Department of Psychology, University of Pennsylvania, Philadelphia, PA, USA.
Psychiatry and Behavioral Sciences, Palo Alto Veterans Affairs Health System, Palo Alto, CA, USA.
Depress Anxiety. 2018 Apr;35(4):330-338. doi: 10.1002/da.22731. Epub 2018 Feb 28.
Dropout rates for effective therapies for posttraumatic stress disorder (PTSD) can be high, especially in practice settings. Although clinicians have intuitions regarding what treatment patients may complete, there are few systematic data to drive those judgments.
A multivariable model of dropout risk was constructed with randomized clinical trial data (n = 160) comparing prolonged exposure (PE) and cognitive processing therapy (CPT) for rape-induced PTSD. A two-step bootstrapped variable selection algorithm was applied to identify moderators of dropout as a function of treatment condition. Employing identified moderators in a model, fivefold cross-validation yielded estimates of dropout probability for each patient in each condition. Dropout rates between patients who did and did not receive their model-indicated treatment were compared.
Despite equivalent dropout rates across treatments, patients assigned to their model-indicated treatment were significantly less likely to drop out relative to patients who did not (relative risk = 0.49 [95% CI: 0.29-0.82]). Moderators included in the model were: childhood physical abuse, current relationship conflict, anger, and being a racial minority, all of which were associated with higher likelihood of dropout in PE than CPT.
Individual differences among patients affect the likelihood they will complete a particular treatment, and clinicians can consider these moderators in treatment planning. In the future, treatment selection models could be used to increase the percentage of patients who will receive a full course of treatment, but replication and extension of such models, and consideration of how best to integrate them into routine practice, are needed.
创伤后应激障碍（PTSD）有效治疗的辍学率可能很高，尤其是在实践环境中。尽管临床医生对患者可能完成的治疗有直觉，但很少有系统数据来驱动这些判断。
使用比较性暴露疗法（PE）和认知加工疗法（CPT）治疗强奸后 PTSD 的随机临床试验数据（n=160）构建辍学风险的多变量模型。采用两步bootstrap 变量选择算法，根据治疗条件确定辍学的调节因素。在模型中使用鉴定出的调节因素，五次交叉验证为每种情况下的每位患者生成辍学概率的估计值。比较接受和未接受模型指示治疗的患者之间的辍学率。
尽管治疗之间的辍学率相当，但接受模型指示治疗的患者相对未接受模型指示治疗的患者辍学的可能性显著降低（相对风险=0.49 [95% CI：0.29-0.82]）。纳入模型的调节因素包括：童年期身体虐待、当前人际关系冲突、愤怒和少数民族身份，所有这些因素都与 PE 比 CPT 更高的辍学率相关。
患者之间的个体差异会影响他们完成特定治疗的可能性，临床医生可以在治疗计划中考虑这些调节因素。未来，治疗选择模型可用于提高接受完整疗程治疗的患者比例，但需要对这些模型进行复制和扩展，并考虑如何将其最好地整合到常规实践中。
