OBJECTIVES: To identify the factors predictive of long-term mortality among a cohort of subjects aged 75 years or older hospitalized from the emergency department.
METHODS: Variables from the standardized geriatric assessment of members of the multicenter SAFES cohort were applied to a Cox model to predict mortality over a 3-year follow-up.
RESULTS: This cohort comprised 1306 patients with a mean age of 85+/-6 years. Half the patients died during the 3 years of follow-up. After adjustment for center, social and demographic variables (age, sex, educational level, and community or institutional residence) and level of comorbidity, the variables that significantly influenced mortality over the next 36 months were: severe malnutrition (p<0.03), dementia (p<0.001) and/or confusion (p<0.001), walking problems (p=0.001) and recent hospitalization (p<0.001). No significant association was found between depression and mortality (p=0.40).
CONCLUSION: A predictive approach to mortality at 3 years is possible with simple scales widely used in geriatrics. Correction of malnutrition, recognition and management of cognitive disorders, and functional rehabilitation must be included in the priorities of care.