Cucunawangsih, Beti Ernawati Dewi,Veli Sungono, Nata Pratama Hardjo Lugito, Bambang Sutrisna, Herdiman T. Pohan,Agus Syahrurachman,Djoko Widodo,Sudarto Ronoatmodjo, Modastri K. Sudaryo, Cicilia Windiyaningsih, T. Mirawati Sudiro
Background: To design a new scoring model to diagnose dengue in the early phase of illness that could be used in primary health care facilities.
Methods and Findings: Cohort design with consecutive sampling of eighty four participants with one/more clinical features similar to dengue illness within 72 hours after onset of fever. Rapid tests of IgM and NS-1 antigen, and RT-PCR were used to confirm dengue infection. Dengue scoring model with sensitivity and specificity for each value was developed using multivariate logistic regression analysis. Performance of the model was assessed using the ROC curve, and the validity was compared to 1997, 2009 and 2011 WHO dengue classification. Presumptive scoring model used days of fever, presence of myalgia, tourniquet test, total WBC count, monocyte count, and platelet count variables, while probable scoring model used monocytes count and NS-1 antigen for laboratory variables. Patients were most likely to have presumptive dengue illness if they had a total score of ≥ 14 with sensitivity, specificity and likelihood ratio positive of 79.7%, 60.0%, and 1.99 respectively. Patients with a total score of ≥ 7 diagnosed as probable dengue with sensitivity, specificity and likelihood ratio positive of 79.7%, 68.0% and 2.48 respectively.
Conclusion: The scoring models could predict dengue illness better than 1997 and 2011 WHO classification. It was easy to implement so that it could help clinicians determine the diagnosis of patients with acute febrile illness.
English 
Spanish 
Russian 
German 
French 
Japanese 
Portuguese 
Hindi