Timely detection of influenza outbreaks in Iran: Evaluating the performance of the exponentially weighted moving average

Abstract

Background: Real time detection of influenza outbreaks is necessary by public health authorities. The aim of this study was to determine the performance of the Exponentially Weighted Moving Average (EWMA) in detection of influenza outbreaks in Iran from January 2010 to December 2015. Methods: The EWMA algorithms were applied to weekly counts of suspected cases of influenza (influenza-like illnesses) to detect real outbreaks which have occurred in Iran from January 2010 to December 2015. The performance of EWMA algorithms was measured using sensitivity, specificity, false alarm rate, likelihood ratios and area under the receiver operating characteristics (ROC) curve. Results: Sensitivity of the EWMA for all of occurred outbreaks from 2010 to 2015 was 40% (95% CI: 29%, 50%). The corresponding value of detection of occurred outbreaks in 2010, 2011, 2013, 2014 and 2015 were 50%, 100%, 76%, 64% and 100% respectively. Among different algorithms, EWMA with λ = 0.5 had the best performance (area under the Curve = 100%) for the detection of occurred outbreaks in 2011. Conclusions: Our findings revealed that the performance of the EWMA in the real time detection influenza outbreak in Iran is appropriate. However, public health surveillance systems need to use different outbreak detection methods for detecting aberrations in influenza-like illnesses activity. Keywords: Public health surveillance, Syndromic surveillance, Exponentially Weighted Moving Average, Influenza, Outbreak, Iran