A dog’s sense of smell can be 10,000 to 100,000 times more sensitive than a human’s because they respond to pheromones and other biological signals in their environment. With this advanced olfactory system, dogs can perceive smells humans cannot even imagine. It’s through this advanced sense of smell that a dog learns about the world around them, how to detect friends from foes, and, in recent years, how they can help medical teams identify asymptomatic patients with diseases.

The human body constantly produces volatile organic compounds (VOCs) through physiological processes, food intake, and metabolism, and dogs can detect these VOCs. When a person is sick, different VOCs are produced with or without symptoms. This has led to dogs becoming the subject of medical research, as they have demonstrated to medical professionals that they can identify cancer and infectious diseases in patients with relatively high accuracy at an early stage. A team from Mexico had eleven dogs trained to detect the COVID-19 VOCs from sweat samples of 379 people. The team also quantified the levels of the COVID-19 virus from the samples by RT-qPCR, the standard method of testing, to determine the sensitivity and specificity of the dogs in diagnosing COVID-19 at different stages of infection.

The current commercial methods of COVID-19 testing involve nasal swabs and analysis by antigen testing or rapid RT-qPCR testing. While these methods are effective, they do come with drawbacks. Performing a nasal swab is intrusive towards the patient and extremely uncomfortable, as the swab must reach deep into the nasal cavities for more accurate measurements. Antigen and rapid tests also require time to manufacture, and some are single-use, generating biologically and ecologically harmful waste. The utilization of dogs' highly developed sense of smell is being proposed as a more sustainable method of screening for COVID-19, as they are easy to train for COVID-19 VOC detection and have comparable accuracy to conventional tests.

The dogs in the Mexico study were the second generation trained for COVID-19 VOC detection using a protein designed to mimic the scent of COVID-19 VOCs. After training, eleven dogs were given sweat samples of 379 people, 90 COVID-19 positive and 289 negative, to determine their sensitivity and accuracy to COVID-19 VOCs. RT-qPCR analysis of the patients' sweat samples and additional nose swabs was used to calculate the approximate viral load to correlate with the dogs’ accuracy. Each dog was assessed on how accurate they were at detecting COVID-19 in the patients, and the team used the viral load to establish a baseline of infection that is perceptible to dogs.

The team wanted to prove two hypotheses. The first hypothesis was that second-generation COVID-19 detection dogs would be more perceptive to infected people. The second hypothesis was that dogs would be more efficient at detecting COVID-19 when the viral load reached certain levels. Their study's results proved the first hypothesis neither wrong nor correct, at least not entirely.

Only five of the eleven second-generation dogs showed sensitivity ranging from 67 to 87 percent, meaning they could correctly identify 60 to 78 of the COVID-19-positive patients from the study based on their sweat. However, the research team observed that the second generation was more accurate than the previous generation, suggesting that their ability to identify COVID-19 VOCs can improve with each generation. On the other hand, the research team’s second hypothesis was proven correct!

To determine the viral load of COVID-19 by conventional rapid testing, RT-qPCR uses fluorescent tags to track the amount of viral genome copied after each replication cycle. The viral load is back-calculated when the fluorescence reaches a detectable level, known as the cycle threshold. After back-calculating the viral loads of the patient samples through the cycle threshold and comparing them to the dogs’ reaction to the sample, the team saw an optimal range of viral loads that the dogs could detect. The dogs could not detect infected samples with cycle thresholds below 18 and above 33 cycles. Instead, the optimal range of viral load had cycle thresholds between 18.5 and 29 cycles. This range is similar to standard COVID-19 RT-qPCR tests, making the dogs as effective.

With this verification of their efficacy and required improvement in sensitivity, the use of dogs in disease screening is gaining traction in medical settings. Red tape must still be crossed for this new method to receive international approval because the WHO has established stringent criteria for a test’s sensitivity for it to be authorized. This is unrealistic for real-life scenarios of infectious disease due to the presence of asymptomatic and symptomatic individuals in the same area. It is also unrealistic for a test to maintain high sensitivity and specificity for extended periods, as tests can lose accuracy through overuse. Instead, the research team argues that the ability to use an accurate test more frequently is more important. In that regard, using dogs for disease screening offers significant advantages: they can screen large groups of people, are non-invasive to the patient, and do not generate harmful waste like conventional tests.

With a greater demand for accurate and immediate tests for emerging infectious diseases like COVID-19, implementing dogs for screening large groups offers accuracy and convenience while other methods become standardized. This study shows that dogs can perceive the VOX profiles of diseases, helping health professionals identify infected individuals, even those without symptoms. From the research team's conclusion, as each new generation of disease-detecting dogs improves on the last, dog units could become as effective as lab tests, without the environmental drawbacks.