Prediction Equations for Digestible and Metabolizable Energy Concentrations Based on Feed Ingredients and Diets for Pigs

The objectives of this study were to develop prediction models for DE and ME concentrations in feedstuffs and diets for pigs and to evaluate the accuracy of the DE prediction equations. A total of 398 data from experiments using marker to marker procedure in the literature were employed to develop equations for predicting DE and ME. The CORR procedure of SAS was used to determine correlation coefficients among variables and the REG procedure was used to generate prediction equations using CP, ether extract (EE), NDF, ash, GE, and DE as independent variables. The equations developed for estimating DE or ME concentrations in feed ingredients or diets for pigs are: DE = 3,968 + 14 × CP + 48 × EE – 21 × NDF – 73 × ash (Eq. 1; R² = 0.723, and P < 0.001); DE = 686 + 0.79 × GE – 25 × NDF – 24 × ash (Eq. 2; R² = 0.729, and P < 0.001); ME = 3,850 + 10 × CP + 49 × EE – 21 × NDF – 61 × ash (Eq. 3; R² = 0.694, and P < 0.001); ME = 1,232 + 0.63 × GE – 23 × NDF – 22 × ash (Eq. 4; R² = 0.678, and P < 0.001); ME = – 70 + 1.00 × DE – 5 × CP (Eq. 5; R² = 0.967, and P < 0.001), where GE, DE, and ME are in kcal/kg DM and all nutrients are in % DM. Data from 56 ingredients reported in NRC (2012) were employed to validate DE equations. To test the accuracy of DE prediction models, a regression analysis of the measured minus predicted DE concentration against the predicted minus average predicted DE concentration was conducted. A validation using Eq. 1 showed that the intercept (–87 ± 47; P = 0.068) representing the mean bias and slope (–0.021 ± 0.088; P = 0.809) representing the linear bias were not different from 0. A validation test of Eq. 2 showed that the intercept (–146 ± 53; P = 0.008) was different from 0 but slope (–0.068 ± 0.100; P = 0.501) was not different from 0. In conclusion, the DE or ME concentrations in feed ingredients or diets for pigs may be predicted using the prediction equations suggested herein.