Effect of narrowing scan area on performance of China Infectious Diseases Automated-alert and Response System on hand, foot and mouth disease cluster detection

Objective To compare the performance of China Infectious Diseases Automated-alert and Response System (CIDARS) for hand, foot and mouth disease (HFMD) outbreak/cluster detection with different scan areas. Methods Information of all the signals generated by CIDARS from January 1,2014 to December 31,2016 in Pudong New District were collected and descriptively analyzed. Performance of CIDARS was evaluated, including sensitivity and false alert ratio, based on real HFMD clusters during the same period. The assessment of the early warning model currently used in CIDARS was conducted by using the time series data of daily reported HFMD cases in three selected communities, and outbreak detection performance with different scan area were compared in each community. Results From 2014 to 2016, a total of 393 signals of HFMD were generated by CIDARS in Pudong, and 182 HFMD clusters were successfully detected. During the same period, 726 HFMD clusters occurred in Pudong. The sensitivity and false alert ratio of CIDAR was 24.97% and 53.69%. The sensitivities to detect large, medium and small clusters were 100.00%, 45.30% and 16.05% respectively. After adjusting the scan area of early warning model in CIDARS from district to community level, the HFMD cluster detection sensitivities in the three selected communities increased obviously from 25.93%, 13.04% and 50.00% to 88.89%, 78.26% and 100.00%, with a slight increase of false alert rate. Conclusions The HFMD cluster detection performance of CIDARS was unsatisfied and differed with cluster size and scan area. Cluster detection performance of CIDARS could be improved through method optimizing, such as narrowing scan area of early warning models.