基于逆概率权重法和贝叶斯结构时间序列模型探讨EV71型手足口病疫苗上市对手足口病发病的影响  

Impact of Introduction of the EV71 Vaccine on the Incidence of Hand,Foot,and Mouth Disease:A Study Based on Inverse Probability Weighting and Bayesian Structural Time Series Models

作　　者：刘天[1] 贺启文 宋开发[1] 向泉 阮德欣 秦周 王霄晔 LIU Tian;HE Qiwen;SONG Kaifa;XIANG Quan;RUAN Dexin;QIN Zhou;WANG Xiaoye(Department for Infectious Disease Control and Prevention,Jingzhou Center for Disease Control and Prevention,Jingzhou,434000,China;Health Emergency Center,Chinese Center for Disease Control and Prevention National Key Laboratory of Intelligent Tracking and Forecasting for Infectious Diseases,Chinese Center for Disease Control and Prevention,Beijing 102206,China)

机构地区：[1]荆州市疾病预防控制中心传染病防治所,荆州434000 [2]中国疾病预防控制中心卫生应急中心、传染病溯源预警与智能决策全国重点实验室,中国疾病预防控制中心,北京102206

出　　处：《病毒学报》2024年第6期1281-1289,共9页Chinese Journal of Virology

基　　金：中国疾病预防控制中心公共卫生应急反应机制的运行(项目号:102393220020010000017),题目:应急综合业务管理。

摘　　要：基于时间序列模型定量计算EV71型手足口病(Hand,foot and mouth disease,HFMD)疫苗上市对EV71型和其他型别手足口病发病的影响。首先利用2012-2019年荆州市手足口病病例资料,采用逆概率权重法构建调整后的EV71型、CoxA16型和其它肠道病毒HFMD时间序列。再对2012-2016年调整后荆州市手足口病时间序列拟合贝叶斯结构时间序列模型并预测2017-2019年发病数,采用反事实推断原则定量估计EV71型HFMD疫苗上后各亚型HFMD发病数变化。对不同年龄组进行亚组分析。相对减少(Relative reduction,RR)用于评价下降作用大小。2011-2019年荆州市累计报告HFMD57286例,其中确诊病例11238例,占比19.62%。建立广义线性模型所有参数检验均有统计学意义(P<0.05),AIC=55615,AUC曲线下面积为0.60。2011-2016年EV71型HFMD重构时间序列拟合贝叶斯结构时间序列模型的平均绝对百分比误差(Mean absolute percentage error,MAPE)为16.61%,R2=0.73。基于模型估计接种EV71型手足口病疫苗2017-2019年荆州市EV71型手足口病减少了3577例(95%CI:2893例~4261例),RR=33.70%,95%CI:29.14%~37.71%。分年份来看,2017年未见明显减少(RR=-141.60%,95%CI:-150.83%~-133.03%);2018年、2019年下降相当,RR依次为84.42%(95%CI:81.60%~86.49%)、81.28%(95%CI:24.54%~89.31%)。CoxA16型及其他肠道病毒未见明显减少(RR<0)。分年龄来看,0~2岁、3~5岁、6岁及以上RR依次为68.28%(95%CI:66.81~69.62)、20.97%(95%CI:19.55%~22.34%)、-478.56%(95%CI:-480.59%~-476.55%)。2017-2019年荆州市EV71型HFMD病例在EV71型手足口病疫苗上市后明显下降。We aimed to quantitatively assess the impact of introduction of the enterovirus(EV)71 hand,foot,and mouth disease(HFMD)vaccine on the incidence of EV71 and other subtypes of HFMD using a time-series model.Utilizing HFMD case data from Jingzhou between 2012 and 2019,an adjusted time series for EV71,Coxsackie virus group A type 16(CoxA16),and other enteric viruses causing HFMD was constructed using inverse probability weighting.Bayesian structural time series models were fitted to the adjusted HFMD time series for Jingzhou City,China between 2012 and 2016,and predictions for the years 2017–2019 were made.Quantitative estimates of changes in HFMD incidence for different subtypes following introduction of the EV71HFMD vaccine were obtained using counterfactual inference principles.Subgroup analyses were conducted for different age groups.Relative reduction(RR)was used to evaluate the magnitude of the decrease.From 2011 to 2019,Jingzhou City in China reported a cumulative total of 57,286 HFMD cases,with 11,238 confirmed cases(19.62%).The generalized linear model parameters were significant(P<0.05),with AIC=55615 and an area under the curve of 0.60.From 2011 to 2016,the mean absolute percentage error for the reconstruction of the EV71-type HFMD time series using Bayesian structural time series models was 16.61%,with R2=0.73.Model-based estimates indicated a reduction of 3,577 cases(95%CI:2,893-4,261)in EV71 HFMD in Jingzhou City,China from 2017 to 2019,with RR=33.70%(95%CI:29.14%-37.71%).There was no significant decrease in 2017(RR=-141.60%,95%CI:-150.83%to-133.03%),while substantial reductions were observed in 2018 and 2019,with RR=84.42%(95%CI:81.60%-86.49%)and 81.28%(95%CI:24.54%-89.31%),respectively.CoxA16 and other enteric viruses did not exhibit a significant reduction(RR<0).Age-specific analysis revealed RR=68.28%(95%CI:66.81%-69.62%)for age 0-2years,20.97%(95%CI:19.55%-22.34%)for age 3-5 years,and-478.56%(95%CI:-480.59%to-476.55%)for age≥6 years.Following introduction of the EV71 HFMD vaccine,there was a si

关 键 词：手足口病 EV71疫苗 模型研究 

分 类 号：R373.9[医药卫生—病原生物学] R-05[医药卫生—基础医学]