Commitment to Treatment Goals in Prediction of Group Cognitive—Behavioral Therapy Treatment Outcome for Women With Bulimia Nervosa Melissa Pederson Mussell Graduate Department of Professional Psychology University of St. Thomas James E. Mitchell Neuropsychiatric Research Institute Ross D. Crosby Neuropsychiatric Research Institute Jayne A. Fulkerson Department of Psychology University of Minnesota Harry M. Hoberman PACIFICA Center John L. Romano Department of Educational Psychology University of Minnesota ABSTRACT The purpose of this study was to investigate potential client variables that predict favorable response to group cognitive—behavioral therapy in a sample of women ( N = 143) seeking treatment for bulimia nervosa. Similar to findings of previous studies, bulimic symptom remission at end of treatment was predicted by baseline degree of bulimic symptom severity but not by depressive symptomatology or perfectionism. After these variables were controlled for, both pretreatment ratings of desire to discontinue bulimic behaviors and expected success significantly added to prediction of treatment outcome. The primary variable found to predict longer term outcome was symptom remission at the end of treatment and at the 1-month follow-up. The body of bulimia nervosa (BN) treatment literature supports the claim that cognitive—behavioral therapy (CBT) is efficacious in reducing bulimic symptom severity for many individuals with BN (see review by Wilson, Fairburn, & Agras, 1997 ). However, heterogeneity of treatment response also has been demonstrated consistently in that many participants remain symptomatic at the end of treatment and therapeutic gains may not be well maintained for some individuals (see reviews by Mitchell, Hoberman, Peterson, Mussell, & Pyle, 1996 , and Wilson, 1996 ). Yet very few controlled CBT trials have reported data on predictors of BN treatment outcome. Furthermore, investigations of client-related prognostic indicators for CBT for BN have yielded few consistent predictors of BN treatment response or relapse (cf. Keel & Mitchell, 1997 ), which may be related in part to several methodological limitations. Differential criteria used to determine treatment success and variations in selection of potential predictor variables make comparisons across studies difficult. Some prediction studies have combined data from disparate treatment conditions, and numerous predictor variables were examined using modest sample sizes, resulting in large variable-to-participant ratios in many studies. Follow-up measures often were not included in the analyses. The limited number of studies available on CBT posttreatment prediction of longer term BN outcome has focused on prediction of relapse ( Fairburn, Peveler, Jones, Hope, & Doll, 1993 ; Freeman, Beach, Davis, & Solyom, 1985 ; Olmsted, Kaplan, & Rockert, 1994 ), suggesting that residual symptoms of BN (e.g., vomiting and body image disturbance) at the end of treatment may be predictive of less favorable long-term outcome; however, data are not Journal of Consulting and Clinical Psychology © 2000 by the American Psychological Association June 2000 Vol. 68, No. 3, 432-437 For personal use only--not for distribution. Page 1 of 10 9/5/2001 http://spider.apa.org/ftdocs/ccp/2000/june/ccp683432.html