The assessment of body composition is often used in clinical practice for nutritional assessment and monitoring. The gold standard method, dual energy X-ray absorptiometry (DXA), is hardly feasible in routine clinical practice. The bioelectrical impedance analysis (BIA) method is a common applicable alternative because it is accessible and easy to use. In addition, methods that use artificial intelligence to assess body composition through photos (AI-Photo) began to emerge as cheaper alternatives than BIA. However, few studies have evaluated the agreement of different methods with DXA. PURPOSE: Evaluate the agreement between one IA-Photo method, three BIA devices and DXA for estimating body fat percent (BF%) in adults. METHODS: A total of 215 Brazilian adults, aged between 18 and 65 years, BMI from 18 to 35 kg/m2 and sedentary and physically active. Volunteers were evaluated by the IA-Photo method (Shaped, Rio de Janeiro, Brasil), bioimpedance scales of the inBody 270 (Biospace, California, USA), inBCA F-500 (Shenzhen, China), Omron HBF-514 (Kyoto, Japan) models and DXA (GE, Madison, USA) at an interval of 1 minute between the methods. The method of Bland-Altman (BA) was used to evaluate the identify the 95% limits of agreement and BIAS between IA-Photo method and BIA devices with DXA. Lin's Concordance Correlation Coefficient (CCC) was utilized to evaluate concordance between each method with DXA. RESULTS: Confidence interval values are represented by the dotted line and the BIAS by the dashed line (Figure 1). The BIAS value of the %error between methods and DXA ranged from 3.33 (Shaped) to 12.2 (inBody-270). CCC was 0.96; 0.93; 0.91 and 0.45 for the Shaped, Omron, inBody-270 and inBCA F-500 methods. The inBody and Omron model BIAs presented a significant p value for the linear regression in the BA plot, indicating a proportion bias in relation to the BF%. CONCLUSIONS: Shaped and Omron had greater agreement with the DXA compared to inBody-270 and inBCA F-500 for estimating BF%.
Read full abstract