论文部分内容阅读
目的利用数据挖掘技术中的4种常见的比例失衡分析方法(Disproportionality Analysis,DPA),分析无细胞百日咳-白喉-破伤风联合疫苗(Diaphtheria,Tetanus and Acellular Pertussis Combined Vaccine,DTa P)预防接种异常反应的可疑信号。方法利用DPA频数法中的比例报告比(Proportional Reporting Ratio,PRR)和报告比值比(Reporting Odds Ratio,ROR)及贝叶斯法中的贝叶斯置信传播神经网络(Bayesian Confidence Propagation Neural Network,BCPNN)和模糊贝叶斯伽马-泊松收缩论(Empirical Bayes Gamma-Poisson Shrinker,GPS)方法,对2011~2013年中国疑似预防接种异常反应信息管理系统中报告发生的有明确诊断的DTa P预防接种异常反应进行分析,检测其预防接种异常反应的可疑信号。结果 20 377例有明确诊断的预防接种异常反应个案参与分析,其中DTa P占16.13%。使用PRR和ROR方法提示,2011~2013年DTa P有6个可疑预防接种异常反应信号,使用BCPNN和GPS方法提示,2011~2013年DTa P有4个可疑预防接种异常反应信号。4种方法发现的可疑预防接种异常反应信号集中在接种DTa P后的血管性水肿和无菌性脓肿。结论无菌性脓肿和血管性水肿为已知的DTa P的预防接种异常反应,利用DPA方法发现的可疑信号是风险的警示,但仅仅提示存在报告数量上的关联,是否是真实的安全性问题,需要经过包括医学评估在内的严谨的信号评价过程。
OBJECTIVE To analyze the abnormality of vaccination against diaphtheria (Tetanus and Acellular Pertussis Combined Vaccine) by using four common methods of DP (Disproportionality Analysis) in data mining, Suspicious signal Methods Based on the DPR frequency method, the ratio of Proportional Reporting Ratio (PRR) to Reporting Odds Ratio (ROR) and the Bayesian Confidence Propagation Neural Network (BCPNN) ) And fuzzy Bayesian Gamma-Poisson Shrinker (GPS) methods were used to evaluate the DTa P prevention in China from 2011 to 2013 Abnormal response to vaccination analysis, detection of abnormal response to vaccination suspicious signals. Results A total of 20 377 cases of vaccination abnormalities with definite diagnosis were involved. Among them, DTa P accounted for 16.13%. Using PRR and ROR methods, there were 6 suspicious immunization abnormalities in DTa P between 2011 and 2013. Using BCPNN and GPS methods, there were 4 suspicious immunization abnormalities in DTa P during 2011-2013. Susceptible vaccinations found in 4 different ways Abnormal response signals focus on vascular edema and aseptic abscess following DTa P inoculation. Conclusions Aseptic abscesses and angioedema are known abnormalities in the vaccination of DTa P, and the suspicious signals detected by the DPA method are risk warnings. However, it is a real safety issue to merely indicate whether there is an association with the number of reports , Need to go through a rigorous process of signal evaluation, including medical assessment.