基于BCPNN法的阿格列汀药物警戒信号的挖掘

Pharmacovigilance signal detection of alogliptin based on Bayesian confidence propagation neural network Method

目的:挖掘、评价阿格列汀上市后不良反应信号,为临床安全用药提供参考。方法:采用贝叶斯置信传播神经网络法对美国药品不良反应(ADR)报告系统中的数据进行分析,信号生成条件定为信息成分IC的95%CI下限> 0且报告数a≥3。结果:共检索到以阿格列汀为怀疑药品或联用药品的ADR报告1 298例,男性患者(53. 85%)多于女性(39. 21%),年龄分布以65～84岁最多(31. 59%)。大部分ADR发生在用药1年内。报告数排名前10的ADR依次为脑梗死、恶心、发热、间质性肺病、贫血、肺炎、病情恶化、腹泻、皮疹和呕吐。共检测到115个警戒信号,涉及21个系统器官分类;有临床参考意义的高风险信号包括间质性肺疾病(n=35,IC=3. 36)、横纹肌溶解症(n=22,IC=2. 57)等。结论:对ADR自发呈报数据库进行数据挖掘,能快速提出警戒信号,为信号验证、评价提供基础。

Objective: To detect and evaluate the safety signals of alogliptin based on spontaneous reporting system to provide reference for clinical use. Methods: The Bayesian confidence propagation neural network method was used to analyze the FDA Adverse Event Reporting System database. The signal generation condition was defined as the lower limit of 95% confidence interval of information component (IC) should be positive and the number of reports in any given drug-adverse reaction pair should be greater than 3. Results: A total of 1 298 reports were retrieved from the database in which alogliptin was used as suspected drug or concomitant drug. Male patients (53. 85%) were more than females (39. 21%),and the largest age group was 65 to 84 year old. Most adverse drug reactions (ADRs) occurred within one year after the beginning of the treatment. The top 10 adverse events reported were cerebral infarction,nausea,pyrexia,interstitial lung disease,anemia,pneumonia,condition aggravated,diarrhea,rash and vomiting. A total of 115 pharmacovigilance signals were detected,distributing in 21 system organ classes;high-risk signals were interstitial lung disease (n = 35,IC = 3. 36),rhabdomyolysis (n =22,IC = 2. 57),et al. Conclusion: The data mining of spontaneous reporting system can detect pharmacovigilance signals efficiently and provide the basis for signal verification and evaluation.