36 Chapter 2 and bulimia scale scores as the dependent variable showed that patients’ bulimia scale scores significantly decreased during treatment: F (1,287) = 582.170, p < .01, η2 = .670. The effect size indicates that the decrease in bulimia scale scores was large. A repeatedmeasures ANOVA with time (pretreatment/post-treatment) as the within-subject factor and BMI as the dependent variable showed that patients’ BMI scores significantly decreased during treatment: F (1,303) = 53.113, p < .01, η2 = .149. The effect size indicates that the decrease in BMI was moderate. Table 2. Pretreatment and post-treatment means (SD) of EDI bulimia scale scores and BMI for treatment completers. Pretreatment Post-treatment EDI bulimia scale scores * 7.70 (4.286) 2.12 (2.933) BMI * 41.95 (6.91) 40.85 (7.30) * p < .010 Note. BMI = body mass index; EDI, Eating Disorder Inventory Prediction of post-treatment bulimia scale scores A hierarchical linear regression analysis was conducted to determine predictors for the post-treatment level of bulimia scale scores. The following pretreatment variables were entered as predictors, using a backward model building approach: EDI bulimia scale scores, the remaining EDI subscales (BD, DT, I, P, ID, IA and MF) and the subscales and total scores of the SCL-90 and the BAT, as well as the NEO-PI-R subscales, BDI, BMI, social embedding, level of education, gender and age. The final model significantly predicted post-treatment bulimia scale scores, and consists of eight relevant predictors. Pretreatment EDI bulimia, EDI ineffectiveness, SCL-90 total score and NEO-PI-R extraversion were significant positive predictors for post-treatment bulimia scale scores. Furthermore, EDI interoceptive awareness was a marginally significant positive predictor: Lower scores on either of these pretreatment predictors were predictive of lower post-treatment bulimia scale scores. Pretreatment SCL-90 depression was a significant negative predictor for post-treatment bulimia scale scores, and EDI drive for thinness and EDI perfectionism were marginally significant negative predictors: Lower scores on either of these pretreatment predictors were predictive of higher post-treatment bulimia scale scores. See Table 3 for an overview of the final model and for test statistics for both the model and the predictors.3 3 When repeating the analysis with only female patients (n = 225), the EDI ineffectiveness scale no longer reached the predetermined p < .10 criterion and was therefore excluded from the final model. Three new predictors did reach the p < .10 criterion and were now included in the model: SCL-90 somatic complaints and SCL-90 distrust (both negative predictors) and BAT total score (positive predictor). All other predictors showed results similar to those reported above (in the data on men and women).
RkJQdWJsaXNoZXIy MjY0ODMw