Validation of different bioimpedance analyzers for predicting cell mass against whole-body counting of potassium (40K) as a reference method

This study compared two different tetrapolar bioimpedance (BIA) devices for estimating body cell mass (BCM), validated them against whole-body counting of K-40 (TBK method), and developed improved prediction equations for estimating BCM from BIA. In 50 healthy volunteers (age 23-65 years, BMI 18.6-27.7 kg/m(2)), BCM was estimated with the BIA devices Nutriguard-M (Data Input, Germany) and Soft-Tissue-Analyzer-STA (Akern, Italy) and by the TBK method. Methods were compared by the Bland-Altman procedure. New prediction equations for BCM were developed by multiple stepwise regression analysis based on a single BIA parallel model. The Akern device gives similar mean estimates of BCM compared to the Data Input device in males (33.5 vs. 33.3 kg, P = 0.789), but higher values in females (24.6 vs. 22.8 kg; P < 0.001). Both BIA devices overestimate mean BCM relative to the TBK method; in males by 5.0 kg (Data Input, P < 0.001) and 5.1 kg (Akern, P < 0.001); in females by 2.3 kg (Data Input, P < 0.001) and 4.1 kg (Akern, P < 0.001). Limits of agreement between BIA and TBK methods are for males +/-4.99 kg (Data Input) and +/-7.16 kg (Akern); for females, +/-4.69 kg (Data Input) and +/-4.12 kg (Akern). New equations were developed for estimating BCM for both BIA analyzers (Data Input, R-2 = 0.91, SEE = 1.46 kg; Akern, R-2 = 0.90, SEE = 1.48 kg). Since estimates of BCM by the present BIA devices do not differ in males, they might be interchangeable. This does not hold true for females. Because both BIA devices overestimate BCM, the newly developed device-specific equations which reduce bias and limits of agreement should be applied. Am. J. Hum. Biol. 16:697-703, 2004. (C) 2004 Wiley-Liss, Inc.