degree of acceptability (OR 1.37, 95% CI 0.73 to 2.56, Z = 0.98, P = 0.33, I² = 0%, k = 8, n = 441, low‐quality evidence) (Analysis 2.4).

Secondary outcomes
2.3 Remission of all anxiety disorders post‐treatment

Five studies provided uncertain evidence on remission of all anxiety disorder diagnoses post‐treatment for CBT versus TAU (Barrington 2005; Sciberras 2018; Sharma 2017; Storch 2013; Storch 2015).

The ITT analyses showed a rate of remission of all anxiety diagnoses of 55.8% for CBT versus 41.4% for TAU (OR 2.74, 95% CI 1.16 to 6.46, Z = 2.30, P = 0.02, k = 5, n = 203, low‐quality evidence) (Analysis 2.5). There was no evidence of significant heterogeneity (Chi² = 3.92, df = 4, P = 0.42; I² = 0%).

Completer analyses showed a similar benefit for CBT versus TAU (OR 2.78, 95% CI 1.18 to 6.55, Z = 2.33, P = 0.02, k = 5, n = 201, I² = 0%) (Analysis 2.6).

2.4 Reduction in anxiety symptoms (child‐report) post‐treatment

We included six studies in the analyses examining post‐treatment child‐reported anxiety symptoms (Barrington 2005; Ginsburg 2012; Sciberras 2018; Shahnavaz 2016; Sharma 2017; Southam Gerow 2010). We found no difference between CBT and TAU in child‐reported anxiety symptoms post‐treatment (SMD −0.15, 95% CI −0.78 to 0.48, Z = 0.46, P = 0.64, n = 214, low‐quality evidence), although there was substantial heterogeneity (Chi² = 23.55, df = 5, P < 0.001; I² = 79%) (Analysis 2.7). This finding was replicated after removal of the one study that used a disorder‐specific symptoms measure (SMD 0.07, 95% CI −0.54 to 0.68, Z = 0.22, P = 0.83, n = 187, I² = 74%) (Analysis 2.8).

2.5 Reduction in anxiety symptoms (parent‐report) post‐treatment

Pooled data from seven studies showed no difference between CBT and TAU in post‐treatment parent‐reported anxiety symptoms (SMD −0.32, 95% CI −0.70 to 0.06, Z = 1.64, P = 0.10, n = 228, low‐quality evidence) (Barrington 2005; Ginsburg 2002; Sciberras 2018; Shahnavaz 2016; Southam Gerow 2010; Storch 2013; Storch 2015), with moderate heterogeneity across studies (Chi² = 11.66, df = 6, P = 0.07; I² = 49%) (Analysis 2.9). Removal of one study that used a disorder‐specific measure did not change this finding (SMD −0.16, 95% CI −0.44 to 0.11, Z = 1.16, P = 0.25, n = 202, I² = 0%) (Analysis 2.10).

2.6 Reduction in depressive symptoms post‐treatment

Only one study reported depressive symptoms post‐treatment for CBT versus TAU (O'Brien 2007), so meta‐analysis was not possible.

2.7 Improvement in global functioning post‐treatment

Only one study reported post‐treatment global functioning for CBT versus TAU (Sharma 2017), so meta‐analysis was not possible.

2.8 Adverse events

No included study reported adverse events in both CBT and TAU groups, so meta‐analysis was not possible.

2.9 Follow‐up outcomes

There was an insufficient number of studies to conduct meta‐analyses examining longer‐term diagnostic or symptom outcomes. Only two studies reported remission from primary anxiety disorders at 12 months post‐treatment (Ginsburg 2019a; Shahnavaz 2016), and one study reported remission from all anxiety disorders at 6 and 12 months (Barrington 2005). Similarly, only three studies reported child‐reported and parent‐reported anxiety symptoms at 6‐month and/or 12‐month follow‐up (Barrington 2005; Ginsburg 2012; Shahnavaz 2016).

Comparison 3: CBT versus attention control

See summary of findings Table 3.

Primary outcomes
3.1 Remission of primary anxiety diagnosis post‐treatment

Ten studies evaluating CBT versus attention control provided data on post‐treatment remission of primary anxiety disorder diagnosis (Ginsburg 2002; Hudson 2009; Kendall 2008; Khanna 2010; Last 1998; Masia Warner 2007; Masia Warner 2016; Ollendick 2009; Rosa Alcazar 2009; Silverman 1999a).

The ITT analysis (484 CBT participants; 338 control participants) provided uncertain evidence for a greater rate of remission of primary anxiety diagnosis post‐treatment for CBT compared to attention controls (48.3% versus 29.3%; OR 2.28, 95% CI 1.33 to 3.89, Z = 3.00, P = 0.003, low‐quality evidence), with moderate heterogeneity across studies (Chi² = 28.77, df = 12, P = 0.004; I² = 58%) (Analysis 3.1).

Subgroup analyses

None of the studies that evaluated CBT versus attention controls used a parent‐only intervention, but there was a difference in the pattern of results for remission of primary anxiety disorder outcomes amongst studies evaluating child‐focused CBT and child‐and‐parent delivered CBT when each was compared to attention controls (Chi² = 7.65, df = 1, P = 0.006; I² = 86.9%) (Analysis 3.1). Summary statistics indicated a benefit for CBT compared to attention controls amongst child‐focused interventions (OR 3.58, 95% CI 1.92 to 6.65, k = 7, n = 509), but not amongst child‐and‐parent delivered interventions (OR 1.12, 95% CI cross 1, k = 4, n = 313).

Subgroup analyses revealed no significant difference in remission of primary anxiety disorder outcomes in studies evaluating individual CBT versus group CBT (Chi² = 1.22, df = 1, P = 0.27; I² = 17.7%) (Chi² = 0.48, df = 1, P = 0.49; I² = 0%) (Analysis 3.2).

Further subgroup analyses were not possible because there were fewer than three studies in at least one group for each planned subgroup analysis.

Completer analysis

In the completer analysis, 53.1% of CBT participants compared to 23.2% of attention control participants were free of their primary anxiety disorder diagnosis post‐treatment, with a significant intervention effect (OR 3.88, 95% CI 2.52 to 5.97, Z = 6.17, P < 0.001, k = 10, n = 752, I² = 21%) (Analysis 3.3).

3.2 Acceptability: loss of participants to post‐treatment assessment

Low‐quality evidence indicated no difference between CBT and attention controls in rate of those lost to post‐treatment assessment, suggesting a similar degree of acceptability (OR 1.00, 95% CI cross 1, Z = 0.01, P = 0.99, n = 797, k = 12, I² = 0%) (Analysis 3.4).

Secondary outcomes
3.3 Remission of all anxiety disorders post‐treatment

We included five studies in analyses comparing CBT versus attention controls for remission of all anxiety disorder diagnoses post‐treatment (Ginsburg 2002; Hudson 2009; Kendall 2008; Masia Warner 2007; Rosa Alcazar 2009).

Both ITT analyses and completer analyses indicated a benefit for CBT compared to attention control conditions (ITT: OR 2.75, 95% CI 1.22 to 6.17, Z = 2.45, P = 0.01, n = 378, I² = 49%, low‐quality evidence; completers: OR 3.88, 95% CI 1.58 to 9.51, Z = 2.96, P = 0.003, n = 359, I² = 48%) (Analysis 3.5; Analysis 3.6).

3.4 Reduction in anxiety symptoms (child‐report) post‐treatment

Fifteen studies provided moderate‐quality evidence for post‐treatment child‐reported anxiety symptoms for CBT and attention control groups (Ginsburg 2002; Herbert 2009; Hudson 2009; Ingul 2013; Kendall 2008; Khanna 2010; Last 1998; Masia Warner 2007; Masia Warner 2016; Muris 2002; Ollendick 2009; Pincus 2010; Rosa Alcazar 2009; Salum 2018; Silverman 1999a). The SMD between CBT and attention controls for child‐reported anxiety symptoms was −0.31 (95% CI −0.51 to −0.11, Z = 3.09, P = 0.002, n = 978) (Analysis 3.7). There was evidence of moderate heterogeneity across studies (Chi² = 40.95, df = 20, P = 0.004; I² = 51%).

After removal of studies that used a disorder‐specific symptom measure, the SMD between CBT and attention controls was reduced to −0.18 (95% CI −0.34 to −0.02, Z = 2.26, P = 0.02, n = 685, k = 11, I² = 0%) (Analysis 3.8).

3.5 Reduction in anxiety symptoms (parent‐report) post‐treatment

Eight studies provided data on post‐treatment parent‐reported anxiety symptoms for CBT and attention control groups (Herbert 2009; Hudson 2009; Kendall 2008; Masia Warner 2007; Masia Warner 2016; Rosa Alcazar 2009; Salum 2018; Silverman 1999a). The analyses indicated no significant differences between CBT and attention controls (SMD −0.25, 95% CI −0.61 to 0.11, Z = 1.37, P = 0.17, n = 638, low‐quality evidence), but with substantial heterogeneity across studies (Chi² = 47.61, df = 11, P < 0.001; I² = 77%) (Analysis 3.9). Removal of studies that used disorder‐specific measures did not change this finding (SMD −0.04, 95% CI −0.26 to 0.18, Z = 0.36, P = 0.72, n = 345, k = 4, I² = 0%) (Analysis 3.10).

3.6 Reduction in depressive symptoms post‐treatment

Ten studies reported post‐treatment depressive symptoms for CBT versus attention controls (Ingul 2013; Kendall 2008; Khanna 2010; Last 1998; Masia Warner 2007; Muris 2002; Ollendick 2009; Pincus 2010; Salum 2018; Silverman 1999a), finding no between‐group difference (SMD −0.18, 95% CI −0.45 to 0.09, Z = 1.32, P = 0.19, n = 613, low‐quality evidence) (Analysis 3.11). There was evidence of moderate heterogeneity across studies (Chi² = 29.19, df = 12, P = 0.004; I² = 59%).

3.7 Improvement in global functioning post‐treatment

Only two studies reported post‐treatment global functioning for CBT versus attention controls (Khanna 2010; Masia Warner 2007), so meta‐analysis was not possible.

3.8 Adverse events

No study reported adverse events in both CBT and attention control groups, so meta‐analysis was not possible.

Remission of primary anxiety disorder diagnoses at follow‐up

Four studies provided data on remission of primary anxiety disorder at ≤ 6 months follow‐up (Hudson 2009; Masia Warner 2007; Masia Warner 2016; Rosa Alcazar 2009). ITT analyses showed a greater rate of remission for CBT groups compared to attention controls (OR 3.06, 95% CI 1.22 to 7.65, Z = 2.39, P = 0.02, n = 341), but with substantial heterogeneity across studies (Chi² = 12.62, df = 5, P = 0.02; I² = 60%) (Analysis 3.12).

The intervention effect increased in the completer analysis (OR 6.52, 95% CI 2.58 to 16.43, Z = 3.97, P < 0.001, k = 4, n = 302, I² = 44%) (Analysis 3.13).

Only one study reported remission of primary anxiety disorder at 12‐month follow‐up, so meta‐analysis was not possible for this time point.

Remission of all anxiety disorder diagnoses at follow‐up

Three studies reported remission from all anxiety disorder diagnoses at ≤ 6 months post‐treatment (Hudson 2009; Masia Warner 2007; Rosa Alcazar 2009). ITT analyses and completer analyses indicated a greater rate of remission of all anxiety disorders for CBT groups compared to attention controls (ITT: OR 5.44, 95% CI 1.00 to 29.60, Z = 1.96, P = 0.05, n = 205; completer analysis: OR 11.19, 95% CI 1.94 to 64.51, Z = 2.70, P = 0.007, n = 182), but with substantial heterogeneity across studies (ITT: I² = 72%, completer analysis: I² = 62%) (Analysis 3.14; Analysis 3.15).

Only one study reported remission of all anxiety disorder diagnoses at 12 months (Rosa Alcazar 2009), so meta‐analysis was not possible for this time point.

Reduction in anxiety symptoms at follow‐up

Six studies reported child‐reported anxiety symptoms at ≤ 6 months post‐treatment (Herbert 2009; Hudson 2009; Masia Warner 2007; Masia Warner 2016; Rosa Alcazar 2009; Silverman 1999a). Pooled data from these studies indicated a benefit for CBT compared to attention controls (SMD −0.44, 95% CI −0.79 to −0.08, Z = 2.40, P = 0.02, n = 434), but there was substantial heterogeneity across studies (Chi² = 23.08, df = 8, P = 0.003; I² = 65%) (Analysis 3.16).

The same six studies provided parent‐reported anxiety symptoms at ≤ 6 months post‐treatment; analyses showed no significant difference between CBT and attention controls (SMD −0.63, 95% CI cross 0, n = 434, I² = 90%) (Analysis 3.17).

Only two studies reported child‐reported anxiety symptoms and parent‐reported anxiety symptoms at 12 months follow‐up (Rosa Alcazar 2009; Silverman 1999a), so meta‐analyses for these outcomes were not possible for this time point.

Comparison 4: CBT versus alternative treatment

See summary of findings Table 4.

Primary outcomes
4.1 Remission of primary anxiety diagnosis post‐treatment

Only two studies that compared CBT to an alternative treatment reported post‐treatment remission of primary anxiety disorder (Creswell 2017; Reigada 2015), therefore data were insufficient to conduct meaningful meta‐analysis or planned subgroup analyses for this outcome.

4.2 Acceptability: loss of participants to post‐treatment assessment

Low‐quality evidence showed no difference between CBT and alternative treatments in rate of those lost to post‐treatment assessment (OR 1.58, 95% CI crosses 1, Z = 0.94, P = 0.35, k = 7, n = 515, I² = 53%), suggesting no difference in degree of acceptability (Analysis 4.1).

Secondary outcomes
4.3 Remission of all anxiety disorders post‐treatment

Pooled data from four studies that reported remission of all anxiety disorder diagnoses post‐treatment showed no difference between CBT and alternative treatment groups in either ITT analyses (OR 0.89, 95% CI crosses 1, Z = 0.25, P = 0.80, n = 401, I² = 73%, low‐quality evidence) or completer analyses (OR 1.52, 95% CI crosses 1, Z = 1.06, P = 0.29, n = 348, I² = 57%) (Analysis 4.2; Analysis 4.3) (Creswell 2017; Lebowitz 2019; Reigada 2015; Silk 2018).

4.4 Reduction in anxiety symptoms (child‐report) post‐treatment

Low‐quality evidence showed no differences between CBT and alternative treatment groups in post‐treatment child‐reported anxiety symptoms (SMD −0.09, 95% CI crosses 0, Z = 0.58, P = 0.56, k = 6, n = 399, I² = 51%) (Analysis 4.4). Removal of one study that used a disorder‐specific symptom measure did not change this finding (Analysis 4.5).

4.5 Reduction in anxiety symptoms (parent‐report) post‐treatment

We also found no significant differences between CBT and alternative treatment groups in post‐treatment parent‐reported anxiety symptoms (SMD −0.13, 95% CI −0.33 to 0.06, Z = 1.31, P = 0.19, k = 6, n = 423, 6 studies, I² = 0%, low‐quality evidence) (Analysis 4.6). The studies all used a broad‐based anxiety measure, so planned sensitivity analysis was not required.

4.6 Reduction in depressive symptoms post‐treatment

Only only study reported post‐treatment depressive symptoms (Salum 2018), so meta analysis was not possible.

4.7 Improvement in global functioning post‐treatment

No studies reported post‐treatment global functioning for CBT versus alternative treatments.

4.8 Adverse events

No included study reported adverse events in both CBT and alternative treatment groups, so meta‐analysis was not possible.

4.9 Follow‐up outcomes

For each of the longer‐term outcomes, data were only available from a maximum of two studies, so meta‐analyses were not possible.

Publication bias

We investigated publication bias in relation to remission of the primary anxiety disorder post‐treatment for CBT versus waitlist controls, and CBT versus attention controls.

CBT versus waitlist/no treatment

There was clear evidence of publication bias from inspection of the funnel plot (Figure 7), which was asymmetric, and the Egger’s test, which was highly significant (t = 5.52, P < 0.000). Furthermore, a contour funnel plot (not included here) indicated that the missing studies were in the areas of low significance, more compatible with publication bias than other explanations of funnel plot asymmetry such as variable study design (Peters 2008).

Figure 7
Funnel plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Funnel plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

CBT versus attention control

There was no evidence of publication bias with a symmetrical funnel plot (Figure 8) and non‐significant Egger’s test (t = 1.99, P = 0.072).

Figure 8
Funnel plot of comparison: 3 CBT vs attention control, outcome: 3.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Funnel plot of comparison: 3 CBT vs attention control, outcome: 3.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Discussion

Summary of main results

Since the last Cochrane Review, James 2015, there has been a considerable increase in the number of studies available for review, from 41 to 88, with more than three times the total number of participants. We found moderate‐quality evidence that CBT leads to greater remission of primary and all anxiety disorders than waitlist/no treatment in the short term. Using conservative ITT criteria, we found a remission rate of 49% for the primary anxiety disorder for CBT versus 18% for waitlist/no treatment controls (OR 5.45, 95% CI 3.90 to 7.60). Although this indicates a slight attenuation in effects from the last Cochrane Review, this is not uncommon as studies accumulate over time. Consistent with the last review, the number needed to treat for an additional beneficial outcome (NNTB) using these conservative ITT data was 3 (95% CI 2.25 to 3.57). Notably, the benefit of CBT compared to waitlist/no treatment controls was shown across all post‐treatment outcomes, albeit with uncertain evidence due to variation in reductions in child‐ and parent‐reported anxiety symptoms, and depressive symptoms, and limited data related to global functioning.

Unfortunately, despite the general increase in studies, there remains a relatively small number of studies comparing CBT to 'active' controls, and where this has been done the control arms included a broad range of interventions (from undefined and variable ‘treatment as usual’ to attention controls, and in a small number of cases treatments that might credibly be delivered in practice). Due to insufficient data, there was low‐quality evidence for each of these comparisons across most outcomes. Encouragingly, the available data did indicate that CBT may be associated with a higher rate of remission of the primary anxiety disorder and all anxiety disorders and a greater reduction in child‐reported anxiety symptoms, compared to attention controls. However, the limited data did not reveal an advantage of CBT compared to treatment as usual across most outcomes, or indicate a benefit of CBT compared to specified alternative treatments. Furthermore, we were unable to compare CBT to medication or CBT in combination with medication to drug placebo due to a lack of available studies.

Unsurprisingly given the high frequency of waitlist control designs, there also continues to be only a very small number of studies that have controlled outcomes at longer‐term follow‐up points. Where outcomes beyond post‐treatment were available, the studies had mixed results, but in terms of remission of the primary anxiety disorder, the current findings support an advantage of CBT compared to waitlist/no treatment and attention controls six months after treatment. However, there is as yet no evidence that this advantage of CBT continues at later time points, and there were insufficient data to determine whether any longer‐term benefits hold up against treatment as usual or alternative treatments.

Previous Cochrane Reviews on this topic have only included studies that involved children older than four years of age; however, in the current review we removed this restriction in order to identify the age of the youngest participants in RCTs of CBT for child anxiety disorders, and to explore relative outcomes for different age groups. We identified studies that included children as young as two years of age. Notably, studies with younger children typically took a parent‐led approach in which parents were supported by a clinician to implement CBT principles in their child’s day‐to‐day life (e.g. Cartwright Hatton 2011).

Given the low number of children with anxiety disorders who access evidence‐based treatments for anxiety disorders (e.g. Reardon 2020), we specifically looked at outcomes for subgroups of studies categorised by the amount of therapist contact time. Relatively few studies involved less than 10 hours of therapist contact time, limiting the amount of data available. However, primary anxiety disorder and anxiety symptom outcomes for CBT compared to waitlist/no treatment did not differ from the more lengthy treatments. The currently available data therefore suggests that services could provide CBT programmes that involve relatively low levels of therapist contact time without a negative impact on post‐treatment outcomes.

The wide variation across trials in relation to a range of study and sample characteristics limits our ability to draw clear conclusions about particular treatment formats or age groups. Amongst waitlist/no treatment comparisons, we found stronger effects for child‐focused CBT than child‐and‐parent and parent‐only interventions. However, parent involvement in trials is related to potentially confounding factors (e.g. child's age, study setting), and even within the child‐focused CBT studies there was considerable heterogeneity in outcomes and variation across study characteristics. Indeed, although we also found variation in outcomes amongst different age groups for CBT versus wait list/no treatment comparisons, this differed according to outcome and may indicate age‐related measurement issues, rather than differences in true effects. We found stronger effects amongst samples of adolescents than samples of children or mixed child/adolescent samples on child‐ and parent‐reported anxiety symptoms, but not in terms of remission of primary or all anxiety disorders. We found a similar picture when looking at subgroups of studies on the basis of whether they delivered CBT with individuals or groups. For example, there was no evidence of different results on diagnostic outcomes, but outcomes for group formats were stronger than individual formats when parent‐ and child‐reported symptoms were assessed (in the waitlist/no treatment comparison). Again, various potential confounders are likely to have influenced these outcomes, including, for example, participant age, study setting (e.g. school versus clinic), and target disorders. Consequently, on the basis of findings relating to treatment format, the currently available literature does not give clear and consistent evidence that one format is better or worse than any other.

Finally, we also set out to assess outcomes specifically for children with autism spectrum disorders (ASD) and intellectual disabilities. Twelve studies evaluated CBT for the treatment of anxiety disorders in children with ASD, with mixed findings across comparisons and outcomes. Notably, the four studies that compared CBT to treatment as usual for children with ASD (where 'treatment as usual' typically meant participants could continue to seek alternative interventions only) did find a consistent advantage of CBT in relation to remission of primary anxiety disorder. However, amongst waitlist/no treatment control comparisons, we only found evidence of a benefit for CBT in parent‐reported anxiety symptoms, not child‐reported symptoms or diagnostic outcomes. No RCTs evaluated the effectiveness of CBT for anxiety disorders amongst children with intellectual disabilities.

Overall completeness and applicability of evidence

As in the previous Cochrane Review, almost all included studies involved participants of both sexes. Whilst a number of countries are represented in this review, it is notable that they were almost exclusively high‐income countries (with two‐thirds of the included studies having been conducted either in the USA or Australia).

Although the overall age of participants ranged from 2 to 18 years, less than 30% of studies focused specifically on children up to age 12 (with less than 6% focusing on children under 8 years of age), and less than 20% focused specifically on adolescents, with the majority including children and adolescents across broad age ranges. This made it difficult to draw conclusions on the basis of specific child ages.

We included participants with comorbid disorders such as depression, other anxiety disorders, ADHD, and conduct disorders, permitting generalised conclusions that appear to be applicable to clinical practice. However, the majority of studies were conducted in outpatient settings, with 41 studies conducted in university‐based research settings. We considered other settings, including primary care and schools; notably, however, no studies were conducted with the most severe cases who receive treatment as day‐ or inpatients, which is perhaps a reflection of the challenges associated with running trials in these settings. Despite the continued growth in studies evaluating CBT for anxiety disorders in general, there were few additional studies that targeted children with ASD since the last Cochrane Review, and the lack of evaluation of CBT for anxiety disorders, including parent‐only approaches, amongst children with intellectual disabilities remains a considerable limitation.

It is important to note that we conducted separate analyses for different diagnostic and child/parent report measures; this meant that given the high levels of inconsistency in measures used and how they were reported, only a subset of studies were included in any one analysis. The most consistent diagnostic outcome measure was remission of the primary diagnosis (included in about two‐thirds of the CBT versus waitlist/no treatment comparisons), whereas less than half of the studies reported remission of all anxiety disorder diagnoses. Some studies did not report remission of anxiety disorders in either of these ways, instead, for example, reporting remission from particular groups of diagnoses, or changes in the total number of different diagnoses. As a result, these studies were not included in our analyses of diagnostic outcomes. Similarly, relevant studies were identified that did not provide data on any of the target outcomes and were therefore not included in the review (e.g. Baer 2005; Flatt 2010; NCT00576719).

Encouragingly, the evidence of the acceptability of CBT was consistent across comparisons, with no difference in the number of CBT participants retained to post‐treatment assessments compared to each of the control groups. However, it is important to acknowledge that rates of attrition may in part reflect acceptability of the research process, and alternative measures of acceptability (e.g. participant perceptions of the intervention) were not assessed in this review.

As in previous Cochrane Reviews of CBT for child anxiety disorders, the number of follow‐up studies remains very limited (four studies reported on our primary outcome at follow‐up assessments), making it difficult to draw conclusions about the medium‐ to long‐term benefits of CBT. The available findings from controlled studies suggest that there is evidence that CBT provides an advantage over waitlist/no treatment six months after treatment, but there is currently no evidence to support longer‐term benefits. However, the recent Child/Adolescent Anxiety Multimodal Study (CAMS) highlights the importance of controlled outcome data at longer‐term follow‐up, based on their findings that over two‐thirds of patients that showed initial treatment gains relapsed over the following four years (Ginsburg 2018).

A major weakness found across studies was the absence of systematic reporting of adverse effects. No adverse effects were reported across all studies; however, in the small number of studies where any reference was made to adverse effects, it was not clear that these were systematically monitored. This issue is not limited to trials of CBT for child anxiety disorders; for example, Duggan 2014 reported that none of the 72 trials of psychological interventions considered in their systematic review mentioned the occurrence of adverse events in their final report, despite the recognition that psychological treatments can introduce a range of negative effects (Barlow 2010).

Given the limited data for treatment as usual, attention control, and alternative treatment comparisons, not all analyses were possible across all comparisons, and many of our subgroup analyses were limited to waitlist/no treatment comparisons.

Quality of the evidence

This review included 87 studies in the quantitative synthesis, with a reasonably large number of study participants (n= 5964). As noted above, we distinguished between baseline and outcome measures using tight criteria in order to ensure we were comparing like with like as far as possible. For example, we distinguished between remission in the primary anxiety disorder and remission of all anxiety disorders, and selected symptom measures that were most consistently reported, prioritising measures of broad anxiety symptoms where possible.

We rated evidence as moderate or low quality across comparisons and outcomes. There was consistent evidence for the benefit of CBT compared to waitlist/no treatment controls, but with moderate to substantial heterogeneity across outcomes, reducing certainty in the results. We assessed the evidence for CBT compared to treatment as usual, attention controls, and alternative treatments mostly as low quality, primarily due to the limited amount of available data.

We noted methodological shortcomings in some studies, although this typically reflected inadequate reporting, rather than evidence of high risk of bias, and was more common amongst older studies. Participant blinding is not possible in research on psychological interventions, which is an obvious potential source of bias; however, the evaluations were typically carried out blind. Nonetheless, we found similar outcomes when we removed studies with weaknesses related to allocation and blinding, supporting the robustness of the results.

Given the increase in number of studies available since previous reviews, we took the opportunity for a stricter approach to our inclusion criteria. We only included studies with robust randomisation procedures, and excluded studies included in other reviews where, for example, children who might be considered at risk were automatically enrolled in the intervention arm (e.g. Nauta 2003). However, we did include studies where assessment points differed between treatment arms, and, in some cases, the period between randomisation and assessment was twice as long in the intervention arm compared to the control arm (e.g. Fujii 2013; Kendall 1994; Kendall 1997), which clearly introduces the potential for substantial bias. The timing of the post‐treatment assessment (the assessment closest to the end of the treatment) also varied across trials.

There was evidence of publication bias (asymmetric funnel plot, significant Egger’s test and contour funnel plot), with a marked absence of non‐significant studies for both CBT versus waitlist/no treatment controls. This is an area that has been richly debated, and remains a significant area of concern when interpreting the results of meta‐analyses, as it is likely that, with the lack of published negative trials, any estimate of the effectiveness of an intervention (CBT in this case) is likely to be inflated. We welcome the call for journals to publish negative trials, and for chief investigators of all registered trials to publish findings, especially negative ones, online within a set time of the trial end date.

Potential biases in the review process

We were able to obtain all the referenced papers; we searched grey literature and reference lists; and made contact with leading researchers in the field. We also included studies that were not in the English language. However, meta‐analyses are limited by the robustness of any search method. The electronic search was thorough and large in scale with broad parameters. Of note, we broadened certain parameters since the last Cochrane Review on this topic, for example removing a restriction on the minimum number of sessions, minimum child age, and including CBT delivered via parents. Nonetheless, there are some limitations to the review. For example, whilst we set a fairly broad definition of ‘cognitive behaviour therapy’, some studies were excluded where the description of the intervention did not include all the core stated components of CBT despite the treatments appearing similar in many ways (e.g. social effectiveness therapy; Beidel 2000; Beidel 2007; behaviour therapy; Rudy 2017; Storch 2019), and it is possible that the brief descriptions of treatments provided in papers may not fully reflect how the treatment is being delivered in practice. We also required some in‐person treatment delivery, and therefore did not include CBT interventions delivered online or purely over the telephone, despite these interventions often being very similar in content. Indeed there is evidence supporting the effectiveness of these remotely delivered approaches (e.g. Lyneham 2006; Pennant 2015).

The analysis of studies with three arms ‐ CBT in two formats versus waiting list controls or active controls ‐ is problematic in that double counting of any group is wrong. We chose a conservative approach to managing this by splitting the control samples into two equal halves between the CBT groups; however, this means that in these cases the odds ratios and confidence intervals appear smaller despite recovery rates that are similar to those found in other studies (e.g. Flannery Schroeder 2000; Thirlwall 2013). We also prioritised ITT analyses and adopted a conservative approach in which drop out in the treatment group was treated as if the participant did not recover, but drop out in the control group was treated as recovery. This conservative approach may have underestimated treatment effects; unsurprisingly, treatment effects were larger in the follow‐up completer analyses. Whilst we took a strict approach to measures that were included (only using specific measures of anxiety symptoms and not, for example, combined measures of emotional and behavioural symptoms), and prioritised broad measures of anxiety symptoms to promote consistency, this may have come at a disadvantage for studies that focused on treating specific disorders, where only a subset of the items on a broad measure might apply to the target of treatment.

Agreements and disagreements with other studies or reviews

Two previous Cochrane Reviews have been conducted on this topic. The first review identified 18 studies with 498 participants and 311 controls, and found a response rate for remission of any measure of anxiety diagnosis using an ITT analysis of 56% for CBT versus 28.2% for controls with the NNTB of 3.0 (95% CI 2.5 to 4.5) (James 2005). The second review identified 41 studies with 1806 participants, with a rate of remission of any measure of anxiety disorder of 58.9% for CBT versus 16% for the waitlist (James 2015). In the current review, we identified a further substantial increase in studies, including 81 studies with 5964 participants. We made a number of changes, most notably distinguishing between remission of the primary anxiety disorder and remission of all anxiety disorders; however, the findings were fairly consistent, with 49.4% of participants being free of their primary anxiety disorder following CBT (versus 17.8% for waitlist controls), and 46.8% being free of all anxiety disorders (versus 19.1% for waitlist controls). It is important to note that studies presenting remission based on primary diagnosis and all anxiety diagnoses only partly overlapped, so these outcomes should not be compared directly. Despite the differences in outcome measures, and our broader inclusion criteria for CBT formats, length, and participant ages, the NNTB has been consistent at NNTB = 3 across all three Cochrane Reviews. These effects are consistent with other meta‐analyses of CBT for anxiety disorders in children and young people (e.g. Cartwright‐Hatton 2004; Compton 2004; Reynolds 2012; Seligman 2011; Silverman 2008; Wang 2017; Warwick 2016).

Our findings are also consistent with previous reviews in terms of the limited available data for comparisons with treatment as usual/alternative treatments (e.g. Wang 2017). Indeed, like others, we have concluded that there is no clear evidence of an advantage of CBT over treatment as usual or treatments that might credibly be used within mental health settings. However, we did find more consistent evidence of an advantage of CBT over attention controls, although the available data are still limited.

However, our conclusions differed from those of a recent network meta‐analysis of psychotherapies for anxiety disorders in children and young people, which concluded that group CBT was more effective that other forms of psychotherapy (including other CBT formats included in this review) (Zhou 2019). Notably, this network meta‐analysis considered outcomes based on symptom questionnaires (prioritising child self‐report). Indeed, when we consider our subgroup analyses, amongst waitlist/no treatment comparisons, we also found stronger effects of group CBT than individual CBT on child‐ and parent‐reported anxiety symptoms; however we did not find significant differences between these subgroups based on our primary outcome (remission of primary anxiety diagnosis) or on our other diagnostic outcome (free of all anxiety diagnoses). Furthermore, studies that differed in terms of treatment delivery format also differed on a number of other characteristics that would likely be associated with outcome (including differences in the study setting, population, and nature of outcome measures). For example, most (5 out of 7) studies that only reported disorder‐specific symptom measures were studies that evaluated treatments delivered in groups, so differences between group versus individual formats may reflect a measurement issue rather than a true difference in effect sizes. As such, we do not feel that the evidence provides clear and consistent support for group CBT having an advantage over other delivery formats.

The same conclusion applies to treatment delivery with children, children and parents, or parents alone. Like previous reviews (e.g. Breinholst 2012; Reynolds 2012; Silverman 2008), we did not find clear evidence of an advantage of any of these delivery formats. Although we found the strongest effects for child‐focused CBT among waitlist/no treatment comparisons, there was considerable variation in findings amongst child‐focused CBT studies, and the treatment delivery subgroups also often differed on other key study and sample characteristics (e.g. child age, setting, comorbidities, treatment duration). Notably, there were relatively few studies that delivered treatment through parents only, and these tended to be more recent studies and included children of younger ages than studies that took a child‐only or child‐and‐parent approach.

Where there were sufficient studies, we generally found fairly consistent findings regardless of whether children in the studies had ASD or not, supporting the evidence from previous meta‐analyses suggesting that CBT can be an effective treatment of anxiety for children with ASD (e.g. Perihan 2019; Ung 2015). Notably, recent reviews have suggested that parental involvement may be particularly important for treating childhood anxiety disorders in the context of ASD, and that longer‐term interventions were associated with larger effects (Perihan 2019). These questions are beyond the scope of the current review; however, it is noteworthy that neither of these sets of findings were replicated in our broader sample, highlighting the need to consider the specific conditions that are needed to optimise CBT approaches amongst particular subgroups of children and young people with anxiety disorders.

Study flow diagram.

Figures and Tables -
Figure 1

Study flow diagram.

Navigate to figure in ReviewOpen in new tab

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figures and Tables -
Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figures and Tables -
Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Figures and Tables -
Figure 4

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.13 Reduction in anxiety symptoms (child report) post‐treatment.

Figures and Tables -
Figure 5

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.13 Reduction in anxiety symptoms (child report) post‐treatment.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.19 Reduction in anxiety symptoms (parent report) post‐treatment.

Figures and Tables -
Figure 6

Forest plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.19 Reduction in anxiety symptoms (parent report) post‐treatment.

Navigate to figure in ReviewOpen in new tab

Funnel plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Figures and Tables -
Figure 7

Funnel plot of comparison: 1 CBT vs waitlist/no treatment, outcome: 1.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Navigate to figure in ReviewOpen in new tab

Funnel plot of comparison: 3 CBT vs attention control, outcome: 3.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Figures and Tables -
Figure 8

Funnel plot of comparison: 3 CBT vs attention control, outcome: 3.1 Remission of primary anxiety diagnosis post‐treatment (ITT).

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 1: Remission of primary anxiety diagnosis post‐treatment (ITT)

Figures and Tables -
Analysis 1.1

Comparison 1: CBT versus waitlist/no treatment, Outcome 1: Remission of primary anxiety diagnosis post‐treatment (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 2: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (individual vs group)

Figures and Tables -
Analysis 1.2

Comparison 1: CBT versus waitlist/no treatment, Outcome 2: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (individual vs group)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 3: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (amount of therapist contact time)

Figures and Tables -
Analysis 1.3

Comparison 1: CBT versus waitlist/no treatment, Outcome 3: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (amount of therapist contact time)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 4: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (age)

Figures and Tables -
Analysis 1.4

Comparison 1: CBT versus waitlist/no treatment, Outcome 4: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (age)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 5: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (ASD vs non‐ASD)

Figures and Tables -
Analysis 1.5

Comparison 1: CBT versus waitlist/no treatment, Outcome 5: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (ASD vs non‐ASD)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 6: Remission of primary anxiety diagnosis post‐treatment (completers)

Figures and Tables -
Analysis 1.6

Comparison 1: CBT versus waitlist/no treatment, Outcome 6: Remission of primary anxiety diagnosis post‐treatment (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 7: Acceptability (number of participants lost to post‐treatment assessment)

Figures and Tables -
Analysis 1.7

Comparison 1: CBT versus waitlist/no treatment, Outcome 7: Acceptability (number of participants lost to post‐treatment assessment)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 8: Remission of all anxiety diagnoses post‐treatment (ITT)

Figures and Tables -
Analysis 1.8

Comparison 1: CBT versus waitlist/no treatment, Outcome 8: Remission of all anxiety diagnoses post‐treatment (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 9: Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (individual vs group)

Figures and Tables -
Analysis 1.9

Comparison 1: CBT versus waitlist/no treatment, Outcome 9: Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (individual vs group)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 10: Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (amount of therapist contact time)

Figures and Tables -
Analysis 1.10

Comparison 1: CBT versus waitlist/no treatment, Outcome 10: Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (amount of therapist contact time)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 11: Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (age)

Figures and Tables -
Analysis 1.11

Comparison 1: CBT versus waitlist/no treatment, Outcome 11: Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (age)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 12: Remission of all anxiety diagnoses post‐treatment (completers)

Figures and Tables -
Analysis 1.12

Comparison 1: CBT versus waitlist/no treatment, Outcome 12: Remission of all anxiety diagnoses post‐treatment (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 13: Reduction in anxiety symptoms (child report) post‐treatment

Figures and Tables -
Analysis 1.13

Comparison 1: CBT versus waitlist/no treatment, Outcome 13: Reduction in anxiety symptoms (child report) post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 14: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures)

Figures and Tables -
Analysis 1.14

Comparison 1: CBT versus waitlist/no treatment, Outcome 14: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 15: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (individual vs group)

Figures and Tables -
Analysis 1.15

Comparison 1: CBT versus waitlist/no treatment, Outcome 15: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (individual vs group)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 16: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (amount of therapist contact time)

Figures and Tables -
Analysis 1.16

Comparison 1: CBT versus waitlist/no treatment, Outcome 16: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (amount of therapist contact time)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 17: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (age)

Figures and Tables -
Analysis 1.17

Comparison 1: CBT versus waitlist/no treatment, Outcome 17: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (age)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 18: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (ASD vs non‐ASD)

Figures and Tables -
Analysis 1.18

Comparison 1: CBT versus waitlist/no treatment, Outcome 18: Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (ASD vs non‐ASD)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 19: Reduction in anxiety symptoms (parent report) post‐treatment

Figures and Tables -
Analysis 1.19

Comparison 1: CBT versus waitlist/no treatment, Outcome 19: Reduction in anxiety symptoms (parent report) post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 20: Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only)

Figures and Tables -
Analysis 1.20

Comparison 1: CBT versus waitlist/no treatment, Outcome 20: Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 21: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (individual vs group)

Figures and Tables -
Analysis 1.21

Comparison 1: CBT versus waitlist/no treatment, Outcome 21: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (individual vs group)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 22: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (amount of therapist contact time)

Figures and Tables -
Analysis 1.22

Comparison 1: CBT versus waitlist/no treatment, Outcome 22: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (amount of therapist contact time)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 23: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (age)

Figures and Tables -
Analysis 1.23

Comparison 1: CBT versus waitlist/no treatment, Outcome 23: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (age)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 24: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (ASD vs non‐ASD)

Figures and Tables -
Analysis 1.24

Comparison 1: CBT versus waitlist/no treatment, Outcome 24: Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (ASD vs non‐ASD)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 25: Reduction in depressive symptoms post‐treatment

Figures and Tables -
Analysis 1.25

Comparison 1: CBT versus waitlist/no treatment, Outcome 25: Reduction in depressive symptoms post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 26: Improvement in global functioning post‐treatment

Figures and Tables -
Analysis 1.26

Comparison 1: CBT versus waitlist/no treatment, Outcome 26: Improvement in global functioning post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 27: Remission of primary anxiety diagnosis at follow‐up (ITT)

Figures and Tables -
Analysis 1.27

Comparison 1: CBT versus waitlist/no treatment, Outcome 27: Remission of primary anxiety diagnosis at follow‐up (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 28: Remission of primary anxiety diagnosis at follow‐up (completers)

Figures and Tables -
Analysis 1.28

Comparison 1: CBT versus waitlist/no treatment, Outcome 28: Remission of primary anxiety diagnosis at follow‐up (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 29: Remission of all anxiety diagnoses at follow‐up (ITT)

Figures and Tables -
Analysis 1.29

Comparison 1: CBT versus waitlist/no treatment, Outcome 29: Remission of all anxiety diagnoses at follow‐up (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 30: Remission of all anxiety diagnoses at follow‐up (completers)

Figures and Tables -
Analysis 1.30

Comparison 1: CBT versus waitlist/no treatment, Outcome 30: Remission of all anxiety diagnoses at follow‐up (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 31: Reduction in anxiety symptoms (child report) at follow‐up

Figures and Tables -
Analysis 1.31

Comparison 1: CBT versus waitlist/no treatment, Outcome 31: Reduction in anxiety symptoms (child report) at follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 32: Sensitivity analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (only studies where sequence generation is clear)

Figures and Tables -
Analysis 1.32

Comparison 1: CBT versus waitlist/no treatment, Outcome 32: Sensitivity analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (only studies where sequence generation is clear)

Navigate to figure in ReviewOpen in new tab

Comparison 1: CBT versus waitlist/no treatment, Outcome 33: Sensitivity analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (only studies where blind assessors used)

Figures and Tables -
Analysis 1.33

Comparison 1: CBT versus waitlist/no treatment, Outcome 33: Sensitivity analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (only studies where blind assessors used)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 1: Remission of primary anxiety diagnosis post‐treatment (ITT)

Figures and Tables -
Analysis 2.1

Comparison 2: CBT versus treatment as usual, Outcome 1: Remission of primary anxiety diagnosis post‐treatment (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 2: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (ASD vs non‐ASD)

Figures and Tables -
Analysis 2.2

Comparison 2: CBT versus treatment as usual, Outcome 2: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (ASD vs non‐ASD)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 3: Remission of primary anxiety diagnosis post‐treatment (completers)

Figures and Tables -
Analysis 2.3

Comparison 2: CBT versus treatment as usual, Outcome 3: Remission of primary anxiety diagnosis post‐treatment (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 4: Acceptability (number of participants lost to post‐treatment assessment)

Figures and Tables -
Analysis 2.4

Comparison 2: CBT versus treatment as usual, Outcome 4: Acceptability (number of participants lost to post‐treatment assessment)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 5: Remission of all anxiety diagnoses post‐treatment (ITT)

Figures and Tables -
Analysis 2.5

Comparison 2: CBT versus treatment as usual, Outcome 5: Remission of all anxiety diagnoses post‐treatment (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 6: Remission of all anxiety diagnoses post‐treatment (completers)

Figures and Tables -
Analysis 2.6

Comparison 2: CBT versus treatment as usual, Outcome 6: Remission of all anxiety diagnoses post‐treatment (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 7: Reduction in anxiety symptoms (child report) post‐treatment

Figures and Tables -
Analysis 2.7

Comparison 2: CBT versus treatment as usual, Outcome 7: Reduction in anxiety symptoms (child report) post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 8: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only)

Figures and Tables -
Analysis 2.8

Comparison 2: CBT versus treatment as usual, Outcome 8: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only)

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 9: Reduction in anxiety symptoms (parent report) post‐treatment

Figures and Tables -
Analysis 2.9

Comparison 2: CBT versus treatment as usual, Outcome 9: Reduction in anxiety symptoms (parent report) post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 2: CBT versus treatment as usual, Outcome 10: Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only)

Figures and Tables -
Analysis 2.10

Comparison 2: CBT versus treatment as usual, Outcome 10: Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 1: Remission of primary anxiety diagnosis post‐treatment (ITT)

Figures and Tables -
Analysis 3.1

Comparison 3: CBT versus attention control, Outcome 1: Remission of primary anxiety diagnosis post‐treatment (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 2: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (individual vs group)

Figures and Tables -
Analysis 3.2

Comparison 3: CBT versus attention control, Outcome 2: Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (individual vs group)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 3: Remission of primary anxiety diagnosis post‐treatment (completers)

Figures and Tables -
Analysis 3.3

Comparison 3: CBT versus attention control, Outcome 3: Remission of primary anxiety diagnosis post‐treatment (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 4: Acceptability (number of participants lost to post‐treatment assessment)

Figures and Tables -
Analysis 3.4

Comparison 3: CBT versus attention control, Outcome 4: Acceptability (number of participants lost to post‐treatment assessment)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 5: Remission of all anxiety diagnoses post‐treatment (ITT)

Figures and Tables -
Analysis 3.5

Comparison 3: CBT versus attention control, Outcome 5: Remission of all anxiety diagnoses post‐treatment (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 6: Remission of all anxiety diagnoses post‐treatment (completers)

Figures and Tables -
Analysis 3.6

Comparison 3: CBT versus attention control, Outcome 6: Remission of all anxiety diagnoses post‐treatment (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 7: Reduction in anxiety symptoms (child report) post‐treatment

Figures and Tables -
Analysis 3.7

Comparison 3: CBT versus attention control, Outcome 7: Reduction in anxiety symptoms (child report) post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 8: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only)

Figures and Tables -
Analysis 3.8

Comparison 3: CBT versus attention control, Outcome 8: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 9: Reduction in anxiety symptoms (parent report) post‐treatment

Figures and Tables -
Analysis 3.9

Comparison 3: CBT versus attention control, Outcome 9: Reduction in anxiety symptoms (parent report) post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 10: Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only)

Figures and Tables -
Analysis 3.10

Comparison 3: CBT versus attention control, Outcome 10: Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 11: Reduction in depressive symptoms post‐treatment

Figures and Tables -
Analysis 3.11

Comparison 3: CBT versus attention control, Outcome 11: Reduction in depressive symptoms post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 12: Remission of primary anxiety diagnosis at follow‐up (ITT)

Figures and Tables -
Analysis 3.12

Comparison 3: CBT versus attention control, Outcome 12: Remission of primary anxiety diagnosis at follow‐up (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 13: Remission of primary anxiety diagnosis at follow‐up (completers)

Figures and Tables -
Analysis 3.13

Comparison 3: CBT versus attention control, Outcome 13: Remission of primary anxiety diagnosis at follow‐up (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 14: Remission of all anxiety diagnoses at follow‐up (ITT)

Figures and Tables -
Analysis 3.14

Comparison 3: CBT versus attention control, Outcome 14: Remission of all anxiety diagnoses at follow‐up (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 15: Remission of all anxiety diagnoses at follow‐up (completers)

Figures and Tables -
Analysis 3.15

Comparison 3: CBT versus attention control, Outcome 15: Remission of all anxiety diagnoses at follow‐up (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 16: Reduction in anxiety symptoms (child report) at follow‐up

Figures and Tables -
Analysis 3.16

Comparison 3: CBT versus attention control, Outcome 16: Reduction in anxiety symptoms (child report) at follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 3: CBT versus attention control, Outcome 17: Reduction in anxiety symptoms (parent report) at follow‐up

Figures and Tables -
Analysis 3.17

Comparison 3: CBT versus attention control, Outcome 17: Reduction in anxiety symptoms (parent report) at follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 4: CBT versus alternative treatment, Outcome 1: Acceptability (number of participants lost to post‐treatment assessment)

Figures and Tables -
Analysis 4.1

Comparison 4: CBT versus alternative treatment, Outcome 1: Acceptability (number of participants lost to post‐treatment assessment)

Navigate to figure in ReviewOpen in new tab

Comparison 4: CBT versus alternative treatment, Outcome 2: Remission of all anxiety diagnoses post‐treatment (ITT)

Figures and Tables -
Analysis 4.2

Comparison 4: CBT versus alternative treatment, Outcome 2: Remission of all anxiety diagnoses post‐treatment (ITT)

Navigate to figure in ReviewOpen in new tab

Comparison 4: CBT versus alternative treatment, Outcome 3: Remission of all anxiety diagnoses post‐treatment (completers)

Figures and Tables -
Analysis 4.3

Comparison 4: CBT versus alternative treatment, Outcome 3: Remission of all anxiety diagnoses post‐treatment (completers)

Navigate to figure in ReviewOpen in new tab

Comparison 4: CBT versus alternative treatment, Outcome 4: Reduction in anxiety symptoms (child report) post‐treatment

Figures and Tables -
Analysis 4.4

Comparison 4: CBT versus alternative treatment, Outcome 4: Reduction in anxiety symptoms (child report) post‐treatment

Navigate to figure in ReviewOpen in new tab

Comparison 4: CBT versus alternative treatment, Outcome 5: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only)

Figures and Tables -
Analysis 4.5

Comparison 4: CBT versus alternative treatment, Outcome 5: Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only)

Navigate to figure in ReviewOpen in new tab

Comparison 4: CBT versus alternative treatment, Outcome 6: Reduction in anxiety symptoms (parent report) post‐treatment

Figures and Tables -
Analysis 4.6

Comparison 4: CBT versus alternative treatment, Outcome 6: Reduction in anxiety symptoms (parent report) post‐treatment

Navigate to figure in ReviewOpen in new tab
Summary of findings 1. CBT compared with waitlist for children and adolescents with anxiety disorders

CBT compared with waitlist for children and adolescents with anxiety disorders

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: waitlist/no treatment

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Waitlist1

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

178 per 1000

541 per 1000
(458 to 622)

OR 5.45 (3.90 to 7.60)

2697
(39 studies)

⊕⊕⊕⊝
moderate2

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 8.57, df = 2, P = 0.01, I² = 76.7%)

  • No difference in outcomes for individual versus group formats (Chi² = 0.90, df = 1, P = 0.34; I² = 0%)

  • No difference in outcomes for interventions with different amounts of therapist contact (Chi² = 0.52, df = 2, P = 0.77; I² = 0%)

  • No difference in outcomes for participants with and without ASD (Chi² = 2.25, df = 1, P = 0.13; I² = 55.5%)

Acceptability (number of participants lost to post‐treatment assessment)

104 per 1000

112 per 1000
(90 to 141)

OR 1.09 (0.85 to 1.41)

3158
(45 studies)

⊕⊕⊝⊝
low3

Remission of all anxiety diagnoses post‐treatment (ITT)

191 per 1000

512 per 1000
(406 to 616)

OR 4.43 (2.89 to 6.78)

2075

(28 studies)

⊕⊕⊕⊝
moderate2

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 8.14, df = 2, P = 0.02, I² = 75.4%)

  • No difference in outcomes for individual versus group formats (Chi² = 0.35, df = 1, P = 0.56; I² = 0%)

  • Difference in outcomes for interventions with different amounts of therapist contact (Chi² = 10.75, df = 2, P= 0.005; I² = 81.4%)

  • Insufficient studies for subgroup analyses examining outcomes for participants with and without ASD

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.67standard deviations lower (0.88 to 0.47 lower).

Moderate effect size

2831
(45 studies)

⊕⊕⊝⊝
low4

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 14.67,df = 2, P < 0.001, I² = 86.4%)

  • Difference in outcomes for individual versus group formats (Chi² =6.47, df = 1, P = 0.01; I² = 84.5%)

  • No difference in outcomes for interventions with different amounts of therapist contact (Chi² = 3.33, df = 2, P = 0.19; I² = 39.9%)

  • No difference in outcomes for participants with and without ASD (Chi² = 0.02, df = 1, P = 0.88; I² = 0%)

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (parent report) in the CBT groups was 0.70standard deviations lower (0.90 to 0.51 lower).

Moderate effect size

2137
(35 studies)

⊕⊕⊝⊝
low4

Subgroup analyses:

  • No difference in outcomes for different delivery formats (child‐focused, child and parent, parent only) (Chi² = 3.43, df = 2, P = 0.18, I² = 41.8%)

  • Difference in outcomes for individual versus group formats (Chi² = 6.79,df = 1, P = 0.009, I² = 85.3%)

  • No difference in outcomes for interventions with different amounts of therapist contact (Chi² = 3.77, df = 2, P = 0.15; I² = 46.9%)

  • No difference in outcomes for participants with and without ASD (Chi² = 1.42, df = 1, P = 0.23; I² = 29.8%)

Reduction in depressive symptoms post‐treatment

The mean depressive symptoms in the CBT groups was 0.34 standard deviations lower (0.51 to 0.17 lower).

Small effect size

1157

(17 studies)

⊕⊕⊕⊝
moderate2

Improvement in global functioning post‐treatment

The mean global functioning in the CBT groups was 1.03 standard deviations higher (0.68 to 1.38 higher).

Large effect size

557

(11 studies)

⊕⊕⊝⊝
low5

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and waitlist/no treatment groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of waitlist groups.
2Downgraded one level due to moderate heterogeneity (inconsistency).
3Downgraded two levels due large variation in treatment effects across studies (inconsistency) and wide confidence intervals (imprecision).
4Downgraded two levels due to substantial heterogeneity (inconsistency).
5Downgraded two levels due to substantial heterogeneity (inconsistency) and assessed and reported in small number of eligible studies (study limitations).

Figures and Tables -
Summary of findings 1. CBT compared with waitlist for children and adolescents with anxiety disorders
Navigate to table in Review
Summary of findings 2. CBT compared with treatment as usual for anxiety disorders in children and adolescents

CBT compared with treatment as usual for anxiety disorders in children and adolescents

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: treatment as usual

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk1

Corresponding risk

Treatment as usual

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

408 per 1000

687 per 1000
(383 to 886)

OR 3.19 (0.90 to 11.29)

487

(8 studies)

⊕⊕⊝⊝
low2

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused versus child and parent) (Chi² = 10.90, df = 1, P < 0.001, I² = 90.8%)

  • Insufficient studies for subgroup analyses examining differences in outcomes for individual versus group formats and interventions with varyingamount of therapist contact time

  • Difference in outcomes for participants with and without ASD (Chi² = 5.71, df = 1, P = 0.02; I² = 82.5%)

Acceptability (number of participants lost to post‐treatment assessment)

93 per 1000

124 per 1000
(70 to 209)

OR 1.37 (0.73 to 2.56)

441

(8 studies)

⊕⊕⊝⊝
low3

Remission of all anxiety diagnoses post‐treatment (ITT)

414 per 1000

660 per 1000
(451 to 820)

OR 2.74 (1.16 to 6.46)

203

(5 studies)

⊕⊕⊝⊝
low3

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.15standard deviations lower (0.78 lower to 0.48 higher).

Cross 0

214

(6 studies)

⊕⊕⊝⊝
low2

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (parent report) in the CBT groups was 0.32standard deviations lower (0.70 lower to 0.06 higher).

Cross 0

228

(7 studies)

⊕⊕⊝⊝
low2

Reduction in depressive symptoms post‐treatment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Improvement in global functioning post‐treatment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and Treatment as Usual groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of treatment as usual groups.
2Downgraded two levels due to at least moderate heterogeneity (inconsistency) and wide confidence intervals and small number of events or participants (imprecision).
3Downgraded two levels due to large variation in treatment effects across studies (inconsistency) and wide confidence intervals and small number of events (imprecision).

Figures and Tables -
Summary of findings 2. CBT compared with treatment as usual for anxiety disorders in children and adolescents
Navigate to table in Review
Summary of findings 3. CBT compared with attention control for anxiety disorders in children and adolescents

CBT compared with attention control for anxiety disorders in children and adolescents

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: attention control

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Attention control1

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

293 per 1000

486 per 1000
(355 to 617)

OR 2.28 (1.33 to 3.89)

822
(10 studies)

⊕⊕⊝⊝
low2,

Subgroup analyses:

  • Difference in outcomes for different delivery formats (child‐focused versus child and parent) (Chi² = 7.65, df = 1, P = 0.006; I² = 86.9%)

  • No difference in outcomes for individual versus group formats (Chi² = 1.22, df = 1, P = 0.27; I² = 17.7%)

  • Insufficient studies for subgroup analyses examining outcomes for interventions with varying amounts of therapist contact, and participants with and without ASD

Acceptability (number of participants lost to post‐treatment assessment)

201 per 1000

201 per 1000
(146 to 272)

OR 1.00 (0.68 to 1.49)

797
(12 studies)

⊕⊕⊝⊝
low3

Remission of all anxiety diagnoses post‐treatment (ITT)

185 per 1000

385 per 1000
(217 to 584)

OR 2.75 (1.22 to 6.17)

378
(5 studies)

⊕⊕⊝⊝
low2

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.31standard deviations lower (0.51 to 0.11 lower).

Small effect size

978
(15 studies)

⊕⊕⊕⊝
moderate4

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (parent report) in the CBT groups was 0.25standard deviations lower (0.61 lower to 0.11 higher).

Cross 0

638
(8 studies)

⊕⊕⊝⊝
low5

Reduction in depressive symptoms post‐treatment

The mean depressive symptoms in the CBT groups was 0.18standard deviations lower (0.45 lower to 0.09 higher).

Cross 0

613
(10 studies)

⊕⊕⊝⊝
low5

Improvement in global functioning post‐treatment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and attention control groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of attention control groups.
2Downgraded two levels due to at least moderate heterogeneity and large variation in treatment effects (inconsistency) and small number of events (imprecision).
3Downgraded two levels due to large variation in treatment effects across studies (inconsistency) and wide confidence intervals and small number of events (imprecision).
4Downgraded one level due to moderate heterogeneity (inconsistency).
5Downgraded two levels due to at least moderate heterogeneity (inconsistency) and wide confidence intervals (imprecision).

Figures and Tables -
Summary of findings 3. CBT compared with attention control for anxiety disorders in children and adolescents
Navigate to table in Review
Summary of findings 4. CBT compared with alternative treatment for anxiety disorders in children and adolescents

CBT compared with alternative treatment for anxiety disorders in children and adolescents

Patient or population: children and adolescents with anxiety disorders

Settings: outpatient clinics/schools

Intervention: CBT

Comparison: alternative treatment

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Alternative treatment1

CBT

Remission of primary anxiety diagnosis post‐treatment (ITT)

See comment

See comment

Not estimable

‐

See comment

Insufficient evidence to estimate effect or conduct subgroup analyses

Acceptability (number of participants lost to post‐treatment assessment)

115 per 1000

171 per 1000
(74 to 350)

OR 1.58 (0.61 to 4.13)

515
(7 studies)

⊕⊕⊝⊝

low2

Remission of all anxiety diagnoses post‐treatment (ITT)

607 per 1000

579 per 1000
(351 to 775)

OR 0.89 (0.35 to 2.23)

401
(4 studies)

⊕⊕⊝⊝

low2

Reduction in anxiety symptoms (child report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.09standard deviations lower (0.40 lower to 0.21 higher).

Cross 0

399
(6 studies)

⊕⊕⊝⊝
low3

Reduction in anxiety symptoms (parent report) post‐treatment

The mean anxiety symptoms (child report) in the CBT groups was 0.13standard deviations lower (0.33 lower to 0.06 higher).

Cross 0

423
(6 studies)

⊕⊕⊝⊝
low4

Reduction in depressive symptoms post‐treatment

Not estimable

Not estimable

‐

See comment

Insufficient evidence to estimate effect

Improvement in global functioning post‐treatment

Not estimable

Not estimable

‐

See comment

No evidence available to estimate effect

Adverse events (randomisation to post‐treatment)

See comment

See comment

Not estimable

‐

See comment

No study reported adverse events in both CBT and Treatment as Usual groups

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CBT: cognitive behavioural therapy; CI: confidence interval; ITT: intention‐to‐treat; OR: odds ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1Control group risk estimates come from pooled estimates of alternative treatment groups.
2Downgraded two levels due to at least moderate heterogeneity (inconsistency) and wide confidence intervals and small number of events (imprecision).
3Downgraded two levels due to moderate heterogeneity (inconsistency) and wide confidence intervals (imprecision).
4Downgraded two levels due to large variation in treatment effects across studies (inconsistency) and wide confidence intervals (imprecision).

Figures and Tables -
Summary of findings 4. CBT compared with alternative treatment for anxiety disorders in children and adolescents
Navigate to table in Review
Table 1. Summary of characteristics of included studies

Study

Comparison 1

Comparison 2

Comparison 3

Broad vs specific anxiety measure

Delivery mode

Delivery mode

ASD vs not ASD

Therapist contact time (hours)

Age

Setting

Disorder‐specific

Other comorbid conditions (required)

Afshari 2014

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

10 to 20

mix

clinic

Arendt 2016

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

20+

mix

clinic (university)

Barrett 1996

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and child+parent

Individual

Not ASD

10 to 20

mix

clinic (university)

Barrett 1998

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Child focused and child+parent

Group

Not ASD

20+

mix

clinic (university)

Barrington 2005

CBT vs TAU

Broad

Child+parent

Individual

Not ASD

10 to 20

mix

clinic (CAMHS)

Berge 2017

CBT vs Wait list / no treatment

Specific

Child focused

Individual

Not ASD

less than 10

mix

clinic ‐dental

Specific Phobia

Cartwright Hatton 2011

CBT vs Wait list / no treatment

Parent only

Group

Not ASD

20+

≤ 12

clinic‐university hospital

Chalfant 2007

CBT vs Wait list / no treatment

Broad

Child+parent

Group

ASD

20+

mix

clinic

Cheung 2016

CBT vs Wait list / no treatment

CBT vs Alternative treatment

Broad

Child focused

Individual

Not ASD

10 to 20

≤ 12

clinic‐CAMHS

Chiu 2013

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

≤ 12

school

Cobham 2017

CBT vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

less than 10

mix

clinic (university)

Cornacchio 2019

CBT vs Wait list / no treatment

Child+parent

Group

Not ASD

20+

≤ 12

clinic

Selective mutism

Creswell 2017

CBT vs Alertnative treatment

Broad

Parent only

Individual

Not ASD

less than 10

≤ 12

clinic (CAMHS)

Dadds 1997

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

10 to 20

mix

school

Flannery Schroeder 2000

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused

Individual and group

Not ASD

10 to 20 and 20+

mix

clinic (university)

Fujii 2013

CBT vs TAU

Child+ parent

Individual

ASD

20+

≤ 12

clinic

Gallagher 2004

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

less than 10

≤ 12

clinic

Social anxiety disorder

Ginsburg 2002

CBT vs Attention control

Broad

Child focused

Group

Not ASD

less than 10

≥ 12

school

Ginsburg 2012

CBT vs TAU

Broad

Child focused

Individual

Not ASD

less than 10

mix

school

Ginsburg 2019

CBT vs TAU

Child focused

Individual

Not ASD

less than 10

mix

school

Hancock 2018

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child +parent

Group

Not ASD

10 to 20

mix

clinic (university)

Herbert 2009

CBT vs Attention control

CBT vs Attention control

Specific

Child focused

Group and Individual

Not ASD

20+ and 10 to 20

≥ 12

clinic (university)

Social anxiety disorder

Hirshfeld Becker 2010

CBT vs Wait list / no treatment

Child + parent

Individual

Not ASD

20+

≤ 12

clinic (university)

Holmes 2014

CBT vs Wait list / no treatment

Broad

Child + parent

Group

Not ASD

10 to 20

≤ 12

clinic (university)

GAD

Hudson 2009

CBT vs Attention control

Broad

Child +parent

Group

Not ASD

20+

mix

clinic (university)

Ingul 2003

CBT vs Attention control

CBT vs Attention control

Broad

Child focused

Individual and Group

Not ASD

10 to 20

≥ 12

schools

Social anxiety disorder

Ishikawa 2019

CBT vs Wait list / no treatment

Broad

Child +parent

Individual

Not ASD

less than 10

mix

clinic

Kendall 1994

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

Kendall 1997

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

Kendall 2008

CBT vs Attention control

CBT vs Attention control

Broad

Child focused and child+parent

Individual

Not ASD

10 to 20

mix

clinic (university)

Kennedy 2009

CBT vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Khanna 2010

CBT vs Attention control

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

Kidd 2018

CBT vs Wait list / no treatment

Broad

Child + parent

Group

ASD

20+

≥ 12

clinic (university)

Last 1998

CBT vs Attention control

Broad

Child and parent

Individual

Not ASD

10 to 20

mix

clinic (university)

School refusal

Lau 2010

CBT vs Wait list / no treatment

Broad

Child focused

Group

Not ASD

10 to 20

≤ 12

clinic

Lau 2017

CBT vs Wait list / no treatment

Broad

Child +parent

Group

Not ASD

10 to 20

≤ 12

child centres and preschools

Lebowitz 2019

CBT vs Alternative treatment

Broad

Child focused

Individual

Not ASD

10 to 20

≤ 12

clinic

Leutgeb 2012

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

less than 10

mix

university

Specific Phobia

Masia Warner 2005

CBT vs Wait list / no treatment

Specific

Child focused

Group focused

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

Masia Warner 2007

CBT vs Attention control

Specific

Child focused

Group focused

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

Masia Warner 2011

CBT vs Wait list / no treatment

Child focused

Individual

Not ASD

10 to 20

mix

primary care/special clinic

functional physical complaints

Masia Warner 2016

CBT vs Attention control

CBT vs Attention control

Specific

Child focused

Group focused

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

McConachie 2014

CBT vs Wait list / no treatment

Broad

Child + parent

Group

ASD

20+

mix

clinic (university)

McNally Keehn 2013

CBT vs Wait list / no treatment

Broad

Child focused

Individual

ASD

20+

mix

clinic (university)

Melfsen 2011

CBT vs Wait list / no treatment

Specific

Child and parent

Individual

Not ASD

20+

mix

clinic

Social anxiety disorder

Muris 2002

CBT vs Attention control

Broad

Child focused

Group

Not ASD

less than 10

≤ 12

school

Murphy 2017

CBT vs Alternative treatment

Broad

Child focused

Individual focused

ASD

10 to 20

≥ 12

CAMHS

O'Brien 2007

CBT vs TAU

Child and parent

Group

Not ASD

10 to 20

mix

clinic

Olivares 2005

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

≥ 12

school

Social anxiety disorder

Olivares 2014

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

≥ 12

school

Social anxiety disorder

Olivares 2019

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Child focused

Group

Not ASD

10 to 20

≥ 12

school

Social anxiety disorder

Ollendick 2009

CBT vs Wait list / no treatment

CBT vs Attention control

Broad

Child focused

Individual

Not ASD

less than 10

mix

clinic

Specific Phobia

Ost 2001

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and child+parent

Individual

Not ASD

less than 10

mix

referrals from clinic and schools

Specific Phobia

Perrin 2019

CBT vs Wait list / no treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

CAMHS

GAD

Pincus 2010

CBT vs Attention control

Broad

Child focused

Individual

Not ASD

less than 10

≥ 12

clinic (university)

Panic Disorder

Rapee 2005

CBt vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

less than 10

≤ 12

preschools

Rapee 2006

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Reaven 2012

CBT vs TAU

Child+parent

Group

ASD

10 to 20

mix

clinic (university)

Reigada 2015

CBT vs Alternative treatment

Specific

Child focused

Individual

Not ASD

10 to 20

mix

clinic (university)

IBD

Rosa‐Alcazar 2009

Cbt vs Attention control

CBT vs Attention control

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

≥ 12

School

Social anxiety disorder

Salari 2018

CBT vs Wait list / no treatment

Broad

Parent only

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Salum 2018

CBT vs Attention control

CBT vs Alternative treatment

Broad

Child+parent

Group

Not ASD

10 to 20

≤ 12

clinic

Sanchez Garcia 2009

CBT vs Wait list / no treatment

Specific

Child focused

Group

Not ASD

10 to 20

mix

School

Social anxiety disorder

Santucci 2013

CBT vs Wait list / no treatment

Broad

Child +parent

Group

Not ASD

20+

≤ 12

clinic (university)

Separation anxiety disorder

Schneider 2011

CBT vs Wait list / no treatment

Broad

Child +parent

Individual

Not ASD

10 to 20

≤ 12

clinic (university)

Separation anxiety disorder

Sciberras 2018

CBT vs TAU

Broad

Child+parent

Individual

Not ASD

10 to 20

≤ 12

clinic ADHD

ADHD

Shahnavaz 2016

CBT vs TAU

Specific

Child+parent

Individual

Not ASD

10 to 20

mix

dental clinics

Specific Phobia

Sharma 2017

CBT vs TAU

Broad

Child focused

Group

Not ASD

20+

mix

clinic (university)

Headaches

Shortt 2001

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

10 to 20

≤ 12

clinic (university)

Silk 2018

CBT vs Alternative treatment

Broad

Child focused

Individual

Not ASD

10 to 20

mix

Silverman 1999a

CBT vs Attention control

Broad

Child+parent

Individual

Not ASD

10 to 20

mix

clinic (university)

Silverman 1999b

CBT vs Wait list / no treatment

Broad

Child + parent

Group

Not ASD

10 to 20

mix

clinic (university)

Simon 2011

CBt vsWait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and parent only

Group

Not ASD

less than 10 and 10 to 20

mix

School

Smith 2014

CBT vs Wait list / no treatment

Broad

Parent only

Individual

Not ASD

10 to 20

mix

clinic (university)

Southam Gerow 2010

CBT vs TAU

Broad

Child focused

Individual

Not ASD

10 to 20

mix

clinics

Spence 2000

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused and child+parent

Group

Not ASD

20+ and 10 to 20

mix

clinic (university)

Social anxiety disorder

Spence 2006

CBT vs Wait list / no treatment

Broad

Child+parent

Group

Not ASD

10 to 20

mix

clinic (university)

Spence 2011

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

Not ASD

10 to 20

≥ 12

clinic (university)

Storch 2013

CBT vs TAU

Broad

Child+parent

Individual

ASD

20+

≤ 12

clinic (university)

Storch 2015

CBT vs TAU

Broad

Child+parent

Individual

ASD

20+

≥ 12

clinic (university)

Thirlwall 2013

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Parent only

Individual

Not ASD

less than 10

≤ 12

clinic

Villabo 2018

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child focused

Individual and Group

Not ASD

10 to 20

mix

clinic

Waters 2009

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Child+Parent and parent only

Group

Not ASD

10 to 20 and 20 +

≤ 12

clinic (university)

Wergelenad 2014

CBT vs Wait list / no treatment

CBT vs Wait list / no treatment

Broad

Child+parent

Individual and group

Not ASD

10 to 20

mix

clinics‐CAMHS

White 2013

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

ASD

20+

≥ 12

clinic (university)

Wood 2009

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

ASD

20+

mix

clinic (university)

Wood 2015

CBT vs Wait list / no treatment

Broad

Child+parent

Individual

ASD

20+

mix

clinic (university)

ASD: autism spectrum disorders
CBT: cognitive behavioural therapy
GAD: generalised anxiety disorder
TAU: treatment as usual

Figures and Tables -
Table 1. Summary of characteristics of included studies
Navigate to table in Review
Comparison 1. CBT versus waitlist/no treatment

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 Remission of primary anxiety diagnosis post‐treatment (ITT) Show forest plot

39

2697

Odds Ratio (IV, Random, 95% CI)

5.45 [3.90, 7.60]

1.1.1 Child focused

19

1184

Odds Ratio (IV, Random, 95% CI)

10.42 [5.84, 18.58]

1.1.2 Child and parent

19

1142

Odds Ratio (IV, Random, 95% CI)

4.08 [2.72, 6.11]

1.1.3 Parent only

4

371

Odds Ratio (IV, Random, 95% CI)

2.83 [1.12, 7.16]

1.2 Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (individual vs group) Show forest plot

39

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.2.1 Individual focused

17

1165

Odds Ratio (IV, Random, 95% CI)

4.53 [2.55, 8.03]

1.2.2 Group focused

25

1532

Odds Ratio (IV, Random, 95% CI)

6.25 [4.45, 8.78]

1.3 Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (amount of therapist contact time) Show forest plot

39

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.3.1 Therapist contact (< 10 hours)

6

513

Odds Ratio (IV, Random, 95% CI)

6.12 [2.08, 18.00]

1.3.2 Therapist contact (≥10 and < 20 hours)

23

1622

Odds Ratio (IV, Random, 95% CI)

5.85 [3.95, 8.64]

1.3.3 Therapist contact (≥ 20 hours)

13

562

Odds Ratio (IV, Random, 95% CI)

4.44 [2.26, 8.72]

1.4 Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (age) Show forest plot

39

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.4.1 ≤ 12 years

11

789

Odds Ratio (IV, Random, 95% CI)

3.93 [2.20, 7.02]

1.4.2 Mixed age range (< 12 and ≥ 12 years)

21

1447

Odds Ratio (IV, Random, 95% CI)

6.43 [3.90, 10.58]

1.4.3 ≥ 12 years

7

461

Odds Ratio (IV, Random, 95% CI)

5.28 [2.83, 9.86]

1.5 Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (ASD vs non‐ASD) Show forest plot

39

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.5.1 Sample with ASD

4

136

Odds Ratio (IV, Random, 95% CI)

1.79 [0.39, 8.16]

1.5.2 Not sample with ASD

35

2561

Odds Ratio (IV, Random, 95% CI)

5.88 [4.19, 8.24]

1.6 Remission of primary anxiety diagnosis post‐treatment (completers) Show forest plot

38

2406

Odds Ratio (IV, Random, 95% CI)

11.55 [8.41, 15.86]

1.7 Acceptability (number of participants lost to post‐treatment assessment) Show forest plot

45

3158

Odds Ratio (IV, Random, 95% CI)

1.09 [0.85, 1.41]

1.8 Remission of all anxiety diagnoses post‐treatment (ITT) Show forest plot

28

2075

Odds Ratio (IV, Random, 95% CI)

4.43 [2.89, 6.78]

1.8.1 Child focused

11

564

Odds Ratio (IV, Random, 95% CI)

8.52 [2.97, 24.39]

1.8.2 Child and parent

15

916

Odds Ratio (IV, Random, 95% CI)

5.19 [3.26, 8.27]

1.8.3 Parent only

7

595

Odds Ratio (IV, Random, 95% CI)

1.89 [0.98, 3.65]

1.9 Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (individual vs group) Show forest plot

28

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.9.1 Individual focused

9

671

Odds Ratio (IV, Random, 95% CI)

3.72 [1.70, 8.12]

1.9.2 Group focused

20

1404

Odds Ratio (IV, Random, 95% CI)

4.92 [2.95, 8.20]

1.10 Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (amount of therapist contact time) Show forest plot

28

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.10.1 Therapist contact (< 10 hours)

5

461

Odds Ratio (IV, Random, 95% CI)

1.42 [0.68, 2.96]

1.10.2 Therapist contact (≥ 10 and < 20 hours)

17

1123

Odds Ratio (IV, Random, 95% CI)

6.59 [3.62, 12.01]

1.10.3 Therapist contact (≥ 20 hours)

9

491

Odds Ratio (IV, Random, 95% CI)

5.03 [2.55, 9.93]

1.11 Subgroup analysis: remission of all anxiety diagnoses post‐treatment (ITT) (age) Show forest plot

28

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.11.1 ≤ 12 years

10

950

Odds Ratio (IV, Random, 95% CI)

3.59 [1.97, 6.56]

1.11.2 Mixed age range (< 12 years and ≥ 12 years)

15

980

Odds Ratio (IV, Random, 95% CI)

4.87 [2.58, 9.21]

1.11.3 ≥ 12 years

3

145

Odds Ratio (IV, Random, 95% CI)

8.48 [0.79, 91.08]

1.12 Remission of all anxiety diagnoses post‐treatment (completers) Show forest plot

27

1871

Odds Ratio (IV, Random, 95% CI)

9.13 [5.78, 14.41]

1.13 Reduction in anxiety symptoms (child report) post‐treatment Show forest plot

45

2831

Std. Mean Difference (IV, Random, 95% CI)

‐0.67 [‐0.88, ‐0.47]

1.13.1 Child focused

24

1239

Std. Mean Difference (IV, Random, 95% CI)

‐1.04 [‐1.41, ‐0.67]

1.13.2 Child and parent

20

1285

Std. Mean Difference (IV, Random, 95% CI)

‐0.45 [‐0.67, ‐0.23]

1.13.3 Parent only

5

307

Std. Mean Difference (IV, Random, 95% CI)

0.04 [‐0.38, 0.46]

1.14 Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures) Show forest plot

38

2459

Std. Mean Difference (IV, Random, 95% CI)

‐0.41 [‐0.57, ‐0.25]

1.15 Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (individual vs group) Show forest plot

45

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.15.1 Individual focused

21

1203

Std. Mean Difference (IV, Random, 95% CI)

‐0.39 [‐0.64, ‐0.15]

1.15.2 Group focused

27

1628

Std. Mean Difference (IV, Random, 95% CI)

‐0.91 [‐1.22, ‐0.60]

1.16 Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (amount of therapist contact time) Show forest plot

45

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.16.1 Therapist contact (< 10 hours)

9

587

Std. Mean Difference (IV, Random, 95% CI)

‐0.36 [‐0.75, 0.03]

1.16.2 Therapist contact (≥ 10 and < 20 hours)

29

1840

Std. Mean Difference (IV, Random, 95% CI)

‐0.81 [‐1.09, ‐0.52]

1.16.3 Therapist contact (≥ 20 hours)

10

404

Std. Mean Difference (IV, Random, 95% CI)

‐0.62 [‐1.06, ‐0.18]

1.17 Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (age) Show forest plot

45

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.17.1 ≤ 12 years

11

663

Std. Mean Difference (IV, Random, 95% CI)

‐0.23 [‐0.54, 0.08]

1.17.2 Mixed age range (< 12 years and ≥ 12 years)

28

1834

Std. Mean Difference (IV, Random, 95% CI)

‐0.62 [‐0.83, ‐0.41]

1.17.3 ≥ 12 years

6

334

Std. Mean Difference (IV, Random, 95% CI)

‐1.78 [‐3.01, ‐0.56]

1.18 Subgroup analysis: reduction in anxiety symptoms (child report) post‐treatment (ASD vs non‐ASD) Show forest plot

45

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.18.1 Sample with ASD

5

181

Std. Mean Difference (IV, Random, 95% CI)

‐0.61 [‐1.54, 0.33]

1.18.2 Not sample with ASD

40

2650

Std. Mean Difference (IV, Random, 95% CI)

‐0.68 [‐0.90, ‐0.47]

1.19 Reduction in anxiety symptoms (parent report) post‐treatment Show forest plot

35

2137

Std. Mean Difference (IV, Random, 95% CI)

‐0.70 [‐0.90, ‐0.51]

1.19.1 Child focused

13

734

Std. Mean Difference (IV, Random, 95% CI)

‐0.87 [‐1.21, ‐0.53]

1.19.2 Child and parent

17

1031

Std. Mean Difference (IV, Random, 95% CI)

‐0.69 [‐0.98, ‐0.39]

1.19.3 Parent only

5

372

Std. Mean Difference (IV, Random, 95% CI)

‐0.37 [‐0.77, 0.04]

1.20 Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only) Show forest plot

32

1952

Std. Mean Difference (IV, Random, 95% CI)

‐0.59 [‐0.77, ‐0.41]

1.21 Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (individual vs group) Show forest plot

35

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.21.1 Individual focused

17

858

Std. Mean Difference (IV, Random, 95% CI)

‐0.43 [‐0.65, ‐0.21]

1.21.2 Group focused

21

1279

Std. Mean Difference (IV, Random, 95% CI)

‐0.92 [‐1.21, ‐0.62]

1.22 Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (amount of therapist contact time) Show forest plot

35

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.22.1 Therapist contact (< 10 hours)

4

336

Std. Mean Difference (IV, Random, 95% CI)

‐0.36 [‐0.81, 0.10]

1.22.2 Therapist contact (≥ 10 and < 20 hours)

22

1402

Std. Mean Difference (IV, Random, 95% CI)

‐0.66 [‐0.89, ‐0.43]

1.22.3 Therapist contact (≥ 20 hours)

10

399

Std. Mean Difference (IV, Random, 95% CI)

‐1.03 [‐1.55, ‐0.52]

1.23 Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (age) Show forest plot

35

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.23.1 ≤ 12 years

11

750

Std. Mean Difference (IV, Random, 95% CI)

‐0.40 [‐0.63, ‐0.17]

1.23.2 Mixed age range (< 12 years and ≥ 12 years)

18

1057

Std. Mean Difference (IV, Random, 95% CI)

‐0.76 [‐1.04, ‐0.49]

1.23.3 ≥ 12 years

6

330

Std. Mean Difference (IV, Random, 95% CI)

‐1.07 [‐1.78, ‐0.37]

1.24 Subgroup analysis: reduction in anxiety symptoms (parent report) post‐treatment (ASD vs non‐ASD) Show forest plot

35

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.24.1 Sample with ASD

7

244

Std. Mean Difference (IV, Random, 95% CI)

‐1.12 [‐1.91, ‐0.34]

1.24.2 Not sample with ASD

28

1893

Std. Mean Difference (IV, Random, 95% CI)

‐0.63 [‐0.82, ‐0.44]

1.25 Reduction in depressive symptoms post‐treatment Show forest plot

17

1157

Std. Mean Difference (IV, Random, 95% CI)

‐0.34 [‐0.51, ‐0.17]

1.26 Improvement in global functioning post‐treatment Show forest plot

11

557

Std. Mean Difference (IV, Random, 95% CI)

1.03 [0.68, 1.38]

1.27 Remission of primary anxiety diagnosis at follow‐up (ITT) Show forest plot

5

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.27.1 Follow‐up (≤ 6 months)

4

153

Odds Ratio (IV, Random, 95% CI)

10.94 [2.33, 51.41]

1.27.2 Follow‐up (> 6 and ≤ 12 months)

3

166

Odds Ratio (IV, Random, 95% CI)

2.80 [0.24, 33.19]

1.28 Remission of primary anxiety diagnosis at follow‐up (completers) Show forest plot

5

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.28.1 Follow‐up (≤ 6 months)

4

151

Odds Ratio (IV, Random, 95% CI)

11.46 [2.40, 54.67]

1.28.2 Follow‐up (> 6 and ≤ 12 months)

3

161

Odds Ratio (IV, Random, 95% CI)

4.04 [0.77, 21.20]

1.29 Remission of all anxiety diagnoses at follow‐up (ITT) Show forest plot

5

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.29.1 Follow‐up (≤ 6 months)

3

130

Odds Ratio (IV, Random, 95% CI)

8.25 [0.69, 98.10]

1.29.2 Follow‐up (> 6 and ≤ 12 months)

3

166

Odds Ratio (IV, Random, 95% CI)

3.75 [0.25, 56.88]

1.29.3 Follow‐up (> 12 months)

2

131

Odds Ratio (IV, Random, 95% CI)

0.48 [0.17, 1.38]

1.30 Remission of all anxiety diagnoses at follow‐up (completers) Show forest plot

5

Odds Ratio (IV, Random, 95% CI)

Subtotals only

1.30.1 Follow‐up (≤ 6 months)

3

128

Odds Ratio (IV, Random, 95% CI)

8.68 [0.72, 104.46]

1.30.2 Follow‐up (> 6 and ≤ 12 months)

3

161

Odds Ratio (IV, Random, 95% CI)

5.01 [0.70, 35.87]

1.30.3 Follow‐up (> 12 months)

2

114

Odds Ratio (IV, Random, 95% CI)

1.25 [0.36, 4.34]

1.31 Reduction in anxiety symptoms (child report) at follow‐up Show forest plot

5

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

1.31.1 Follow‐up (≤ 6 months)

5

179

Std. Mean Difference (IV, Random, 95% CI)

‐1.92 [‐2.95, ‐0.89]

1.32 Sensitivity analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (only studies where sequence generation is clear) Show forest plot

20

1476

Odds Ratio (IV, Random, 95% CI)

4.84 [3.03, 7.73]

1.33 Sensitivity analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (only studies where blind assessors used) Show forest plot

31

2037

Odds Ratio (IV, Random, 95% CI)

5.54 [3.76, 8.18]
Figures and Tables -
Comparison 1. CBT versus waitlist/no treatment
Navigate to table in Review
Comparison 2. CBT versus treatment as usual

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

2.1 Remission of primary anxiety diagnosis post‐treatment (ITT) Show forest plot

8

487

Odds Ratio (IV, Random, 95% CI)

3.19 [0.90, 11.29]

2.1.1 Child focused

3

315

Odds Ratio (IV, Random, 95% CI)

0.61 [0.18, 2.01]

2.1.2 Child and parent

5

172

Odds Ratio (IV, Random, 95% CI)

8.56 [3.10, 23.66]

2.2 Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (ASD vs non‐ASD) Show forest plot

8

Odds Ratio (IV, Random, 95% CI)

Subtotals only

2.2.1 Sample with ASD

4

142

Odds Ratio (IV, Random, 95% CI)

11.25 [3.11, 40.79]

2.2.2 Not sample with ASD

4

345

Odds Ratio (IV, Random, 95% CI)

1.14 [0.29, 4.47]

2.3 Remission of primary anxiety diagnosis post‐treatment (completers) Show forest plot

8

440

Odds Ratio (IV, Random, 95% CI)

4.87 [1.51, 15.72]

2.4 Acceptability (number of participants lost to post‐treatment assessment) Show forest plot

8

441

Odds Ratio (IV, Random, 95% CI)

1.37 [0.73, 2.56]

2.5 Remission of all anxiety diagnoses post‐treatment (ITT) Show forest plot

5

203

Odds Ratio (IV, Random, 95% CI)

2.74 [1.16, 6.46]

2.6 Remission of all anxiety diagnoses post‐treatment (completers) Show forest plot

5

201

Odds Ratio (IV, Random, 95% CI)

2.78 [1.18, 6.55]

2.7 Reduction in anxiety symptoms (child report) post‐treatment Show forest plot

6

214

Std. Mean Difference (IV, Random, 95% CI)

‐0.15 [‐0.78, 0.48]

2.8 Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only) Show forest plot

5

187

Std. Mean Difference (IV, Random, 95% CI)

0.07 [‐0.54, 0.68]

2.9 Reduction in anxiety symptoms (parent report) post‐treatment Show forest plot

7

228

Std. Mean Difference (IV, Random, 95% CI)

‐0.32 [‐0.70, 0.06]

2.10 Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only) Show forest plot

6

202

Std. Mean Difference (IV, Random, 95% CI)

‐0.16 [‐0.44, 0.11]
Figures and Tables -
Comparison 2. CBT versus treatment as usual
Navigate to table in Review
Comparison 3. CBT versus attention control

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

3.1 Remission of primary anxiety diagnosis post‐treatment (ITT) Show forest plot

10

822

Odds Ratio (IV, Random, 95% CI)

2.28 [1.33, 3.89]

3.1.1 Child focused

7

509

Odds Ratio (IV, Random, 95% CI)

3.58 [1.92, 6.65]

3.1.2 Child and parent

4

313

Odds Ratio (IV, Random, 95% CI)

1.12 [0.65, 1.92]

3.2 Subgroup analysis: remission of primary anxiety diagnosis post‐treatment (ITT) (individual vs group) Show forest plot

10

Odds Ratio (IV, Random, 95% CI)

Subtotals only

3.2.1 Individual focused

5

469

Odds Ratio (IV, Random, 95% CI)

2.04 [1.06, 3.91]

3.2.2 Group focused

5

353

Odds Ratio (IV, Random, 95% CI)

3.10 [1.14, 8.45]

3.3 Remission of primary anxiety diagnosis post‐treatment (completers) Show forest plot

10

752

Odds Ratio (IV, Random, 95% CI)

3.88 [2.52, 5.97]

3.4 Acceptability (number of participants lost to post‐treatment assessment) Show forest plot

12

797

Odds Ratio (IV, Random, 95% CI)

1.00 [0.68, 1.49]

3.5 Remission of all anxiety diagnoses post‐treatment (ITT) Show forest plot

5

378

Odds Ratio (IV, Random, 95% CI)

2.75 [1.22, 6.17]

3.6 Remission of all anxiety diagnoses post‐treatment (completers) Show forest plot

5

359

Odds Ratio (IV, Random, 95% CI)

3.88 [1.58, 9.51]

3.7 Reduction in anxiety symptoms (child report) post‐treatment Show forest plot

15

978

Std. Mean Difference (IV, Random, 95% CI)

‐0.31 [‐0.51, ‐0.11]

3.8 Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only) Show forest plot

11

685

Std. Mean Difference (IV, Random, 95% CI)

‐0.18 [‐0.34, ‐0.02]

3.9 Reduction in anxiety symptoms (parent report) post‐treatment Show forest plot

8

638

Std. Mean Difference (IV, Random, 95% CI)

‐0.25 [‐0.61, 0.11]

3.10 Sensitivity analysis: reduction in anxiety symptoms (parent report) post‐treatment (broad anxiety measures only) Show forest plot

4

345

Std. Mean Difference (IV, Random, 95% CI)

‐0.04 [‐0.26, 0.18]

3.11 Reduction in depressive symptoms post‐treatment Show forest plot

10

613

Std. Mean Difference (IV, Random, 95% CI)

‐0.18 [‐0.45, 0.09]

3.12 Remission of primary anxiety diagnosis at follow‐up (ITT) Show forest plot

4

Odds Ratio (IV, Random, 95% CI)

Subtotals only

3.12.1 Follow‐up (≤ 6 months)

4

341

Odds Ratio (IV, Random, 95% CI)

3.06 [1.22, 7.65]

3.13 Remission of primary anxiety diagnosis at follow‐up (completers) Show forest plot

4

Odds Ratio (IV, Random, 95% CI)

Subtotals only

3.13.1 Follow‐up (≤ 6 months)

4

302

Odds Ratio (IV, Random, 95% CI)

6.52 [2.58, 16.43]

3.14 Remission of all anxiety diagnoses at follow‐up (ITT) Show forest plot

3

Odds Ratio (IV, Random, 95% CI)

Subtotals only

3.14.1 Follow‐up (≤ 6 months)

3

205

Odds Ratio (IV, Random, 95% CI)

5.44 [1.00, 29.60]

3.15 Remission of all anxiety diagnoses at follow‐up (completers) Show forest plot

3

Odds Ratio (IV, Random, 95% CI)

Subtotals only

3.15.1 Follow‐up (≤ 6 months)

3

182

Odds Ratio (IV, Random, 95% CI)

11.19 [1.94, 64.51]

3.16 Reduction in anxiety symptoms (child report) at follow‐up Show forest plot

6

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

3.16.1 Follow‐up (≤ 6 months)

6

434

Std. Mean Difference (IV, Random, 95% CI)

‐0.44 [‐0.79, ‐0.08]

3.17 Reduction in anxiety symptoms (parent report) at follow‐up Show forest plot

6

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

3.17.1 Follow‐up (≤ 6 months)

6

434

Std. Mean Difference (IV, Random, 95% CI)

‐0.63 [‐1.33, 0.08]
Figures and Tables -
Comparison 3. CBT versus attention control
Navigate to table in Review
Comparison 4. CBT versus alternative treatment

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

4.1 Acceptability (number of participants lost to post‐treatment assessment) Show forest plot

7

515

Odds Ratio (IV, Random, 95% CI)

1.58 [0.61, 4.13]

4.2 Remission of all anxiety diagnoses post‐treatment (ITT) Show forest plot

4

401

Odds Ratio (IV, Random, 95% CI)

0.89 [0.35, 2.23]

4.3 Remission of all anxiety diagnoses post‐treatment (completers) Show forest plot

4

348

Odds Ratio (IV, Random, 95% CI)

1.52 [0.70, 3.27]

4.4 Reduction in anxiety symptoms (child report) post‐treatment Show forest plot

6

399

Std. Mean Difference (IV, Random, 95% CI)

‐0.09 [‐0.40, 0.21]

4.5 Sensitivity analysis: reduction in anxiety symptoms (child report) post‐treatment (broad anxiety measures only) Show forest plot

5

378

Std. Mean Difference (IV, Random, 95% CI)

‐0.00 [‐0.21, 0.20]

4.6 Reduction in anxiety symptoms (parent report) post‐treatment Show forest plot

6

423

Std. Mean Difference (IV, Random, 95% CI)

‐0.13 [‐0.33, 0.06]
Figures and Tables -
Comparison 4. CBT versus alternative treatment
Navigate to table in Review