Sniffing Out Coronavirus: Working Dogs Can Detect COVID-19

When it comes to dogs, the nose knows. That’s why, early on in the pandemic when COVID-19 tests were scarce and the US Centers for Disease Control and Prevention (CDC) was still tinkering with mask guidance amid a critical shortage of personal protective equipment, small groups of researchers around the world started searching for canine alternatives to traditional COVID-19 screening.

by Tony McReynolds

WHEN IT COMES TO DOGS, THE NOSE KNOWS. That’s why, early on in the pandemic when COVID-19 tests were scarce and the United States’ Centers for Disease Control and Prevention (CDC) was still tinkering with mask guidance amid a critical shortage of personal protective equipment, small groups of researchers around the world started searching for canine alternatives to traditional COVID-19 screening.

Using dogs to sniff out disease isn’t a new idea. One study in 2004 found that dogs could identify patients with bladder cancer based on the smell of their urine. In a 2006 study, dogs were able to detect lung cancer by smelling patients’ breath. And sniffer dogs have long been used at airports, sporting events, and border crossings to screen people for explosives and illegal drugs.

Would it be possible to train those same dogs to detect COVID-19? Researchers in the United States, Great Britain, France, and Finland, among other countries, thought the answer might be yes. they set out to develop a fast-casual version of a COVID-19 test using dogs to do the testing.

In Finland, Kössi, a Spanish Galgo, and Lucky, a golden retriever, were old hands (or noses) at sniffng out sickness. they’d previously been trained to detect breast cancer and prostate cancer based on scent profiles. Last spring, they were given a new challenge by researchers at the University of Helsinki (UH): sniffng out SARS-CoV-2, the virus that causes COVID-19. After collecting urine specimens from family members of Finnish patients who had tested positive for COVID-19, lead researcher Anna Hielm-Björkman, PhD, associate professor of animal clinical research at UH, and her team trained the dogs to differentiate between positive and negative SARS-CoV-2 samples. In an unexpected development, the dogs were able to detect the presence of SARS-CoV-2 even before patients displayed any signs of illness; in this case, family members who gave urine specimens that either hadn’t been tested or had previously tested negative. When the dogs picked up on the presence of the virus in those samples, the researchers asked the donors to test again using the standard methods.

In every case, the donors retested positive. But how far in advance of the retesting did the dogs know? “We found that the dogs could see that a person was getting sick about four to five days before they got the disease,” Hielm-Björkman said. “that was really encouraging because it means that the sensitivity of the dogs is better than the tests.”

At the time, the COVID-19 tests typically returned results in two to three days, she added. the dogs, however, provided instant results.

In France, researchers were coming up with similar results to those of the Finnish team but chose to focus on a different bodily fluid. the French researchers trained dogs to scent SARS-CoV-2 in human sweat—specifically, armpit sweat.

Lead researcher Dominique Grandjean, DVM, PhD, HDR, head of the equine and carnivores clinical sciences department at the National Veterinary School of Alfort, said his team opted for the armpit samples because human sweat contains a strong chemical signal indicating a possible pathogen in the body, and it’s easy to collect. It has the added advantage of ensuring that the dogs wouldn’t be exposed to the actual virus, which, as far as scientists know, can’t be transmitted through perspiration.

The team trained 18 experienced detection dogs, including 8 Belgian Malinois shepherds who’d previously been trained to successfully nose out explosives and colon cancer. The dogs were trained to sniff sweat samples taken from the armpits of 360 infected and uninfected participants.

For the training, jars containing samples of perspiration were placed in a line, and then funnels were inserted into the jars to allow the dogs to put their noses close to the sample. During the trials, each dog identified between 15 and 68 samples. Four of the dogs achieved a perfect score of 100%, while the rest had an accuracy rate between 83% and 94%.

As in Finland, the French dogs were able to indicate a positive result for two samples that supposedly came from people who were not infected by COVID-19. Those people were immediately retested by traditional laboratory methods, and both results came back positive. In both cases, the dogs were able to detect the virus a week before the people showed positive in the lab test.

The French study found that “COVID-19-positive people produce [underarm] sweat that has a different odor for the detection dog than COVID-19-negative people.” Overall, the dogs averaged a 96% accuracy rate. So how does that compare to standard tests?

According to Grandjean, “the accuracy of lab tests depends on the type of [test] and the quality of the sample.” In their study, accuracy ranged from 70% to 95% for positive results and from 35% to 70% for negative results. “So it seems dogs are more accurate [than traditional lab tests] in our trial.”

Meanwhile, scientists at the University of Pennsylvania’s School of Veterinary Medicine were training dogs to detect COVID-19-positive saliva, urine, and sweat in a pilot training program launched in late April of 2020. Like their European counterparts, the Penn researchers reasoned that it might be possible that dogs (in this case, golden retrievers) could provide a noninvasive, four-legged method to screen people in airports, businesses, or hospitals—places where testing poses particular challenges—without depleting limited human testing resources.

Ensuring the safety of the dogs in all three studies was a primary concern. At the time, it was unclear if or how easily dogs could become infected with SARS-CoV-2. There had been several reports of pets—including several dogs—testing positive for the novel coronavirus, although such reports were rare and remain so.

Cynthia Otto, DVM, PhD, DACVECC, director of the Working Dog Center at Penn Vet, led the study. She said keeping the dogs (and handlers) safe from infection was top of mind, and the team based their safety protocols on the latest science-based guidance from the American Veterinary Medical Association and the CDC. According to the guidance at the time, Otto said, there was “no evidence that animals play a significant role in spreading the virus that causes COVID-19. That said, we’re taking all precautions to minimize risk and monitor for exposure.”

F1 Nose 2.pngThe most important precaution taken was that the virus used in the Penn trial was inactive, and the samples were kept in devices that prevented access, so there was no chance that dogs or humans would be exposed. In Finland, Hielm-Björkman said protecting the dogs’ health was one of the main reasons they used urine samples. “We’ve done enough research to say that dogs can’t get [COVID-19] and their handlers can’t get [COVID-19] from exposure to urine.”

So what’s next for COVID-19-sniffng dogs? Grandjean said that, following the successful results of their most recent trials with patients at the Greater Paris University Hospitals, “the French Ministry of Health wants to deploy dogs throughout the country.” Grandjean’s team is currently setting up several field tests at airports, shipping ports, and public events to figure out how that’s going to work logistically.

Speed is going to be an important factor, Grandjean added. “We are also working . . . on a way to train the dogs faster.” Currently, training an experienced sniffer dog to detect COVID-19 takes six weeks.

Although results for most of the studies aren’t in— Grandjean’s team has three papers in review and hopes to publish soon—sniffer dogs trained by his team have been deployed to detect COVID-19 in a number of countries,including Lebanon and the United Arab Emirates. In Florida, sniffer dogs are screening fans coming to Miami Heat basketball games.

Are we jumping the gun? Maybe.

Otto said that based on her team’s preliminary findings— their proof-of-concept study was published last April in the journal PLOS ONE—it’s uncertain how long it will be before dogs can reliably be counted on to accurately screen people for COVID-19. “It depends on the strategy for how people are screened. They may be able to reliably screen samples from people if they’re trained on suffcient numbers of both positive and negative samples and tested in a double-blind fashion on novel samples at regular intervals.”

There’s still a lot of work to be done before we can reliably count on dogs to screen live humans for COVID-19, Otto said. “We know there’s a learning curve for dogs to transition from [COVID-19] samples to screening humans. The next biggest challenge will be the availability of dogs and handlers to do the screening.”

As for deploying Penn’s COVID-19-detection dogs before all the facts are in, Otto had reservations. “We would not feel comfortable deploying dogs from our program without clear documentation of their sensitivity and specificity in double-blind testing with novel samples in the operational setting.” She said she’d also like to see plans for regular, ongoing training sessions for both dogs and their handlers, “just like is done with bomb and drug dogs.”

Which begs the question: How do the dogs do it? Hielm-Björkman freely admitted she has no idea. “We’re really not sure what they’re detecting at this point,” she said. Possibly they’re scenting something people with the virus are metabolizing in their urine, but that’s speculation. All she knows for sure is that it seems to be working so far—and better than any machine. “They’re not even comparable,” she said. “There’s no machine on Earth that can even come close.”

Hielm-Björkman also noted that additional research is needed to confirm whether the dogs are specifically detecting the presence of SARS-CoV-2 instead of another virus, but she said the preliminary results were very promising—so promising that the Finnish government deployed COVID-19-sniffng dogs at the Helsinki airport last September.

F1 Nose 3.pngIn addition to helping with the fight against COVID-19, Hielm-Björkman sees amazing potential for employing sniffer dogs during future pandemics to detect coronavirus at border crossings and in concentrated living facilities like nursing homes.

Grandjean has no problem deploying dogs trained by his team to act as prescreeners, with people who “smelled” positive undergoing traditional lab testing afterward. “the dog is reliable day after day, people can be sniffed without any intrusion, [and] it does not cost anything [besides] the dogs’ training and kibble.” Moreover, he said, dogs are ideal for smaller communities and countries that can’t afford mass testing.

And despite a lack of peer-reviewed research with definitive results, Grandjean remains committed to the idea of COVID-19 screening via sniffer dogs.

His reasoning is simple: “It works.”

Tony McReynolds
Tony McReynolds is AAHA’s NEWStat writer and editor.


Photo credits: MirasWonderland/iStock via Getty Images Plus, Aleksandr Zotov/iStock via Getty Images Plus



Subscribe to NEWStat