Association between separate items of the Mini Nutritional Assessment instrument and mortality among older adults: A prospective cohort study introducing a trimmed MNA version.

Abstract

The predictive ability of the separate items of the full Mini Nutritional Assessment (MNA-F) instrument in relation to all-cause mortality (ACM) in older adults is unclear. The present study aimed to examine the magnitudes of the independent associations between separate MNA-F items and ACM among people aged ≥65 years old admitted to hospital. A secondary aim was to construct a trimmed MNA instrument (MNA-T) consisting of MNA-F items that contributed independently to predicting ACM, and compare its predictive ability with that of MNA-F and MNA short forms (MNA-SF). This prospective cohort study included 1768 individuals aged ≥65 years old admitted to a Swedish hospital who answered MNA-F in 2008. They were followed-up for ACM after two years using national registers. Associations with ACM were calculated using Cox regression analysis. Predictive abilities for ACM were calculated using an R2-type measure. After a mean follow-up time of 1.7 years (3006 person-years), 455 (25.7%) participants had died. In adjusted analyses, nine MNA-F items contributed independently to predicting ACM (P<0.20) and were included in MNA-T. Five items (food intake, independent living, > 3 prescription drugs, fluid intake, self-assessed health status) had significant contributions (P<0.05). MNA-T had a better predictive ability for ACM than MNA-F and MNA-SF. MNA-F could possibly benefit from excluding less important and more time-consuming items when used for predicting the ultimate consequence of malnutrition, an early death. MNA-T should be a more reliable instrument to use for this purpose.