Comparison of bioimpedance spectroscopy and dual energy X-ray absorptiometry for assessing body composition changes in obese children during weight loss

• Published in: European journal of clinical nutrition 2021-01, Vol.75 (1), p.73-84
• Main Authors: Vermeiren, Eline, Ysebaert, Marijke, Van Hoorenbeeck, Kim, Bruyndonckx, Luc, Van Dessel, Kristof, Van Helvoirt, Maria, De Guchtenaere, Ann, De Winter, Benedicte, Verhulst, Stijn, Van Eyck, Annelies
• Format: Article
• Language: English
• Subjects: 692/163/2743/393; 692/699/1702/393; 692/700/1720; Agreements; Bias; Body composition; Body fat; Body weight loss; Bone densitometry; Children; Clinical Nutrition; Comparative analysis; Dual energy X-ray absorptiometry; Epidemiology; Evaluation; Fat-free body mass; Impedance spectroscopy; Internal Medicine; Medicine; Medicine & Public Health; Metabolic Diseases; Methods; Monitoring; Obesity; Obesity in children; Public Health; Spectroscopy; Spectrum analysis; Telemedicine; Weight control; Weight loss
• ISSN: 0954-3007; 1476-5640
• DOI: 10.1038/s41430-020-00738-9

Background Obesity and age influence the reliability of dual energy X-ray absorptiometry scanning (DEXA) and bioimpedance spectroscopy (BIS). Both are used in clinical settings, but have not been compared for measurements in obese children. We compared DEXA and BIS for evaluating body composition and inherent changes in obese children before and after a 10-month weight loss programme. Methods DEXA and BIS were used to evaluate 130 patients at baseline and 75 at follow-up. We tested agreement between the two techniques using Bland–Altman plots and proportional bias using Passing–Bablok regressions. Results The Bland–Altman plots showed wide agreement limits before and after weight loss and when monitoring longitudinal changes. At baseline, the Passing–Bablok regressions revealed a proportional bias for all body compartments. After significant weight loss no proportional bias was found for fat mass and percentage, although BIS systematically underestimated fat mass by 2.9 kg. Longitudinally, no proportional bias was found in the measured changes of absolute fat, fat-free mass and fat-free percentage between both methods, although BIS systematically underestimated fat and fat-free mass by 2.6 and 0.7 kg, respectively. Conclusion While BIS and DEXA are not interchangeable at baseline, the agreement between the two improved after significant weight loss. Proportional changes in fat mass, fat-free mass and fat-free percentage were similar for both techniques. BIS is a viable alternative to DEXA for future paediatric obesity studies measuring treatment effect at group levels, but is not superior to DEXA and cannot be used for monitoring individual changes due to wide limits of agreement.