Also, all 14 studies showed the post intervention attrition rate. Two studies that each had three branches in the original research were respect ively divided into two comparisons, and they were identi fied as Clarke 2005a and Clarke 2005b in the former, and as Titov 2009a and Titov 2009b in the later, thereby giving a total of 16 comparisons from 14 ori ginal free copy studies for the meta analysis. Primary and secondary outcomes We included 16 comparisons in terms of the reduction in depression symptoms between CCBT and controls following treatment. The pooled SMD was ?0. 48, indicat ing a significant moderate effect. Reduction in depres sion symptoms at long follow up The pooled SMD from the five studies that pro vided long term follow up data was ?0. 05, indicating no significant difference between the groups.
Improvement in function at post treatment The pooled SMD from the twelve compari sons that provided data on function at post Inhibitors,Modulators,Libraries treatment was ?0. 05, indicating no signifi cant difference between two groups. In addition, the pooled relative risk from all trials providing data on attrition at post treatment was 1. 68, indicating a significant difference between the two groups. Subgroup analyses We attempted to analyse the differences in effects between studies where the control was a waiting list and those where it was TAU. The control Inhibitors,Modulators,Libraries was a waitlist in nine comparisons, while it was TAU in seven comparisons. The pooled SMD for waitlist controlled trials was ?0. 63, indicating a moderate effect. By contrast, the pooled SMD for TAU controlled trials was ?0.
23, indicating a small effect. There was a sig nificant difference between two groups. Sensitivity analyses Sensitivity analyses were conducted as show Inhibitors,Modulators,Libraries below. Rating scales except BDI Inhibitors,Modulators,Libraries For the seven comparisons that employed neither BDI I nor II as the primary Inhibitors,Modulators,Libraries outcome measure, the SMD was ?0. 32, indicating a small clinical effect in favour of CCBT. Acceptable attrition rate For the seven comparisons that were reported to have an acceptable dropout rate, the SMD was ?0. 59, showing a moderate clinical effect in favour of CCBT. Imputation techniques For the nine comparisons without a significant differ ence in attrition rate between intervention and control at post treatment, the SMD was ?0. 50, showing a moderate clin ical effect in favour of CCBT.
On the other hand, for the eight comparisons from seven trials with modern imput ation techniques, the SMD was ?0. 34, indicating a small clinical effect sellckchem in favour of CCBT. Publication bias We explored publication bias, using a Funnel plot. The plot can be seen to be asymmetric, indicating a rela tionship between intervention effect and study design. In particular, this asymmetry suggests a publication bias to ward larger effect size in smaller studies, since there was a marked concentration of studies shown on the left side in the lower part of the plot.