What does digestibility measure?
Digestibility describes the proportion of an ingested nutrient that is no longer recovered at a specified point in digestion. The trials below mainly measured apparent total-tract digestibility (ATTD): intake compared with what is recovered in faeces. It is a useful measure of diet utilisation, within the conditions of a feeding trial.
“Apparent” matters. Material in faeces includes more than undigested food: there are also losses from the animal and its gut microbes. Total-tract disappearance is not the same measurement as absorption at the end of the small intestine, nor does it show how much of each essential amino acid the dog can use. Measurements made with different methods should not be treated as interchangeable.
Protein, fat, dry matter, organic matter and energy have separate digestibility estimates. A food can perform differently across these outcomes. Energy digestibility concerns energy lost in faeces; it is not automatically the same as metabolisable energy, which also accounts for other energy losses. A label’s calories and a trial’s percentage energy digestibility answer different questions.
A small stool is not a direct digestibility test. Feeding amount, dietary water, fibre and fermentation can affect faecal output. Likewise, a change in the relative abundance of a gut bacterium does not by itself establish a benefit or harm to the dog.
Why the formulation matters as much as the format
If two foods differ in meat sources, fat, fibre and starch as well as cooking, a difference between them has several possible explanations. Calling one “raw” and the other “dry” does not remove those differences. A more informative processing experiment starts with a common recipe or batch and varies the treatment. Even then, measured nutrient composition, added processing aids, temperature, pressure, drying and packaging need checking.
The 2026 study is the closest reported input match here. Its abstract and lay summary were accessible; final analyses, numerical coefficients and formal disclosures remain unchecked. Common starting inputs do not establish identical final diets.
“Cooked” is also too broad to function as a single experimental treatment. Steaming and sous vide differ from retorting a sealed pack at high temperature; extrusion combines several processing conditions. Raw formats also differ: freezing, freeze-drying and high-pressure processing are distinct treatments. Results for one of these preparations cannot be assigned to every food using the same marketing category.
What the dog feeding trials actually compared
The table separates experiments with common inputs from comparisons of different retail recipes. Sample sizes are total dogs, not dogs per treatment in a crossover. These are separate experiments; we have not combined their percentages into an average or ranked food categories.
| Primary study | Design and diets | Relevant findings | Funding, interests and limits | Access reviewed |
|---|---|---|---|---|
| Geary et al., 2026 (opens in a new tab) | 10 healthy adult female Beagles; replicated 5 × 5 Latin square, five 28-day periods. Common-batch inputs: retort, sous vide, steamed, high-pressure processed (HPP), freeze-dried. | Protein ATTD: retort lower than the other four diets. Fat ATTD: freeze-dried lower than the other four. Dry/organic matter: HPP and steamed exceeded retort and freeze-dried; HPP exceeded sous vide, steamed did not. Energy: HPP/steamed exceeded retort and freeze-dried; sous vide exceeded freeze-dried but not retort. | Primal-affiliated co-author identified in publisher indexing; formal funding/disclosure unverified. Small, short trial; full diet analyses and numerical coefficients unchecked. | Publisher abstract and lay summary only, published 31 August 2026. |
| Cai et al., 2022 (opens in a new tab) | 18 healthy adult Beagles (nine male, nine female), six per parallel group; 56 days after acclimation. Same listed inputs: raw, 80°C/20-minute pasteurisation, 121°C/30-minute autoclaving. Some inputs were already processed powders. | Results text reports higher protein ATTD with high-temperature treatment than raw or pasteurised food. The abstract’s ambiguous percentage/comparator wording is not used here. There was no kibble arm. | Shanghai agricultural-college and academy grants; authors declared no competing interests. One recipe and six dogs/group; short-term and limited mechanistic evidence. | Full paper, methods and results. EP study appraisal. |
| Algya et al., 2018 (opens in a new tab) | Eight healthy adult female Beagles total; replicated 4 × 4 Latin square, four 28-day periods. Extruded, two roasted/refrigerated and raw commercial foods. Different recipes. | Grain-free roasted/refrigerated food had higher protein ATTD than extruded and raw; the other roasted food exceeded extruded. Raw exceeded extruded and grain-inclusive roasted on fat. Dry/organic-matter ATTD did not differ. | Freshpet funded; co-author Lynn Lye affiliated with Freshpet. Very small, healthy-dog crossover; recipe and processing differences confounded. | Full text checked for this synthesis. Existing EP record remains labelled full-text appraisal pending. |
| Geary et al., 2024 (opens in a new tab) | 10 healthy adult female Beagles; replicated 5 × 5 Latin square, five 35-day periods. Extruded, fresh, frozen raw, hybrid freeze-dried raw and freeze-dried raw commercial foods. Different recipes. | Protein and fat ATTD differed among the tested diets. Protein favoured frozen and freeze-dried raw over the other diets; the extruded food had lower fat ATTD than the other foods. | Primal Pet Foods funding; co-author James Templeman employed by Primal Pet Group. Small trial and differing ingredients/composition prevent a processing-only interpretation. | Full text checked. Primary open paper: publisher full text (opens in a new tab). |
How to read apparently conflicting findings
The trials do not test one universal raw diet against one universal cooked diet. They ask different questions under different conditions. A commercial recipe comparison can find high digestibility in a raw food while an experiment using shared inputs finds an advantage for a particular heat treatment. Both observations can be true without establishing a general hierarchy.
The common-input experiments are especially relevant when the question is about processing. Their results still belong to the tested ingredients and treatments. A result from autoclaving one recipe does not estimate the effect of gently cooking another. The 2026 trial’s differences between raw formats also show why removing heat does not make all processing equivalent.
Statistical comparisons need care. “No significant difference” does not prove exact equivalence, especially in a small experiment. A significant before-and-after change in one group is not evidence that its change was significantly different from another group’s change. We report between-diet findings where verified and leave uncertain numerical interpretations out.
Laboratory and other-animal evidence belongs in a separate box
A 2025 Geary et al. rooster assay (opens in a new tab) examined amino-acid digestibility and nitrogen-corrected true metabolisable energy. It is useful mechanistic evidence, but it did not feed dogs. Methods describe 30 cecectomised and 30 conventional roosters, five per diet per assay; the lay summary gives conflicting counts. The retorted diet included 0.3% guar gum while the other diets did not. It should not be described as a perfectly identical-formulation dog trial.
In-vitro digestion systems and precision-fed bird assays can help investigate mechanisms or protein quality. They do not replace a target-species feeding study. We have not pooled their outcomes with canine ATTD, and we have not counted a conference presentation as an additional independent dog cohort.
What does this tell an owner?
Ask what was measured and whether the study tested the exact recipe, life stage and outcome relevant to your dog. A digestibility claim is more informative when it gives a published method, comparator, nutrient-specific result and clear connection to the food being sold. A general statement that a category is “easier to digest” omits much of the information needed to assess it.
- Digestibility does not establish nutritional completeness. A highly digestible diet can still have an unsuitable balance of essential nutrients.
- It does not establish long-term health. Short trials in healthy adult dogs cannot settle outcomes across a lifetime or in dogs with disease.
- It does not establish microbiological safety. These digestibility findings do not quantify pathogen risk or household exposure.
- It does not establish a category winner. An effect in a tested food is not evidence that every food with its format is superior.
For day-to-day comparison, first check species, life stage, complete or complementary status, feeding guidance and storage. Compare declared nutrients on compatible bases and use reliable calorie information where available. If moisture or energy is missing, keep that gap visible. See how to compare two pet foods and what complete pet food means.
Food choice for a dog with persistent digestive signs or another medical condition needs individual assessment. The studies here do not test a treatment for that dog. Nor can ingredient research establish that an untested finished commercial food will have the same effect.
Search, selection and review note
This is a targeted narrative synthesis, not a systematic review or meta-analysis. On 24 October 2026 we searched combinations of dog/canine, raw, cooked, digestibility, retort, extrusion, frozen and freeze-dried, then checked exact titles and DOIs. We used primary publisher pages, PubMed and open full texts, alongside EP’s existing study records. We prioritised dog feeding trials with direct processing or format contrasts and checked whether recipes were matched.
The four core dog studies are identified above. A relevant 2021 fresh-versus-kibble paper (DOI 10.1093/tas/txab057 (opens in a new tab)) was located but not fully extracted in this pass and contributes no numerical conclusion here. Observational owner reports, microbiome-only comparisons, other-species assays and in-vitro work were not used as interchangeable estimates of dog digestibility. This search may miss relevant papers and does not support a claim that the table exhausts the literature.
We checked designs, populations, comparison scope and reported outcomes, with funding and author interests where accessible. Funding is context, not a reason by itself to accept or dismiss a result. The 2026 full text and formal disclosures remain an identified gap. We have not performed an independent statistical reanalysis or laboratory verification.
Primary references
- Geary EL, Wang T-L, Vogel CL, et al. Journal of Animal Science. 2026;104:skag283. DOI: 10.1093/jas/skag283 (opens in a new tab).
- Cai X, Liao R, Chen G, Lu Y, Zhao Y, Chen Y. PLOS ONE. 2022;17:e0262284. DOI: 10.1371/journal.pone.0262284 (opens in a new tab).
- Algya KM, Cross TWL, Leuck KN, et al. Journal of Animal Science. 2018;96:3670–3683. DOI: 10.1093/jas/sky235 (opens in a new tab).
- Geary EL, Oba PM, Templeman JR, Swanson KS. Translational Animal Science. 2024;8:txae163. DOI: 10.1093/tas/txae163 (opens in a new tab).
- Geary EL, et al. Journal of Animal Science. 2025;103:skaf334. Rooster assays. DOI: 10.1093/jas/skaf334 (opens in a new tab).
Existing EP appraisal statuses and study IDs are preserved. Material corrections should name the affected claim, source and update date. See how EP structures an evidence record.