Serum albumin and total cholesterol as prognostic factors of mortality in very old patients hospitalized by acute illness

Backgrounds and aims: To analyze whether serum cholesterol and albumin may be factors associated with early and late mortality, independent from other variables such as comorbidity, functional and mental status, in very old patients admitted with acute illness. Methods: Prospective study of all patients admitted for medical causes in the year 2009. The primary outcome was the mortality at 6 and 36 months in relation to serum levels of albumin (categorized as > 3.5 g/dL; 3-3.5 and < 3 g/dL) and total cholesterol (grouped as quartiles). Multivariate Cox regression analysis adjusted for age, sex, diagnosis at admission, premorbid functional and mental status, functional decline at admission and comorbidity was performed to determine whether these serum parameters were associated with mortality at 6 and 36 months. Results: One thousand two hundred and eighty-eight patients were studied, mean age 86.7 years (SD 6.69), 66.1% female. During follow-up, 848 patients died. In Cox regression multivariate analysis, decreased albumin was significantly associated with increased mortality at 6 months [HR: 1.55 (CI95% = 1.03-2.34) for 3-3.5 g/dL albumin, and HR: 1.82 (CI95% = 1.15-2.86) for albumin < 3 g/dL] and after 3 years [HR 1.39 (CI95% = 1.07-1.80) for 3-3.5 albumin; and HR 1.66 (CI95% = 1.22-2.66) for albumin < 3 g/dL]. Decreased cholesterol was associated with increased mortality, losing significance when this was adjusted for variables that influence prognosis in the elderly. Conclusions: In very old patients admitted for acute medical conditions decreased levels of albumin may be associated with increased mortality. (C) 2015 Elsevier Masson SAS and European Union Geriatric Medicine Society. All rights reserved.