To identify factors associated with death among patients with laboratory-confirmed severe dengue in Espírito Santo, Brazil, from 2020 through 2024, including exploratory age-related differences.
Approach:
Study Design: A population-based retrospective cohort study using the state’s compulsory disease surveillance database.
Outcome Assessment: Researchers classified 251 patients as fatal or nonfatal cases according to recorded dengue-related death.
Statistical Analysis: Multivariable Poisson regression with robust variance estimated risk ratios. Exploratory analyses compared patients aged <60 years and ≥60 years.
Key Findings:
Among 251 patients, 117 (46.6%) died. Fatal cases were significantly older than survivors.
Weak or undetectable pulse, late-stage arterial hypotension, progressive hematocrit increase, and altered consciousness were associated with higher mortality in the overall multivariable model.
Fever and a sudden platelet count decline were associated with lower mortality.
Weak or undetectable pulse and late-stage hypotension remained associated with higher mortality in both age groups.
Hematocrit increase and sudden platelet decline were associated with mortality only in patients aged <60 years; fever was associated with lower mortality only in those aged ≥60 years.
Interpretation:
Mortality is predominantly associated with advanced hemodynamic instability. Associations with lower mortality do not establish protective effects, and the findings do not demonstrate early predictive value.
Limitations:
Surveillance data may be affected by underreporting or misclassification.
The timing of individual manifestations relative to death was unavailable.
Information on time to care, fluid management, medications, and socioeconomic factors was unavailable.
Age-stratified findings are exploratory; formal interaction tests were not performed, and the age cutoff was selected using the study population.
Conclusion:
Established hemodynamic collapse is associated with mortality in severe dengue. Prospective studies with precisely timed clinical data are needed to investigate potential age-related differences.
by Creuza Rachel Vicente, João Paulo Cola, Ana Paula Brioschi dos Santos, Theresa Cristina Cardoso, Crispim Cerutti Junior, Eng Eong Ooi, Stefano Gobbi, Julia Luch dos Santos, Angelica Espinosa Miranda, Kuan Rong Chan
Standardized preprocessing, multicenter datasets, external validation, and interpretable models may matter more than further gains in classification accuracy