Dogs can discriminate between people infected with SARS-CoV-2 from those uninfected, although their results vary depending on the settings in which they are exposed to infected individuals or samples of urine, sweat or saliva. This variability likely depends on the viral load of infected people, which may be closely associated with physiological changes in infected patients. Determining this viral load is challenging, and a practical approach is to use the cycle threshold (Ct) value of a RT-qPCR test. The hypothesis was that dogs should have a specific Ct range at which they could detect people infected with SARS-CoV-2. Therefore, the objective was to determine this Ct range. Sweat samples and epidemiological data were collected from 89 infected and 289 non-infected individuals at real life settings (e.g. health centers, offices, football fields). To determine each person's infection status, the Norgen Biotek kit for RT-qPCR was used; targeting the N1 and N2 regions of the SARS-CoV-2 nucleocapsid N gene. The performance of 11 trained dogs was evaluated on sweat samples of 379 individuals to determine their sensitivity and specificity (± 95% Confidence Intervals; CI) in detecting SARS-CoV-2 infections. Additionally, the SARS-CoV-2 viral load was calculated from Ct values using a reference curve, and the Ct range at which dogs showed optimal performance was determined. Six dogs exhibited a marginal performance, as their sensitivity 95% CI overlapped with the region of randomness (50%). The remaining five dogs demonstrated sensitivity values between 67% and 87%, with none of their 95% CIs overlapping the randomness region. Regarding specificity, three dogs showed values between 87% and 92%, while all other dogs exhibited values of ≥ 90%. Dogs demonstrated higher detection accuracy in a range of Ct values between 18.49 and 29.17 for the N1 region and between 24.07 and 26.69 for the N2 region of the SARS-CoV-2 nucleocapsid gene. Detection significantly decreases for Ct values greater than 30 or less than 16, indicating an optimal range in which dogs are most effective. These performance values concur well with those reported for commercial rapid antigen tests for detecting SARS-CoV-2. Consequently, it is considered that using properly trained animals could offer a viable option to supplement existing diagnostic methods, allowing for rapid diagnosis while optimizing time and economic resources. Moreover, this approach is ecologically sustainable, as it generates less waste compared to the use of rapid tests, while continuing to confirm positive cases.
Read full abstract