基于Apriori算法分析2021年山东省医疗器械不良事件的关联性

Association analysis of medical device adverse events in Shandong Province in 2021:Apriori algorithm

目的 医疗器械不良事件监测是医疗器械上市后风险管理的重要手段。本研究旨在基于Apriori算法分析2021年山东省医疗器械不良事件的关联性。方法 对2021年山东省各监测机构上报的63 041起不良伤害事件,按广义医疗器械分类划分为三类(无源医疗器械、有源医疗器械以及体外诊断试剂)医疗器械不良事件进行描述分析。采用关联规则挖掘中的Apriori模型,挖掘出与不良事件相关的器械类别、使用科室、医院类别、是否超期使用以及上报单位所属地区,探索医疗器械不良事件关联风险。结果 不良事件中包含有源医疗器械20 564起、无源医疗器械42 181起及体外诊断试剂296起。其中,无源医疗器械不良事件发生最多的地级市为烟台市(5 711起)、科室为手术室(835起)、医院类别为二级综合医院(5 320起);有源医疗器械不良事件发生最多的地级市为济南市(2 271起)、科室为手术室(196起)、医院类别为三级综合医院(1 108起);体外诊断试剂不良事件发生最多的地级市为烟台市(42起)、科室为儿科(6起)、医院类别为一级医院(42起)。根据关联规则可知,一级医院中卫生院使用未超期无源器械关联规则支持度最高,而在超期使用产品中,日照市的三级综合医院重症监护室使用有源器械发生不良事件支持度最高。结论 各不良事件发生与各级别医院及使用科室存在强关联,而在超期产品使用中也存在类似问题,这可为各监测单位及医疗机构深化管理医疗器械提供指导。

Objective Monitoring of adverse events related to medical devices is an important means of post-marketing risk management of medical devices. The purpose of this study is to analyze the correlation of adverse events related to medical devices in Shandong Province in 2012 based on the Apriori algorithm. Methods A descriptive analysis was conducted on 63 041 adverse injury events reported by monitoring institutions in Shandong Province in 2021, which were classified into three categories based on the classification of generalized medical devices, including passive medical devices, active medical devices and in vitro diagnostic reagents. The Apriori model for association rule mining was used to investigate the risks associated with adverse events related to medical devices by mining the device category, department of use, hospital category, use of expired devices, and the region to which the reported hospital belonged. Results Of the 63 041 adverse events, 20 564 were related to active medical devices, 42 181 to passive medical devices, and 296 to in vitro diagnostic reagents. As for the adverse events related to passive medical devices, the city, department and hospital which had the most adverse events were Yantai(n=5 711), operating room(n=835), and secondary general hospitals(n=5 320). As for the adverse events related to active medical devices, the city, department and hospital which had the most adverse events were Jinan(n=2 271), operating room(n=196), and tertiary general hospitals(n=1 108). As for the adverse events related to in vitro diagnostic reagents, the city, department and hospital which had the most adverse events were Yantai(n=42), the pediatric department(n=6), and primary hospitals(n=42). According to the association rule, the use of unexpired passive devices in health centers among primary hospitals received the most support, while for the use of expired devices, the most support was found in the use of active devices in the intensive care units of tertiary general hospitals in Rizhao. Conclusion The occurrence of adverse events is strongly associated with hospitals and departments at all levels, and similar problems also exist in the use of expired products. Our findings provide references for monitoring units and medical institutions to impprove the management of medical devices.