USF Program for Comparative Research Effectiveness, Division of Evidence Based Medicine, Department of Internal Medicine, Morsani College of Medicine, University of South Florida, Tampa, FL, USA; H Lee Moffitt Cancer Center & Research Institute, Tampa, FL, USA.
Eur J Clin Invest. 2015 May;45(5):485-93. doi: 10.1111/eci.12421. Epub 2015 Mar 9.
The threshold model represents an important advance in the field of medical decision-making. It is a linchpin between evidence (which exists on the continuum of credibility) and decision-making (which is a categorical exercise - we decide to act or not act). The threshold concept is closely related to the question of rational decision-making. When should the physician act, that is order a diagnostic test, or prescribe treatment? The threshold model embodies the decision theoretic rationality that says the most rational decision is to prescribe treatment when the expected treatment benefit outweighs its expected harms. However, the well-documented large variation in the way physicians order diagnostic tests or decide to administer treatments is consistent with a notion that physicians' individual action thresholds vary.
We present a narrative review summarizing the existing literature on physicians' use of a threshold strategy for decision-making.
We found that the observed variation in decision action thresholds is partially due to the way people integrate benefits and harms. That is, explanation of variation in clinical practice can be reduced to a consideration of thresholds. Limited evidence suggests that non-expected utility threshold (non-EUT) models, such as regret-based and dual-processing models, may explain current medical practice better. However, inclusion of costs and recognition of risk attitudes towards uncertain treatment effects and comorbidities may improve the explanatory and predictive value of the EUT-based threshold models.
The decision when to act is closely related to the question of rational choice. We conclude that the medical community has not yet fully defined criteria for rational clinical decision-making. The traditional notion of rationality rooted in EUT may need to be supplemented by reflective rationality, which strives to integrate all aspects of medical practice - medical, humanistic and socio-economic - within a coherent reasoning system.
阈值模型是医学决策领域的重要进展。它是证据（存在于可信度连续体上）和决策（是一种分类练习——我们决定采取行动或不采取行动）之间的关键。阈值概念与理性决策问题密切相关。医生何时应该采取行动，即开诊断测试或开治疗处方？阈值模型体现了决策理论的合理性，即当预期治疗收益超过其预期危害时，最合理的决策是开治疗处方。然而，医生开诊断测试或决定进行治疗的方式存在大量有据可查的差异，这与医生的个体行动阈值存在差异的观点一致。
我们进行了叙述性综述，总结了关于医生使用阈值策略进行决策的现有文献。
我们发现，观察到的决策行动阈值的变化部分归因于人们对收益和危害的综合方式。也就是说，对临床实践变化的解释可以归结为对阈值的考虑。有限的证据表明，非期望效用阈值（非 EUT）模型，如基于后悔和双重处理模型，可以更好地解释当前的医学实践。然而，纳入成本以及认识到对不确定治疗效果和合并症的风险态度，可能会提高基于 EUT 的阈值模型的解释和预测价值。
何时采取行动的决策与理性选择问题密切相关。我们得出结论，医学界尚未完全定义合理临床决策的标准。基于 EUT 的理性传统观念可能需要通过反思理性来补充，反思理性努力将医学实践的所有方面——医学、人文和社会经济——整合到一个连贯的推理系统中。
