Skip to main content
. 2023 Aug 3;66(1):e63. doi: 10.1192/j.eurpsy.2023.2437

Table 2.

Pooled effects of psychotherapies for adult depression

k g 95% CI I2 95% CI 95% PI NNT
All comparisons (combined) 353 0.71 0.64; 0.79 82 80; 84 −0.48; 1.91 4.26
One ES/study (lowest) 300 0.63 0.55; 0.71 82 80; 84 −0.53; 1.78 4.95
One ES/study (highest) 300 0.72 0.64; 0.80 83 81; 84 −0.49; 1.93 4.2
Outliers removed 248 0.62 0.58; 0.65 23 9; 35 0.39; 0.84 5.04
Influence analysis 343 0.62 0.56; 0.68 74 72; 77 −0.22; 1.45 5.05
Only risk of bias > 3 115 0.45 0.36; 0.54 73 68; 78 −0.3; 1.20 7.31
Three-level model (CHE) 410 0.68 0.61; 0.76 88 −0.51; 1.88 4.49
Publication bias correction
– Trim-and-fill method 457 0.38 0.29; 0.48 88 87; 89 −1.45; 2.21 8.82
– Limit meta-analysis 353 0.25 0.14; 0.36 93 −0.95; 1.45 14.23
– Selection modell 353 0.58 0.47; 0.70 88 −0.88; 2.05 5.38

Abbreviations: CHE, “correlated and hierarchical effects” model; CI, confidence interval; ES, effect size; g, Hedges’ g; I2, level of heterogeneity; k, number of comparisons; NNT, number-needed-to-treat; PI, prediction interval.