A trained dog takes part in a Covid-19 sniffing test.
CNN  — 

Dogs can be trained to detect Covid-19 by sniffing human sweat, according to a proof-of-concept study published on Thursday.

Detection dogs are already a common sight in airports and some other public places – usually, they’re sniffing out drugs, weapons or explosives.

But specially trained dogs have also been taught to detect infections and diseases, including colon cancer, malaria and Parkinson’s disease.

Now, many countries worldwide are exploring the possibility of using dogs as a rapid, reliable and relatively cheap way to prescreen people for Covid-19 or perform rapid checking in certain circumstances, like at airports.

In the United Kingdom, a team from the London School of Hygiene and Tropical Medicine are training six dogs in the hope that they will be able to detect Covid-positive people – even if they have no symptoms.

In Finland, a group of sniffer dogs trained to detect Covid-19 began working at Helsinki Airport in September in an effort to identify those who have contracted the virus and in Chile police dogs are being trained to sniff out Covid-19 in humans.

Scent detection of Covid-19

In this new study, researchers based in France and Lebanon took sweat samples from the underarms of a total of 177 patients from four hospitals in Paris and one in Beirut – 95 had tested positive for Covid-19, while 82 had tested negative. It was important that both the negative and positive samples came from the same hospitals so the dogs wouldn’t be picking up on a particular “hospital odour,” the researchers said.

Using some of the sweat samples, they trained 14 dogs that had been working as explosive detection dogs, search and rescue dogs or colon cancer detection dogs to take part in the study. The scientists did not use dogs trained to detect illicit drugs because they could not rule out whether study patients who had provided sweat samples had used prohibited substances.

Six dogs – Guess, Maika, Gun, Oslo, Bella and Jacky – had their sniffing abilities formally tested in the study. Five of them were Belgian Malinois, a common breed for French working dogs, and Jacky was a Jack Russell terrier.

Shown here is one of the detection dogs that took part in the study.

The researchers asked the animals to detect a positive Covid-19 sweat sample in one olfactory cone from a lineup of three to four cones that contained negative, or mock, samples. The scientists believe the dogs pick up on a specific scent produced by volatile organic compounds that are generated by catabolites, substances produced by the replication of the virus that escape the body through sweat.

During the testing period, the dogs did dozens of trials, with a success rate of between 76% to 100%. Jacky and Bella, the two dogs that specialized in detecting colon cancer, had a 100% success rate in the 68 tests they completed.

Moreover, during the testing, two samples collected from individuals who tested negative were repeatedly marked by two of the dogs. The relevant hospitals were informed, and in subsequent tests these individuals tested positive.

The study was published in the journal PLOS One on Thursday.

Two companies – DiagNose and Cynopro Detection Dogs – provided dogs that can detect explosives, free of charge for the study, said lead author Dominique Grandjean, a professor and head of the equine and carnivores clinical sciences department at the National Veterinary School of Alfort in France. Of the 44 people listed as authors on the study, six had their salaries paid by these and two other companies involved in training dogs or supplying dogs for security purposes. Another co-author worked for a company that supplied some materials used to take samples, Grandjean said.

When asked about potential conflict of interest or competing interests, he said these individuals were employees of these companies and volunteered their help and expertise for the study. These companies did not have any role in the design of the study, collecting data or analysis. And the study said there were no patents, products in development or marketed products associated with this research.

‘Complementary tool’

By filming the trials, the researchers were able to understand why sometimes the dogs were unsuccessful. In one case, a horse walked close to a testing site that was situated at a veterinary school. However, the reasons for detection failure weren’t always obvious – but the researchers said this wasn’t necessarily a negative.

“Even if trained dogs are able to correctly discriminate symptomatic Covid-19 positive individuals from asymptomatic negative ones, they should not be considered a perfect diagnostic test – but rather a complementary tool,” the study said.

The study, the researchers said, was a “promising first step” in providing some evidence that dogs may be able to detect Covid-19 samples collected from sweat. However, the scientists said their study had some limitations, including that sweat samples had to be reused and researchers couldn’t rule out that the dog was memorizing the scent – although they didn’t think memory played a major role. Further research was needed, they said.

Get CNN Health's weekly newsletter

Sign up here to get The Results Are In with Dr. Sanjay Gupta every Tuesday from the CNN Health team.

Susan Hazel, a senior lecturer in animal behavior, welfare and ethics at the School of Animal and Veterinary Science at the University of Adelaide, said dogs would never replace diagnostic test, but some animals in training are showing promise, especially in detecting individuals who have Covid-19 and don’t show symptoms. She wasn’t involved in this research but is part of an international collaboration led by Grandjean and is involved in a project with Australia’s Border Force training 14 dogs to sniff out Covid-19. The results from those trials are expected to be published in early 2021.

This latest study is only the second to be published in a peer-reviewed journal that looks at dogs’ ability to sniff out Covid-19. More work is needed on larger samples, experts say. “We need to take a scientific approach,” Hazel added.

“It’s important not to go out too early with grand claims and small data sets,” James Logan, a professor and head of the department of disease control at the London School of Hygiene and Tropical Medicine, told Nature last month.

In a few cases, dogs have been known to get infected with the coronavirus, the study said, but, according to the US Centers for Disease Control, there is no evidence that animals, including dogs, play any significant role in spreading Covid-19. For safety, the researchers did not use any samples for training or testing the dogs within 24 hours of collection.

Editor’s note: This story has been updated with details about the companies that supplied some of the dogs and the limits of their involvement with the study. This version also includes outside comment from Susan Hazel on the study’s findings.