Appendicular Lean Soft Tissue: Development and Cross-Validation of Predictive Models for Older Men and Women.

Abstract

Sarcopenia may explain, in a large proportion, physical disability, falls and fractures, especially in aged elderly. However, a diagnosis in an operationally systematic, simple and low cost way is extremely important, particularly for home-based, epidemiological studies. The purpose of this study was to develop and validate predictive equations of appendicular lean soft tissue (ALST) in elderly older than 80 years. A validation study was performed in 106 elderly (men and women) aged 80 years and older. Body weight, height, circumference (arm, midcalf, hip and waist) and triceps skinfold were measured in the elderly. ALST were measured using as the reference method dual-energy X-ray absorptiometry (DXA). Two models were predicted. The first model (ALST, in kg = 0.074*height + 0.277*weight - 0.144*triceps skinfold - 0.103*waist circumference + 1.831*gender -0.966), which considered all possible variables in stepwise multiple regression, presented better statistical performance (r2 = 0.82; SEE = 1.67 kg), compared to the second model (ALST, in kg = 0.138*height + 0.103*weight + 3.061*gender - 12.489), a more practical equation, due to a lesser quantity of predictive variables (r2 = 0.75; SEE = 1.94 kg). Both models were validated, however, it was verified trend (p<0.05) for overestimation of predicted ALST. In summary, two models for predicting ALST in men and women with age ≥ 80 years were developed and cross-validated. Model 1, with a greater number of predictive variables, presented a better accuracy than did the model with only three variables (height, weight, and gender). Validation studies are needed to test the usefulness of both models in other populations.