Factors predicting discharge outcomes of sepsis patients admitted to intensive care unit in a major tertiary care hospital: A retrospective study from Palestine

Sepsis remains a leading cause of morbidity and mortality among critically ill patients, particularly in resource-limited settings where diagnostic capacity and therapeutic options are constrained. In this study, we aimed to identify clinical, laboratory, and treatment factors that independently predict in-hospital mortality among adult sepsis patients admitted to a tertiary care intensive care unit (ICU) in the West Bank of Palestine. We conducted a retrospective cohort study of 326 adult patients (aged 18-80 years) admitted with sepsis to the medical ICU of a major tertiary referral hospital between January 2018 and December 2023. In-hospital mortality was 41.4% (n = 135). Predictors of mortality were assessed using a multivariable logistic regression model. Multivariable logistic regression identified advancing age (OR 1.03 per year; 95% CI: 1.01-1.06; p = 0.010), cardiovascular disease (OR 2.66; 95% CI: 1.17-6.04; p = 0.020), elevated heart rate (OR 1.03 per beat/min; 95% CI: 1.01-1.04; p < 0.001), reduced urine output (OR 1.00 per mL; 95% CI: 1.00-1.00; p = 0.035), elevated serum lactate (OR 1.15 per mmol/L; 95% CI: 1.01-1.30; p = 0.037), prolonged ventilator days (OR 1.15 per day; 95% CI: 1.09-1.21; p < 0.001), lower PaO2/FiO2 ratio (OR 1.00 per unit; 95% CI: 1.00-1.00; p = 0.006), and shorter ICU length of stay (OR 0.91 per day; 95% CI: 0.87-0.96; p < 0.001) as independent predictors of in‑hospital mortality. These findings highlight the prognostic importance of simple bedside measures, core laboratory indices, and markers of illness trajectory. Together, they form a pragmatic panel of universally available variables that reliably stratify mortality risk among septic ICU patients in Palestine. Embedding these predictors into admission checklists and electronic health record alerts could strengthen early risk recognition, guide triage decisions, and optimize allocation of scarce resources in resource‑limited settings.