Objective: To test the predictive power of comorbidity and of the interaction between age and comorbidity in geriatric patients with acute medical illness.
Design: Prospective observational study.
Setting: Medical and geriatric wards of an acute-care hospital.
Subjects: Three hundred and seventy patients over 70 years of age consecutively admitted in an 18-month period.
Main outcome measure: In-hospital mortality.
Method: On admission a multidimensional assessment was performed, and a comorbidity index and an age-comorbidity index developed on a comparable training population were calculated. The comorbidity index is based upon a scoring system that quantifies the prognostic weight of individual diseases, while the age-comorbidity index corrects the former for the age-related increase of the risk of death. The predictive power of variables univariately correlated with the outcome was tested by logistic regression.
Results: Death was independently predicted by clinical diagnosis of malnutrition (odds ratio = 1.87, confidence limits CL = 1.20-2.86), age-comorbidity index > 7 (odds ratio = 1.77, CL = 1.15-2.72), preadmission impairment in activities of daily living (odds ratio = 1.74, CL = 1.13-2.69), lymphocytopenia (odds ratio = 1.74, CL = 1.15-2.61). A weaker predictive model was obtained by substituting the comorbidity index for the index of age-comorbidity. Excluding comorbidity from the logistic regression greatly weakened the predictive model.