Two years of COVID-19 mortality in a government hospital in the capital of Bangladesh: a retrospective study

Background: Bangladesh experienced multiple waves of COVID-19 infection. A comprehensive understanding of mortality patterns within government healthcare facilities is essential for evidence-based healthcare planning and resource allocation in resource-constrained settings. This study systematically examines two years of COVID-19-associated mortality data from a major tertiary government hospital in Dhaka, Bangladesh. Methods: A retrospective observational analysis was conducted examining 693 laboratory-confirmed COVID-19-related deaths occurring between 2020 and 2022. Patient demographic characteristics, pre-existing comorbidities, intensive care unit (ICU) utilization, length of hospital stay (LOS), vaccination status, and temporal mortality trends were systematically examined. Multivariable logistic regression modelling was employed to assess independent predictors of prolonged LOS and ICU admission. Results: The median age of the deceased was 66 years, and 65% were male. Most (78%) were aged 60 years or older. Diabetes (55%) and hypertension (51%) were the most common comorbidities. Of all deaths, 35% occurred after ICU admission and 65% outside the ICU. Hypertension increased the likelihood of ICU admission (OR 1.43), while chronic kidney disease (CKD) decreased it (OR 0.50). No significant predictors of prolonged LOS were identified, and model performance was poor (AUC≈0.50). Only five deaths occurred among vaccinated individuals. Conclusions: COVID-19 mortality at this government hospital primarily affected older males with metabolic comorbidities. ICU use varied by wave, reflecting changes in resource availability and triage. Hypertension and CKD influenced ICU admission but not survival duration. Enhancing critical-care capacity and clinical data systems is essential for future pandemic preparedness.