Skip to Main Content

It seemed like such a great idea. But at the end of the day, as they say, those dogs didn’t hunt. Or at least not well enough.

New research from Toronto throws into question the notion that canines could be used in hospitals to detect patients with a debilitating and highly contagious form of diarrhea caused by the bacterium Clostridium difficile, or C. diff.


Earlier infection control teams in the Netherlands and Vancouver, Canada, had reported success in using dogs to sniff out patients with C. difficile from those with other types of diarrhea, or to find the bacterium’s hard-to-eradicate spores in hospital rooms being cleaned after housing a C. diff patient. In each case, the journal reports were based on the performance of a single trained dog.

But the new research revealed problems when the detective powers of two dogs were compared. To put it succinctly, Chase, a border collie, and Piper, a German shepherd, didn’t agree enough of the time for the researchers to feel confident the dogs were reliable enough to use in the hospital.

“They weren’t always wrong the same way and they weren’t always right the same way. And when I see that variability as a researcher or even as a clinician, it scares me as far as generalizability moving forward,” said Dr. Jeff Powis, medical director of infection prevention and control at Toronto’s Michael Garron Hospital, and senior author of the paper.


“If they could tell somebody had C. diff without an actual stool specimen, that would have saved us an enormous amount of time,” said Powis, who had been excited about the potential of using dogs to detect C. diff. But the study’s findings, published this week in the journal Open Forum Infectious Diseases, has led Powis and his team to conclude canine C. diff detectives aren’t the answer.

“From our assessment, we’re stopping here,” he said.

Diarrhea can be caused by a multitude of pathogens — both bacterial and viral — as well as diet and even a reaction to some medications. But C. diff diarrhea is a particular concern in hospitals because of its potential to spread from patient to patient. Incoming patients who are suspected of having C. diff are typically put in isolation, with hospital staff required to don gowns and gloves every time they enter the room, said Maureen Taylor, a physician assistant on Powis’ team who led the study.

There is no point-of-care test that will rapidly allow hospital staff to rule C. diff in or out, and it takes a day or two to get test results from the laboratory. Getting a diarrhea specimen to send off to the lab for testing can take as long. “Nobody is anxious to take on that job,” Taylor said.

Vancouver General Hospital and VU University Medical Center in Amsterdam had each trained a single dog — a beagle named Angus at the former and a springer spaniel named Cliff at the latter — to detect C. diff, showing some impressive results they had published in medical journals.

But while those studies suggested that the famed olfactory powers of dogs — already used to detect illicit drugs, bombs, and even bedbugs — could be useful, research based on a single dog can’t answer the question of how reliable dogs are at this work in general.

The Michael Garron Hospital team worked with an animal specialist who found and trained Chase and Piper, using metal scent boxes that contained sponges on which stool had been applied. Some of the stool samples were from patients who had C. diff diarrhea. The dogs were given a reward when they successfully found the scent boxes that contained samples contaminated with C. diff.

In the study, dogs were easily distracted by food and water. Michael Garron Hospital

The work was done in a decommissioned ward, to acclimatize the dogs to a hospital environment.

After extended training came the study. While the dogs were reasonably good at finding C. diff specimens and ignoring specimens that didn’t contain the bacteria, they were occasionally wrong — missing some samples that contained C. diff and mistaking some negative samples as positive. More problematic, though, was that they didn’t make the same mistakes. If they’d both missed specimen 12, for instance, it might mean the concentration of C. diff spores in the stool sample was too low. But that was not the case.

And Chase, who had been at the work longer and had been quite accurate in training, underwhelmed.

“We were impressed at first with Chase. [But] by the time the study actually happened two years later and Piper came on, Chase did not perform in the study as well as Piper did,” Taylor said. She added: “I don’t know dogs very well, but I got the sense by the time we did the study Chase was kind of saying to us: ‘You know what? I’ve moved on.’”

There was also the issue of distractions.

“If there was a breakfast tray on that patient’s bed, [Chase] was way more interested in the breakfast tray than he was in trying to find C. difficile spores,” Taylor noted. Another problem: “The dogs found it hard to pass a toilet without drinking out of it.”

“I don’t know dogs very well, but I got the sense by the time we did the study Chase was kind of saying to us: ‘You know what? I’ve moved on.’”

Maureen Taylor, physician assistant on Powis’ team who led the study

Dr. Marije Bomers, the lead author of the Dutch study, said the Toronto research shows that some dogs are better at this work than others.

“This of course does not mean dogs in general are unable to detect C. difficile at the bedside,” she said, noting Cliff, the spaniel her group used, was very accurate.

But variability in the performances of individual dogs could make it difficult to organize a program of this type, acknowledged Bomers, an infectious diseases specialist. “This study further illustrates that the capability of one dog cannot simply be extrapolated to other dogs, complicating practical implementation of C. diff sniffer dogs on a larger scale.”

VU Medical Center encountered a different problem with its program — the hospital doesn’t have enough C. diff cases to make the dog program cost-effective, Bomers said.

Cliff, the Dutch C. diff sniffer dog, has been retired.

So have Chase and Piper, both rescue dogs. For Chase, all that work learning to detect C. difficile samples hasn’t gone unrewarded. He’s been adopted, Taylor said, by one of the hospital’s pharmacists.

  • These fluffy articles get a lot of media attention, because dogs are cute. This is typical of what they call Healthcare journalism lately. As long as the reporting requirements are easy for clever healthcare Industry providers to avoid, we have a serious problem. Hospital acquired infections and antibiotic resistance are killing a lot of people. Instead of fact based information articles with cure little doggies, get more coverage, and mislead the public. Perhaps these clever journalists should look at the number of reported deaths, they are low compared to the actual deaths, and connect the dots. Hospitals resort to blaming patients and silencing them, patient dumping and attributing the deaths to underlying causes. The Hospital Industry lobbyists told Congress that counting the cases was “government overreach.”
    This is how they Gas Light us folks, I bet this cute feel good story went viral on Facebook. The Facts are inconvenient and unprofitable. Thanks to this fluffy journalism, most Americans have no idea what their odds are of contracting one of these diseases. They can look it up, but the reporting requirements are vague by design. The Facts were bad for business!

  • Ignoring food should be a part of training. And the trainers should keep water handy for the dog because they need moisture to smell to the best of their ability. That knowledge should be part of the handler’s training so the dog doesn’t need to drink water from a toilet.

    I suspect the problem lies in isolating the C. diff itself from the ever-changing background odor of stool, or finding enough different training samples so the dog can learn which particular smell will be rewarded. More training in ignoring distracting odors like breakfast is needed, too.

    I attended a medical alert dog training seminar from someone who pioneered several innovative techniques and several “firsts” among sniffer dogs. (I won’t describe her in more detail without her permission.)

    She told a story about demonstrating an explosives detection dog for a group of dignitaries who were frankly skeptical of her or the dog’s abilities. They set up the demo in a restaurant that had a full meal at every place setting. And the dignitaries were gleefully lining the room, prepared for a great show.

    She brought the dog in to search. He went right to work, methodically and quickly checked the room, looking for the scent of a small tin filled with gunpowder that had been planted somewhere. I should note that he showed no interest in the food at all while working. Then, as he passed a particular table, he did a classic double-take, turned, and sat facing the table, looking at her. “Show me,” my friend said. The dog pointed his nose at a plate on the table, and looked back at her.

    My friend picked up the plate (still filled with food) and found the tin under the plate. She opened it. Someone had dumped almost all the powder out of it so there were only a few grains left. She looked around, said “Oh, someone must have dropped it by accident. It is almost empty.” She snapped the tin shut and said, “Any questions?”

    • The person who wrote this comment, has a perfect understanding of what the TRAINER needs to work on, as well as what the DOG needs to work on. The dog is only as good as its handler. A well trained dog will do what the dog in the restaurant did. And they will do it until they drop from exhaustion. Anyone who works in Search and Rescue trains their dog breed of choice to ignore, oh a rabbit went by here, a deer did too oh and a porcupine. They teach that dog to follow a scent that is played out in a cone shape. That is why, for every S & R, you will always see someone as a base operator, and at least 6 dogs fan out from that site. These dogs, like Certified Service dogs, will work with any handler other than their regular one. You ALWAYS have water for the dog. Why? 1. You don’t want your dog getting sick and 2. They need it to search correctly.

    • Sorry, I made a typo in my first statement. It should read :
      They teach the dog to follow a scent that is layed out in a cone shape.
      And I would like to add, it is a SPECIFIC SCENT that they are looking for, that they have been trained to follow. In S & R, that scent changes case per case, but it is still, biologically the same scent, the human scent.

    • Thank you, Jennifer. I’m still learning. The trainer in the scenario I described often says that the humans are harder to train than the dogs. But training the dogs correctly takes an almost obsessive attention to detail and record-keeping. Did I say “almost”? Scratch that.

  • Sounds to me that the dogs weren’t trained very well. And not all dogs are created equal. Why not go to a few dog clubs/organizations and identify dogs with proven track records for scent detection. Create an simple obedience test to identify the dogs that won’t be distracted by food or toilet water. Select 10 dogs and train them to detect C.diff.

  • “If there was a breakfast tray on that patient’s bed, [Chase] was way more interested in the breakfast tray than he was in trying to find C. difficile spores,” Taylor noted. Another problem: “The dogs found it hard to pass a toilet without drinking out of it.”

    Thanks for a good laugh! Who could have seen that complication coming? Only every dog owner! 🙂

Comments are closed.