论文部分内容阅读
目的:探索清热类中药注射剂不良反应发生的流行病学特点。方法:通过系统检索1979~2015年期刊文献,采集清热类中药注射剂不良反应详细个案报告,在此基础上应用Microsoft ACCESS构建数据库,进而以Clementine 12.0数据挖掘平台,综合应用贝叶斯网络、神经网络、关联规则Apriori算法和决策树(CART)算法开展数据挖掘研究。结果:共检索符合纳入条件的清热类中药注射剂不良反应详细个案1 315例,其中321例不良反应发生在用药开始5min内,有993例不良反应发生在用药开始30min内。不良反应涉及人体多个系统,皮肤损害占29.27%,过敏性休克占27.30%,呼吸系统损害占16.42%,神经系统损害占9.80%,消化系统损害占8.59%,循环系统损害占3.57%,眼部损害占3.42%等。数据挖掘结果显示不同的注射剂不良反应临床表现相似,不良反应临床表现也与原发疾病类、过敏史、配液用量有关。结论:清热类中药注射剂不良反应发生是在临床用药因素、患者体质因素和药品自身因素共同作用下的复杂表现,其发生规律尚需更大样本量的数据分析验证。
Objective: To explore the epidemiological characteristics of adverse reactions of heat-clearing traditional Chinese medicine injection. Methods: Through systematically searching the literature from 1979 to 2015, we collected the detailed case reports of adverse reactions of Qingrejiedu injection. On the basis of this, we used Microsoft ACCESS to build a database, and then use Clementine 12.0 data mining platform to integrate Bayesian network, neural network , Association rules Apriori algorithm and decision tree (CART) algorithm to carry out data mining research. Results: A total of 1 315 cases of adverse reactions of Chinese traditional medicine injections were obtained. The adverse reactions occurred in 321 cases within 5 minutes after the start of medication and in 993 cases within 30 minutes after the start of medication. Adverse reactions involving multiple human systems, skin damage accounted for 29.27%, 27.30% of anaphylactic shock, respiratory system damage accounted for 16.42%, nervous system damage accounted for 9.80%, accounting for 8.59% of digestive system damage, circulatory damage accounted for 3.57%, eye Ministry of damage accounted for 3.42% and so on. Data mining results showed that different clinical manifestations of adverse reactions of injection similar, the clinical manifestations of adverse reactions and the primary disease, allergy history, with the amount of fluid. Conclusions: Adverse reactions of heat-clearing traditional Chinese medicine injections are complicated behaviors under the combination of clinical medication factors, patient’s constitutional factors and drug’s own factors. The occurrence of adverse reactions still needs a larger sample of data analysis and verification.