Delivery formats of cognitive behavior therapy in adults with eating disorders: a network meta-analysis

Although CBT has been found to be effective in the treatment of eating disorders, it is not clear if there are differences between treatment formats. We conducted a network meta-analysis (NMA) of randomized trials of broadly defined CBT comparing individual, group, guided self-help (GSH) and unguided self-help (USH) with each other or with a control condition. The NMA used a frequentist graph-theoretical approach and included 36 trials (53 comparisons; 3,136 participants). Only one trial was aimed at anorexia nervosa. All formats resulted in large, significant effects when compared to waitlists, with no significant difference between formats (group: g = 1.08, 95% CI: 0.84; 1.31; GSH: g = 0.94, 95% CI: 0.75; 1.13; individual: g = 1.06, 95% CI: 0.77; 1.36; USH: g = 0.62, 95% CI: 0.30; 0.93). No significant difference was found between any format and care-as-usual. Analyses limited to binge eating disorder supported the effects of individual, group and GSH formats, with no significant differences between them. Few trials with low risk of bias were available. CBT for eating disorders can probably be delivered effectively in any format, without significant differences between the formats. These results should be considered with caution because of the non-significant differences when compared to care-as-usual and the considerable risk of bias.