真实世界数据质量评价指标研究

打开文本图片集

【关键词】真实世界研究；真实世界数据；质量评价；数据管理；数据治理【中图分类号】 【文献标识码】A

【Abstract】 Objective To develop an index system for the quantitative evaluation of real-world data quality. Methods Computerized searches were conducted acros literature databases，including PubMed， Europe PMC，WanFang Data，and CNKI，as wellas the official websites of 13 international academic organizations and regulatory agencies.Based on predefined inclusion and exclusion criteria，literature， guidelines，and standards related to real-world data （RWD） quality assessment， were ultimately selected. Through thematic induction and content summarization，core elements from the selected documents and guidelies were extracted to construct an initial set of indicators.These indicators were subsequently revised and refined using expert consultation methods.Results A total of 35 publications were included.Incorporating theresearch team's practical experience，this paper explored data management processes for prospective data and data governance processes for retrospective data. Corresponding data quality evaluation indicators were established accordingly.In both frameworks，data quality was categorized into three primary dimensions： raw data quality， processquality，and outcome quality. These dimensions were further organized according to the chronological sequence of clinical research.For the data management component，15 secondary indicators and 43 tertiary indicators were developed.For the data governance component， 13 secondary indicators and 29 tertiary indicators were formulated.Conclusions The evaluation indicators developed through literature analysis and expert consultation method demonstrate a considerable degree of scientific rigor and feasibility. They can serve as valuable references for regulatoryauthorities，sponsors，and researchers involved in the quality assessment of real-world data.

【KeyWords】Real world study; Real world data; Quality evaluation; Data management; Data governance

真实世界数据（realworlddata，RWD）指从多种来源收集的与研究对象健康状况和/或常规诊疗及保健相关的数据[1-2]，其核心是在实际医疗环境下形成和收集的数据[3]。（剩余15230字）