Raw Feeding and Service Dogs: The Public Access Question Nobody Asks

Walk into a VA hospital with a therapy dog and that dog was bathed within the last twenty-four hours. Her nails were trimmed before she left the house. There is a clean towel in the handler’s bag to put between the dog and any patient she sits with. And no raw meat has been fed to her — or to any other animal in that handler’s home — because the organization she volunteers for prohibits it.

Walk into the same building with a service dog and none of that applies. Not one line of it. Nobody may ask what your dog ate, and nothing in federal law would let them.

Both dogs are healthy. Both dogs are clean. Only one of them has a rule.

I did not know this existed until a therapy dog handler explained it to me in a VA waiting room, and she was clear about something I want to pass on exactly as she said it: the bathing and the no-raw-meat were not her personal preference. They were her organization’s requirements, and she could recite them. (If the distinction between the two jobs is new to you, we cover it in detail in service dog versus therapy dog.)

The service dog beside her — mine, yours, anyone’s — is held to nothing. That is what this article is about. Not whether raw feeding is good for a dog. Whether it is a fair thing to bring into a room full of people who cannot choose.

The three sets of rules in one building

This is the part worth understanding before anything else, because almost nobody knows the layers exist. In a single hospital corridor, three completely different standards are operating at once.

Read the middle row against the top row. The private nonprofit is stricter than the federal government — considerably stricter — and the handler who signed up for it did so voluntarily.

Now read the bottom row. A service dog has the broadest access rights of the three. He goes where the therapy dog goes and also into restaurants, grocery stores, courthouses and school classrooms, by federal right, without an invitation and without a screening. And there is no organization above him, no diet standard, and no lawful question.

That is not carelessness by anyone. The Americans with Disabilities Act regulates access, not animal husbandry, and it was written long before raw feeding became a mass-market category. The same two-question limit that protects you when a business tries to deny your dog access is the reason nobody screens him either. But the practical result stands:

The rule that governs the therapy dog exists because somebody measured the risk. The absence of a rule for the service dog is not evidence that the risk is different. It is evidence that nobody wrote the rule.

Why the therapy dog rule exists at all

The rules did not appear from nowhere. They came from a study of therapy dogs specifically, and it is worth knowing what it found.

Researchers at the Ontario Veterinary College followed 200 healthy therapy dogs across Ontario and Alberta for eighteen months, collecting stool samples every two months and recording what each dog ate. About one in five ate raw meat at some point during the study.

The authors’ recommendation became the industry standard: “Dogs fed raw meat should be excluded from AAI programmes, particularly when programmes involve interaction with humans at high risk of infection.”

Three honest caveats belong with that, and I would rather give them to you than have you find them yourself later. This was an observational study, not a trial — raw-fed dogs may differ in other ways too. The confidence interval around that 22.7 is enormous, running from 3 to 59, which means the effect is real and large but not precisely pinned down. And it found no association with three of the organisms hospitals worry most about, including MRSA. That last one matters. An article that left it out would be selling you something.

The finding almost nobody quotes is the second one. Cephalosporin-resistant E. coli, odds ratio 17.2, with a much tighter confidence interval than the Salmonella figure. A hospital’s real fear is not a dog with a stomach upset. It is a resistant organism arriving in a building full of people on immunosuppressants.

And every one of those 200 dogs was healthy. You cannot look at a dog and tell.

Lefebvre SL, Reid-Smith R, Boerlin P, Weese JS. Evaluation of the risks of shedding Salmonellae and other potential pathogens by therapy dogs fed raw diets in Ontario and Alberta. Zoonoses and Public Health, 2008.

What is actually in the food

The FDA ran a two-year sampling program from October 2010 through July 2012, testing more than a thousand pet food samples in one laboratory, under one protocol. Raw pet food was added in the second year: 196 samples of commercially available raw dog and cat food.

Roughly one raw sample in thirteen carried Salmonella. One in six carried Listeria. Every other category was essentially clean. This is the cleanest single comparison in the whole literature precisely because it was one lab doing one thing at one time.

Two independent surveys since have looked at product on shelves rather than in a federal lab.

A Dutch study analyzed 35 commercial frozen raw products from eight brands. Listeria monocytogenes was in 54 percent of them. Salmonella in 20 percent. E. coli O157:H7 — the Shiga toxin producer — in 23 percent. Extended-spectrum beta-lactamase-producing E. coli, which is to say antibiotic-resistant, in 80 percent. Two products contained Toxoplasma gondii. The authors note that freezing handles the parasites; it does nothing about the bacteria.

A Swiss study at the University of Zurich tested 51 raw pet food samples. Seventy-three percent exceeded the recommended level for enterobacteria. Sixty-one percent carried ESBL-producing bacteria. And twice, they found E. coli carrying mcr-1 — the gene that confers resistance to colistin.

That last one deserves its own sentence. Colistin is a last-resort antibiotic in human medicine, an old and fairly toxic drug kept in reserve for infections that nothing else will touch. The gene that defeats it moves easily between bacteria. It was found in dog food, in a country with some of the strictest veterinary antibiotic rules in the world.

Van Bree FPJ, Bokken GCAM, et al. Zoonotic bacteria and parasites found in raw meat-based diets for cats and dogs. Veterinary Record, 2018. · Nüesch-Inderbinen M, Treier A, Zurfluh K, Stephan R. Raw meat-based diets for companion animals. Royal Society Open Science, 2019.

What comes back out of the dog

Contamination in a package is a handling problem. You can, in principle, manage a handling problem with bleach and discipline — that is the same lesson the dog food recall history teaches about every other contamination event. What you cannot manage is the dog.

A Swedish study kept the design simple enough to hold in your head: fifty dogs in one municipality, twenty-five fed raw and twenty-five fed kibble, one stool sample each.

Thirteen dogs out of twenty-five were shedding an antibiotic-resistant organism. Not sick. Shedding.

A Brazilian group compared 46 raw-fed dogs to 192 on commercial dry food and found the raw-fed dogs almost thirty times more likely to test positive for Salmonella, including multidrug-resistant strains. They also isolated toxigenic Clostridioides difficile from raw-fed dogs only.

But the part of that study I keep coming back to is not the bacteriology. It is the survey they ran on the owners.

• 69.3 percent chose the diet because they considered it more natural
• 87.9 percent believed it posed no health risk to their animals
• 98.8 percent believed it posed no health risk to humans
• 34.8 percent — more than a third — had at least one person at high risk of infection in contact with the dog

More than a third of those households contained someone vulnerable, and virtually none of the owners had been given any reason to think there was a human risk at all.

That is not a study about bacteria. That is a study about what people have been told, and by whom.

Runesvärd E, Wikström C, Fernström LL, Hansson I. Presence of pathogenic bacteria in faeces from dogs fed raw meat-based diets or dry kibble. Veterinary Record, 2020. · Viegas FM, Ramos CP, et al. Fecal shedding of Salmonella spp., Clostridium perfringens, and Clostridioides difficile in dogs fed raw meat-based diets in Brazil. PLoS One, 2020.

Four people in England

In 2021, UK public health investigators published a cluster of Shiga toxin-producing E. coli O157:H7. Four confirmed cases, median age six years.

“All four cases reported diarrhoea, three of which had bloody diarrhoea accompanied by vomiting. Three cases were hospitalised and one case developed HUS and subsequently died.”

The vehicle was raw tripe fed to the household dogs. The same strain was found in the tripe.

I am going to state the limits of that in the same breath, because it deserves them. Four cases. One country. One product. The authors themselves say raw pet food was a potential source. This is not a mortality rate and it does not tell you the odds of anything. It is a demonstrated route, from a bag of dog food to a child, published in a peer-reviewed journal by a national health agency.

Kaindama L, Jenkins C, Aird H, Jorgensen F, Stoker K, Byrne L. A cluster of Shiga Toxin-producing Escherichia coli O157:H7 highlights raw pet food as an emerging potential source of infection in humans. Epidemiology & Infection, 2021.

“But mine is a complete commercial raw diet”

This is the sentence I hear most, and it is the one place where the evidence is bluntest.

In December 2025, researchers analyzed 33 pre-prepared raw dog foods from six manufacturers, every one of them labeled complete, and measured what was actually in them.

Their conclusion: “None of the preprepared raw dishes covered the mineral requirements of dogs.”

Two limits, stated plainly. These were products on the Hungarian market, measured against European recommendations rather than the AAFCO profiles used here. So this is not proof that American raw foods fail. What it is: thirty-three products sold as complete, none of which met the standard they were sold against — and no equivalent survey of the US market has been published at all, which is its own answer to the question.

Home-prepared raw does no better. A German veterinary nutrition service analyzed 95 home-prepared raw rations and found that 60 percent had at least one nutrient imbalance. One in ten supplied less than a quarter of the calcium a dog needs. About half were short on iodine. More than half were short on copper and zinc.

If you want to see what a mineral this small does to a dog that big, our article on calcium in large breed puppy food walks through the mechanism, and the nutrition pillar covers what a working dog’s requirements actually are.

There is also a hazard peculiar to meat-based feeding that has nothing to do with bacteria. Fourteen dogs turned up in one case series with thyroid hormone levels far above normal, and imaging showed their own thyroid glands had shut down — the hormone was coming from outside. The cause was thyroid tissue left attached to neck meat and gullet in all-meat foods and treats. Every dog returned to normal within about four weeks of the food being withdrawn. That is a hazard of the ingredient, not the temperature, and it is a good illustration of what “just meat” can quietly include.

Moravszki L, Krizsán J, Freiler-Nagy Á, et al. Assessment of mineral adequacy in preprepared raw dog foods labeled as complete. Scientific Reports, 2025. · Dillitzer N, Becker N, Kienzle E. Intake of minerals, trace elements and vitamins in bone and raw food rations in adult dogs. British Journal of Nutrition, 2011. · Broome MR, Peterson ME, et al. Exogenous thyrotoxicosis in dogs. JAVMA, 2015.

Bones, and the thing pet studies never measure

Here the honest answer is less dramatic than either camp claims.

A study of 129 dogs that had swallowed bones found most of them did fine. Forty-two of forty-five esophageal bones were cleared without surgery. Sixty-two of eighty-four bones that reached the stomach were simply left there to dissolve, with no reported complications. The real risk factor was time — erosion became far likelier the longer a bone stayed lodged.

So no, bones are not routinely fatal, and I am not going to tell you they are.

But look at what a good outcome actually consists of. An emergency visit. Sedation, or an endoscopy. A dog who is not working for several days at least, and a bill.

For a pet, that is a bad week. For a handler whose dog is their balance, their alerting, their ability to leave the house — a bad week is something else entirely. That is the cost no pet-focused study measures, because a pet owner who loses their dog for six days loses their dog for six days. A handler loses the thing that lets them go outside. It is worth weighing before you hand over something that can lodge.

Barash NR, Lashnits E, Kern ZT, Tolbert MK, Lunn KF. Outcomes of esophageal and gastric bone foreign bodies in dogs. Journal of Veterinary Internal Medicine, 2022.

The honest case for raw, and the honest case against the rules

I am not going to pretend the other side has nothing.

Raw diets are measurably more digestible than kibble. A 2024 study at the University of Illinois fed ten adult beagles five different food formats and found that protein digestibility was higher for frozen raw and freeze-dried raw than for the others, while the extruded kibble came out lowest on dry matter, protein, fat and energy digestibility of anything tested. That is a real, measured advantage and you deserve to know it exists.

What it is not: evidence of a health benefit. Ten dogs, short feeding periods, and the study measured digestibility and stool quality — not disease, not longevity, not anything you would take a dog to a vet about. The authors do not claim otherwise. The World Small Animal Veterinary Association’s summary of the whole benefit question is that “there is no evidence that raw meat-based diets provide health benefits over commercial or balanced homemade cooked diets.”

And there is one more piece of evidence that cuts against everything above, which I would rather hand you myself. The CDC’s own guideline on animals in health care facilities says: “No evidence suggests that animals pose a more significant risk of transmitting infection than people,” and “No reports have been published regarding infectious disease that affects humans originating in service dogs.”

That is a fair quote and it is in a federal guideline. Here is why it is weaker than it looks. It was written in 2003 — before the Ontario therapy dog cohort, before the FDA’s sampling study, before the Dutch and Swiss product surveys, before the English cluster. And “no reports have been published” is not the same as “no cases occurred.” There is no surveillance system anywhere that would connect a hospital-acquired Salmonella case back to a service dog, because nobody records what the dog ate. It is absence of evidence in a system with no detector.

That is the fairest statement this article can make, and it is the one I am making.

Who profits from which answer

Every position in this argument is held by somebody with something to gain, and you should know which is which before you weigh any of it.

So apply the test to the studies rather than the opinions. The FDA sampling program was a federal laboratory. The Ontario cohort was a veterinary college. The Dutch and Swiss surveys were universities. The English cluster was a national health agency. None of them sell dog food, and they point the same direction.

The one study in this article that shows raw in a favorable light came from a university lab that publishes work funded by manufacturers across the whole category — including the kibble makers. That does not make it wrong. It makes it worth saying out loud, which is the same standard I applied to everything else here.

When people with opposite financial interests disagree, look for the finding neither of them paid for.

Where this leaves you

I am not going to tell you what to feed your dog. That is not what this article is for, and anyone who gets to that conclusion in one page was never really weighing it.

What I will say is that the decision is genuinely yours, in a way that most decisions about service dogs are not. Nobody will stop you at the door. Nobody may ask. There is no organization above you, no renewal form, no evaluator. The therapy dog handler I met has all of those, and she still knew every rule by heart.

So here is what the evidence actually supports, and nothing more:

• A raw-fed dog is substantially more likely to shed Salmonella and antibiotic-resistant E. coli than a dog on cooked food, and you cannot tell by looking. The 200 dogs in the Ontario study were all healthy.
• The risk lands hardest on the groups the CDC names for raw pet food: children under 5, adults 65 and older, people with weakened immune systems, and pregnant women. On a site whose readers are disproportionately disabled, often immunosuppressed by the medication that lets them function, and in medical buildings more often than most — that is not a footnote. That is us, and the people sitting next to us.
• Two independent bodies that had to write a real rule, the hospital epidemiologists at SHEA and the international animal-assisted intervention body IAHAIO, both landed on 90 days between the last raw meal and a visit. Neither shows its working. It is a margin of safety rather than a measured shedding time, and it is the only published answer to the question raw feeders actually ask.
• Careful handling does not solve it. Every kitchen protocol in the world addresses the counter. The dog is the vector. That is the whole reason Pet Partners bars an animal from the program if raw is fed to any animal in the household.

And if you are weighing this, the conversation worth having is with your own veterinarian, about your own dog, before anything changes. Not with a forum, and not with anyone selling either answer. We keep our own standards for how any of this gets recommended public for the same reason.

The dogs in this article did not choose any of it. Neither did the people in the waiting room.

Frequently asked questions

Does the ADA allow a business or hospital to ask what my service dog eats?

No. Under 28 CFR §36.302(c), staff may ask only two questions: whether the dog is required because of a disability, and what work or task it has been trained to perform. There is no third question. A service animal may be excluded only if it is out of control or not housebroken — neither of which describes a healthy dog shedding bacteria. The decision about diet is entirely the handler’s, which is precisely why it is worth thinking about carefully.

Why are therapy dogs banned from raw food when service dogs are not?

Because therapy dogs are governed by private organizations that can set any standard they like, and service dogs are governed by a federal access law that does not address animal care at all. Pet Partners prohibits raw protein diets for registered therapy animals and excludes animals from households where raw is fed. No equivalent body exists for service dogs, and the ADA was written to protect access rather than to regulate husbandry.

How long does a dog shed bacteria after eating raw food?

Nobody has published a definitive answer. What exists are two exclusion periods set by bodies that had to write a workable rule: the Society for Healthcare Epidemiology of America recommends excluding any animal fed raw or dehydrated animal-origin food, chews or treats within the past 90 days, and IAHAIO sets a 90-day suspension from animal-assisted intervention programs. Both are margins of safety rather than measured shedding times, and neither publishes its reasoning.

Can freezing raw dog food make it safe?

Freezing addresses parasites. It does not address bacteria. The Dutch study that found Toxoplasma gondii in commercial raw products specifically noted that parasitic infection becomes a risk if non-frozen meat is fed — while the same frozen products carried Listeria monocytogenes in 54 percent of samples and antibiotic-resistant E. coli in 80 percent. Freezing is not a kill step for bacteria.

Are commercial raw diets labeled complete and balanced nutritionally adequate?

Not reliably. A 2025 analysis of 33 pre-prepared raw products labeled complete found that none of them met mineral requirements — selenium was below the minimum in every single food, while calcium, phosphorus and iodine exceeded maximums in a third to a half of products. That study was conducted on the European market, and no equivalent survey of American raw products has been published. Our article on what “complete and balanced” actually means explains what the phrase does and does not guarantee.

Should I stop giving my service dog bones and animal-origin chews?

That is a decision for you and your veterinarian, but the categories are broader than most handlers realize. The infection-control standards that apply to therapy animals cover bones, rawhides, pig ears, bully sticks, freeze-dried treats and raw-coated kibble — anything of animal origin that has not been cooked, high-pressure pasteurized or irradiated. Separately, a study of 129 dogs found most swallowed bones resolved without surgery, but the ones that caused damage were the ones that stayed lodged longest.

What does the CDC recommend about raw pet food?

The CDC’s position is direct: it does not recommend feeding raw pet food or treats to dogs and cats. It names children under 5, adults 65 and older, people with weakened immune systems, and pregnant women as being at higher risk of illness, and advises washing hands before and after handling raw pet food, cleaning any surface it touches, storing it sealed and separate, thawing in the refrigerator, and discarding leftovers left at room temperature.

Do veterinary organizations agree about raw feeding?

Broadly, yes, and it is worth knowing they have financial entanglements worth weighing. The AVMA discourages feeding raw or undercooked animal-source proteins to dogs and cats. AAHA states that it cannot support or advocate raw protein diets and singles out therapy animals as especially important. WSAVA finds no evidence of health benefits over cooked diets. A Canadian veterinary review applying evidence-based grading concluded there is a lack of large cohort studies on either risk or benefit, while still finding enough evidence to warrant the conversation — which is a fair summary of where the science actually stands.

This article names no product and recommends no brand. Nothing on this site is sponsored, gifted, or affiliate-driven where a health claim is involved. Sources are linked in full above so you can read them yourself rather than take my word for any of it.