Objectives. We develop an epidemiological approach to determine the emergency patients at high risk of severe dengue (SD), known as hemorrhagic fever (DHF) or dengue shock syndrome (DSS), and to facilitate timely appropriate medical treatment reducing the number of complications and death by the scoring system. Methods. This retrospective study was conducted during 2015 dengue outbreak in southern Taiwan. An efficient scoring system has been created to differentiate SD by clinical and laboratory features. Type A was defined as cases who were less than 65 years of age, and have none of the dengue related underlying diseases (DM, CKD or ESRD, liver cirrhosis, COPD, CHF, and neoplasm) according the WHO guidelines. Type B were those aged 65 years and older, and with any of the dengue-related underlying diseases. Significant coefficients of parameters were transformed into SD scores and constructed by division of each coefficient with the smallest coefficient of the multivariate logistic regression models by the backward stepwise selection. The accuracy and validity of SD scores are tested and predicted. Results. The severity scores ranged from 1 to $3and the sum of total scores were 10. Significant parameters were used to construct possible SD models to distinguish a higher critical odds ratio when there was an underlying diseases [Adjusted OR, AOR (AOR: 22.68, 95% CI: 6.86-74.96)], tachypnea (AOR: 2.9$195% CI: 1.29-6.56), shock (AOR: 15.5$295% CI: 3.97-60.65), and aPTT prolonged (AOR: 4.48, 95% CI: 2.05-9.78), and Hepatitis (AOR: 2.48, 95% CI: 1.07-5.74). An AUC of 0.82 was obtained which confirmed the good performance of the model. Positive likelihood ratio (LR+) of 10.74 indicates that the result greater than or equal to the severe score of 8 has a large effect on increasing the probability of SD presence. Conclusions. Our study has demonstrated an efficient scoring system to predict SD. The further validation of SD application should be confirmed.
展开▼
