Abstract
Objective: Multilevel growth analysis was used to establish the mean growth trajectory (shape of change) for cognitive behavioral therapy (CBT) for youth with anxiety disorders. Two-level growth analysis was conducted to identify important between-youth predictors of session-by-session symptom change. Method: Fifty-five youth (ages 7-17; 50.9% male) and their parents participated in a 16- to 20-week CBT that emphasized affective, cognitive, and exposure-based exercises. Multilevel growth models (MLMs) were estimated to model session symptom data taking into account an "anxiety spike" hypothesized to occur at initiation of exposure sessions. Three models were compared: a cubic curve, a log-linear curve plus an exposure covariate, and a linear curve plus exposure covariate. Two-level MLM examined the effect of demographic traits (sex, age, race/ethnicity), pretreatment symptom severity, comorbid school refusal, early treatment factors (use of selective serotonin reuptake inhibitor medication, therapeutic alliance, treatment attrition), and pretreatment coping (engagement, disengagement, and involuntary coping). Results: Fit indices provided support for the cubic growth model using either parent or youth anxiety data. Level 2 analysis identified youth age, symptom severity (anxiety, externalizing), early attrition, and engagement and disengagement coping as significant predictors of symptom trajectories. Predictors accounted for 34%-37% of between-youth variance in midtreatment anxiety scores. Conclusions: Findings suggest that the symptom course of CBT, and the effect of between-youth factors on treatment outcomes, is more complex than previously thought. Educating therapists and clients about findings can aid treatment expectations and dissemination efforts of empirically supported treatments.