Threshold level of Influenza-Like-Illness (ILI) as the early warning signal for seasonal influenza and early detection of outbreak in clusters, Thailand

Influenza are highly infectious disease, often cause the outbreak in clusters and epidemic. Timely surveillance to provide early warning has always been challenging. Influenza-Like-Illness (ILI) had been initiated and implemented in Thailand, to monitor Influenza epidemic and to provide an early warning. However, data on evaluation of ILI for epidemic threshold are limited. We analyzed information to determine the optimal epidemic threshold using 3 data sources: (a) National Diseases Notification (b) Electronic reporting, via SMS, for the proportion of ILI among Out Patient Services in nation-wide network of hospitals in Thailand and (c) Reporting of outbreak in clusters in database of outbreaks investigation and notification of clusters into event-based surveillance from nation-wide of Surveillance and Rapid Response Team (SRRT). This study determined the ILI level at 5.56% with 95.00% confidence interval 5.41 and 5.71, as the epidemic threshold for early warning of outbreaks. For practical purpose, this study recommends using the cut-off level of ILI at 5.00%. At this level, the sensitivity associated with detecting clusters of 3 cases of Influenza or more was 85.40% and the specificity of the test was 79.50%. Use ILI with event-based surveillance increase the sensitivity and specificity of Influenza surveillance.