A classification tree model for decision-making in clinical practice: An application based on the data of the german multicenter study on eating disorders, Project TR-EAT

To bridge the gap between science and practice in mental health care, scientific methods need to be developed that have direct applications in clinical practice. Studies that report "average" effects document the efficacy of psychotherapy but have limited implications for decision making in clinical practice. Although it is very unlikely that statistical models will ever perfectly predict treatment outcomes, models that identify and describe changes in a patient's risk of treatment failure may be developed to support decisions during ongoing treatment. In the current study, the classification and regression tree (CART) method was applied to data from the German multicenter study on the relationship between treatment duration/intensity and outcome in inpatient treatment of eating disorders (Project TR-EAT). Within a sample of patients diagnosed with bulimia nervosa (n = 630), data on admission, treatment strategy, and treatment response were used to develop a CART decision tree. This has potential for treatment decision support because the tree structure identifies changes in risk status associated with available treatment options and is sensitive to a patient's unique presenting characteristics and observed treatment response.