Identifying and reducing cognitive biases created by decision aids

描述(由申请人提供)：决策辅助(DA)为患者提供有关偏好敏感决策的信息，在这些决策中，患者的偏好可能在确定最佳决策方面发挥关键作用，并允许他们比在典型的临床会诊中有更多的时间来处理这些信息。一个国际专家小组，国际患者决策辅助标准(IPDAS)合作，建立了一个内容清单，鼓励决策辅助开发人员将其放入DAS中。然而，它没有就如何呈现此类内容提供指导，也没有提供如何确保此类内容不会造成决策偏见的指导。我们最初的拨款旨在识别和减少DAS造成的认知偏差。我们确定了DAS可能造成的多种认知偏差，并开发了几种实用的技术来减少或消除这些偏差。但我们的工作还没有完成。为了扩大我们的成就并回应IPDAS的建议，我们提议针对DAS可能造成的其他几种偏见进行研究，并开发消除或减少这些偏见的技术。此外，我们建议进行研究，不仅关注认知在人们决策中的作用，还关注情感的作用。因此，我们计划确定当患者使用决策辅助工具时可能出现的决策偏差，无论这些偏差是通过认知过程还是情感过程产生的，并确定减少这些偏差的方法。具体地说，我们的目标是探索与决策辅助相关的四个偏差来源：(1)模糊概率的呈现(2)关于概率的上下文信息的呈现(3)人们对健康结果的情感反应，以及(4)人们在单一决策中整合概率和效用的方法。对于每个来源，我们将确定影响患者决定的多个因素。我们建议确定每个问题的范围，描述产生偏差的情况，并测试减少或消除偏差的方法。我们所有的研究都将使用来自先前存在的决策辅助工具的材料。我们的方法包括两个截然不同的研究阶段。在第一阶段，我们使用迭代随机调查实验来探索哪些决策辅助因素会产生偏差。我们将根据真实的决策辅助材料，向人们提供假设性的医疗决定，并在不同的人之间改变决策任务的不同方面，以测试关于如何将决策偏差降至最低的假设。我们的初步研究将考虑相关因素的子集，后续研究将深入研究在我们的第一次实验中具有重要意义的因素，并测试由于功率和复杂性原因而未包括的其他因素。在第二阶段，我们将展示我们的第一阶段结果与患者护理的相关性。对于每一种偏倚来源，我们的研究将以对患者群体进行去偏倚技术的随机试验来达到顶峰。相关声明我们的研究旨在改进患者决策辅助工具，即为患者提供有关医疗决策的信息并帮助他们做出最符合自己喜好的决定的小册子、网站或视频。虽然决策辅助工具的内容存在国际标准，但我们的结果将帮助决策辅助工具开发人员知道如何最好地在决策辅助工具中呈现信息，以避免认知和情感上的偏差。这项工作对于确保患者得到充分的信息并能够参与有关其医疗护理的决策至关重要。

DESCRIPTION (provided by applicant): Decision aids (DAs) provide patients with information about preference sensitive decisions in which patients' preferences are likely to play a key role in determining what the best decision is and allow them more time to process the information than they would in a typical clinical encounter. An international group of experts, the International Patient Decision Aids Standards (IPDAS) Collaboration, has established a list of content that decision aid developers are encouraged to put into DAs. However, it provided no guidance about how to present such content, or how to make sure such content does not create decision biases. Our original grant aimed to identify and reduce cognitive biases created by DAs. We identified multiple cognitive biases that could be created by DAs and have developed several practical techniques for reducing or eliminating these biases. But our work is not done. To extend our accomplishments and respond to the IPDAS recommendations, we are proposing to conduct research targeting several additional biases that are potentially created by DAs and developing techniques to eliminate or reduce them. In addition, we propose to conduct studies that focus not only on the role that cognition plays in people's decisions, but also the role of affect. We thus plan to identify the kind of decision biases likely to arise when patients use decision aids, regardless of whether those biases arise through cognitive or affective processes, and identify ways to reduce them. Specifically, we aim to explore four sources of biases relevant to decision aids: (1) presentation of ambiguous probabilities (2) presentation of contextual information about probabilities (3) people's affective reaction to health outcomes, and (4) people's methods of integrating probability and utility in a single decision. For each source, we will identify multiple factors which bias patient decisions. We propose to identify the scope of each problem, delineate the circumstances under which biases result, and test methods for reducing or eliminating the biases. All our studies will use materials from pre-existing decision aids. Our approach includes two distinct research stages. In the first stage, we use iterative randomized survey experiments to explore which decision aid elements generate bias. We will present people with hypothetical medical decisions, based on material from real decision aids, and vary aspects of the decision task across people, to test hypotheses about how to minimize decision biases. Our initial studies will consider a subset of the relevant factors, and follow-up studies will drill down on factors that are significant in our first experiments and also test additional factors not included due to power and complexity reasons. In the second stage, we will demonstrate the relevance of our first-stage results for patient care. For each source of bias, our research will culminate with a randomized trial of a debiasing technique delivered to a patient population. Relevance Statement Our research aims to improve patient decision aids, which are booklets, websites, or videos that provide patients with information about medical decisions and help them to make the decision that best matches their own preferences. While international standards exist for the content of decision aids, our results will help decision aid developers know how to best present the information in decision aids so as to avoid cognitive and emotional biases. This work is essential to ensuring that patients are fully informed and able to participate in decision making regarding their medical care.