Abstract
Dengue infection can be challenging to diagnose early in the course of infection before severe manifestations develop, but early diagnosis can improve patient outcomes and promote timely public health interventions. We developed age-based predictive models generated from 2 years of data from an enhanced dengue surveillance system in Puerto Rico. These models were internally validated and were able to differentiate dengue infection from other acute febrile illnesses with moderate accuracy. The accuracy of the models was greater than either the current World Health Organization case definition for dengue fever or a proposed modification to this definition, while requiring the collection of fewer data. In young children, thrombocytopenia and the absence of cough were associated with dengue infection; for adults, rash, leucopenia, and the absence of sore throat were associated with dengue infection; in all age groups, retro-orbital pain was associated with dengue infection.