Interdevice variability in percent fat estimates using the BOD POD

Abstract

Objective: To evaluate interdevice reliability in body density (Db) and percent body fat (%BF) using air-displacement plethysmography, the BOD POD (BP) body composition system.

Design and setting: Duplicate body composition tests were performed in immediate succession on 50 adults (26 M, 24 F; 21-53 y) using two BP units located in the same body composition laboratory.

Results: Mean Db and %BF between BP1 and BP2 did not differ significantly for men (DeltaDb = 0.0003+/-0.0008 g/ml, P = 0.632; Delta%BF = 0.1+/-1.3, P = 0.665), while for women, there were small but significant differences in Db and %BF between BP1 and BP2 (DeltaDb = 0.0018+/-0.0003 g/ml, P = 0.001; Delta%BF = 0.8+/-1.1, P = 0.001). The regression between %BF by BP1 and BP2 did not deviate significantly from the line of identity for both men and women (R2 = 0.95, standard error of estimate (s.e.e.) = 1.23 %BF for men; R2 = 0.97, s.e.e = 1.13 %BF for women). Individual variations in %BF estimates between the two BP units were within acceptable ranges (95% limits of agreement = -2.5-2.7 %BF for men; -1.4-3.0 %BF for women), and there was no trend in individual differences as %BF varied (r = -0.19, P = 0.359 for men; r = 0.09, P = 0.677 for women). Other subject characteristics, including age, body mass, height, and body mass index, did not significantly contribute to the differences in %BF estimates by the two BP units.

Conclusions: No clinically significant differences in Db and %BF estimates exist between the BP units, and the interdevice variability of the BP has minimal impact on %BF estimates. Further, test-to-test reliability between BP units appears to be as good as within one unit.