Optimizing Allocation to Telehealth and In-Person Prolonged Exposure for Women Veterans with Military Sexual Trauma: A Precision Medicine Approach

Abstract

Military sexual trauma-related post-traumatic stress disorder (PTSD) is highly prevalent and costly among women veterans, making the need for effective and accessible treatment of critical importance. Access to care is a key mechanism of mental health disparities and might affect differential response to treatment. The goal of this study was to estimate an individualized treatment rule based on readily available individual characteristics related to access to care to optimize allocation to in-person vs. telehealth delivery of prolonged exposure for PTSD in military sexual trauma survivors. The following variables were used as prescriptive factors: age, race, disability status, socioeconomic status, rural vs. urban status, and baseline PTSD level. The rule was estimated using a machine-learning approach, Outcome Weighted Learning. The estimated optimal rule outperformed a one-size-fits-all rule where everyone is universally assigned to telehealth; it led to markedly lower mean PTSD levels following 6 months from treatment (Vdopt - VTelehealth = -14.55, 95% CI: -27.24, -1.86). However, the rule did not significantly discriminate for in-person therapy (Vdopt - VIn-person = -11.86, 95% CI: -25.83, 2.12). Upon further validation with larger and more diverse samples, such a rule may be applied in practice settings to aid clinical decision-making and personalization of treatment assignment.