Department of Healthcare Policy, Harvard Medical School, Boston, Massachusetts, USA.
Curr Opin Psychiatry. 2018 Jan;31(1):32-39. doi: 10.1097/YCO.0000000000000377.
To review progress developing clinical decision support tools for personalized treatment of major depressive disorder (MDD).
Over the years, a variety of individual indicators ranging from biomarkers to clinical observations and self-report scales have been used to predict various aspects of differential MDD treatment response. Most of this work focused on predicting remission either with antidepressant medications versus psychotherapy, some antidepressant medications versus others, some psychotherapies versus others, and combination therapies versus monotherapies. However, to date, none of the individual predictors in these studies has been strong enough to guide optimal treatment selection for most patients. Interest consequently turned to decision support tools made up of multiple predictors, but the development of such tools has been hampered by small study sample sizes. Design recommendations are made here for future studies to address this problem.
Recommendations include using large prospective observational studies followed by pragmatic trials rather than smaller, expensive controlled treatment trials for preliminary development of decision support tools; basing these tools on comprehensive batteries of inexpensive self-report and clinical predictors (e.g., self-administered performance-based neurocognitive tests) versus expensive biomarkers; and reserving biomarker assessments for targeted studies of patients not well classified by inexpensive predictor batteries.
回顾开发临床决策支持工具以实现重度抑郁症（MDD）个体化治疗的进展。
多年来，各种个体指标，从生物标志物到临床观察和自我报告量表，都被用于预测 MDD 治疗反应的各个方面。这些工作大多集中在预测抗抑郁药物与心理治疗、某些抗抑郁药物与其他药物、某些心理治疗与其他治疗、联合治疗与单一治疗的缓解率。然而，迄今为止，这些研究中的任何单一预测指标都不足以指导大多数患者的最佳治疗选择。因此，人们的兴趣转向了由多个预测指标组成的决策支持工具，但由于研究样本量小，这些工具的开发受到了阻碍。本文为未来的研究提供了设计建议，以解决这个问题。
建议包括使用大型前瞻性观察研究，然后是实用试验，而不是较小、昂贵的对照治疗试验，初步开发决策支持工具；这些工具基于综合的、廉价的自我报告和临床预测指标（例如，自我管理的基于表现的神经认知测试），而不是昂贵的生物标志物；并将生物标志物评估保留给通过廉价预测指标组合分类不佳的患者的靶向研究。
