Parkhurst S M, Blaser M J, Laxson L B, Wang W L
Am J Infect Control. 1985 Feb;13(1):7-15. doi: 10.1016/0196-6553(85)90003-3.
We describe the development of a surveillance system that uses thresholds for detecting nosocomial infections and the potential for nosocomial outbreaks based on data from microbiology laboratory records at our hospital from 1980 to 1982. These records were monitored weekly to determine the number of positive isolates by the identity of the organism and by the site of the culture. A mean of 225 specimens was processed weekly, with 60 of these yielding bacteria or fungi. The average number of organisms isolated per positive culture was 1.46. Two methods of establishing thresholds were compared, one based on percentiles of ranked isolates, the other based on the mean plus intervals of standard error. The system using thresholds established by the standard error method was consistently more useful to highlight weeks for which there was high risk of a problem occurring in less time than were the surveillance techniques traditionally employed.
我们描述了一种监测系统的开发,该系统基于我院1980年至1982年微生物实验室记录的数据,使用阈值来检测医院感染及医院感染暴发的可能性。每周对这些记录进行监测,以根据生物体的种类和培养部位确定阳性分离株的数量。每周平均处理225份标本,其中60份培养出细菌或真菌。每份阳性培养物分离出的生物体平均数量为1.46。比较了两种建立阈值的方法,一种基于排序后的分离株百分位数,另一种基于均值加标准误区间。与传统使用的监测技术相比,使用标准误方法建立阈值的系统在更短时间内突出显示存在问题高风险的周数方面始终更有用。