Appendicular Skeletal Muscle Mass in Older Adults Can Be Estimated With a Simple Equation Using a Few Zero-Cost Variables

Abstract

Background and Purpose: Assessing appendicular skeletal muscle (ASM) mass is crucial for the diagnosis of numerous pathologies related to the decline of muscle mass in old age, such as sarcopenia, malnutrition, or cachexia. The dual-energy X-ray absorptiometer (DEXA) radiological technique, which is the gold standard for its assessment, is particularly costly and not routinely used in clinical practice. The aim of this study was to derive computationally simple equations capable of estimating the DEXA-measured ASM at zero cost in older adult populations. Methods: We used the cross-sectional data collected by the National Health and Nutrition Examination Survey (NHANES) over 7 years (1999-2006). The study sample included 16,477 individuals aged 18 years and over, of which 4401 were over 60 years old. We considered 38 nonlaboratory variables. For the derivation of the equations, we employed the Brain Project, an innovative artificial intelligence tool that combines genetic programming and neural networks. The approach searches simultaneously for the mathematical expression and the variables to use in the equation. The derived equations are useful to estimate the DEXA-measured ASM. Results and Discussion: A simple equation that includes the body weight of the patient as the sole variable can estimate the outcome of DEXA with an accuracy equivalent to previously published equations. When used to identify individuals over 60 years old with muscle mass loss, it achieved an area under the curve (AUC) value of 0.85 for both males and females. The inclusion of sex and anthropometric data (thigh and arm circumference) improved the accuracy for male individuals (AUC 0.89). The model is also suitable to be applied to the general adult population of 18 years of age or older. Using more than 3 variables does not lead to better accuracy. Conclusions: The newly proposed equations have better diagnostic accuracy than previous equations for the estimation of DEXA-measured ASM. They are readily applicable in clinical practice for the screening of muscle mass loss in the over 60-year-old population with nearly zero-cost variables. The most complex model proposed in this study requires only the inspection of a simple diagnostic chart to estimate the status of muscle mass loss.