Twenty-seven placebo-controlled RCTs were included (n=1,496). Eight studies scored 3 on the Jadad scale and the others scored 1 or 2. Few studies provided intention-to-treat data (six reported intention-to-treat continuous measures of anxiety disorders and eight reported intention-to-treat response rates).
There was no statistically significant difference in attrition rates between CBT and placebo (23% versus 22%, p=0.26).
Completer analysis: Compared to placebo, CBT was associated with a statistically significant medium to large effect on anxiety disorder severity compared to placebo (g=0.73, CI: 0.56 to 0.90, p<0.001), a significant increase in treatment response rate (odds ratio 4.06, CI: 2.78 to 5.92, p<0.001) and a small to medium statistically significant effect on depression symptom severity (g=0.45, CI: 0.25 to 0.65, p<0.001).
Intention-to-treat analysis: Similar results were obtained for anxiety severity and treatment response rates, but effect sizes were smaller.
There was no evidence of publication bias (fail-safe N ranged from 183 for depression severity to 829 for anxiety severity).
Pairwise comparisons showed that anxiety disorder severity effect sizes for acute stress disorder were significantly greater (p<0.05) than those for all other disorders apart from obsessive compulsive disorder. There were no significant differences in depression severity effect sizes between different types of disorders.
Anxiety disorder effect sizes were not influenced by any of the moderator factors examined.