Abstract
The Centers for Disease Control and Prevention (CDC) defines influenza-like illness (ILI) for its sentinel providers as fever (temperature >=100.5[degrees]F or 37.8[degrees]C) and a cough and/or a sore throat in the absence of a known cause other than influenza. For electronic disease surveillance systems, classifying ILI with clinical data that identify only individual aspects of the case definition may add excessive levels of unwanted noise to the system; however, the capability to analyze a patient's body temperature along with other available clinical data (International Classification of Diseases, Ninth Revision codes) could improve diagnostic precision and more accurately classify cases of ILI in a syndromic surveillance system. Developing Boolean algorithms to properly classify true cases of influenza plays an important role toward understanding accurate levels of disease in a community and can also be a key tool for allocating urgent prophylaxis such as antiviral medications during severe outbreaks and pandemics. Results for this study show that elevated body temperature was 40% efficient in correctly predicting laboratory-positive confirmations of influenza (sensitivity) but at the same time was 76% efficient in ruling out influenza (specificity) in the group of sampled members who were tested for influenza.