Comparison of air displacement plethysmography to hydrostatic weighing for estimating total body density in children

The purpose of this study was to examine the accuracy of total body density and percent body fat (% fat) using air displacement plethysmography (ADP) and hydrostatic weighing (HW) in children.

Sixty-six male and female subjects (40 males: 12.4 ± 1.3 yrs, 47.4 ± 14.8 kg, 155.4 ± 11.9 cm, 19.3 ± 4.1 kg/m²; 26 females: 12.0 ± 1.9 yrs, 41.4 ± 7.7 kg, 152.1 ± 8.9 cm, 17.7 ± 1.7 kg/m²) were tested using ADP and HW with ADP always preceding HW. Accuracy, precision, and bias were examined in ADP with HW serving as the criterion method. Lohman's equations that are child specific for age and gender were used to convert body density to % fat. Regression analysis determined the accuracy of ADP and potential bias between ADP and HW using Bland-Altman analysis.

For the entire group (Y = 0.835x + 0.171, R² = 0.84, SEE = 0.007 g/cm³) and for the males (Y = 0.837x + 0.174, R² = 0.90, SEE = 0.006 g/cm³) the regression between total body density by HW and by ADP significantly deviated from the line of identity. However in females, the regression between total body density by HW and ADP did not significantly deviate from the line of identity (Y = 0.750x + 0.258, R² = 0.55, SEE = 0.008 g/cm³). The regression between % fat by HW and ADP for the group (Y = 0.84x + 3.81, R² = 0.83, SEE = 3.35 % fat) and for the males (Y = 0.84x + 3.25, R² = 0.90, SEE = 3.00 % fat) significantly deviated from the line of identity. However, in females the regression between % fat by HW and ADP did not significantly deviate from the line of identity (Y = 0.81x + 5.17, R² = 0.56, SEE = 3.80 % fat). Bland-Altman analysis revealed no bias between HW total body density and ADP total body density for the entire group (R = 0.-22; P = 0.08) or for females (R = 0.02; P = 0.92), however bias existed in males (R = -0.37; P ≤ 0.05). Bland-Altman analysis revealed no bias between HW and ADP % fat for the entire group (R = 0.21; P = 0.10) or in females (R = 0.10; P = 0.57), however bias was indicated for males by a significant correlation (R = 0.36; P ≤ 0.05), with ADP underestimating % fat at lower fat values and overestimating at the higher % fat values.

A significant difference in total body density and % fat was observed between ADP and HW in children 10–15 years old with a potential gender difference being detected. Upon further investigation it was revealed that the study was inadequately powered, thus we recommend that larger studies that are appropriately powered be conducted to better understand this potential gender difference.