Andy Zunz
Vice President of Media Strategy, NAVC

A dog’s incredibly sensitive sniffer could play a big role in coronavirus response efforts.
The University of Pennsylvania’s School of Veterinary Medicine launched a training program to determine if dogs can identify people with COVID-19 using scent detection, a useful tool in potentially screening asymptomatic carriers.
The program is designed to first determine if trained dogs can successfully identify COVID-19 positive samples in a research environment, and then if that ability can be transferred to a human in a real-life scenario.
“There’s strong evidence that this is going to work,” says Penn Vet Working Dog Center Director Cynthia Otto, DVM, PhD. “Our biggest question is: what are the confounders that makes it more challenging? Does influenza smell like coronavirus? We don’t know. What we’re hoping is that through the training, we can help the dogs hone in on a very unique odor, but until we have a lot of samples and have tested a lot of people we won’t know that.”

Dr. Cynthia Otto works with Casey at the Penn Vet Working Dog Center in a program unrelated to the COVID-19 training. Photo by Eric Sucar/Penn Vet
Here’s how the research will work, according to Otto:
• Learn the wheel “The very first step is working with an amazing trainer—Pat Nolan at Tactical Directional Canine—who’s currently pre-training these dogs. These dogs have learned how to find a synthetic odor, a training odor, and they’ve already learned how to use the scent wheel. So, once we get the samples and are ready to go, these dogs only have to learn the odors and not the process.”
• Obtain the virus “The very first point that we’re going to address is the proof of concept. Is there a unique odor associated with COVID-19 infection? We do that by taking COVID-19 positive samples that are collected at the Hospital of the University of Pennsylvania and the Children’s Hospital of Philadelphia and processed to inactivate any live virus. That’s important because we to be very safe.”
• Set the stage “Those samples are put in specialized containers that were developed by our colleagues at the Army Chem Bio Center, and they allow the odor to come out without the substance within coming out. Those are put on a scent wheel, which has another seal on it so the dogs can’t get to the container.”
• Find the scent “We start by allowing the dogs to sniff the container from a COVID-positive patient. That happens and suddenly a bell goes off, treats descend, and it’s a party. So, the dog’s like: whoa, what did I do to earn all of these amazing treats? But they have no idea. So we do that again and again with positive samples. Then, we introduce a sample that is not COVID-19 positive. They sniff, and think: where’s my treat? What’s different about this? They start to realize that there is something unique about the samples from COVID-19 patients. We don’t know what that is, and we may never know although our chemists are working to find out what odors are different.”
• Hit a moving target “So, that’s our first proof of concept, if the dogs can indeed do that. And we have pretty strong confidence based on all the evidence from other diseases that they can do that. That pulls the trigger for two different things to happen: one, our chemists and our colleagues who are developing an electronic nose move forward to see if they can develop a tool for screening. So, they will develop a chemistry-based screening mechanism and an electronic screening mechanism. Then, we will move forward to see if the dogs can recognize that odor when a person is holding it. What we set up in our initial proof of concept is a really isolated environment without a lot of distractions, so they had a very clear scent of the odor. When we move to a person, we have to make sure the dog is not overwhelmed or confounded by the odor of the person, so we need to do that with several people, in different positions, and moving.”
• Test new scenarios “If all of that works, then we think about how we can move to an operational setting. What does an operational setting look like? We equate to how it works in an airport with our explosive detection dogs. You walk through a security screening and you walk by a dog, and if you have something they are going to alert. They freeze or, depending on what is best in that environment, we might ask them to sit or lie down.
If all goes well, Otto plans to document the proper training technique to allow others around the country to train working dogs to identify the scent.
“What’s important is that they’re able to identify individuals who are positive for COVID-19, and that going to take a lot more time,” says Otto. “We need to make sure we’re dealing with dogs that are resilient, reliable, and comfortable in that environment, and that we’re putting all of those safety measures in place.”
Otto’s team at the Working Dog Center has learned from its research in the field of medical detection, particularly how dogs can detect ovarian cancer.
“We’ve also learned from our foray into diabetes alert, and currently we have a project where we’re doing infection detection, looking at the ability of dogs to detect biofilms,” says Otto. “All of this has helped us understand how we might apply the wonderful skill of the dog’s nose to see if we can somehow advance the screening for COVID-19.”
But Otto’s optimism comes from research being done across the pond, where Medical Detection Dogs UK has worked on malaria detection.
“They’ve even been able to confirm that the dogs can pick up the odor in asymptomatic patients, and that is really encouraging,” says Otto.
The scent detection would likely be used as a pre-screening if successful; dogs could quickly scan people in a crowded environment or waiting to enter a building and identify individuals who would need further testing. But Otto will be following the Scientific Working Group on Dog and Orthogonal detector Guidelines in order to maintain a high standard.
“But we know for some things the dogs are more accurate than the confirmatory test,” says Otto. “If we can document the dogs are reliable, maybe they will become an FDA-approved test. But until then, we would use them for screening.”

Poncho, 1 of the dogs in the Penn Vet COVID-19 research program. Photo by Pat Nolan/Penn Vet
The initial research is using 8 Labrador retrievers ranging from 1 to 1.5 years old that are “green,” or new to detection training.
It will only take a few weeks for these 8 dogs to determine whether COVID-19 has a traceable scent, but several more months after that for the full operational training to be completed.
“The other thing to consider is that we’re training a dog: there are many places that may potentially benefit from this. So, it’s just a drop in the bucket. How do we address the scaling of this out? We’re hoping to develop the techniques that will allow people to safely train dogs to do this,” says Otto.
She estimates that if all goes well, qualified working dogs trainers might be able to get other dogs up to speed in about 4 to 6 months after this program is completed.
The field of canine medical detection has great growth potential, Otto says, but opportunities for funding and attention at this scale do not come along often. She hopes to expand this canine detection tool in the future.
“We’re about to start a project with chronic wasting disease in deer. That’s a prion disease, so we’re very excited to think about how the dogs could be used to identify deer that are affected or areas that are affected,” says Otto. “We did a proof of concept looking at an invasive species, spotted lanternflies, where we trained the dogs to see if they could identify where the eggs are being laid by this invasive pest. And, indeed they were able to.
“My dream would be to have a proof of concept center where we could take an emerging disease and test to see if the dogs could do this and provide resources for others to train dogs at a scientific standard.”
More COVID-19 Coverage
• Dr. Jennifer Ogeer Discusses Antech’s SARS-CoV-2 Surveillance Program
• VetWatch Offers Veterinary Business Insight During Coronavirus Pandemic
• The Challenge of Determining Essential Veterinary Services Amid the Coronavirus Crisis