Skinfold-based-equations to assess longitudinal body composition in children from birth to age 5 years

• Published in: Clinical nutrition (Edinburgh, Scotland) Scotland), 2023-07, Vol.42 (7), p.1213-1218
• Main Authors: van Beijsterveldt, Inge A.L.P., Dorrepaal, Demi J., de Fluiter, Kirsten S., de Ridder, Maria A.J., Hokken-Koelega, Anita C.S.
• Format: Article
• Language: English
• Subjects: adiposity; anthropometric measurements; Anthropometrics; childhood; children; clinical nutrition; densitometry; dual-energy X-ray absorptiometry; Fat mass; Infant; prediction; Prediction equation; puberty; regression analysis; risk
• ISSN: 0261-5614; 1532-1983
• DOI: 10.1016/j.clnu.2023.04.024

In order to identify children at risk for excess adiposity, it is important to determine body composition longitudinally throughout childhood. However, most frequently used techniques in research are expensive and time-consuming and, therefore, not feasible for use in general clinical practice. Skinfold measurements can be used as proxy for adiposity, but current anthropometry-based-equations have random and systematic errors, especially when used longitudinally in pre-pubertal children. We developed and validated skinfold-based-equations to estimate total fat mass (FM) longitudinally in children aged 0–5 years. This study was embedded in the Sophia Pluto study, a prospective birth cohort. In 998 healthy term-born children, we longitudinally measured anthropometrics, including skinfolds and determined FM using Air Displacement Plethysmography (ADP) by PEA POD and Dual energy X-ray Absorptiometry (DXA) from birth to age 5 years. Of each child one random measurement was used in the determination cohort, others for validation. Linear regression was used to determine the best fitting FM-prediction model based on anthropometric measurements using ADP and DXA as reference methods. For validation, we used calibration plots to determine predictive value and agreement between measured and predicted FM. Three skinfold-based-equations were developed for adjoined age ranges (0–6 months, 6–24 months and 2–5 years), based on FM-trajectories. Validation of these prediction equations showed significant correlations between measured and predicted FM (R: 0.921, 0.779 and 0.893, respectively) and good agreement with small mean prediction errors of 1, 24 and −96 g, respectively. We developed and validated reliable skinfold-based-equations which may be used longitudinally from birth to age 5 years in general practice and large epidemiological studies.