Changes in weight, height and skin temperature account for 73% of between-day variability in BIA measurements

Although bioelectrical impedance analysis (BIA) is reproducible to 1% within a single day, day to day variations are larger and the factors that contribute to this variation are presently unknown. This study assessed the reproducibility of BIA measurements performed in a clinical research center on 2 separate occasions, 3 weeks apart, in 9 healthy young adults (5 females and 4 males), after controlling for all known measurement variables. Standing height (Ht) and weight (Wt) wearing indoor clothing was measured after an overnight fast. Distal tetrapolar electrode placement was applied using previously described anatomical sites (wrist and ankle). Resistance was measured using a BIA analyzer (Xitron Technologies Inc, San Diego, CA) and skin temperatures (ST) were measured at the site of electrode placements with an infrared temperature scanner. All measurements were made immediately upon recumbancy to control for orthostatic changes. Even under controlled measurement conditions, predicted total body water (TBW) by multiple regression analysis (42.9 ±13.9 L vs 42.8 ±13.8 L) differed by -0.18 ±1.29 L (CV 3%). The difference in measured Ht, Wt, and ST on the 2 occasions was 0.09 + 0.51 cm, 0.39 ±1.02 kg, and -0.53 ±0.69 °C respectively. 12% of the variability in predicted TBW was accounted for by discrepancies in measured Ht. Wt accounted for 52% of the variability seen in predicted TBW. ST, dependent on ambient room temperature, was responsible for 8% of the variability in predicted TBW. After correcting for the difference in Ht, Wt and ST, the precision in predicted TBW between visits improved to ±0.68 L, reducing the CV to 1.5%. Changes in Ht and ST, though small, are not insignificant measurement variables and improve reproducibility of the method by 20% if corrected or controlled. Although 73% of the measurement variability can be explained, the remaining 27% is unaccounted for in this study. This variability could be due to analytical variability in the instrument or due to other presently unknown factors.