Statistical and Time-Dependent Survival Analysis of Mortality in Acute Febrile Illness with Thrombocytopenia
Year: 2026
Type: conference-paper
Study design: Observational Study
Abstract: The high burden of infectious diseases in many regions has created significant diagnostic and prognostic difficulties and requires the need for early detection of predictors of mortality, allowing for timely intervention. This study analyzes acute febrile illness with thrombocytopenia as a broad clinical syndrome and investigates temporal patterns of risk for mortality via both statistical and survival-based approaches. The study has reviewed a retrospective cohort of 292 hospitalized patients from a single tertiary care hospital using a range of statistical analyses, including descriptive statistics, Mann-Whitney U tests, chi-square tests, Fisher's exact tests for comparison of groups, followed by logistic regression to identify independent predictors of mortality. Additionally, Kaplan-Meier estimates were employed to assess survival, and Cox proportional hazard models with time-dependent covariates were used to characterize changes in hazard over the course of the illness. Parametric modeling techniques were also applied to assess phase-specific changes in hazard, and SMOTE was used where necessary to address class imbalance. The study found that older age, respiratory distress, renal insufficiency, severe thrombocytopenia, elevated liver enzymes, low albumin levels, prolonged coagulation parameters and comorbidities were all statistically significant independent predictors of mortality. Findings also indicated that the pattern of risk of mortality is non-constant over the illness period, with a drastic increase in the likelihood of mortality occurring in the second week of the illness, which indicates that this is a critical period for providing an intervention and/or improvement in clinical outcomes.
Language: English
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W7203662098