腰围
人体测量学
周长
一致性
医学
线性回归
瘦体质量
一致相关系数
共线性
数学
回归分析
体质指数
统计
内科学
体重
几何学
作者
Leonardo Pozza Santos,Maria Cristina Gonzalez,Silvana Paiva Orlandi,Renata Moraes Bielemann,Thiago Gonzalez Barbosa-Silva,Steven B. Heymsfield
摘要
Abstract Background Low appendicular skeletal muscle mass (ASM) is associated with negative outcomes, but its assessment requires proper limb muscle evaluation. We aimed to verify how anthropometric circumferences are correlated to ASM and to develop new prediction equations based on calf circumference and other anthropometric measures, using dual‐energy X‐ray absorptiometry (DEXA) as the reference method. Methods DEXA and anthropometric information from 15,293 adults surveyed in the 1999–2006 NHANES were evaluated. ASM was defined by the sum of the lean soft tissue from the limbs. Anthropometric data included BMI and calf, arm, thigh, and waist circumferences. Correlations were assessed by Pearson's correlation, and multivariable linear regression produced 4 different ASM prediction equations. The concordance and the overall 95% limits of agreement between measured and estimated ASM were assessed using Lin's coefficient and Bland‐Altman's approach. Results Calf and thigh circumferences were highly correlated with ASM, independent of age and ethnicity. Among the models, the best performance came from the equation constituted solely by calf circumference, sex, race, and age as independent variables, which was able to explain almost 90% of the DEXA‐measured ASM variation. The inclusion of different anthropometric parameters in the model increased collinearity without improving estimates. Concordance between the four developed equations and DEXA‐measured ASM was high (Lin's concordance coefficient >0.90). Conclusion Despite the good performance of the four developed equations in predicting ASM, the best results came from the equation constituted only by calf circumference, sex, race, and age. This equation allows satisfactory ASM estimation from a single anthropometric measurement.
科研通智能强力驱动
Strongly Powered by AbleSci AI