医保真实世界数据质量评估研究——以谈判药品“纳入支付范围的药品费用”为例  

Research on Quality Assessment of Real World Data on Medical Insurance——Taking the Example of“Drug Costs Included in the Payment Scope”of Negotiated Drugs

作　　者：刘雨欣 侯宜坦 左后娟 罗毅[1] 钟刘琪 文小桐 马露[4,5] 杨莹[6] 崔丹[4,5] 毛宗福[5,7] 

机构地区：[1]华中科技大学同济医学院附属同济医院,武汉430030 [2]郑州大学政治与公共管理学院,郑州450001 [3]新疆医科大学公共卫生学院,乌鲁木齐830017 [4]武汉大学公共卫生学院,武汉430071 [5]武汉大学全球健康研究中心,武汉430071 [6]华中科技大学同济医学院护理学院,武汉430030 [7]武汉大学董辅礽经济社会发展研究院,武汉430071

出　　处：《中国医疗保险》2025年第1期34-41,共8页China Health Insurance

基　　金：国家自然科学基金“医保战略性购买视角下慢性病多层次门诊用药保障的经济效应与健康效应研究”(72404098);中国博士后科学基金项目“价值导向下慢性病门诊用药保障与老年人健康产出:作用机制与政策优化”(2024M761028);湖北省博士后创新人才培养项目(2024HBBHCXB019);国家资助博士后研究人员计划(GZC20240534)。

摘　　要：目的:以谈判药品“纳入支付范围的药品费用”的统计过程为例,分析归纳医保数据质量问题特征,为医保数据清洗与应用提供参考。方法:通过某地级市医保结算系统采集2018年1月至2024年9月间国家谈判准入协议期内产品结算记录(231305条)为目标数据集,辅以网络检索多源形成若干辅助校验数据集,构建包含完整性、规范性、一致性三个维度的目标数据质量评估框架。结果:数据完整性维度,主要表现为疾病诊断编码、名称缺失,所涉及结算记录条目数占比29.6%和29.7%。规范性维度,药品编码与名称存在一对多的不规范问题,结算条目数占比66.2%;医药机构存在编码内容异常(0.7%)、编码与名称对应关系不规范(43.3%)的问题;疾病诊断中60.3%结算条目存在编码样式不符合规范;患者身份中包含48名非基本医疗保险参保人,涉及405结算条目。一致性维度,对照国谈支付标准发现4.9%结算条目存在支付价格超出±10%范围;药店购药记录中1.4%条目对应的零售药店超出“双通道”资质认定范围;78条记录不符合医保限定支付范围对患者年龄或开具处方医院条件的规定。结论:医保数据可能存在数据缺失、医保信息业务编码应用不规范、患者身份统计混淆以及疑似医保支付不当导致的逻辑错误等质量问题。需加快探索相适应、统一、可操作化的统计规范与数据清洗规则,推动医保数据的决策效能发挥。Objective:The paper analyzed and summarized the characteristics of medical insurance data quality by taking the statistical process of“drug costs included in the payment scope”of negotiated drugs as an example,providing references for medical insurance data cleaning and application.Methods:Through the medical insurance settlement system of a city,this study collected medical insurance settlement records of national negotiated products within the agreement period(231305 items)from January 2018 to September 2024 as the target dataset,supplemented with several auxiliary datasets by Internet search.A data quality assessment framework covering completeness,standardization,and consistency was constructed.Results:In the completeness dimension,the main data problems were missing disease diagnosis codes and names,involving 29.6%and 29.7%of record items.In the standardization dimension,66.2%of record items had non-standard one-to-many correspondence between drug codes and names.There were problems of abnormal medical institution coding content(0.7%)and irregularities in the correspondence between medical institution codes and names(43.3%).60.3%of record items had disease diagnosis codes that did not conform to standards.There were 48 non-basic medical insurance participants,involving 405 record items.In the consistency dimension,4.9%of record items had payment prices that exceeded the payment standard by±10%;1.4%of pharmacy purchase data items corresponded to retail pharmacies outside the“dual-channel”recognition scope;78 record items did not qualify for medical insurance coverage of patient age or prescription hospital conditions.Conclusion:There are quality issues in medical insurance data,such as missing data,irregular application of medical security information coding system,confusion of patient identity,and logic error due to suspected improper payment of medical insurance.It is suggested to accelerate the exploration of appropriate uniform and operational statistical and cleaning rules which are suitable for

关 键 词：医保数据 数据质量 医保谈判 医保支付范围 真实世界数据 

分 类 号：F840.684[经济管理—保险]