Objective: To assess the accuracy of air displacement plethysmography (ADP) and dual energy X-ray absorptiometry (DXA) per cent body fat (%BF) estimations in comparison with a reference five-compartment (5C) model used as the reference method.
Design: Cross-sectional study.
Setting: Outpatient University Laboratory, Lisbon, Portugal.
Methods: A total of 32 girls (age: 15.1+/-0.3 years) and 46 boys (age: 15.3+/-1.2 years) athletes were measured. Body volume was assessed by ADP; bone mineral content was assessed by DXA; and total-body water was assessed by deuterium dilution. Statistical analyses included examination of the coefficient of correlation (r), standard error of estimation (s.e.e.), slope, intercept, and pure error (p.e.) and the agreement between models.
Results: For boys and girls, differences between the 5C model and ADP %BF were 0.2 and 1.7% (r = 0.86 and 0.98, s.e.e.= 2.50 and 1.55%, p.e.=2.77 and 2.23%), respectively. Differences between the 5C model and DXA %BF were -1.0 and -3.7% (r = 0.85 and 0.91, s.e.e.= 2.60 and 2.91%, p.e.= 2.90 and 4.66%), for boys and girls, respectively. For girls, regression between ADP and DXA against the reference method did not differ from the line of identity (P > 0.05) while for boys differences were found (P < 0.05). Dual energy X-ray absorptiometry overestimated %BF, particularly in girls. For both genders, large limits of agreement were found between the reference method and both techniques, with the exception of ADP in female athletes.
Conclusion: We conclude that the two techniques were not precise for individual %BF prediction, though ADP revealed better agreement for girls. However, considering all performance criteria for the mean group, our analysis highlights ADP as a valid and nonbiased tool for the evaluation of body composition in adolescent athletes.