Stojan Jennifer N, Daniel Michelle, Morgan Helen K, Whitman Laurie, Gruppen Larry D
J.N. Stojan is assistant professor, Departments of Internal Medicine and Pediatrics, University of Michigan Medical School, Ann Arbor, Michigan. M. Daniel is assistant dean for curriculum and assistant professor, Departments of Emergency Medicine and Learning Health Sciences, University of Michigan Medical School, Ann Arbor, Michigan; ORCID: http://orcid.org/0000-0001-8961-7119. H.K. Morgan is clinical assistant professor, Departments of Obstetrics and Gynecology and Learning Health Sciences, University of Michigan Medical School, Ann Arbor, Michigan. L. Whitman is program manager, Standardized Patient Program, and standardized patient educator, University of Michigan Medical School, Ann Arbor, Michigan. L.D. Gruppen is professor, Department of Learning Health Sciences, and director, University of Michigan Master of Health Professions Education Program, Ann Arbor, Michigan; ORCID: http://orcid.org/0000-0002-2107-0126.
Acad Med. 2017 Nov;92(11S Association of American Medical Colleges Learn Serve Lead: Proceedings of the 56th Annual Research in Medical Education Sessions):S43-S47. doi: 10.1097/ACM.0000000000001909.
Learning to make decisions under uncertain conditions is a critical component of diagnostic and therapeutic reasoning. This study sought to determine treatment decisions medical students make when presented with different thresholds of diagnostic uncertainty and whether they appropriately adjust diagnostic probabilities with test information.
Two classes (2015, 2016) of fourth-year students (N = 342) were presented a patient with viral pneumonia and given 10%, 20%, or 50% pretest probabilities of that patient having a superimposed bacterial infection. Students decided to not treat, order a diagnostic test to guide management, or treat without testing based on these probabilities. The 2015 class was provided a posttest probability of 10% or 50% and asked to adjust their initial treatment decision.
When given a low (10%) pretest probability, students were less likely to decide to treat (6%) and more likely to decide not to treat (36%). The percentage of students deciding to treat increased as the pretest probability of a superimposed infection increased from 10% to 50%, while the percentage of students not wanting to treat decreased. Interestingly, at 10%, 20%, and 50% pretest probability levels, most students were unable to decide and chose to order another test (57%, 67%, and 64%, respectively). When provided low and high posttest probabilities, students appropriately adjusted their decision making, but 29% to 32% still wanted additional testing.
Students adjusted treatment decisions to reflect different levels of diagnostic uncertainty, but varied considerably in their individual thresholds to make decisions, possibly contributing to unnecessary testing.
学会在不确定条件下做出决策是诊断和治疗推理的关键组成部分。本研究旨在确定医学生在面对不同诊断不确定性阈值时所做出的治疗决策，以及他们是否根据检测信息适当地调整诊断概率。
向两个班级（2015级、2016级）的四年级学生（N = 342）展示一名病毒性肺炎患者，并给出该患者发生叠加细菌感染的预检概率为10%、20%或50%。学生们根据这些概率决定不治疗、进行诊断性检测以指导治疗，或不进行检测直接治疗。向2015级学生提供10%或50%的检测后概率，并要求他们调整初始治疗决策。
当预检概率较低（10%）时，学生决定治疗的可能性较小（6%），而决定不治疗的可能性较大（36%）。随着叠加感染预检概率从10%增加到50%，决定治疗的学生比例增加，而不想治疗的学生比例下降。有趣的是，在预检概率为10%、20%和50%时，大多数学生无法做出决定，选择进行另一项检测（分别为57%、67%和64%）。当提供低和高检测后概率时，学生适当地调整了他们的决策，但仍有29%至32%的学生希望进行额外检测。
学生调整治疗决策以反映不同水平的诊断不确定性，但在个人决策阈值方面差异很大，这可能导致不必要的检测。
