Introduction: Relatively little research has assessed predictors of both treatment outcomes and attrition in binge eating disorder (BED) treatment. Even fewer studies did so for digital forms of...Show moreIntroduction: Relatively little research has assessed predictors of both treatment outcomes and attrition in binge eating disorder (BED) treatment. Even fewer studies did so for digital forms of therapy. This study thus aims to contribute to the current pool of knowledge by examining the predictive value of various variables in a recently developed digital BED-treatment: BED-online. Methods: This study was part of a RCT into the effects of BED-online therapy. Participants were over the age of 18, Dutch-speaking and diagnosed with the DSM-V BED. A total of 180 participants were found to be eligible, of whom 40 (22.2%) dropped-out before the last session. Post-treatment measurements from an interview (EDE) and a self-report questionnaire (EDE-Q) were used to determine the immediate treatment effects. A follow-up (24 weeks post-treatment) EDE-Q measurement determined the long-term effects. These variables served as the dependent variables in three different hierarchical linear regression analyses. A fourth logistical regression analysis was conducted, where treatment-related drop-out formed the dependent variable. The following eight predictor variables were chosen based on literature research: ethnicity, age, gender, educational level, comorbidity, frequency of binge eating episodes, levels of BED pathology and treatment condition. The predictor variables served as the independent variables. Results: BED pathology at baseline was found to be the only significant predictor of treatment outcomes as measured by the EDE (β=.41, t=3.71, p<.001), EDE-Q (β=.57, t=8.23, p<.001) and at follow-up (β=.47, t= 6.24, p< .001). Attrition could be predicted by both ethnicity and gender, where males were 5.63 times more likely to discontinue treatment prematurely (OR=5.63, 95%CI [1.81, 17.53]) and participants born abroad were 3.91 times more likely to discontinue treatment (OR=3.91, 95%CI [1.43, 8.76]). Other independent variables did not significantly contribute to the final regression models. Discussion and conclusion: All predictor variables are discussed in detail, reflecting on the results found and comparing them to findings of previous studies. Limitations are elaborated in depth. Due to these limitations, and the modest number of previous studies, further research is encouraged, exploring both the current and other dependent variablesShow less