Using Electronic Records to Detect and Learn from Ambulatory Diagnostic Errors

描述(由申请人提供)：应用程序的总体目标是利用电子健康记录(EHR)中的数据来检测初级保健中的诊断错误，了解其原因，并为制定未来的预防策略奠定基础。诊断错误可能是初级保健中最常见的错误类型。它们也是最昂贵的，也是医疗事故索赔的主要依据。尽管诊断错误很重要，但在患者安全方面，诊断错误是一个没有得到重视和研究的领域，部分原因是它们很难被发现。帮助发现诊断错误的策略对于提高初级保健服务的质量和安全性至关重要。诸如错误报告系统和随机图表检查等用于识别诊断错误的检测方法是有限的。近年来，计算机化的触发技术已经被用来通过选择特定的图表进行更详细的审查来识别其他环境中的不良事件。在我们的初步工作中，我们开发并测试了两个计算机化的触发工具，以确定可能包含初级保健诊断错误证据的图表。我们认为，改进这些工具，并将它们与其他变量相结合，可能会导致更高的诊断错误检测率。在另外的前期工作中，我们测试了一种计算机化的方法，该方法可以作为一种新的触发工具来检测与异常检测结果后续相关的诊断错误。我们现在建议验证这一潜在的触发工具。我们还建议将我们的研究从退伍军人管理局扩展到德克萨斯州中部一个以实践为基础的研究网络中的大型初级保健网络。这些背景将包括内科和家庭医学、学术和非学术实践、城市和农村患者，以及显著的种族、性别、民族、年龄和社会经济多样性。我们的具体目标是：1)应用和改进基于访问模式的计算机化触发器，以检测、测量和学习不同初级保健环境中的诊断错误。2)测试计算机跟踪可能丢失到随访中的异常测试结果的方法是否可以被用作在初级保健中识别诊断差错的触发器。在目标1中，我们将使用触发器查询临床资料库，并以电子方式收集有关初级保健就诊的额外临床数据(变量)。为了测试触发器的效用，我们将它们的阳性预测值(PPV)与对照进行比较，后者是不符合触发条件的随机访问样本。受过培训的盲表审查员将验证这两个子集中是否存在诊断错误。为了改进我们的触发器，我们将使用Logistic回归模型来测试将触发器与特定的独立临床变量整合在一起的加性PPV。在目标2中，我们将根据一个新的潜在触发因素，以电子方式跟踪和识别患者的记录，以供进一步的图表审查。我们将通过比较触发子集和对照子集的PPV来测试该触发工具在检测与后续丢失的测试结果相关的诊断错误方面的有效性。细化和验证不同初级保健环境中诊断错误的触发因素，将为在全国范围内使用它们进行质量测量和改进活动奠定基础。

DESCRIPTION (Provided by the Applicant): The overall goal of the application is to utilize data from electronic health records (EHR) to detect diagnostic errors in primary care, understand their causes, and lay groundwork to formulate future prevention strategies. Diagnostic errors are likely the most common types of errors in primary care. They are also the most expensive and are the leading basis for malpractice claims. Despite their importance, diagnostic errors are an underemphasized and understudied area of patient safety, in part because they are difficult to detect. Strategies to help detect diagnostic errors are critical to improving quality and safety of primary care delivery. Detection methods such as error reporting systems and random chart reviews to identify diagnostic errors are limited. In recent years, computerized trigger techniques have been used to identify adverse events in other settings by selecting specific charts for more detailed review. In our preliminary work, we have developed and tested two computerized trigger tools to identify charts that may contain evidence of diagnostic errors in primary care. We believe that refining these tools and integrating them with additional variables could lead to a higher detection rate for diagnostic errors. In additional preliminary work, we tested a computerized method that could be used as a new trigger tool to detect diagnostic errors related to abnormal test result follow-up. We now propose to validate this potential trigger tool. We also propose to expand our research beyond the VA to a large primary care network in a practice based research network in Central Texas. These settings will include internal medicine and family medicine, academic and nonacademic practices, urban and rural patients, and significant racial, gender, ethnic, age, and socioeconomic diversity. Our specific aims are: 1) To apply and improve computerized triggers based on visit patterns to detect, measure, and learn from diagnostic errors in diverse primary care settings. 2) To test whether a method of computerized tracking for abnormal test results that are potentially lost to follow-up can be used as a trigger to identify diagnostic near-misses in primary care. In Aim 1 we will query clinical repositories with triggers and electronically collect additional clinical data (variables) about primary care visits. To test the utility of triggers, we will compare their positive predictive values (PPV) with controls, a random sample of visits that do not meet trigger criteria. Trained, blinded chart reviewers will verify the presence of diagnostic errors in the two subsets. To improve our trigger we will use a logistic regression model to test the additive PPV of integrating the trigger with specific independent clinical variables. In Aim 2, we will electronically track and identify records of patients for further chart reviews based on a new potential trigger. We will test the validity of this trigger tool for detecting diagnostic errors related to test results lost to follow-up by comparing the PPVs of the triggered and control subsets. Refining and validating triggers for diagnostic errors in diverse primary care settings would set the stage for their use nationally in quality measurement and improvement activities.