Identification of dengue patients with high risk of severe disease, using early clinical and laboratory features, in a resource-limited setting

Only a minority of dengue infections lead to plasma leakage (critical phase [CP]). Early identification of the risk for CP is helpful for triage of patients. This study aimed to identify early clinical predictors of CP that will aid in patient triage during early illness. A retrospective, case-record-based analysis was performed on all microbiologically confirmed (NS1-antigen- or dengue-IgM-antibody-positive), dengue patients (n = 697), admitted to our unit from 01.01.2017 to 30.06.2017. Bivariate analysis was performed to identify clinical and laboratory parameters that predicted CP. Stepwise multivariate logistic regression with backward elimination (p < 0.05) was used to identify independent risk factors for CP. CP developed in 226 (32.4%) patients. Mortality was 1.0%. Predictors for CP (p < 0.05) within the first three days included age category 41-50 years (OR = 1.96), females (OR = 2.09), diabetes (OR = 1.30), persistent vomiting (OR = 2.18), platelet count < 120,000/mm^-3 (OR = 1.91) and AST > 60 IU/L (OR = 3.72). On multivariate analysis, other variables except diabetes remained significant. Elevated transaminase levels remained the strongest independent predictor of CP (OR 2.83). The absence of all five risk factors excluded CP (negative predictive value: 97.2%). Age 41-50 years, female gender, persistent vomiting, thrombocytopenia, and elevated transaminases were early predictors of CP in dengue fever. The absence of these can be used to identify patients who may not require hospital admission. Elevated transaminase was the strongest predictor of CP during early illness.