Predictors of in-hospital mortality in critically ill Geriatric patients

Abstract

Objective: To determine the predictors of in-hospital mortality in critically ill older patients. Participants and Methods: A prospective cohort study including 305 critically ill older adults (age ≥60 years) who were admitted to High Dependency Units (HDUs) in a tertiary care university hospital from March 22, 2019 to January 4, 2020. Demographic, clinical and laboratory data of participants were collected thought a review of medical records and clinical observation at HDUs till either death or discharge alive. Statistical analysis included univariate analysis of selected potential predictors such as various comorbidities, C- reactive protein / Albumin ratio (CAR), Neutrophil-lymphocyte ratio (NLR), Red cell distribution width (RDW), Po2/Fio2 ratio (PFR), Simplified acute physiology score II (SAPS II), Charlson comorbidity index (CCI) and multidrug-resistant Gram-negative bacterial (MDR-GNB) infection to ascertain their association with mortality, followed by multivariable logistic regression to derive the final prediction model. The discriminative ability of the model was evaluated by using the receiver operating characteristic (ROC) curve. RESULTS: Overall mortality was 53.1%. Multivariate regression analysis revealed independent predictors of mortality including dementia in clinical history with an odds ratio (OR) of 4.86 (95% CI: 1.28-18.34), total protein with an OR of .53 (95% CI: .30-.95) and the use of mechanical ventilation (MV) and/or intravenous cardiovascular support with an OR of 148.34 (95% CI: 34.28-641.77), formulating a novel prognostic model with an area under the ROC curve of .93 (95% CI: .89-.96, P.000). CONCLUSION: History of dementia, total protein and the use of MV and/or intravenous cardiovascular support are predictors of mortality in critically ill geriatric patients. It provides a novel prognostic model which needs validation in other multicenter prospective studies.