Raw Diet for a Dog With IBD: What the Trial Bowl Delivers

A dog with IBD can be fed a raw diet, and food is now the first thing most clinicians will change. The 2026 ACVIM-endorsed consensus statement on canine chronic inflammatory enteropathy puts a dietary treatment trial ahead of invasive testing, as the preferred first step. The reason is blunt: 38 to 89 percent of affected dogs turn out to be food-responsive, many of them holding remission on diet alone. Raw is a legitimate way to run that trial. It is not, by itself, the treatment.

What almost nobody puts a number on is the second half of the deal. An elimination trial works by stripping the diet down to one protein for weeks at a time, and stripping a diet down has nutritional consequences you can calculate in advance. Run a strict single-protein raw bowl through the NRC 2006 adult allowances and the shortfalls are not subtle. On wild rabbit alone, a dog gets 2 percent of its iodine allowance, 8 percent of its manganese, 12 percent of its copper, 49 percent of its zinc – and no vitamin A at all. Calcium, meanwhile, runs to roughly 1.6 times the ceiling FEDIAF sets for an adult diet. Those are the numbers this article works through.

What changed in 2026

The guidance most raw-feeding pages are still arguing with is sixteen years old. The consensus statement published in the Journal of Veterinary Internal Medicine on 21 January 2026 replaced the 2010 document, and three of its changes matter to anyone feeding a dog through this.

The first is the name. The panel suggests avoiding inflammatory bowel disease in favour of chronic inflammatory enteropathy, on the grounds that the canine condition is similar to but not identical with the human disease the term describes. The second change matters more in practice: the 2010 statement defined IBD by excluding the dogs that got better on a diet change, which quietly framed diet as the thing you rule out before the real work starts. The 2026 statement does the opposite and recognises food-responsive disease as a phenotype inside the spectrum – a diagnosis in its own right, not a failed one. The third is that empirical antibiotics are now discouraged: dogs relapse soon after the course ends, and the panel notes that antibiotics can induce substantial long-lasting intestinal dysbiosis, so they are reserved for cases that have failed everything else.

Read together, that is a large shift in where the bowl sits. Diet used to be step four. It is now step one, and the panel is explicit that it belongs there only while the dog is clinically stable and still eating – a dog that is anorexic or crashing needs the diagnostics first, not a slower diet plan.

The trial is a nutrition decision, not only an allergy test

An elimination trial asks one question: does this dog get better when it stops meeting the proteins it has been meeting. Asking it cleanly takes three things: a protein the dog has genuinely never eaten, an ingredient list short enough to read, and exclusive feeding – nothing else in the dog at all. The 2026 statement puts the minimum at two weeks of exclusive feeding per diet, with a clinical response usually visible within 10 to 14 days, and suggests being prepared to work through at least three different diets before concluding the dog is not food-responsive. All of that is sound, and it is the same logic behind an elimination trial run for a suspected food allergy.

What gets lost is that the same restriction is also a formulation change. Every protein you remove takes its mineral and vitamin profile with it, and a diet built to answer an immunological question was never designed to hit a nutrient target. Over three or four weeks that is irrelevant. The moment it stops being irrelevant is written into the guidance itself: a diet that induces remission should be fed for at least 12 weeks before anyone tries to transition away from it, and dogs with protein-losing disease may stay on it indefinitely. Three months on a deliberately narrowed bowl, potentially followed by years, is exactly the window in which the gaps a restricted raw diet tends to leave start to matter. The consensus statement itself flags malnutrition as a live risk in these dogs and asks clinicians to score body and muscle condition and watch albumin, particularly in protein-losing cases.

What one protein actually puts in the bowl

Take the reference dog NRC 2006 uses for its own worked examples: a 15 kg adult eating about 1,000 kcal a day. Put it on the strictest version of a novel-protein raw trial – whole wild rabbit, meat and bone, nothing else – and the Raw & Well IngredientDatabase, which is what the calculator runs on, says the day's food weighs about 925 g. Here is what that delivers on the eight minerals where the arithmetic is least forgiving, measured against the NRC 2006 adult recommended allowances – and what changes when you add the one thing almost every raw feeder adds next: 5 percent beef liver, about 46 g.

Four rows are worth reading twice. Iodine at 2 percent of allowance is not a rounding error; land animal muscle and bone carry almost none of it, and a diet with no fish, no egg and no kelp in it has no meaningful iodine source at all. Manganese behaves the same way. Zinc lands at roughly half, which is the shortfall that shows up on the dog before the bloodwork does – crusty muzzle, dull coat, slow healing.

And calcium runs to about 10.2 g per 1,000 kcal. FEDIAF sets the adult nutritional maximum at 6.25 g per 1,000 kcal, so this bowl sits over that fence by more than half again, with a calcium-to-phosphorus ratio of exactly 2.0:1 – the top of FEDIAF's adult band, with nothing left in hand. That is not a rabbit problem. It is the same effect described in how much calcium that bone is really carrying: a whole carcass is 28 percent bone by weight, and bone percentage is a poor predictor of the calcium underneath it. FEDIAF adds the sting in a footnote – as calcium approaches the nutritional maximum, zinc and copper may need to go up, because high calcium interferes with absorbing them. This bowl runs calcium above the maximum and zinc at half allowance simultaneously.

What liver fixes, and what it leaves exactly where it was

Compare the two columns. Adding 46 g of beef liver takes copper from 12 percent of allowance to 312 percent – it does not close that gap so much as vault over it, which is what liver does to copper in any diet. It also rescues the nutrient missing from the table entirely: whole rabbit carries no vitamin A in the database at all, and 46 g of beef liver brings in roughly 2,290 µg of retinol, comfortably past an adult dog's daily requirement and into the territory where how much liver a bowl can carry becomes its own question.

The trace minerals are where it stops helping. Manganese goes from 8 to 19 percent of allowance. Iodine goes from 2 to 4. Zinc creeps from 49 to 59. So the reflex fix is a real fix for copper, vitamin A and the B vitamins, and close to no fix at all for manganese, iodine and zinc. Liver is not a multivitamin; it is a specific and rather violent correction to a specific set of nutrients, and it should be added knowing which set.

It is also the point where the trial's logic and the diet's logic collide, because beef liver is a second protein. In a dog being tested for beef reactivity, adding it is not a nutritional decision at all – it is contaminating the experiment. Where the gaps genuinely cannot be closed inside a one-protein bowl, closing a gap food can't close with a measured supplement keeps the trial clean in a way that a second meat does not.

The volume problem nobody warns you about

There is a plainer problem sitting on top of the mineral arithmetic, and it catches people out first. Wild rabbit carries about 21.5 g of fat per 1,000 kcal – genuinely lean, which is often why it is chosen for a gut case. Lean food is bulky food. To reach 1,000 kcal that 15 kg dog has to eat about 925 g of it, which is 6.2 percent of its body weight in a day, two to three times the portion most raw feeders are used to weighing out.

Feed the customary 2 to 3 percent of body weight instead and the dog gets somewhere between 324 and 486 kcal – a third to a half of what it needs. A dog with chronic enteropathy that is already absorbing poorly, being fed a third of its calories in the name of a clean trial, will lose weight, and the weight loss will be read as the disease progressing rather than the plan under-feeding. The fix is not exotic: raise the calorie density with a fattier cut of the same protein, or feed the volume properly and split it across more meals. But it has to be a decision someone made, not an accident of choosing the leanest protein on the shelf.

What should you not feed a dog with IBD?

During a trial, the answer is short: anything the dog has eaten before, and anything you cannot account for. That means no treats, no flavoured chews, no table scraps, no flavoured medications or toothpaste, and no second protein slipped in for nutritional reasons – each one is a variable that makes the result unreadable. Outside the trial, there is no universal banned list for these dogs; fat is the ingredient most often worth restricting where pancreatic involvement is suspected, and rapid change is worth avoiding in every case. Chicken is not inherently a trigger. It is simply the protein most dogs have already met, which disqualifies it as a novel protein and nothing more.

Does pumpkin help a dog with IBD?

Pumpkin is a mild source of soluble fibre and it can firm a loose stool, which is why it is recommended so widely, but it is not a treatment for chronic enteropathy and it will not change the underlying inflammation. Two cautions apply during a trial. It is another ingredient, so it is another variable, and adding it mid-trial muddies the answer you are paying weeks to get. And it is bulk without much energy – useful when a dog is over-condition, unhelpful when the problem is a dog that cannot eat enough. Introduce it, if at all, once the trial has given you its result.

The raw-specific caveat your vet will raise

This part belongs in an article on a raw feeding site precisely because it is inconvenient. The 2026 consensus statement recommends that faecal culture – a test it otherwise advises against running routinely – be considered specifically for dogs with defined risk factors, and it names a raw meat-based diet as one of them. The panel's wording is that support for an infectious bacterial cause of chronic enteropathy is lacking, but that feeding a raw meat-based diet is a risk factor, for Campylobacter among others, and carries zoonotic risk.

That is not a reason to abandon a raw trial. It is a reason to expect the question, to answer it honestly, and to be scrupulous about handling in a household where an immunosuppressed dog is being managed and someone may also be immunocompromised. It also means something practical. If your vet asks for a faecal culture, they are following current guidance, not expressing an opinion about raw feeding. The fastest route through that conversation is to hand over the ingredient list.

Coming off the trial

A trial is only worth the weeks it costs if the end of it is planned as carefully as the start. If the dog improved, the diet has answered the diagnostic question and the job changes. The same bowl now has to hold as a long-term diet. That is the point where the iodine, manganese and zinc figures in the table above – and the vitamin A that a meat-and-bone bowl simply does not contain – stop being a footnote and become the thing to fix. If the dog did not improve, the value is in having eliminated diet cleanly, which is only true if nothing crept in.

Either way, one change at a time from here. Reintroduce a single ingredient, hold it long enough to mean something, and write down what the stool did. Where loose stool that starts after a diet change settles within days, it is usually adaptation; where it arrives with weight loss, blood, or a dog that seems unwell, it is not, and it is not a stage to wait out.

When the food question stops being the question

Some things end the arithmetic and start the phone call:

• Blood in the stool, black tarry stool, or persistent vomiting
• Weight loss continuing despite a plan the dog is actually eating
• Fluid retention or a swollen abdomen, which can signal the low albumin of protein-losing disease
• A dog that stops eating – the point at which the consensus statement says a diet trial is no longer the right first step

Between those episodes, running the bowl against the NRC 2006 allowances is genuinely useful work, and it is work an owner can do. During them, it is the wrong tool, and the honest version of this article is that a spreadsheet has never once stopped a flare.