Evaluating the Performance of an Outbreak- Detection Algorithms using Semi-Synthetic Approach: Cumulative Sum Algorithm

• Published in: Iranian journal of epidemiology Vol. 9; no. 2; pp. 29 - 38
• Main Authors: M Karami, H Soori, Y Mehrabi, AA Haghdoost, MM Gouya, N Esmailnasab
• Format: Journal Article
• Language: English; Persian
• Published: Tehran University of Medical Sciences 2013
• Subjects: CUSUM; Iran; Measles; Outbreak Detection Method; Semi-synthetic Simulation; Surveillance System
• ISSN: 1735-7489; 1735-7489

Background & Objectives: Timely response to emerging diseases and outbreaks are a major public health and health systems priority. There are few published studies that evaluate the performance of cumulative sum (CUSUM) on identical data using semi- synthetic simulation approach. This study was undertaken to determine the performance of the CUSUM in timely detection of 831 days of simulated outbreaks. Methods: We evaluated the performances of the CUSUM as an outbreak detection method on simulated outbreaks injected to daily counts of suspected cases of measles as baseline data in Iran between 21 March 2008 till 20 March 2011. Data obtained from the Iranian national surveillance system. The performance of algorithms was evaluated using sensitivity, false alarm rate, likelihood ratios and Area under the Receiver Operating Characteristic (ROC) curve. Results: Generally the sensitivity of the CUSUM algorithm in detecting simulated outbreaks was 50% (95% CI: 47- 54). The corresponding values are disaggregated according to outbreak size, shape and duration. The CUSUM algorithm detected the half of outbreaks after 13.84 days on average. Conclusion: We concluded that CUSUM algorithm performed good in detection of large outbreaks with short periods and poorly in detecting long period outbreaks, particularly those simulated outbreaks that did not begin with a surge of cases.