New bioelectrical impedance analysis equations for children and adolescents based on the deuterium dilution technique

Background and aims: Body composition in childhood is not only a marker of the prevalence of obesity, but it can also be used to assess associated metabolic complications. Bioelectrical impedance analysis (BIA) shows promise as an easy to use, rapid, and non-invasive tool to evaluate body composition. The objectives of this study were to: (a) develop BIA prediction equations to estimate total body water (TBW) and fat-free mass (FFM) in European children and early adolescents and to validate the analysis with the deuterium dilution as the reference technique and (b) compare our results with previously published paediatric BIA equations.

Methods: The cohort included 266 healthy children and adolescents between 7 and 14 years of age, 46% girls, in five European countries: Bosnia and Herzegovina, Latvia, Montenegro, North Macedonia, and Portugal. TBW and FFM were the target variables in the developed regression models. For model development, the dataset was randomly split into training and test sets, in 70:30 ratio, respectively. Model tuning was performed with 10-fold cross-validation that confirmed the unbiased estimate of its performance. The final regression models were retrained on the whole dataset.

Results: Cross-validated regression models were developed using resistance index, weight, and sex as the optimal predictors. The new prediction equations explained 87% variability in both TBW and FFM. Limits of agreement between BIA and reference values, were within ±17% of the mean, (-3.4, 3.7) and (-4.5, 4.8) kg for TBW and FFM, respectively. BIA FFM and TBW estimates were within one standard deviation for approximately 83% of the children. BIA prediction equations underestimated TBW and FFM by 0.2 kg and 0.1 kg respectively with no proportional bias and comparable accuracy among different BMI-for-age subgroups. Comparison with predictive equations from published studies revealed varying discrepancy rates with the deuterium dilution measurements, with only two being equivalent to the equations developed in this study.

Conclusions: The small difference between deuterium dilution and BIA measurements validated by Bland-Altman analysis, supports the application of BIA for epidemiological studies in European children using the developed equations.