Using Gross Energy Improves Metabolizable Energy Predictive Equations for Pet Foods Whereas Undigested Protein and Fiber Content Predict Stool Quality

TL;DR

In a retrospective review of 558 AAFCO-based digestibility studies, predictive equations for metabolizable energy (ME) in dog and cat foods were accurate when using modified Atwater or NRC factors, but equations incorporating measured gross energy (GE) plus moisture, protein, fat, and fiber improved accuracy and precision. Stool quality was better when less undigested protein and more fiber reached the large intestine.

Key findings

• Modified Atwater factors predicted ME accurately in dog foods (r² = 0.97) and cat foods (r² = 0.98); NRC equations also gave r² = 0.97 for both dogs and cats. Average predicted ME differed from measured ME by 0.16% and 2.24% in dogs using modified Atwater and NRC equations, respectively, and by 1.57% and 1.80% in cats.
• New equations using measured GE and new coefficients for moisture, protein, fat, and fiber yielded r² = 0.99 in dogs and 1.00 in cats, with less than 0.01% difference between average measured and predicted ME; the absolute difference between measured and predicted ME was reduced by approximately 50% in dogs and 60% in cats.
• In dogs, stool score was higher when less protein entered the large intestine (4.3 ± 0.04 vs. 4.1 ± 0.04, P≤0.05) and when more fiber entered the large intestine (4.2 ± 0.04 vs. 4.1 ± 0.04, P≤0.05). In cats, stool score was higher when less protein entered the large intestine (4.1 ± 0.05 vs. 3.7 ± 0.05, P≤0.05) and when more fiber entered the large intestine (4.0 ± 0.07 vs. 3.8 ± 0.06, P≤0.05).
• There was no correlation between dietary protein concentration and stool quality.

Methods

• Retrospective review of 558 digestibility studies conducted over 7 years: 331 studies in healthy adult Beagle dogs (124 dogs; mean age 6.7 years, range 2–12) and 227 studies in healthy adult domestic short hair cats (138 cats; mean age 8.0 years, range 1–15). Each test used six adult dogs or cats and followed the AAFCO quantitative collection protocol, with a pre-collection phase of at least 7 days and a 5-day (120-hour) total fecal collection phase.
• Foods included dry and canned dog and cat foods, both commercial and non-commercial, all meeting AAFCO requirements for complete and balanced adult dog or cat foods. Food energy, moisture, protein, fat, fiber, and ash were analyzed by a commercial laboratory using AOAC methods; ME was measured with AAFCO methods and compared with modified Atwater, NRC, and new study-derived predictive equations.
• Statistical analyses used SAS general linear models; multivariate regression identified equations minimizing variation between estimated and actual ME. Stool quality was scored on a 1–5 scale by trained, masked graders. Nutrient digestibility effects on stool quality were evaluated with species and form (dry or canned) as class variables and measured dietary ME as a covariate; high/low classifications for protein or fiber entering the large intestine were based on median split, and mean separation used Duncan’s least significant difference test with significance at P≤0.05.

Limitations

• The paper does not state an explicit limitations section.
• Evident from the text: the new ME equations were derived from and evaluated against the same set of 558 digestibility studies, with no independent validation reported, so the high accuracy and precision may be optimistic. The studies were conducted at a single commercial sponsor (Hill’s Pet Nutrition) using animals owned by the funders, which may limit generalizability.
• Evident from the text: stool quality relied on subjective 1–5 grading, though graders were trained, previously evaluated, and masked to foods. The study population was healthy adult dogs and cats fed AAFCO-complete foods, so findings may not extend to other ages, health states, or non-AAFCO diets.