Comparison of 3-Compartment Model Body Fat Predictions Via Bioimpedance Devices and the Influence of Hydration in Female Athletes

Document Type


Publication Date



Background: Proper hydration status is not only important in sport performance, but can be helpful when attempting to estimate body fat percentage in athletes. Such methods of monitoring hydration status include deuterium oxide dilution (D2O), which may not be readily available and can be costly and time consuming. Therefore, alternative means of estimating total body water (TBW) have been suggested, such as single and multi-frequency bioelectrical impedance devices. However, the accuracy of these methods has yet to be validated in female athletes.

Purpose: The purpose of this investigation was two-fold: 1) examination of the agreement between 3-C model estimates of BF%, which utilized D2O as a criterion, and six bioelectrical impedance devices and methods, using an estimation of TBW; and 2) determine the validity of BF% estimations when accounting for different levels of fat-free mass hydration status.

Methods: Fifty-two females (age = 24 ± 8.9 years, height = 166.3 ± 5.7 cm, weight = 62.5 ± 8.3 kg) volunteered to participate in this investigation. Participants completed multiple laboratory measures to determine body fat percentage via 3-C model calculations. D2O was used as the criterion measure for TBW, while hydrostatic weighing served as the criterion for body volume. Additionally, TBW was estimated via the use of three laboratory bioimpedance devices (bioimpedance spectroscopy [BIS], multi-frequency bioimpedance [MFBIA], and single-frequency bioimpedance [SFBIA]) and three home-use bioimpedance scales (HS1, HS2, HS3). Furthermore, the accuracy of the six bioimpedance devices were compared within the entire sample (n = 52), as well as in a sub-sample of subjects (n = 28) who were characterized as having a FFM hydration value that was either > 74.8% or < 72.2%.

Results: Results of this investigation demonstrated a sample (n = 52) mean (±SD) BF% of 23.53 ± 5.39% for the 3-C (criterion) model, 23.34 ± 4.91% for BIS, 23.17 ± 5.79% for MFBIA, 27.10 ± 4.77% for SFBIA, 23.43 ± 4.47% for HS1, 24.15 ± 4.57% for HS2, and 23.72 ± 6.28% for HS3. The bias and limits of agreement (bias ± 1.96*SD) for each of the predictive measures in the sample (n =52) were as follows: BIS 0.19 ± 4.78, MFBIA 0.36 ± 4.71, SFBIA -3.57 ± 4.84, HS1 0.09 ± 4.81, HS2 -0.62 ± 5.18, HS3 -0.19 ± 6.23. After accounting for improper hydration status, the sub-sample (n = 28) had a mean BF% of 24.01 ± 5.09% for the 3-C model (criterion), 23.76 ± 5.17% for BIS, 23.25 ± 6.16% for MFBIA, 26.92 ± 5.18% for SFBIA, 23.42 ± 4.48% for HS1, 24.11 ± 4.96% for HS2, and 23.62 ± 6.72% for HS3. The limits of agreement (bias ± 1.96*SD) for each of the predictive measures in the sub-sample (n =28) were 0.25 ± 4.71 for BIS, 0.75 ± 5.46 for MFBIA, -2.92 ± 4.98 for SFBIA, 0.59 ± 5.13 for HS1, -0.11 ± 5.18 for HS2, and 0.38 ± 7.13 for HS3.

Conclusion: Results of this investigation demonstrate that BIS and HS2 provided the smallest mean differences and limits of agreement for the entire sample; while BIS and HS1 provided the smallest differences and limits for the sub-sample with abnormal FFM hydration. Additionally, SFBIA significantly overestimated BF% as compared to the criterion 3-C model.

Practical Application: Practitioners should take note that abnormal hydration (i.e., >74.8% or < 72.2%) status may alter 3C BF% results when using TBW from certain bioimpedance methods in female athletes. The BIS-based 3C model produced the best individual agreement and was not affected by abnormal hydration and therefore is suggest for use when measuring TBW in 3C model in female athletes.


National Strength and Conditioning Association National Conference (NSCA)


Las Vegas, NV