Effect of body surface area calculations on body fat estimates in non-obese and obese subjects.

The purpose of the present study was to compare body surface area (BSA) estimates using two equations (Dubois and Dubois versus Livingston) and their respective effects on per cent body fat (%BF) obtained with two molecular approaches of body composition analysis, two-compartment (2C) and five-compartment (5C) models. Body composition data using the 2C model were studied in healthy adults, 432 women (body mass index (BMI): 28.3 +/- 4.4 kg m(-2)) and 147 men (BMI: 26.8 +/- 3.9 kg m(-2)), while another sample of 126 women (BMI: 30.4 +/- 3.7 kg m(-2)) was evaluated using the 5C model. Measures of body volume (BV) assessed by air displacement plethysmography, bone mineral content by dual energy x-ray absorptiometry (DXA) and total-body water by deuterium dilution were used to estimate %BF with the 5C model. Comparison of means and linear regression analysis was performed. Using BSA(Dubois), either in 2C and 5C models, BV and %BF estimates were significantly underestimated compared to results obtained using BSA(Livingston) (p < 0.05). BMI was strongly associated with %BF differences using BSA(Dubois) and BSA(Livingston) in both 2C (men: r = 0.90; women: r = 0.88) and 5C models (r = 0.88). Though %BF(Dubois) and %BF(Livingston) were strongly associated (r(2) = 1.000), some variability was observed on %BF differences using BSA(Dubois) and BSA(Livingston). These findings suggest that BSA calculation is critical in BF estimation, supporting the use of a more accurate equation for non-obese and obese subjects.