Development and validation of predictive workable weight equations for Brazilian older adult residents in long-term care institutions.

Abstract

Weight measurement is important in the nutritional anthropometric monitoring of older adults. When this measurement is not possible, estimates may be used. Developing and validating weight predictive equations for older adult residents in long-term care institutions in Brazil. The sample comprised 393 older adult residents in long-term care institutions. Data were collected in two stages, with 315 older adults in the first and 78 in the second. We have measured the arm, calf, and waist circumferences, as well as the triceps and subscapular skinfold and knee height. Multiple linear regression was used to develop the equations, which were evaluated through the coefficient of determination, standard error of estimation, Akaike information criterion, intraclass correlation coefficient (ICC), and Bland-Altmann plot. Five models with different anthropometric measurements were developed, (1) arm circumference as a discriminant variable (ICC: 0.842); (2) best statistical fit for men and women (ICC: 0.874) and its stratification by sex (3) (ICC: 0.876); (4) easy-to-perform measurement for men and women (ICC: 0.842) and its stratification by sex (5) (ICC: 0.828). Five models for estimating the weight of older adult residents in long-term care institutions were developed and validated. The choice to use the models should be based on the physical capacity of the older adults to be evaluated.