This research aimed to show that using different cutoff points for women and men in nutritional assessment through the Mini Nutritional Assessment (MNA) and different anthropometric variables, improves the diagnostic accuracy of malnutrition among people over 65 years. A cross-sectional study was carried out in a sample of 240 patients of 65 to 104 years in Córdoba (Spain) between 2013 and 2015. Excluded were those with disabling diseases that prevented informed consent. We assessed Nutritional status using the MNA and the Chang Method (Gold Standard). Global cut points per sex were determined for each independent variable using Receiver Operating Characteristic curves, choosing the best of each according to the highest Youden´s index (J). None of the variables studied showed high precision in the diagnosis of malnutrition in the elderly. Different cutoff points were found for all variables by sex, as well as variations in their validity indexes. The number of lymphocytes was more accurate in women (J = 0.58), being the muscular circumference of the arm the best in men (J = 0.8). Regarding MNA, the Long Form version showed higher sensitivity (S) in women (S = 76%) and short versions in men (S = 71% and 59%) at different cutoff points. The introduction of the sex variable seems determinant to achieve a more accurate diagnostic of malnutrition, avoiding biases derivate from a global study without considering its effect in the development of diseases.
Keywords: Aged; malnutrition; nutritional assessment; sex bias.