Performance evaluation of CIDARS on dengue fever outbreak detection, China

Abstract

Performance evaluation of CIDARS on dengue fever outbreak detection, China Honglong Zhanga#, Zhongjie Lia#, Shengjie Laia, Archie Clementsb, Wenbiao Hub*, Weizhong Yanga* a Key Laboratory of Surveillance and Early-warning on infectious Disease, Division of Infectious Diseases, Chinese Center for Disease Control and Prevention, Beijing 102206 b School of Population Health, The University of Queensland, Brisbane 4006 QLD, Australia #: These authors contributed equally *Correspondence? Weizhong Yang: yangwz@chinacdc.cn Wenbiao Hu: w.hu@sph.uq.edu.au Background For more effective prevention and control of dengue fever (DF), a China Infectious Disease Automated-alert and Response System (CIDARS) was used to identify possible early warning signals of DF outbreaks at country level in China. Aims This study aims to evaluate the performance of the CIDARS early warning system for DF. Methods In CIDARS a time series moving percentile method was used to detect the outbreak of DF. If the cases of reported DF are over 50 percentile of historical data, an early warning signal will be generated and sent to the local county CDC staff who are responsible for certifying and investigating the signals. In this study, descriptive analysis was used to display the spatial and temporal patterns of DF reported cases and early warning signals of DF during the study period. Sensitivity, specificity and timeliness were employed to assess the performance of DF early warning in CIDARS. Results There were 651 reported DF cases in China between 1st January 2009 and 31st December 2011. 58 signals were mostly occurred between August and December and distributed in Guangdong (35), Zhejiang (7) and Guangxi (4) provinces. 93.10% signals were responded within 24 hours. There were 9 outbreaks of DF recorded during study period with 8 (88.9%) outbreaks detected by CIDARS. The average time of outbreaks detection by CIDARS was 0.63 day. Conclusions CIDARS is a useful platform to detect the DF outbreak at early stage. More accurate algorithms could be tested to reduce the negative signals.