Moderators and mediators of outcome in treatments for anorexia nervosa and bulimia nervosa in adolescents: A systematic review of randomized controlled trials.

OBJECTIVE

To critically appraise papers reporting on moderators and mediators of recommended psychological treatments for anorexia nervosa (AN) and bulimia nervosa (BN) in adolescents.

METHOD: A systematic search of databases was conducted including PsycINFO, Embase, MEDLINE, AMED, CINAHL, and the Cochrane Library. Studies were included where a randomized controlled trial (RCT) compared therapies for AN or BN and reported on moderators or mediators of treatment effect. Twenty-one eligible papers were included, all based on data from eight RCTs.

RESULTS

Family therapies were dominant in the literature. Individual or separated treatment appeared superior for families with more difficult relationships, whereas conjoint family treatment appeared to be superior where good family relationships were reported. Where there was greater eating disorder psychopathology in AN, including eating disorder-related obsessions and compulsions, the response was better to a family approach than to individual therapies. There was some evidence that a family treatment was superior for those engaging in purging behaviors in BN. Measures of family relationships, parental self-efficacy, and early change emerged as possible mediators; however, the quality of evidence was mixed and the findings, in some cases, arguably circular. Moderator and mediator analyses were underpowered in all studies, with multiple, and post-hoc, analyses being run, and a broad range of outcome measures used.

DISCUSSION: This review recommends that emerging findings are explored further in adequately powered trials of the different recommended therapies, with a move toward focusing on effect sizes. A consensus on acceptable definitions of outcome, including remission and recovery, would benefit future research.