Assessing the accuracy of bioelectrical impedance analysis for fat-free mass estimation in growing pigs using...

This study aimed to assess the accuracy of bioelectrical impedance analysis (BIA) in estimating fat-free mass in growing pigs using dual-energy X-ray absorptiometry (DXA) as the reference method. A longitudinal evaluation was conducted on thirty-nine immunocastrated Landrace × Large White male pigs, with initial and final body weights of 23.1 ± 1.5 kg and 127.4 ± 10.5 kg, respectively, across four growth stages. Lean and fat-free mass were determined by DXA using the Lunar enCORE software (version 8.10.027), whereas BIA measured body resistance (Rs) and reactance (Xc), from which fat-free mass was predicted using established models. The agreement between methods was assessed using Pearson correlations, ANOVA, and mean square prediction error (MSPE) decomposition into errors of central tendency (ECT), regression (ER), and disturbance (ED). Furthermore, non-significant differences (P > 0.05) indicate that the models are in agreement and provide no evidence of predictive bias, supporting the reliability of the predictions. All BIA models exhibited a high correlation with the DXA-derived values (r > 0.93), although the predictive accuracy varied. The Swantek, Marchello, Tilton, and Crenshaw (1999) model demonstrated the best performance for lean and fat-free mass (P > 0.05), with MSPE values up to 45 % lower than those of the other available models, indicating superior predictive reliability for lean and fat-free mass estimation. For protein + water, the Swantek, Crenshaw, Marchello, and Lukaski (1992) model was the most accurate (P > 0.05), with MSPE values 69 % lower than those of Swantek et al. (1999) and over 96 % lower than those of Marchello, Berg, Swantek, and Tilton (1999), suggesting enhanced predictive reliability for protein + water estimation. To our knowledge, this is the first longitudinal study to evaluate multiple BIA models in pigs across a wide weight range. These findings demonstrate that BIA is a practical and non-invasive method for estimating body composition in swine, reinforcing the need for updated prediction equations.