Predicting treatment dropout among veterans receiving prolonged exposure therapy.

Objective: To examine whether dropout from prolonged exposure (PE) therapy can be predicted from demographic and outcomes data that would typically be available to clinicians. Methods: Dropout was examined in 2,606 patients treated by clinicians in the U.S. Veterans Health Administration PE Training Program. PE typically consists of 8-15 sessions, with 8 sessions being considered a minimum therapeutic dose for most patients. Logistic regression was used to assess the impact of demographics, depression, trauma history, and PE target trauma on risk for dropout. Growth mixture modeling was used to study how posttraumatic stress disorder symptom patterns during the first 5 treatment encounters predicted dropout. Results: In total, 782 patients (30.0%) completed fewer than 8 sessions of PE. Younger veterans were more likely to drop out of PE; odds ratio (OR) per year of age = 0.97, p <.01. Controlling for other factors, veterans who focused on childhood trauma were less likely to drop out than those focusing on combat trauma (OR = 0.51, p <.05). Dropout was unrelated to symptom course or symptom worsening between sessions. Nevertheless, clinicians attributed dropout to distress or avoidance in 45% of the patients who dropped out, citing other factors in 37% of dropout cases. Conclusions: Treatment dropout was predicted by age but not by initial symptom severity or symptom course early in treatment. Symptom exacerbation was rare and did not increase risk of dropout. Nonetheless, clinicians often attributed dropout to patients not tolerating PE.