Derivation of a Simple Risk Scoring Scheme for Prediction of Severe Dengue Infection in Adult Patients in Thailand

Background/Objectives: Severe dengue infection remains a major public health burden in Thailand, where timely identification of high-risk patients is essential for effective clinical management. Existing predictive models are often complex and less feasible in routine practice. This study aimed to develop a simple risk scoring system to predict dengue severity based on patient characteristics and routine clinical data. Methods: Retrospective data of adult dengue patients from nine general hospitals in Thailand from 2019 to 2022 were reviewed. Dengue infection was classified into two groups using the WHO 2009 modified criteria: non-severe (n = 577) and severe (n = 107). Demographic data, clinical characteristics, and laboratory findings were analyzed using logistic regression. Regression coefficients of significant predictors of severe dengue were converted into weighted item scores. Total scores were categorized into three risk levels based on probability distribution cut-off points. Results: The severity score stratified patients into three risk groups with significantly different prognoses: ≤2.0 points (low risk), 2.5–5.0 points (moderate risk), and ≥5.5 points (high risk). The positive likelihood ratios for low-, moderate-, and high-risk groups were 0.12, 1.05, and 28.76, respectively. The distribution of severity scores differed significantly between non-severe and severe cases. The scoring system discriminated between non-severe and severe dengue with an area under the receiver operating characteristic curve (AUROC) of 88.04% (95% CI, 83.99–92.08). Conclusions: The derived dengue severity scoring system classified patients into low, moderate, and high risk with excellent discriminatory performance, effectively distinguishing non-severe from severe dengue infection.