Hervera M, Baucells M D, Torre C, Buj A, Castrillo C
Animal Nutrition, Management and Welfare Research Group, Departament de Ciència Animal i dels Aliments, Facultat de Veterinària, Universitat Autònoma de Barcelona, Barcelona, Spain.
J Anim Physiol Anim Nutr (Berl). 2008 Jun;92(3):253-9. doi: 10.1111/j.1439-0396.2007.00740.x.
The proposal of National Research Council (NRC), based on the use of modified Atwater factors, is nowadays the widely used method to estimate digestible energy (DE) content of pet foods. Recently, alternative methods have been suggested for predicting energy content of commercial canine dry food. Factorial equations including food fibre content as estimator, in vitro digestions methods or near-infrared spectroscopy (NIRS) techniques have been considered as good approaches to predict the energy content of dog foods. The aim of this study was to compare the accuracy of some of those estimation methods. Seventeen samples of commercial extruded dog food were used to validate and compare some estimation methods of energy digestibility (Ed, %) and DE value [MJ/kg dry matter (DM)]. The apparent Ed and DE of each food were previously determined by in vivo trials. In vivo Ed and DE of foods ranged from 79.30% to 91.05% and from 16.25 to 21.82 MJ/kg DM, respectively, and their crude fibre (CF) content ranged from 0.72% to 3.28% (in DM base). The % Ed of each sample was estimated by the factorial equation (% Ed = 91.2 - 1.43 x CF %) and by the in vitro digestion method [% Ed(in vitro) = -2.45 + 0.98 organic matter (OM) disappearance(in vitro)%]. The set of samples also was analysed by NIRS, using a calibration equation developed from a set of 69 samples of commercial extruded dog food (0.76 and 0.89 cross-validation r(2) and 2.33 and 0.61 cross-validation SE for Ed and DE respectively). The in vitro method gave better estimations of Ed in vivo than NIRS and factorial methods, although all the methods assessed showed a very good and similar accuracy in the prediction of DE value. These three methods showed a slight better accuracy than that previously proposed by the NRC. To consider constant digestibility values of nutrient content of food can result in bias and error in the estimated energy values. The alternative prediction methods used in this study take into account differences of ingredient composition and availability of nutrients of different extruded dog foods thus could be better systems of valuating energy content in a wider range of different kind of foods than in use method.
美国国家研究委员会（NRC）基于改良阿特沃特因子提出的方法，是目前广泛用于估算宠物食品可消化能量（DE）含量的方法。最近，有人提出了预测商业犬用干粮能量含量的替代方法。包含食物纤维含量作为估算指标的析因方程、体外消化方法或近红外光谱（NIRS）技术，都被认为是预测狗粮能量含量的好方法。本研究的目的是比较其中一些估算方法的准确性。使用了17个商业挤压狗粮样品来验证和比较能量消化率（Ed，%）和DE值[兆焦/千克干物质（DM）]的一些估算方法。每种食物的表观Ed和DE先前已通过体内试验确定。食物的体内Ed和DE分别在79.30%至91.05%和16.25至21.82兆焦/千克DM范围内，其粗纤维（CF）含量在0.72%至3.28%（以DM计）范围内。每个样品的Ed%通过析因方程（%Ed = 91.2 - 1.43×CF%）和体外消化方法[Ed（体外）% = -2.45 + 0.98有机质（OM）体外消失率%]进行估算。还使用从一组69个商业挤压狗粮样品开发的校准方程，通过NIRS对这组样品进行了分析（Ed和DE的交叉验证r²分别为0.76和0.89，交叉验证标准误分别为2.33和0.61）。体外方法在体内对Ed的估算比NIRS和析因方法更好，尽管所有评估方法在预测DE值方面都显示出非常好且相似的准确性。这三种方法的准确性略高于NRC先前提出的方法。认为食物营养成分的消化率值恒定可能会导致估算能量值出现偏差和误差。本研究中使用的替代预测方法考虑了不同挤压狗粮成分组成和营养成分可利用性的差异，因此与现行方法相比，在评估更广泛的不同种类食物的能量含量方面可能是更好的系统。
