In both a randomized trial and a real-life study, trained dogs were able to detect positive samples of the coronavirus with similar accuracy to PCR tests, Molly Walker reports for MedPage Today.
Researchers find scent detection dogs have high accuracy with Covid-19
In spring 2020, researchers from the University of Helsinki trained four dogs to detect positive samples of the coronavirus, either from among negative samples or from confounding samples, which included other viruses, cancer, and diabetes.
Of the four dogs, two were male, and two were female, and their ages ranged from four to eight years. Three dogs were Labrador retrievers, and all had previously worked a scent detection role. Two dogs had detected "dangerous goods," one had detected narcotics, and one had detected canine cancer.
Once the dogs achieved an 80% success rate in detecting the coronavirus in training, the researchers moved on to a randomized validation trial. During the trial, a total of 420 skin swab samples, 306 negative and 114 positive, were used, and each dog was exposed to 140 "scent tracks." Some tracks had no positive samples, which the researchers said "mimics better the real-life situation in low-prevalence settings."
In the trial, the dogs had a 92% accuracy rate, with 92% sensitivity and 91% specificity rates, compared with a RT-PCR test. According to the researchers, there was only "minor variation between the dogs" when it came to accuracy, but they noted that the dogs were "less accurate" at detecting the alpha variant (36%) than the original wild-type coronavirus (89%).
However, the researchers noted that the dogs had only been trained with the wild-type coronavirus, and being able to detect the alpha variant even part of the time suggested that they could be trained to detect different variants in the future. "Our preliminary observations suggest that dogs primed with one virus type can in a few hours be retrained to detect its variants," the researchers wrote.
After the validation trial, the researchers tested the dogs' ability using a real-life cohort of 303 samples from passengers at the Helsinki-Vantaa International Airport. Overall, the dogs correctly identified 296 of the samples correctly, as well as 296 out of 300 RT-PCR negative samples.
There was one positive case among the samples, which the dogs did not detect. Due to this low prevalence of positive cases, the researchers also provided the dogs with 155 novel RT-PCR "spike" samples, and they were able to accurately identify 98.7% of them as positive.
According to a post-hoc analysis of the findings, if the 155 spike samples had been included with the 303 real-life samples, the dogs would have achieved a sensitivity of 97% and a specificity of 99% when it came to detecting the coronavirus.
Overall, the researchers said using scent detection dogs may be a "particularly appealing" way to screen Covid-infected individuals in public areas or using mass transit. (Walker, MedPage Today, 5/16; Kantele et al., BMJ Global Health, 3/14)