Application of prospective spatiotemporal rearrangement scan statisticto early warning of mumps clustering epidemic in Shenzhen city
ZHOU Zhi-feng1, LIAO Yu-xue2, LI Xue-yun1, YIN Ling3, XU Yu-cheng1, LIANG Jing1, MEI Shu-jiang2
1. Futian District Center for Disease Control and Prevention, Shenzhen, Guangdong 518040, China; 2. Shenzhen Municipal Center for Disease Control and Prevention, Shenzhen, Guangdong 518055, China; 3. Shenzhen Institutes of Advanced Technology, Shenzhen, Guangdong 518055, China
Abstract:Objective To analyze the effect of prospective spatiotemporal rearrangement scan statistic on the early warning of mumps epidemic in Shenzhen city, and to provide a scientific basis for improving infectious disease surveillance and warning system. Methods Taking Shenzhen city and the street administrative areas as the research scale, the prospective spatiotemporal rearrangement scan model of the SaTScan software was used to conduct daily simulated early warning on the incidence of mumps in Shenzhen city from January 1 to December 31 in 2012. According to the case information reported by the National Disease Surveillance Information Report Management System (hereinafter referred to as the pandemic system) and clustering epidemic events reported by Shenzhen Disease Control Information Management System (hereinafter referred to as the Shenzhen system), the authenticity of simulated early warning information was verified, and the sensitivity, error warning rate, outbreak detection time and advance warning time of the model were calculated. We compared the differences in warning effects of the spatiotemporal rearrangement scan model as well as the time series modeland the space-time sequence modelof China Infectious Disease Automated-alert and Response System (CIDARS). Results A total of 29 mumps aggregation events were reported by the Shenzhen system in 2012. A total of 9,756 mumps analysis cases and 366 times of cumulative analysis from December 3, 2011 to December 31, 2012 in Shenzhen city were enrolled in this study, of which the number of warning signals with P<0.05 was 88, and 16 times were identified as positive. The sensitivity of the model was 55.17% (16/29), and the error warning rate was 81.82% (72/88). The outbreak detection time ranged from 0 to 22 days, with an average of 6.25 days and a median of 4 days. The advance warning time ranged from -12 to 41 days, with an average of 9.75 days and a median of 5.5 days. The time series warning signals were sent out for 319 times, and the positive ones were judged for 2 times, with a sensitivity of 6.90% (2/29) and an error warning rate of 99.37% (315/317). The spatial and temporal sequence warning signals were sent out for 208 times and judged to be positive for 3 times, with a sensitivity of 10.34% (3/29) and an error warning rate of 98.56% (205/208). The number of warningsignals in the spatiotemporal rearrangement scan model was 72.41% and 57.69% lower than that in the time seriesmodel and the space-time sequence model of the pandemic system, respectively. The sensitivity of the spatiotemporal rearrangement scan model was significantly higher than those of the time series model (P<0.001) and thespace-time sequencemodel (P<0.001) of the pandemic system. The error warning rate of the spatiotemporal rearrangement scan model was significantly lower than those of the time series model (P<0.001) and the space-time sequence model of CIDARS (P<0.001). Conclusions The prospective spatiotemporal rearrangement scan model has a better early warning effect in mumps epidemic in Shenzhen city.