Hershberger P J, Part H M, Markert R J, Cohen S M, Finger W W
Department of Veterans Affairs Medical Center (DVAMC), Dayton, OH 45428.
Acad Med. 1994 Oct;69(10):839-42. doi: 10.1097/00001888-199410000-00014.
To develop an inventory to measure the influence of cognitive biases on medical decisions.
The Inventory of Cognitive Biases in Medicine (ICBM) consists of 22 medical scenarios in which respondents choose between alternatives that represent bias-prone or statistically based decisions. In 1992 and 1993 the ICBM was administered to 102 medical students and residents and 318 physician faculty at the Wright State University School of Medicine.
The statistically correct decision was made in fewer than half the ICBM scenarios for both groups (students and residents, 41%; faculty, 49%); and only the faculty scored better than chance. Acceptable reliability (KR 20 = .62 for the faculty administration) and item discrimination (mean = .49) were achieved. Further, the results suggest that the ICBM has both content and construct validity.
The rather poor performances of the students and residents and the faculty on the ICBM suggest that cognitive biases constitute an important detraction from reliance on logical and statistical strategies. The ICBM shows promise for use as a tool in both instructional and research endeavors attempting to minimize and elucidate this phenomenon.
编制一份量表以衡量认知偏差对医疗决策的影响。
医学认知偏差量表（ICBM）包含22个医疗场景，受访者需在代表易产生偏差或基于统计学的决策的选项中进行选择。1992年和1993年，ICBM被应用于赖特州立大学医学院的102名医学生和住院医师以及318名医师教员。
两组（医学生和住院医师组为41%，教员组为49%）在ICBM场景中做出统计学上正确决策的比例均不到一半；只有教员组的得分高于随机水平。量表达到了可接受的信度（教员管理组的KR 20 = 0.62）和项目区分度（均值 = 0.49）。此外，结果表明ICBM具有内容效度和结构效度。
学生、住院医师和教员在ICBM上的表现相当差，这表明认知偏差严重影响了对逻辑和统计策略的依赖。ICBM有望作为一种工具，用于教学和研究工作，以尽量减少并阐明这一现象。
