Table 1.
Model | No. of studies (pairwise comparisons) | PT | EW+ | EW | NW | τ2 | I2 (%) | Inconsistency (p value) |
---|---|---|---|---|---|---|---|---|
Baseline (all studies in) | 44 (54) | −0.09 (−0.27 to 0.10) | −0.12 (−0.23 to −0.02) | −0.06 (−0.14 to 0.02) | −0.05 (−0.13 to 0.02) | 0 | 0 | 0.46 |
Only end-of-treatment | 34 (36) | −1.03 (−1.49 to −0.56) | −0.90 (−1.19 to −0.62) | −0.40 (−0.81 to 0.01) | −0.36 (−0.81 to 0.09) | 0.17 | 73.8 | 0.005 |
Only long-term | 26 (36) | −0.73 (−1.11 to −0.35) | −0.74 (−1.09 to −0.38) | −0.42 (−0.69 to −0.16) | −0.37 (−0.65 to −0.09) | 0.08 | 71.3 | <0.001 |
Longest available FU (all studies in) | 44 (54) | −0.78 (−1.10 to −0.46) | −0.81 (−1.02 to −0.61) | −0.43 (−0.65 to −0.21) | −0.37 (−0.61 to −0.14) | 0.08 | 67.6 | <0.001 |
Sensitivity analyses (longest available FU): | ||||||||
comparisons EW+ v. WL excluded | 35 (45) | −0.52 (−0.84 to −0.19) | −0.41 (−0.72 to −0.10) | −0.29 (−0.49 to −0.09) | −0.22 (−0.43 to −0.001) | 0.05 | 55.8 | 0.03 |
Studies with imputed data excluded | 40 (48) | −0.82 (−1.19 to −0.45) | −0.85 (−1.08 to −0.61) | −0.51 (−0.79 to −0.24) | −0.44 (−0.74 to −0.14) | 0.12 | 70.5 | <0.001 |
Studies with high indirectness excluded | 35 (45) | −0.60 (−0.95 to −0.24) | −0.74 (−1.02 to −0.46) | −0.33 (−0.54 to −0.12) | −0.25 (−0.48 to −0.03) | 0.06 | 59.8 | 0.002 |
Studies with observer-rated outcomes excluded | 41 (51) | −0.69 (−1.01 to −0.36) | −0.73 (−0.93 to −0.54) | −0.39 (−0.60 to −0.19) | −0.33 (−0.55 to −0.11) | 0.06 | 63.5 | <0.001 |
Studies with experimental PTSD psychotherapies excluded | 42 (50) | −0.82 (−1.21 to −0.42) | −0.82 (−1.03 to −0.61) | −0.43 (−0.65 to −0.20) | −0.35 (−0.59 to −0.11) | 0.08 | 67.8 | <0.001 |
Studies reporting only enhanced PTSD symptoms (no diagnosis) excluded | 19 (23) | −0.99 (−1.49 to −0.49) | −0.95 (−1.30 to −0.60) | −0.60 (−1.17 to −0.03) | −0.26 (−0.89 to 0.37) | 0.16 | 71.7 | 0.02 |
EW, expressive writing; EW+, enhanced writing; FU, follow-up; NW, neutral writing; PT, psychotherapy; WL, waiting list.
SMDs below 0 indicate superiority of a comparator over WL. SMDs above 0 indicate superiority of WL over a comparator. Statistically significant results are printed bold.