
NEW YORK —Trained dogs can detect SARS-CoV-2 among both symptomatic and asymptomatic individuals who are infected with the virus with high accuracy, a new study has found.
Currently, SARS-CoV-2 is most commonly diagnosed by nasopharyngeal sampling, which is invasive and uncomfortable, for either RT-PCR or antigen testing. Results from RT-PCR testing, for which saliva samples are also sometimes used, can also take time.
Previously, dogs have been trained to detect volatile organic compounds associated with conditions like cancer and degenerative diseases, as well as infectious diseases. In a new study, researchers from University Paris-Est in France and elsewhere examined whether dogs could be trained to detect SARS-CoV-2 infections. As they reported in PLOS One on Wednesday, they found that dogs could detect positive samples with high sensitivity and specificity, as compared to RT-PCR approaches, among both symptomatic and asymptomatic individuals. This suggests that such an approach could be adopted in situations where mass testing might be needed or where nasopharyngeal sampling is difficult.
"This type of detection can be very useful first on persons that cannot be tested with PCR samples … as dogs are noninvasive [and] give immediate results," first author Dominique Grandjean from the École Nationale Vétérinaire d’Alfort said in an email, referring in particular to elderly individuals with Alzheimer's or Parkinson's disease or children with autism or other disorders. She added that trained dogs could also be used in mass detection efforts, such as at airports or events.
She and her colleagues collected paired sweat and nasopharyngeal samples from 355 patients as part of the SALICOV-APHP study, which is examining new approaches to detect or diagnose COVID-19. About 41 percent of the participants were seeking SARS-CoV-2 testing because they were symptomatic, while others were undergoing testing because they had been in close contact with someone with COVID-19, or for travel or other reasons.
Of these 355 patients, 109, or nearly a third, tested positive for SARS-CoV-2 on nasopharyngeal RT-PCR testing.
Sweat samples from these patients — collected by holding surgical compresses under the arm for two minutes — were also tested using sniffer dogs. The dogs used in the study were from either French fire departments or the Ministry of the Interior of the United Arab Emirates and had undergone training. For dogs already trained in odor detection, training took about three weeks, but dogs new to the field needed about five to six weeks of training, though it depended on the individual dog's ability. Just what compounds the dogs are detecting is not yet clear, though Grandjean noted they are likely volatile organic compounds that arise due to the virus's activity in cells as well as the viral proteins themselves.
In the study — the dog handlers were blinded to each sweat sample's status — the dogs detected 126 positive cases. Of those, 106 were positive and 20 were negative by nasopharyngeal RT-PCR testing. The dogs did not detect SARS-CoV-2 in 209 cases, three of which had positive nasopharyngeal RT-PCR tests and 206 of which were negative by RT-PCR.
Overall, the dogs detected SARS-CoV-2 with 97 percent sensitivity and a specificity of 91 percent. For asymptomatic individuals, the dogs had a sensitivity of 100 percent and specificity of 94 percent.
A portion of the patients in the study also underwent antigen testing for SARS-CoV-2. Canine testing had higher sensitivity but lower specificity than antigen testing, the researchers found, suggesting that it was comparable to nasopharyngeal-based antigen testing but less invasive, and that trained dogs could represent an alternative approach to antigen testing.
However, they noted that dog-based testing would be limited by the need to certify that the dogs have been properly trained, as well as by the number of people who may be afraid of dogs.
Grandjean and her colleagues are now also investigating whether the dogs can detect the seven main SARS-CoV-2 variants without additional training.