Background: A single definition of influenza-like illness (ILI) has been recommended by the Australian Influenza Pandemic planning committee for influenza surveillance systems throughout Australia.
Objectives: To examine combinations of clinical symptoms and determine which combination was most likely to predict laboratory-confirmed influenza in adult patients with ILI.
Study design: Sentinel general-practices in Western Australia and Victoria, 1998 and 1999. Univariate analysis and stepwise logistic regression were used to determine significant independent clinical predictors of influenza in adults. Sensitivity, specificity and positive predictive value (PPV) were calculated for various symptom complexes.
Results: The combination of cough, fever and fatigue was both sensitive (43.5-75.1%) and specific (46.6-80.3%) with PPVs ranging from 23.3 to 59.7% in the surveillance data sets from both states. The symptom complex of cough, fever and fatigue was more likely to predict laboratory-confirmed influenza than the three different surveillance case definitions actually used in those years.
Conclusions: We recommend the symptom complex of cough, fever and fatigue as a simple case definition for ILI in influenza surveillance. Accurate identification of influenza activity still requires laboratory confirmation in at least a proportion of cases.