Syndromic surveillance for influenza in Tianjin, China: 2013–14

Abstract

Background: Diverse sources of syndromic surveillance including over-the-counter (OTC) drug sales, hospital and school-based influenza-like illness (ILI) and Baidu search queries estimate influenza activity in Tianjin, China. The purpose of this study was to determine which syndromic surveillance systems had the strongest correlation with laboratory-confirmed influenza activity. Methods: Data were obtained from sentinel hospitals and laboratories; sentinel hospitals also reported percentage of ILI. OTC sales and school-based ILI absentee data were provided by public pharmacies and schools. Baidu search queries for influenza surveillance were analyzed. Spearman correlation analysis examined correlations of syndromic systems with laboratory-confirmed data. Results: Syndromic data for hospital ILI%, OTC sales and school-based ILI correlated well with laboratory data (r = 0.732, 0.490 and 0.693, respectively; P < 0.05). Baidu, the predominant Chinese Internet service, searches for 'influenza', 'cough' and 'fever' correlated best with laboratory-confirmed activity; queries for 'fever' were strongest (r = 0.924, P < 0.001). Correlations between school-based ILI and laboratory-confirmed influenza increased from 0.693 to 0.795 after a 1-week lag (P < 0.05). Conclusions: A Baidu query of 'fever' provided the strongest correlation to laboratory surveillance. School-based ILI absence reporting detected influenza virus activity 1 week earlier than laboratory confirmation. Use of diverse syndromic surveillance systems in conjunction with traditional surveillance systems can improve influenza surveillance.