Sixteen studies (1,686 participants, range 12 to 360) were included. Eleven studies used an intention-to-treat analysis. Drop-out rates ranged from 0% to 52% in treatment groups and 0% to 56% in comparison groups. There was evidence of publication bias.
Compared to comparison conditions, there was a small but statistically significant reduction in post-traumatic stress symptoms for group treatments (d 0.24, 95% CI 0.09 to 0.39). There was no evidence of statistical heterogeneity. There was a medium and statistically significant reduction in symptoms for group treatments compared to waiting list comparison (d 0.56, 95% CI 0.31 to 0.82; six trials) but a moderator analysis showed no significant differences in symptoms reported when group treatment was compared with active comparisons (d 0.09, 95% CI -0.03 to 0.22; 10 trials).
The type of trauma and type of comparison moderated the treatment effects on post-traumatic stress disorder for the between-group analyses. Studies that included participants who had been sexually abused as children reported smaller effects than those that included mixed or other trauma participants.
Results of within-group analyses were reported.