An investigation into the validity of percentage body fat estimations by a new bioelectrical impedance analyser

Abstract

This study was designed to assess the validity of a new method of estimating per cent body fat (%BF) by bioelectrical impedance (BIA). Both models of the new analyser, Tanita TBF 305 and TBF 511 were used to estimate %BF in a sample of active Caucasian adults: 16 male and 23 female. Males: mean age 45.37+ 9.16 years, mean weight 62.10±8.30 kg. Females: mean age 41.96±8.01 years; mean weight 62.10±8.30kg. The new analysers were validated against estimates of %BF by Densitometry (D), compared with sum of skinfolds (SS), BIA Bodystat 1500 (BS) and calculated %BF from BMI. Paired 't' tests were conducted on all methods and compared to D. Bias and 95% Limits of Agreement were calculated using the method of Bland and Altman. Results showed that all methods overestimated %BF significantly in relation to D, except with BS in men. Statistical analysis using paired 't' test and the Bland & Altman method of calculating bias and 95% Limits of Agreement was TBF305 1.18 ± 7.50 %BF for men and 7.40 ± 12.84 %BF for women; TBF 511, 2.14 ± 7.54 %BF for men and 8.40 ± 11.72 %BF for women. It was concluded that because the bias and 95% Limits of Agreement suggested that the new method will overestimate significantly in comparison to D that it could not be recommended as a valid method of estimating %BF and that its use even as a comparative measure of %BF in large epidemiological studies is limited.