A Comparison of Prolonged Exposure Therapy, Pharmacotherapy, and Their Combination for PTSD: What Works Best, and for Whom

Posttraumatic Stress Disorder (PTSD) remains a salient and debilitating problem, in the general population and for military veterans in particular. Several psychological and pharmacological treatments for PTSD have evidence to support their efficacy. However, the lack of comparative effectiveness trials for PTSD treatments remains a major gap in the literature, which limits conclusions that can be drawn about which of these treatments work best. In particular, trials directly comparing efficacious psychotherapies and pharmacotherapies are needed to inform clinical decision making for patients and providers. To address this gap, the proposed study will aim to compare the effectiveness of PTSD treatments with the strongest evidentiary support - Prolonged Exposure (PE) therapy and pharmacotherapy with paroxetine or venlafaxine - as well as the combination of these two treatments. A randomized trial in proposed with a large, diverse sample of veterans with PTSD (N = 300) recruited from 6 Veterans Affairs Medical Centers throughout the US (in Philadelphia, Coatesville, Milwaukee, Dallas, San Diego, and Palo Alto). Treatments conditions will reflect "real world" practice in these settings, and minimal exclusion criteria related to safety will be adopted, to maximize external validity. Participants will be permitted to complete treatment sessions in person or via telehealth (based on evidence for equivalent outcomes across these modalities), to maximize patient access, recruitment, and generalizability. Participants will complete baseline assessments, followed by 14 weeks of active treatment (involving up to 14 sessions of PE and/or medication management) with mid and posttreatment assessments after 7 and 14 weeks respectively, and then follow-up assessments at 27 and 40 weeks. Primary outcomes will include PTSD severity, depression symptoms, quality of life and functioning, assessed via clinical ratings and self-report measures. Further, a range of demographic and clinically relevant variables (e.g., trauma type/number, physiological arousal) will be collected at baseline and examined as potential predictors or moderators of treatment response, addressing another key gap in the PTSD treatment literature. Specifically, these data will be used to develop algorithms from predicting the optimal treatment for individual patients (i.e., "personalized advantage indices"; PAIs), a statistical approach which has advanced the depression treatment literature but has only been used in a limited capacity in PTSD research. The project will include an Advisory Board composed of clinician and patient representatives, in order to obtain stakeholder feedback at every stage of the study (from implementation to dissemination of findings). The effectiveness of the treatments will be compared using multilevel modeling. PAIs will be developed by conducting bootstrapped analyses to select variables that predict or moderate outcomes (Clinician Administered PTSD Scale severity ratings at Week 14), followed by leave-one-out cross-validation (i.e., jackknife) analyses to determine the magnitude of the predicted difference that results in each analysis representing that individuals "predicted advantage" (of one treatment over the others). We hypothesize that individuals who receive PE will have better outcomes than those who receive pharmacotherapy alone, based on existing data (e.g., cross study effect size comparisons), but have planned the study and sample to maximize statistical power in all analyses.