采用报告比值比法挖掘非甾体抗炎药相关肝损伤信号

Analysis of risk signals of liver injuries related to nonsteroidal anti-inflammatory drugs using reporting odds ratio

目的：探讨非甾体抗炎药（NSAID）相关肝损伤信号的挖掘及报告比值比（ROR）法在信号挖掘中的应用。方法以“肝”“、胆”为关键词检索国家药品不良反应（ ADR）监测系统2013年1月1日至12月31日接收到的北京市上报的ADR报告，将筛选后的报告中药物与肝损伤因果关系为“肯定”“、很可能”“、可能”的病例纳入肝损伤组，其他所有病例均纳入非肝损伤组。以NSAID为目标药物，其他所有药物为非目标药物，根据ROR计算公式计算NSAID相关肝损伤的ROR及其95%置信区间（ CI），95%CI下限〉1提示出现ADR信号。结果经删重后共14657例患者纳入研究。肝损伤组626例，其中35例为NSAID相关肝损伤；非肝损伤组14031例。在35例NSAID相关肝损伤患者中，与NSAID单方制剂相关者30例，与复方制剂相关者5例，NSAID单、复方制剂及NSAID总体相关肝损伤ROR及其95%CI分别为1.78（1.22-2.61）、1.80（0.78-4.15）、1.76（1.24-2.51）；其中，NSAID单方制剂和总体相关肝损伤95%CI下限〉1，出现ADR信号。35例肝损伤共涉及37种药物，其中单方制剂帕瑞昔布、阿司匹林和复方制剂布洛伪麻、氨酚烷胺致肝损伤ROR及其95%CI分别为8.00（2.03-27.78）、2.45（1.43-4.21）、22.00（1.40-359.32）、和11.22（1.02-123.94），均出现ADR信号。结论对NSAID相关肝损伤的风险应予关注。ROR法有助于挖掘ADR信号，为药物的安全使用提供预警。

Objective To analyze risk signals of liver injuries related to nonsteroidal anti-inflammatory drugs（ NSAID）and application of reporting odds ratio（ ROR）in data mining. Methods A search of adverse drug reaction（ ADR ） reports in Beijing of national monitoring system of adverse drug reaction from January 1st 2013 to December 31st 2013 was conducted using keywords ＂liver＂ and ＂gallbladder＂ . Of these filtered reports,the cases whose causal relationship of drugs and liver injuries was judged as＂positive＂,＂likely＂,or ＂possible＂ were enrolled into liver injury group and the rest cases were all enrolled into non-liver injury group. Using NSAID as the target drugs and the other drugs as non-target, ROR and its 95% confidence interval（ CI）of liver injuries related to NSAID were calculated according to formula of ROR. The lower limit of 95% CI〉1 was regarded as suggestive of ADR signal. Results After removing duplication,14 657 patients were enrolled in the study,which comprised liver injury group 626 patients including 35 cases of liver injuries due to NSAID and non-liver injury group 14 031 patients. Of the 35 patients with liver injuries related to NSAID,30 patients were associated with single-preparation NSAID and 6 patients were associated with compound-preparation. The ROR levels and their 95% CI of single-preparation NSAID,compound-preparation NSAID,and general NSAID were respectively 1. 78 （1. 22-2. 61）,1. 80（0. 78-4. 15）,and 1. 76（1. 24-2. 51）. Both of the lower limits of 95% CI of live injuries related to single-preparation NSAID and general NSAID were higher than 1 and there were ADR signals. Thirty-seven kinds of drugs were involved by 35 cases of liver injuries. Of them,the ROR levels and their 95% CI of single-preparation parecoxib, single-preparation aspirin, compound-preparation ibuprofen and pseudoephedrine,and compound-preparation parace-tamol and amantadine hydrochloride were respectively 8. 00（2. 03-27. 78）,2. 45（1. 43-4. 21）,22. 00（1. 40-359. 32）,and 11. 22（1. 02-123. 94） and there were ADR signals. Conclusions Attention should be paid to liver injuries related to NSAID. The method of ROR can help analyze ADR signals and provide useful early-warning of drug safety.