Abstract
BACKGROUND: We evaluated dengue presentation by age, the performance of the 2015 Pan American Health Organization (PAHO) case criteria in identifying dengue cases, and variables to improve specificity. METHODS: Patients with fever ≤7 days (N = 10 408) were recruited from 2 emergency departments from May 2012 through December 2015. Serum samples were tested for dengue, chikungunya, and nasopharyngeal swabs for respiratory viruses. Smoothing splines assessed differences in the frequencies of signs/symptoms by age. Least absolute shrinkage and selection operator regressions identified the variables that best predicted dengue. RESULTS: Among 985 dengue cases, children aged <5 years were least likely to have leukopenia, but most likely to have rash and petechiae. Adults had the highest odds of aches/pains and headaches/retro-orbital pain. The 2015 PAHO criteria had sensitivity of 93% and specificity of 25%. Specificity could be improved by requiring at least 2 of the following criteria: vomiting/nausea, petechiae, rash, or leukopenia (specificity 68%, sensitivity 71%) or by using 2015 PAHO criteria plus either (1) aspartate aminotransferase >50 IU/L or platelet count <100 000 platelets/μL (specificity 81%, sensitivity 56%) or (2) itchy skin or absence of rhinorrhea or cough (specificity 51%, sensitivity 82%). CONCLUSIONS: The 2015 PAHO dengue case criteria had excellent sensitivity but poor specificity. This can be improved by adding signs/symptoms associated with dengue diagnosis.