Anthropometry-Based Prediction Equation of Body Composition in a Population Aged 12-88 Years

Background: Direct anthropometric measurements to assess body composition have limited precision and/or specificity. Objectives: The study aimed to develop a prediction equation based on anthropometric measurements for estimating total body water (TBW) in a healthy population.Methods: TBW was assessed using the deuterium dilution method in 398 Moroccan participants (235 women and 163 men) aged 11.6-88 y, then compared to the estimations made by previously published anthropometric equations. By sex, participants were randomly assigned into development (n = 199) and validation subgroups (n = 199). A new anthropometric equation was developed in the development subgroup from data obtained on body volume (BV), weight, and sex, where the BV was calculated from participants' height. Then the equation was validated in the validation subgroup using the Bland and Altman procedure, bias, and pure error. External validation was performed using a sample from Tunisia (n = 220, 51.8% female, 18-65 y).Results: Bias in predicting TBW showed unacceptable value for all previously published equations as it was significantly overestimated. The following new equation using anthropometric measurements [TBW (kg) = -5.249 + 107.502 BV (L) + 0.289 weight (kg) + 2.015 sex (male: 1, female: 0); (R2 = 0.91, RMSE = 1.885 kg)] was developed, and its internal validation was confirmed. The generated bias and pure error values were 0.047 kg (95% CI: -0.235, 0.330) and 2.02 kg (95% CI: 1.73, 2.31), respectively. The external validation of the new TBW prediction equation on a Tunisian sample showed a bias and pure error values of 0.07 kg (95% CI: -0.289, 0.429) and 2.36 kg, respectively.Conclusions: The results indicate that the new anthropometric-based prediction equation provides a good estimation of TBW and other body composition compartments. It could be recommended as an alternative method to assess body composition for epidemiological and clinical studies.