Key Points

• Selective serotonin reuptake inhibitor (SSRIs) improved primary anxiety symptoms (clinician and parent report), function, remission, and clinical response, compared to pill placebo (moderate to high SOE).
• Serotonin–norepinephrine reuptake inhibitors (SNRIs) improved primary anxiety symptoms, compared to pill placebo (only clinician report) (high SOE).
• Tricyclic antidepressants marginally improved clinical response, compared to pill placebo (low SOE).
• Benzodiazepines did not show significant improvement in anxiety symptoms over pill placebo (low SOE).

Discussion

Nineteen RCTs^{7, 22, 49–72} compared medications to pill placebo, including atomoxetine, clonazepam, clomipramine, duloxetine, fluvoxamine, fluoxetine, imipramine, paroxetine, sertraline, and venlafaxine. Overall, 2,498 patients were included with a mean age of 11.6 years old and 54.1 percent male. 13 studies^{49–52, 54, 56, 58, 60–65} (68.4%) included patients without any comorbidity. 6 studies ^{7, 22, 53–55, 57, 66, 67, 69–73} included children with anxiety and comorbidity (attention deficit hyperactivity disorder (ADHD), autism, ODD, obsessive compulsive disorder (OCD) and other internalizing disorders). Details of the included studies can be found in Appendix Table E.1. We were unable to evaluate publication bias due to small number of studies (n<20) included in each comparisons.

As a class, SSRIs improved primary anxiety symptoms (clinician and parent report), function, remission, and clinical response, compared to pill placebo (moderate to high SOE). SNRIs improved primary anxiety symptoms (clinician report, high SOE). TCAs marginally improved clinical response (low SOE), whereas benzodiazepines did not show significant improvement in anxiety symptoms over pill placebo (low SOE). Results of the comparisons of drug classes with pill placebo and associated SOE are presented in Table 5. Results of the individual drugs comparisons with pill placebo are presented in Table 6.

Table 5

Strength of evidence for drug classes versus pill placebo.

Table 6

Strength of evidence for individual drugs versus pill placebo.

Key Points

• Only two RCTs conducted head-to-head comparison.
• Compared to clomipramine, fluoxetine was more effective in improving primary anxiety symptoms (child report) (low SOE).
• No significant difference was found between venlafaxine and atomoxetine on primary anxiety symptoms (child, parent, and clinician reports) (low SOE).

Discussion

One RCT^{54, 62} compared fluoxetine to clomipramine⁵⁴ and another RCT compared venlafaxine to atomoxetine.⁶² Overall, 39 patients were included with age range of 6–17 years old and 52.2% male. These 2 studies^{54, 62} included patients without any comorbidity. Details of the included studies can be found in Appendix Table E.2. We were unable to evaluate publication bias due to small number of studies (n<20) included in each comparisons.

Compared to clomipramine, fluoxetine was more effective in improving primary anxiety symptoms (child report, low SOE); while venlafaxine and atomoxetine were not significantly different in improving primary anxiety symptoms (low SOE). Results of the individual drugs comparisons with drugs are presented in Table 7.

Table 7

Strength of evidence for drugs versus drugs.

Key Points

• Only two RCTs compared CBT to SSRIs.
• CBT reduced primary anxiety symptoms and improved function more than fluoxetine (moderate SOE).
• CBT was more likely to increase remission than sertraline (moderate SOE).

Discussion

One RCT compared CBT to fluoxetine.^{51, 63} Overall, 102 patients were included with a mean age of 11.6 years old and 51.5 percent male. Details of the included study can be found in Appendix Table E.3. CBT was more effective in improving primary anxiety symptoms (clinician report), function, and secondary anxiety measures (moderate SOE). Table 8 includes summary of the results and assessment of SOE.

Table 8

Strength of evidence for drugs versus CBT.

One RCT of 272 patients with a primary diagnosis of social anxiety disorder, generalized anxiety disorder, or social anxiety disorder (mean age: 10.7 years), compared CBT to sertraline.^{7, 67, 69–73} Patients were randomized to receive either 14 sessions of CBT or sertraline (up to 200 mg per day). Details of the included study can be found in Appendix Table E.3. CBT was more likely to increase remission (moderate SOE). There were no other significant differences in other outcomes (low SOE) (Table 8).

Key Points

• Compared to pill placebo, CBT improved secondary anxiety measures (low SOE).
• Compared to waitlisting or no treatment, CBT improved primary anxiety symptoms (clinician, child, and parent report), function, remission, and clinical response (low to moderate SOE).
• Compared to attention control or treatment as usual, CBT reduced primary anxiety symptoms (child report) (moderate SOE).

Discussion

Eighty-four RCTs and 4 non-randomized comparative studies compared CBT to controls. 29 RCTs^{7, 33, 38, 41, 63, 67–72, 75–101} and 1 non-randomized comparative study ¹⁰² compared CBT to attention control/treatment as usual, 60 RCTs ^{33, 34, 36, 39–42, 44, 84, 86, 89, 100, 101, 103–153} and 3 non-randomized comparative study ^154–156 compared CBT versus waitlisting/no treatment, and 3 RCTs compared CBT versus pill placebo.^{7, 51, 63, 67–72} Overall, 6,978 patients were included with a mean age of 11.2 years and 47.9 percent male. 59 studies^{33, 34, 36, 38–42, 44, 51, 63, 75–81, 83, 85–87, 89, 90, 92, 94, 96, 100, 102–105, 107–109, 110, 113, 117, 118, 120, 123–126, 130, 131, 133, 135, 137, 139–145, 147, 154–156} (67.0%) of included patients without any comorbidity. 30 studies^{7, 67–73, 82, 84, 88, 91, 93, 95, 97–99, 101, 106, 111, 112, 114–116, 119, 121, 122, 127–129, 134, 136, 138, 146, 148–150, 152, 153} included children with anxiety and other comorbidities. 8 studies^{39, 42, 44, 89, 98, 99, 104, 113, 133} (9.1%) didn’t provide enough quantitative data and were excluded from meta-analyses. Details of the included studies can be found in Appendix Tables E.4 to E.6. We found indications of potential publication bias when CBT was compared to waitlisting on primary anxiety symptoms (Appendix Figures H.1 to H.3). We were unable to evaluate publication bias due to small number of studies (n<20) included in other comparisons.

Compared to pill placebo, CBT improved secondary anxiety measures (low SOE). Compared to waitlisting or no treatment, CBT improved primary anxiety symptoms (clinician, child, and parent report), function, remission, and clinical response (low to moderate SOE). Compared to attention control or treatment as usual, CBT reduced primary anxiety symptoms (child report, moderate SOE). Table 9 includes summary of the results and assessment of SOE.

Table 9

Strength of evidence for CBT versus pill placebo, waitlisting/no treatment, or attention control/treatment as usual.

Key Points

• Compared to CBT alone, the combination of imipramine and CBT reduced primary anxiety symptoms (child report) and function (moderate SOE).
• The combination of fluoxetine and CBT was found to have lower remission rate compared to CBT alone (low SOE).
• The combination of sertraline and CBT reduced primary anxiety symptoms (clinician report), improved function, and increased clinical response, compared to CBT alone (moderate SOE).
• The combination of sertraline and CBT improved primary anxiety symptoms (clinician report), function, and clinical response (moderate SOE), compared to sertraline alone (moderate SOE).

Discussion

One RCT with 63 patients compared the combination of imipramine and CBT to CBT alone¹⁵⁹. All patients had major depressive disorder and at least one anxiety disorder. The mean age of the included patients was 13.9 years and 90.5 percent were Caucasians. Details of the included study can be found in Appendix Table E.7. Compared with CBT alone, adding imipramine to CBT reduced primary anxiety symptoms (child report) and improved function (moderate SOE).

One RCT of 41 anxious school refusing adolescents compared fluoxetine plus CBT to CBT¹⁶⁰. Details of the included study can be found in Appendix Table E.7. Patients in the CBT and fluoxetine group had lower remission than CBT alone (low SOE).

One RCT of 272 patients compared the combination of CBT and sertraline to CBT, or to sertraline^{7, 67, 69–73}. Patients (7–17 years old; mean age: 10.7; primary diagnosis of social anxiety disorder, generalized anxiety disorder, or social anxiety disorder) were randomized to receive either 14 sessions of CBT or sertraline (up to 200 mg per day). Details of the included study can be found in Appendix Table E.7. Compared to CBT alone, adding sertraline reduced primary anxiety symptoms (clinician report), improved function, and improved clinical response (moderate SOE). The addition of CBT to sertraline (compared to sertraline alone) improved primary anxiety symptoms (clinician report), function, and likelihood of clinical response (moderate SOE) (Table 10).

Table 10

Strength of evidence for CBT combined with drugs.

Key Points

• Treatment effects observed immediately post intervention were larger than those observed after a period of followup.
• Individual-based CBT had statistically significantly more improvement on function than group-based CBT.
• Relaxation and cognitive strategies in CBT were not associated with improvements on primary anxiety symptoms, function, secondary measures, social function, and clinical response; while exposure statistically significantly reduced primary anxiety symptoms (parent report).
• Compared to waitlisting or no treatment, CBT was found to have more improvement on functioning in age group 13–18 than age group 7–12.

Discussion

We were not able to conduct a large number of the planned subgroup analyses, including those based on race/ethnicity, parent education level, family income, disease severity (measured by CGI), treatment sequence, and provider. This was due to studies not providing sufficient stratified data per subgroup variable. The results of the feasible exploratory analyses are reported in Appendix Tables G.1 to G.12 and were summarized as follows:

• Age: when CBT compared to waitlisting or no treatment, we found statistically significantly more improvement in function in age group 13–18 than age group 7–12.
• Comorbidity: when CBT compared to pill placebo, patients without comorbidity had statistically significantly more improvement in secondary anxiety measures than patient with any comorbidity. However, the finding was limited by the fact that CBT delivered to children with comorbidities was different in children without comorbidities. Inference from subgroup analyses evaluating comorbidities is less reliable.
• ADHD: when fluvoxamine compared to pill placebo, we found no statistically significant difference on primary anxiety symptoms (clinician report).
• Autism: When CBT was compared to waitlisting or no treatment, we did not find statistically significant difference in outcomes (primary anxiety symptoms, clinician, child, and parent report), function, or clinical response in patients with autism than patients without autism.
• School refusal: when CBT compared to pill placebo, patients without school refusal were found to have statistically significant better outcome (secondary anxiety measures) than patient with school refusal.
• Diagnosis: when CBT compared to attention control/treatment as usual, patients with social anxiety disorder were found to have more improvement on secondary anxiety measures than patients with panic disorder.
• Treatment settings: when CBT compared to attention control or treatment as usual, we found statistically significantly more improvement in secondary anxiety measures in school settings than mental health clinic.
• Length of follow-up: when CBT compared to waitlisting or no treatment, post intervention response rate was significantly higher than those reported at less than 6-month followup. Post intervention reduction of primary anxiety symptoms (child report) and remission rate were also significantly larger than those reported after more than 6 month followup.
• Exposure sessions in CBT: Compared with non-exposure CBT, exposure sessions statistically significant reduced primary anxiety symptoms (parent report only).
• Cognitive strategies in CBT: CBT with cognitive strategies were found to have statistically significant less improvement in primary anxiety symptoms (parent report) than CBT without cognitive strategies. No other significant differences were found on primary anxiety symptoms (clinician and child report), function, secondary measures, social function, or clinical response.
• Relaxation strategies in CBT: We found no statistically significant differences between CBT with relaxation and CBT without relaxation on primary anxiety symptoms (clinician, child, and parent report), function, secondary measures, social function, and clinical response.
• Individual-based CBT versus group-based CBT: we found that individual-based CBT had statistically significantly more improvement on function than group-based CBT.
• Treatment intensity: we found no statistically significant difference in any outcome based on treatment intensity.