Using risk and errors to make less errors in clinical decision-making.

Monica Ortendahl

摘要

Dear Editor, I read with interest the study by Vanagas and Kinduris [1]on assessing the validity of cardiac surgery risk stratification systems for CABG patients, using patient-relatedfactors to predict mortality and postoperative morbidity. Their findings deserve consideration, as manyjudgments and decisions are made in clinical work, where the assessment of risk is necessary. There isrisk involved in the choice of tests to be used to reach a diagnosis. There is also an uncertainty andrisk in how to interpret results from tests used. Taken this uncertainty into consideration how shouldinformation from clinical and biomedical knowledge be combined to reach a diagnosis [2]? With a diagnosisobtained with some certainty or uncertainty what treatment should be chosen? In all these proceduresthere is risk involved. An issue is how clinical inferences generally are arrived at when making judgmentsand decisions. Theories have been provided about how doctors could include relevant information to improvedecision-making [3]. However, a reasoning error could be made in clinical inference, as it is characterizedby backward reasoning, where diagnosticians attempt to link observed effects to prior causes [4]. Incontrast to this post hoc explanation, statistical prediction entails forward reasoning, because it isconcerned with forecasting future outcomes given observed information. Clinical inference utilizes informationfrom prior periods to make a statement about today, and tends to consider error as a nuisance variable.The statistical approach, on the other hand, accepts error as inevitable, and in so doing probably makesfewer errors in prediction for periods extending over a relatively long time [5]. Moreover, the statisticalapproach is based upon group data to arrive at a conclusion. The situation is different in clinical inferenceand decision-making, where group data concerning risk constitute the basis for diagnostic and treatmentchoices concerning the individual patient. It has also been found that doctors exhibit an interindividual,as well as an intraindividual variation in judgments [6]. One example in practical work is the outcomeof clinical examinations that could vary between doctors. Another example is the interpretations of radiologicalpictures that could exhibit a variation between doctors. Many people tend to overestimate how much theyknow, even about the easiest knowledge tasks [7]. Overconfidence (i.e., greater certainty than circumstanceswarrant) leads to overestimation of the importance of occurrences that confirm their hypothesis. It impedeslearning from environmental feedback, and hence results in deleterious effects on future predictions.In many decision settings, inexperienced practitioners and even naive laboratory subjects perform aswell (or as poorly) as performers with more experience [8]. The performance of the patient could be asgood or bad as these subjects. The daily work with patients implies considering risks at many stagesof the decision process. How to convey this information about risk and error to the patients, being anunavoidable condition in clinical work, in order to reach a mutual agreement on treatment judgments anddecisions? By being aware of errors that can be made some of the errors can be counteracted. Therefore,it is a challenge in clinical practice to include different features of risk, and engage providers andpatients in present and future health. Sincerelly, Monica Ortendahl MD, PhD, Malma Backe 3 H, 756 47Uppsala, Sweden, e-mail: monicaortendahl@hotmail.com References: 1.Vanagas G, Kinduris S: Assessing thevalidity of cardiac surgery risk stratification systems for CABG patients in a single center. Med SciMonit, 2005;11: CR215-CR218 2.Gonzales-Clemente JM, Galdon G, Mitjavila J et al: Translation of the recommendationsfor the diagnosis of diabetes mellitus into daily clinical practice in a primary health care setting.Diabetes Res Clin Pract, 2003; 62: 123-29 3.Patel VL, Arocha JF, Chaudhari

机译：尊敬的编辑，我感兴趣地阅读了Vanagas和Kinduris [1]的研究，该研究使用与患者相关的因素来预测死亡率和术后发病率，从而评估CABG患者的心脏手术风险分层系统的有效性。他们的发现值得考虑，因为在临床工作中需要进行风险评估的许多判断和决策。选择要用于诊断的测试存在风险。如何解释所用测试的结果也存在不确定性和风险。考虑到这种不确定性，应如何结合临床和生物医学知识中的信息来进行诊断[2]？有了确定性或不确定性的诊断后，应该选择哪种治疗方法？在所有这些程序中，都涉及风险。一个问题是在做出判断和决策时通常如何得出临床推论。提供了有关医生如何将相关信息包括在内以改善决策制定的理论[3]。但是，临床推理中可能会出现推理错误，因为它的特征是向后推理，在这种情况下，诊断医生试图将观察到的效果与先前的原因联系起来[4]。与这种事后解释相反，统计预测需要进行正向推理，因为它与在给定观察信息的情况下预测未来结果有关。临床推论利用以前时期的信息来陈述今天的情况，并倾向于将误差视为令人讨厌的变量;另一方面，统计方法则认为误差是不可避免的，因此这样做可能会使预测期延长的误差减少相对较长的时间[5]。此外，统计方法是基于组数据得出结论的。在临床推论和决策中情况有所不同，其中有关风险的分组数据构成了针对单个患者的诊断和治疗选择的基础。还发现医生在判断方面表现出个体差异和个体差异[6]。实际工作中的一个例子是临床检查的结果，医生之间可能会有所不同。另一个例子是放射图像的解释，这些图像可能在医生之间表现出差异。许多人倾向于高估他们知道的知识，即使是最简单的知识任务[7]。过度自信（即确定性强于情况保证）导致高估了证实其假设的事件的重要性。它阻碍了从环境反馈中学习，并因此对未来的预测产生有害影响。在许多决策环境中，经验不足的从业人员甚至幼稚的实验室对象在具有更多经验的执行者中也表现出色（或表现不佳）[8]。患者的表现可能与这些受试者的好坏相同。与患者的日常工作意味着在决策过程的许多阶段都要考虑风险。在临床工作中不可避免的情况下，如何将有关风险和错误的信息传达给患者，以便就治疗的判断和决定达成共识？通过了解可能发生的错误，可以抵消某些错误。因此，在临床实践中挑战包括不同的风险特征，并使提供者和患者参与当前和未来的健康。此致Monica Ortendahl医学博士，博士，马尔玛·贝克（Halma Backe）3小时，电话756 47瑞典乌普萨拉，电子邮件：monicaortendahl@hotmail.com参考文献：1.Vanagas G，金杜里斯S：评估CABG患者心脏手术风险分层系统的有效性单中心。 Med SciMonit，2005; 11：CR215-CR218 2.Gonzales-Clemente JM，Galdon G，Mitjavila J等人：在初级卫生保健机构中将诊断糖尿病的建议转化为日常临床实践。DiabetesRes Clin Pract， 2003; 62：123-29 3.Patel VL，Arocha JF，Chaudhari

Using risk and errors to make less errors in clinical decision-making.

摘要

著录项

相似文献

相关主题

期刊订阅