摘要
目的评价累积和(Cumulative Sum,CUSUM)预警模型应用于医院感染性腹泻暴发预警中的效果,确定最优阈值。方法通过2016-2017年上海市徐汇区医院感染性腹泻症状监测数据(1 532 030人·天数)建立CUSUM C1、C2、C3预警模型,将预警信号与经流行病学调查确定为暴发事件数据比较,以灵敏度、特异度、阳性预测值、及时性等指标评价最优阈值下的预警效果。结果监测点医疗机构共报告感染性腹泻病例306例,发病率为0.20‰(1.99/万人·天),暴发的事件14起。H=0.80是C1模型的最优阈值(Se=93.00%,Sp=92.00%,PPV=21.00%),H=0.40是C2模型的最优阈值(Se=93.00%,Sp=87.00%,PPV=22.00%),H=3.20是C3模型的最优阈值(Se=93.00%,Sp=92.00%,PPV=53.00%),均可及时预警,重复预警信号剔除的计算规则能在不降低灵敏度的情况下提高特异度。结论 CUSUM预警模型可用于医院感染性腹泻暴发预警,为及早发现暴发疫情及时采取控制措施提供预警信息。
OBJECTIVE To evaluate the prediction effects of CUSUM algorithms for the outbreak of acute infectious gastroenteritis in health-care facilities.METHODS Outbreak of acute infectious gastroenteritis between 2016 and 2017 were confirmed by epidemiological survey,and the timing of identified outbreaks was compared with signals generated from three different types of CUSUM-based automated monitoring algorithms（C1,C2 and C3）.The best threshold was selected by comparing the sensitivity,specificity,positive predictive value and time to detection.RESULTS 306 nosocomial gastroenteritis cases were reported from 2016 to 2017 with an incidence rate of 1.99 per 10,000 inpatient-day,and 14 defined outbreaks occurred.The optimized thresholds of C1,C2,C3 were 0.8（Se=93.00%,Sp=92.00%,PPV=21.00%）,0.4（Se=93.00%,Sp=87.00%,PPV=22.00%）,3.2（Se=93.00%,Sp=92.00%,PPV=53.00%）.Time to detection was less than 2 days.The method to reduce false flags was useful to enhance specificity without changing the sensitivity.CONCLUSIONCUSUM model can appropriately predict the outbreak of acute infectious gastroenteritis in health-care facilities,and can be used to detect outbreak at an early time.
作者
钱子煜
张越文
马小越
严冰清
王琰
QIAN Zi-yu;ZHANG Yue-wen;MA Xiao-yue;YAN Bing-qing;WANG Yan(Xuhui Center for Disease Control and Prevention,Shanghai 200237,China)
出处
《中华医院感染学杂志》
CAS
CSCD
北大核心
2018年第19期3011-3015,共5页
Chinese Journal of Nosocomiology
基金
上海市徐汇区科委医学科研基金资助项目(SHXH201429)
关键词
CUSUM模型
医院感染
感染性腹泻
暴发
预警
CUSUM algorithms
Nosocomial infection
Acute Gastroenteritis
Disease Outbreaks
Forecasting
