05 TSAI_04 LORD_c

Objective: To compare the mortality-predictive ability of the fulland short-form (SF) Mini Nutritional Assessment (MNA). Design: A prospective cohort study. Setting: Population-representative sample. Participants: 2872 ≥65-year old men and women. Measurements: The study analyzed 1999 and 2003 datasets of the Taiwan Longitudinal Survey on Aging (TLSA). Subjects were graded for nutritional status with the fullMNA and MNA-SF of a Taiwanese-specific version (T2, containing calf circumference instead of BMI) at baseline (1999) and tracked their survival status for 4 years. Mortality-predictive abilities of the full-MNA and MNA-SF were compared using Cox regression analysis and Net Reclassification Improvement (NRI). Results: The full-MNA and MNA-SF have comparable abilities in predicting follow-up 4-year mortality risk according to the hazard ratios (all p<0.001) and Akaike information criterion (AIC). It also showed a slight improvement (not significant) if the full-MNA in a predictive model was replaced by the MNA-SF (NRI=0.09%, p=0.956). Conclusion: The MNA-SF has at least comparable or even slightly better ability in predicting follow-up 4-year mortality risk of elderly Taiwanese. Results suggest that MNA-SF with calf circumference may possess some basic characteristics of a comprehensive and universal geriatric screening scale.