Terminology Services to Reduce Avoidable CT Imaging

 描述（由申请人提供）：本项目的目标是为自动警报系统创建基础，该系统将在订购计算机断层扫描（CT）扫描之前通知提供者，如果患者之前在任何参与健康信息交换（HIE）的设施进行过类似扫描。实现这一目标将有助于避免重复CT扫描，这将解决一个主要的患者安全问题，并降低医疗保健成本。该项目所面临的挑战是，当重复的、可能避免的CT存在于其他站点时，这是困难的，并且目前不可能建立警报系统，因为HIE中的不同提供者，包括医院和放射设施，使用不同的专有CT代码。如果我们要了解CT扫描重复的频率并创建一个减少冗余的系统，则将这些代码映射到常见的受控放射学术语至关重要。具体目标是：1）评估将参与HIE的40个研究中心的专有CT代码映射到新组合的LOINC/RadLex术语的可靠性和全面性; 2）创建/调整身体区域的解剖框架，并将所有LOINC CT代码映射到该框架; 3）评价使用HIE范围数据检测CT“交叉”率。将通过测量来自Healthix HIE的去识别数据集的kappa评定者间可靠性，评估用于创建一致术语框架的映射可靠性。将测量概念类型和概念标记覆盖率，以评估合并的LOINC/RadLex代码是否全面覆盖每个研究中心和整个Healthix HIE的CT术语。在目标2中创建的映射表将在互联网上自由传播，该映射表将所有CT代码映射到经LOINC/Radlex放射学委员会验证的解剖框架。将在目标3中评估这项工作的潜在影响，目标3将测量有多少独特的患者在HIE的多个部位具有相同的CT、相似的CT（相同的身体区域）或邻近的CT（邻近的身体区域），并证明与每个提供者的“部位特定”数据相比，使用HIE范围的数据时检测这些类别的潜在重复CT的能力增加。

 DESCRIPTION (provided by applicant): The goal of this project is to create the foundations for an automated alerting system that will notify providers before ordering a computer tomography (CT) scan if a patient has had prior similar scans at any of the facilities participatin in a health information exchange (HIE). Achieving this goal will help avoid duplicate CT scans, which will address a major patient safety issue and decrease healthcare costs. The challenge addressed by this project is that it is difficult when duplicate, potentially avoidable CTs exist a other sites, and a building an alerting system is currently not possible because the diverse providers in the HIE, including hospitals and radiology facilities, use different proprietary CT codes. Mapping these codes to a common, controlled radiology terminology is essential if we are to understand how often CT scans are duplicated and create a system to reduce redundancy. The specific aims are 1) Assess the reliability and comprehensiveness of mapping proprietary CT codes from 40 sites participating in an HIE to a newly combined LOINC/RadLex terminology; 2) Create/adapt an anatomical framework for body regions and map all LOINC CT codes to it; 3) Evaluate the use of HIE-wide data to detect CT "crossover" rates. The reliability o mapping to create a consistent terminology framework will be assessed by measuring kappa inter-rater reliability on a de-identified dataset drawn from the Healthix HIE. Concept-type and concept-token coverage will be measured to assess whether the combined LOINC/RadLex codes comprehensively cover CT terms at each site and across the entire Healthix HIE. The mapping table created in aim 2, which will map all CT codes to an anatomical framework validated by the LOINC/Radlex Radiology Committee, will be freely disseminated on the Internet. The potential impact of this work will be assessed in Aim 3, which will measure how many unique patients have same CT, similar CT (same body region), or proximate CT (adjacent body region) across multiple sites in the HIE, and demonstrate the increased ability to detect these categories of potentially duplicate CTs when using HIE-wide data compared to each provider's "site-specific" data.