Dengue Severity Prediction in a Hyperendemic Region in Colombia

Background: Early detection of severe dengue (SD) is crucial in preventing life-threatening complications. Despite its importance, comprehensive knowledge about these early indicators is still limited. This study aimed to identify predictors of SD in a hyperendemic region of Colombia. Methods: A cross-sectional analysis was conducted using data from 2018 to 2022, encompassing 233 patients. By utilizing the 2009 World Health Organization dengue classifications, cases were differentiated between severe dengue (SD) and non-severe dengue (non-SD). Among these, 47 were confirmed as SD. Associations between clinical, demographic, and laboratory data and disease severity were examined using Fisher’s exact tests or the Mann–Whitney U test (<i>p</i> < 0.05). Profiles for SD and non-SD cases were established through multiple correspondence analysis, and a logistic regression-based predictive model was validated using training and test sets. The model’s performance was evaluated using the area under the receiver operating characteristic curve (AUC-ROC), accuracy, sensitivity, F1-score, and precision. Results: Differences in place of residence, comorbidities, type of infection, and signs and symptoms were observed between the severe dengue (SD) and non-severe dengue (non-SD) groups. Median levels of platelets, white blood cells (WBC), aspartate aminotransferase (AST), and alanine aminotransferase (ALT) were found to be higher in the SD group compared to the non-SD group. Neutrophils, leukocytes, platelets, AST, and primary infection were significant predictors of SD. The model demonstrated an area under the receiver operating characteristic curve (AUC) of 0.91 (95% CI, 0.85–0.96). Conclusions: The developed predictive model provides significant assistance to clinicians in assessing SD risk and optimizing triage, which is particularly crucial during dengue outbreaks.