Performance of nutritional screening tools in predicting poor six-month outcome in hospitalised older patients.

Abstract

Malnutrition is a major problem in hospitalised older people. Many nutrition screening tools are available for malnutrition identification, however little is known about their prognostic ability. This prospective, observational study investigated the prognostic value of three nutritional screening tools in a Geriatric Evaluation and Management Unit: the Geriatric Nutritional Risk Index (GNRI), the Mini Nutritional Assessment (MNA) and the Mini Nutritional Assessment short form (MNA-SF), incorporating either body mass index or calf circumference. Poor six- month outcome was defined as new admission to higher level residential care or mortality at six months post-discharge. Predictive ability of poor outcome was assessed by logistic regression models, adjusting for age, gender, cognition and co-morbidity. Predictive accuracy was determined by area under Receiver Operator Characteristic curves, sensitivity, specificity, predictive values and Youden Index. One hundred and seventy-two consecutive patients with a mean (SD) age=85.2 (6.4) years were included in the study. Malnutrition was identified in 31% of patients using the MNA and was associated with a higher risk of poor six-month outcome when identified by the MNA (OR, 95% CI=3.29, 1.17-9.23) and the GNRI (OR, 95% CI=2.84, 1.31-6.19), but not by the MNA-SF. All screening tools lacked discriminative power for outcome prediction. The MNA and GNRI were useful clinical predictors of poor six-month outcome, although their accuracy of prediction was low. Nutritional screening remains a priority in the routine assessment of hospitalised older people.