Prognostic Models in Patients with Dengue: A Systematic Review

Area of Science:

• Infectious Diseases
• Epidemiology
• Biostatistics

Background:

• Predictive models for dengue prognosis show uncertain clinical utility.
• Accurate prognosis is crucial for effective dengue patient management.

Purpose of the Study:

• To systematically review and evaluate prognostic models for dengue severity, hospitalization, intensive care unit (ICU) admission, and mortality.

Main Methods:

• Systematic literature search in PubMed, Embase, and LILACS up to May 2023.
• Inclusion of case-control and cohort studies developing or validating multivariable prognostic models for dengue.
• Narrative synthesis of model performance measures and risk of bias assessment.

Main Results:

• 35 studies with 43 prognostic models were included; 35 were developmental, 8 external validation.
• Models primarily targeted severity (30) and mortality (10).
• C-statistics ranged from 0.70-0.95 (severity) and 0.83-0.99 (mortality), indicating acceptable discrimination, but calibration was poorly reported. High risk of bias was noted for all models.

Conclusions:

• Multiple prognostic models exist for dengue, particularly for severity and mortality prediction.
• Despite acceptable discrimination, poor reporting of calibration and high risk of bias limit current clinical applicability.
• Further high-quality research is needed to develop reliable prognostic tools for dengue.