Objective: To test the prognostic role of nutritional variables as a component of geriatric multidimensional assessment and to study the effect of hospitalization on nutritional status.
Design: Validation cohort study: multidimensional assessment on admission and at discharge and a weekly nutritional assessment.
Setting: General Medicine and Geriatrics wards in an acute-care university hospital.
Patients: A consecutive sample of 302 patients aged 79 +/- 6 years, range 70-96 years.
Main outcome measures: Mortality, longstay (> 29 days), loss of lean body mass as expressed by a negative change in mid-arm muscle circumference (MAMC).
Results: Incidence of mortality, longstay, and decreased MAMC was 6.9%, 24.8%, and 64.2%, respectively. According to logistic regression analysis, mortality was independently predicted by preadmission dependency in at least one Activity of Daily Living (odds ratio = 2.08, confidence limits = 1.19-3.65), clinical diagnosis of malnutrition (OR = 1.89, CL = 1.11-3.21), serum albumin < 3.5 g/dL (OR = 1.82, CL = 1.06-3.14). This predictive model allowed us to recognize 75% of the patients at risk of death by targeting 23% of the population. Longstay was independently predicted by stroke (OR = 1.54, CL = 1.01-2.35), clinical diagnosis of malnutrition (OR = 1.41, CL = 1.04-1.93), and more than five comorbid diseases (OR = 1.39, CL = 1.01-1.94). Dependency in at least one ADL was the only independent predictor of decreased MAMC (OR = 1.71, CL = 1.27-2.30).
Conclusions: Nutrition variables are a cardinal component of multidimensional assessment in the acute-care setting. Nutritional status deteriorates during the hospital stay, mostly in physically dependent patients.