The elimination diet, done properly: an 8-week protocol for dogs

An elimination diet is the only reliable way to diagnose a food-responsive skin or gut problem in a dog. There is no blood test, no saliva test, and no hair test that will do it for you — those have been studied repeatedly and they return positives for foods the dog has never eaten.

The protocol is simple. Executing it is not. Here's the version that survives contact with a real household.

Step 0: rule out the cheaper explanations first

Do not start an eight-week trial on a dog with an active skin infection or uncontrolled fleas. You'll spend two months and learn nothing, because something else is driving the itch.

Before day one you want: reliable flea control on every animal in the house, any ear or skin infection treated, and a rough sense of whether the problem is seasonal. If it disappears every November, it's environmental, not dietary.

Step 1: choose the diet

Two valid options, and one that isn't.

Novel protein. A protein and carbohydrate source the dog has demonstrably never eaten — venison, rabbit, kangaroo, ostrich. The catch is that "never eaten" is a high bar, and over-the-counter novel-protein foods are frequently contaminated with proteins not on the label. Multiple studies have found beef, chicken, or soy DNA in bags declaring none.

Hydrolysed protein. Protein broken into fragments too small for the immune system to recognise. Prescription-only, more expensive, tastes worse, and considerably more reliable — this is what most dermatologists choose when the answer actually matters.

Grain-free. Not an elimination diet. Grains are rarely the culprit; the common reactive ingredients in dogs are beef, dairy, chicken, wheat, and lamb — proteins, mostly. A grain-free bag with chicken in it changes nothing for a chicken-reactive dog.

Home-cooked trials are legitimate and sometimes necessary, but they must be balanced by a veterinary nutritionist if they'll run more than a few weeks. Growing dogs especially.

Step 2: eight weeks of absolute strictness

This is where trials die. Three mistakes account for nearly all of them:

1. Flavoured medication. Chewable heartworm preventatives, flavoured NSAIDs, pill pockets, dental chews. All are food. Swap to unflavoured formulations for the duration — ask your vet, there's almost always an option.
2. The other pet's bowl. If there's a second animal in the house, feed separately, in different rooms, and pick the bowls up. A cat's leftovers are a food trial's most common invisible failure.
3. The person who doesn't know. The neighbour who gives a biscuit, the child who shares toast, the walker with a pocket of treats. One conversation with everyone, on day one, and a note on the fridge.

Everything the dog eats gets logged. Everything. This is the reason a shared, timestamped log beats memory: in a two-adult household, neither person can tell you what the other fed on a Tuesday three weeks ago. PawSolids syncs across the household, so both of you are writing into the same timeline instead of two half-timelines that never meet.

Treats during the trial: use kibble from the trial diet itself, or a single-ingredient treat made from the same novel protein. Nothing else.

What each phase looks like

• Weeks 1–2. Frequently worse before better, especially if you're also tapering steroids. Stool may loosen while the gut adjusts to the new diet. Score the itch daily anyway — the baseline matters more than the noise.
• Weeks 3–4. GI signs usually improve first if food is a player. Soft stool firms up, gas settles, the ear rubbing eases off.
• Weeks 5–8. Skin is slow. The full response takes the entire eight weeks, and in some cases up to twelve, because inflamed skin needs a full turnover cycle to normalise. Judging at week four is the second most common way to reach a false negative.

By week eight you want a clean comparison: average itch score for weeks 1–2 versus weeks 7–8, plus stool grades across the whole run. That comparison is the trial result.

Step 3: the rechallenge nobody does

If the dog improved, you are not finished. You have one hypothesis and no proof — dogs improve during trials for reasons unrelated to food, including seasonal change, a course of antibiotics, or simply the removal of a high-fat diet.

Rechallenge: return the dog to the original food. If food was the driver, signs typically return within 3 to 14 days. If nothing happens in two weeks, food was not your answer, and you've saved yourself a lifetime of expensive prescription diets bought on a coincidence.

If signs do return, you have a confirmed diagnosis. From there you can reintroduce single ingredients one at a time — two weeks each, one protein at a time — and build a list of what's actually safe rather than avoiding an entire imaginary category forever.

Yes, deliberately triggering a flare feels wrong. It's a few uncomfortable days in exchange for certainty that shapes the next decade of feeding.

Talk to your vet before starting if your dog is under a year old, is pregnant or nursing, has a diagnosed condition requiring a therapeutic diet (kidney disease, pancreatitis, urinary stones), or is on long-term medication. Elimination diets interact with all of these.

The paperwork that makes it count

An elimination diet produces one deliverable: a defensible before/after. Daily itch scores, daily stool grades, the complete food list, the deviations you know about, and the rechallenge result.

Export it as a single PDF and bring it to the appointment. A dermatology consult that starts with eight weeks of scored data is a different conversation than one that starts with "we tried a different food and I think it helped a bit."

---

PawSolids is a health journal and pattern detector, not a diagnostic tool. Elimination diets are a clinical protocol — run yours with your veterinarian's input, especially the diet choice.