BackgroundHabits are behavioral routines that are automatic and frequent, relatively independent of any desired outcome, and have potent antecedent cues. Among individuals with anorexia nervosa (AN), behaviors that promote the starved state appear habitual, and this is the foundation of a recent neurobiological model of AN. In this proof-of-concept study, we tested the habit model of AN by examining the impact of an intervention focused on antecedent cues for eating disorder routines.MethodsThe primary intervention target was habit strength; we also measured clinical impact via eating disorder psychopathology and actual eating. Twenty-two hospitalized patients with AN were randomly assigned to 12 sessions of either Supportive Psychotherapy or a behavioral intervention aimed at cues for maladaptive behavioral routines, Regulating Emotions and Changing Habits (REaCH).ResultsCovarying for baseline, REaCH was associated with a significantly lower Self-Report Habit Index (SRHI) score and significantly lower Eating Disorder Examination-Questionnaire (EDE-Q) global score at the end-of-treatment. The end-of-treatment effect size for SRHI was d = 1.28, for EDE-Q was d = 0.81, and for caloric intake was d = 1.16.ConclusionsREaCH changed habit strength of maladaptive routines more than an active control therapy, and targeting habit strength yielded improvement in clinically meaningful measures. These findings support a habit-based model of AN, and suggest habit strength as a mechanism-based target for intervention.