A predictive model to differentiate dengue from other febrile illness

Area of Science:

• Tropical medicine
• Epidemiology
• Infectious diseases

Background:

• Dengue fever is a significant public health concern in tropical and subtropical regions.
• Clinical diagnosis of dengue is often based on symptoms due to delayed laboratory confirmation.
• Dengue shares symptoms with other febrile illnesses, complicating early identification.

Purpose of the Study:

• To identify clinical, hematological, and demographical predictors for dengue fever in patients with febrile illness.
• To develop a statistical model for predicting dengue cases based on patient data.

Main Methods:

• Retrospective cohort study of 548 patients with febrile syndrome in Honduras.
• Utilized clinical, laboratory, and demographic data.
• Employed statistical modeling, including univariable and multivariable logistic regression, for prediction.

Main Results:

• 390 out of 548 patients were confirmed with dengue.
• Key predictors for dengue included retro-ocular pain, petechiae, and gingival bleeding.
• The predictive model achieved 86.2% sensitivity and 69.2% overall accuracy, with low specificity (27.2%).

Conclusions:

• Specific symptoms like retro-ocular pain and gingival bleeding are associated with dengue in Honduran febrile patients.
• The developed predictive model has limitations in accuracy and specificity.
• Validation of the model in diverse populations with similar dengue transmission patterns is necessary.