基于电子病例报告表在医院医疗信息平台自动抓取儿童肺炎真实世界数据的方法学探讨

Methodological discussion on automatically capturing real-world data on pediatric pneumonia from electronic case reports form based on the hospital medical information platform

背景在医院医疗信息中自动抓取真实世界数据用于临床研究和管理备受关注。目的构建电子病例报告表(eCRF)数据采集平台,实现真实世界数据服务于临床研究和疾病管理。设计数据建模。方法系统文献检索和提取儿童肺炎变量形成变量池,审核专家根据变量池确定核心变量,开发临床数据交换标准协会eCRF表单。验证和修正变量后,构建和完善平台数据中采集测试模型,测试数据采集能力,并行测试数据采集时效性和系统安全性分析。主要结局指标儿童肺炎采集数据的应答率。结果审核专家选择了7部指南、4部诊疗建议、10部专家共识和9部经典专著作为变量的来源,确定了18个域及其335个可自动采集的儿童肺炎变量,经5轮测试评估和改进修正结果。人口学信息模块(8个变量)第1~5轮均100%达标。结构化信息模块(175个变量)第1轮89.7%达标,第2~5轮均100%达标。文本数据模块(152个变量)第1轮50.0%达标,未达标变量数21个,未出现变量数70个;第5轮90.1%达标,未达标变量数7个,未出现变量数8个。总体达标率为95.5%(320/335)。检测变量均能够在数据产生的24 h内实现自动采集,均可实现数据的核查和锁定功能。结论构建的儿童肺炎eCRF数据采集平台可以实现在医院医疗信息中自动抓取真实世界数据,服务于临床研究和疾病管理。

Background Automatic capture of real-world data from hospital medical information system for clinical research and management has always been paid much attention.Objective To construct an electronic case report form(eCRF)data collecting platform,and to apply the real-world data to clinical research and disease management.Design Data modeling.Methods Variables for pediatric pneumonia were extracted from the literature sought by a systematic search in databases to form a variable pool.Experts for reviewing were required to identify core variables from the pool for the development of eCRF following the rule of Clinical Data Interchange Standards Consortium(CDISC).After variable verification and correction,data collection test model was built and improved by testing the data collection ability,and the timeliness and system security were analyzed in parallel.Main outcome measures Response rate of eCRF for pediatric pneumonia data.Results Based on the selected 7 guidelines,4 articles on diagnosis and treatment recommendations,10 papers of expert consensus and 9 classic monographs as the original source of variables,experts for reviewing identified 18 domains and 335 pediatric pneumonia variables for automatic collection.The results were refined after 5 rounds of testing and evaluation.The module of demographic information(8 variables)was 100%up to the standard in round 1 to 5.The module of structured information(175 variables)was 89.7%up to the standard in the first round,and 100%in the second to fifth rounds.The module of text data(152 variables)was 50.0%up to the standard in round 1 with 21 variables failing to meet the standard and 70 variables failing to be captured and in round 5,the rate was 90.1%,and the number of variables failing to meet the standard and to appear decreased to 7 and 8,respectively.The overall compliance rate was 95.5%(320/335).The detected variables could be automatically collected within 24 hours of data generation.Data verification and locking could also be done at the same time.Conclusion The eCRF data collection platform can automatically capture real-world data from hospital medical information system to facilitate clinical research and disease management.