(PS7-1) A Bayesian Network Analysis to Treatment Fidelity in Family-based Treatment

Background: Treatment fidelity refers to the degree to which an intervention is delivered as intended and comprises three parts: adherence, therapist competence, and treatment differentiation. Fidelity is crucial to intervention development and implementation; however, its dynamic nature creates challenges in conceptualization, measurement, and intervention. Examining the relationship between fidelity and treatment outcomes may suggest specific procedures to train therapists, optimize clinical outcomes, and enhance dissemination and implementation. This would be particularly beneficial for low weight eating disorders (LWED), for which the quality of community care may vary substantially from provider to provider. Family-Based Therapy (FBT) for adolescents with LWEDs explicitly targets weight restoration, causing weight gain to be considered the field standard for assessing treatment efficacy. Therefore, the current exploratory analysis aimed to characterize the relationship between treatment fidelity and weight-gain in adolescents with a LWED receiving 6 sessions of FBT-Interoceptive Exposure (FBT-IE). 

Methods: FBT-IE sessions of 30 adolescents with a LWED were videotaped and coded for treatment fidelity using the Family Therapy Fidelity and Adherence Check (FBT-FACT). The treatment outcome measure was percent ideal body weight (%IBW). Bayesian belief network models were utilized to characterize the relationship between variables with unidirectional relationships, allowing for causal inferences about these relationships. Specifically, the model intended to predict the relationship between treatment fidelity and pre-post changes in weight. Bootstrap model searches were then used to confirm the hypothesized network structure, and k-fold validation was used to assess the reliability of the network. We hypothesized that treatment fidelity would predict an increase in %IBW at 6-weeks. 

Results: The hypothesized model with serial positive relationships from Competence to Adherence and Differentiation and indirectly through both to %IBW indicated that competence drove improvements in weight through influence over both adherence and differentiation. The results of the incumbent model significantly predicted individual level %IBW. 

Discussion: We present evidence that a conceptual model of fidelity which positions competence as a dynamic driving factor of adherence and differentiation can predict clinical changes in primary outcome of %IBW in brief FBT-IE. This pilot study suggests that larger applications of the model could prove informative in treatment development or help guide training efforts at competency, which dictates the degree of adherence and differentiation applied to the individual needs of a patient or family.