A multidimensional meta-analysis of pharmacotherapy for bulimia nervosa: summarizing the range of outcomes in controlled clinical trials.

The empirical literature on pharmacotherapy for bulimia nervosa reveals mixed results. We examined the results of controlled clinical trials of pharmacotherapies for bulimia published from 1980 to 1999. To do this, we employed a multidimensional meta-analysis, a method for aggregating a range of clinically meaningful indicators of outcome (including but not limited to effect-size estimates) across studies. We found that pharmacotherapy for bulimia yields a moderate initial effect. However, only a small minority of patients recover, and the average patient continues to meet full DSM-IV criteria for the disorder. Combined pharmacotherapy and short-term psychotherapy appears to produce better results, although most patients continue to show symptoms at termination, and few data are available on sustained recovery over time. In accordance with recent calls in the medical literature for standardization of reporting practices in clinical trials, we suggest that investigators and meta-analysts report a range of indices that bear on efficacy and generalizability to clinical practice. These include exclusion rates and reasons for exclusion, percentage recovered, percentage improved, percentage remaining improved or recovered at follow-up, and percentage seeking additional treatment at follow-up, as well as outcome data for both completer and intent-to-treat samples.