RESUMO
BACKGROUND: The study objectives were to identify the main predictive factors for long hospital stays and to propose new and improved methods of risk assessment. METHODS: This prospective cohort study was conducted in the clinics and surgical wards of a tertiary hospital and involved 523 elderly patients over 60 years of age. Demographic, clinical, functional, and cognitive characteristics assessed between 48 and 72 h after admission were analyzed to investigate correlations with lengths of stay greater than 10 days. Univariate and multivariate analyses were performed, and in the final model, long-term probability scores were estimated for each variable. RESULTS: Of the 523 patients studied, 33 (6.3%) remained hospitalized for more than 10 days. Multiple regression analysis revealed that both the presence of diabetes and the inability to perform chair-to-bed transfers (Barthel Index) remained significant risk predictors. Diabetes doubled the risk of prolonged hospital stays, while a chair-to-bed transfer score of 0 or 5 led to an eight-fold increase in risk. CONCLUSIONS: In this study, we propose an easy method that can be used, after external validation, to screen for long-term risk (using diabetes and bed/chair transfer) as a first step in identifying hospitalized elderly patients who will require comprehensive assessment to guide prevention plans and rehabilitation programs.