Differences in Cognitive Behavioral Therapy Dropout Rates between Bulimia Nervosa Subtypes Based on Drive for Thinness and Depression

Dropout rates from therapy in bulimia nervosa (BN) are variable and premature [1,2,3,4]. Despite the proven effectiveness of cognitive behavioral therapy (CBT) for BN, attrition rates in routine clinical practice may be as high as 50%, though in selected populations they appear to be lower with around 20% [1,2]. This variability may suggest that depending on the population under study, a higher or lower number of BN patients may detect that this treatment is not suitable for them, which may be associated with their pretreatment characteristics [3]. In keeping with this, non-adherence to CBT in BN has been consistently related to depression [2,3,4]. This is important because CBT for BN [5], as other brief interventions [6], is a focal and practical method of treatment that from the very first sessions explicitly addresses specific problems such as reducing extreme dietary restraint to stop the vicious binge eating-purging cycle, but it does not address mood disturbances [5]. However, while dropouts from CBT would be expected to be higher in BN patients with higher pretreatment depressive levels, contradictory findings exist because no differences were found between patients classified by cluster analysis into one of two subtypes, a mixed ‘dietary-depressive’ and a pure ‘dietary’ with low levels of depression [7].This apparent contradiction might be partially resolved by using the expanded version of the original etiologic and maintenance model of binge eating [7], based on Eating Disorders Inventory – drive for thinness (EDI-DT) and Symptom Checklist – depression [8]. This new model proposed that besides the two above-mentioned types with high dietary restraint, two other types characterized by either low levels of DT and depression (‘mild’) or low to moderate levels of DT but high depression (‘depressive’) seem to exist among eating-disordered patients [8]. Therefore, it is hypothesized that those individuals seeking relief of the symptoms that CBT addresses (high dietary restraint or DT without depression) will find it more suitable and will adhere more to CBT than those with the opposite symptoms (depression without high DT) or than those without them. Similarly to previous studies [7], those in the mixed ‘DT-depressive’ subtype versus the pure ‘DT’ type, despite higher pretreatment levels of depression, will be likely to find some relief for the active problem of dietary restraint and present no differences in dropout.In the present study, a total of 154 adult females meeting DSM-IV-TR [9] criteria for purging BN (BN-P) participated. They were consecutively admitted for group-based outpatient treatment with no other psychotherapy at the time at the Eating Disorders Unit (University Hospital of Bellvitge). Treatment consisted of 22 (8–10 outpatient) 90-min sessions over 20 weeks (twice weekly for the first 2 weeks). These comprised 6 initial sessions of brief psychoeducation [6] followed by 16 sessions of CBT [5]. The Institutional Ethics Committee approved this study and informed consent was obtained from all participants.Patients were assessed with well-validated questionnaires as detailed elsewhere [8], such as the Eating Disorders Inventory-2 (EDI-2) and the Symptom Checklist-90-Revised (SCL-90-R). For their subtyping, the scores on the EDI-DT and SCL-90-R depression subscales were used. The EDI-DT specifically looks at preoccupation with weight showing high correlation with common dietary restraint measures. Since an EDI-DT cut-off score of 14 is used to detect individuals at risk of an eating disorder, all individuals scoring ≤14 were considered as presenting low DT as opposed to those scoring above, who were considered to present high DT. The SCL-90-R depression subscale highly correlates with the commonly used Beck Depression Inventory and differentiates non-clinical and mild to moderate depression from severe depression (scores >2). Therefore, all patients scoring ≤2 were considered as having low depression and those scoring above as having high depression. By using these pretreatment cut-off scores, four subtypes emerged (table 1), similar in proportions to those previously found for BN-P with rates of 14, 25.1, 24 and 37% of the BN-P corresponding to ‘mild’, pure ‘DT’, ‘depressive-mild DT’ and ‘dietary-depressive’ subtypes, respectively [8]. The rate of total attrition (33%) was also very similar to rates found for group CBT [10] or individual CBT [11]. Attrition rates differed between subtypes in keeping with the hypothesis (table 1): dropout rate was higher (60.5%) in the ‘depressive’ subtype than in the rest, and lower (18.5%) in the ‘pure DT’ subtype than in the two subtypes with mild DT scores. No differences were found between the two subtypes with high DT.Thus, this study supports and extends previous findings by proposing a new strategy to classify patients into four different subtypes throughout the use of worldwide valid cut-off scores of DT and depression [8] without needing to submit them to cluster analysis as in previous studies [7,8]. In addition, it provides for the first time predictive validity for this subtyping scheme with regard to dropout from CBT in an unselected BN-P population seeking treatment.Future studies are needed to replicate these findings in BN-P or other binge eating-related disorders. Particularly, these should control for timing of dropout and pharmacotherapy during CBT, which are limiting factors of the present study. For example, antidepressant drugs such as fluoxetine might explain a substantial percentage of those women within the depressive subtypes that did not drop out from treatment in the present study.In sum, the present results suggest that this expanded subtyping model [8] based on DT and depression present predictive validity as an instrument to identify BN-P patients at risk of CBT failure, and could be used at initial assessment to lower dropout rates.Financial support was received from the Fondo de Investigación Sanitaria (PI11/210; PI081714) and Agència de Gestió d’Ajuts Universitaris i de Recerca (2009SGR1554; FI-DGR 2008). CIBER Fisiopatología de la Obesidad y Nutrición is an initiative of Instituto Salud Carlos III. Gobierno de Extremadura-European Union-FEDER-BS10023.